Citation
The analytics of multibank holding company behavior

Material Information

Title:
The analytics of multibank holding company behavior
Creator:
Boczar, Gregory Edward, 1943-
Publication Date:
Copyright Date:
1973
Language:
English
Physical Description:
xiii, 143 leaves. : ; 28 cm.

Subjects

Subjects / Keywords:
Assets ( jstor )
Bank holding companies ( jstor )
Banking ( jstor )
Consumer loans ( jstor )
Federal Reserve Bank ( jstor )
Holding companies ( jstor )
Mathematical variables ( jstor )
Personal income ( jstor )
Riverbanks ( jstor )
Statistical models ( jstor )
Bank holding companies -- Florida ( lcsh )
Dissertations, Academic -- Economics -- UF
Economics thesis Ph. D
City of Miami ( local )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis -- University of Florida.
Bibliography:
Bibliography: leaves 139-143.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
022836390 ( AlephBibNum )
14125030 ( OCLC )
ADB0940 ( NOTIS )

Downloads

This item has the following downloads:


Full Text

THE ANALYTICS OF MULTIBANKC
HOLDING COMPANY BEHAVIOR








By




GREiGORY EDWARD BOCZAR


A DISSERTATION PRESENTED TO THE GRADUATE
COUNCIL OF THE UNIVERSITY OF PLORIDA IN PARTIAL
FULFILLMENT OF 'HE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY


UNIVERSITY OF FLORIDA
1973





UNIVEAITYO




























Copyright by
Gregory Edward Boczar
1973








ACKNOWLEDGMENTS


There are many individuals who have assisted in the com-

pletion of the dissertation. Speciial mention needs to be made of the

assistance provided by Professor Ralph H. Blodgett, the chairman of

my dissertation committee. He was a steady source-of encouragement,

advice and tolerance. His willingness to smooth over some of the

obstacles to getting an education was essential to my getting this far.

I am appreciative of the support in various forms provided

by the Board of Governors of the Federal System. Thanks go to Bob

Creller and Joe Holmes for very competent programming assistance.

Several of my colleagues were most generous with their time and advice

including Steve Rhoades, Bud Talley, Art Fraas and Bob Lawrence.

Steve Rhoades especially forced me: time and again -to clarify my

thinking and to improve my writing. The Research Library at the

Federal Reserve Board always provided professional service beyond the

call of duty. Both Ann Marie-Martin and Judy Back deserve special

mention. Statistical and clerical assistance were competently provided

by Anita Barley, Gloria Battle, Jackie Rupp, Ronnie Mc~illiams and

Lucy Nelson among others. Typing of numerous drafts was done by

Janet Breen, Ann Marie Cowan and Joan Albamonte.

The person who deserves the most, thanks is my wife Marilyn.

She has helped in every conceivable way assuming family responsibilities

so that I could work, acting as a sounding board- for my ideas, and

encouraging me when I was discouraged.









TABLE OF CONTENTS


Page
ACKNOWLEDGMENTS**********-********....... iv


LIST OF TABLES**--***************** ..................... .........v~~,iii

LIST OF FIGURES............,~ .....n ,n x


ABSTRACT***************-**************************........4..~~~~/~ i


Chapter

1. PURPOSE AND DESIGN OF THE STUDY....................,.....,... 1
Purpose of the Study.............. .,................ ... 1
Design of the Study: A Homogeneous Population. ,......... 1
Design of the Study: The Extreme Instance. ............ 5
Design of the Study: Decisions and Independent.,
Decision Makers...................................., 7
Organization of the Study. ....,......., .,,........... 7
Definitions and Terminology. ...............,,, .......... 8

2. A HISTORICAL AND LEGAL REVIEW OF BANK HOLDING COMPANIES....., 11
Historical Review Up to 1956.......................... 11
Regulation of Bank Holding Companies............. ..,,, 14
Banking Act of 1933 ............,............. ..,.,, 14
The Bank Holding Company Act of 1956. ............. ,. 16
The 1966 Amendments................................, 18
The 1970 Amendments.................................. 20
Development of Bank Holding Companies After 1956....... 21

3. A CRITICAL REVIEW OF THE LITERATURE. ..........;............. 28
A Selective and Critical Review: An Aid to
SModel Building............... ........,..,....,....o. 28
Fischer: Bank Holding Companies. ...........,,...... 28
Lawrence: The Performance of Bank Holding Companies.....; 31
McLeary: Bank Holding Companies--Their Growth
and Performance...............................o, 34
Weiss: Bank Holding Companies and Public Policy........., 35
Talley: The Effects of Holding Company Acquisitions
on Bank Performance...,................ .~............. 36
Piper: The Economics of Bank Acquisitions
by Holding Companies. ...............;.............. 38
Possible Efficiencies of the Holding Company Structure.. 41-
Schweitzer: Economies of Scale ald Holding Company
Affiliation................................:.. 42
Upson and Jessup: Returns from Bank Ho'lding Companies... 43
Alhadeffs: Recent Bank Mergers.........................., 44
The Relative Standing of,Banks.........;.,.........o... 45
Summary ..........,...............o.,..... .....~..o~nn, 46





Chapter Page

4. BANK HOLDING COMPANIES: PROFIT MAXIMIZERS OR SIZE
MAXIMIZERS ........... ,.............................. 48
External Expansion and Profit Maximization. ........... 49
The Alternative Hypothesis: Size Maximization. ........ 50
Profit Maximization Versus Size Maximization. ......... 52
Application of the Profit Maximization Hypothesis..... 53

5. A MIODEL OF LEAD BANK STATUS............................... 55
Definition of Lead Bank Status........................ 55
The Set of Potential Lead Banks.....................;.. 56
Lead Banks and Chains................................. 58
The Time Framework of the Study. .................... 59
A Framework for the Empirical Model...............,, 60
The Regulatory Position. ..........,.................. 61
Individual Bank Characteristics....................... 62
Environment. ............. ............. ............. 67
The Statistical Model. ..........,,.................. 69

6, THE DETERMINANTS OF LEAD BANK STATUS. ................... 78
Interpreting the Probit Statistical Model............. 78
The Likelihood Ratio Test............................. 79
A Stepwise Testing Procedure..........;................ 80
The Results for Single Predictors..................... 81
The Results for Several Predictors.................... 85

7. .A MDDEL OF HOLDING COMPANY ACTIITY....................... 93
The. Profit Maximization Hypothesis and Holding
Company Activity..............,..................... 93
Measures of Holding Company Activity............ ...... 94
The Dependent Variable and Tobit Analysis............. 96
Metropolitan Markets..............................;... 97
The Time Period. ................... ................... 99
A Framework for the Explanatory Variables............. 102
The Regulatory Environment. .......................... 102
Expected Profitability of the Market.................. 103
The Snowball Effect. ................... ............... 106
Alternative Measures of Explanatory Variables......... 108
The Model of Holding Company Activity. ................ 108

8. THE DETERMINANTS OF HOLDING COMPANY ACTIVITY .............. 109
Interpreting the Tabit Statistical Model.............. 109
The Results for Single Predictors..................... 111
The Results for Several Predictors. ................. 120

9. CONCLUSIONS FOR CASE ANALYSIS AND DIRECTIONS FOR
FUTURE RESEARCH. ......... .............. ............. 130
A Sumrmary of the Results................. 130
Conclusions for Cases Analysis........................11









Chapter Page

Directions for Future Research. ................... .... 134

APPENDIX............. ................. .'.................... .... 135

BIBLIOGRAPHY. ............... ................................. 139








LIST OF TABLES

Table Page

1.1 Multibank Holding Companies in Florida as of
December 29, 1972...................................... 3

1,2 Deposit-Size Distribution of Banks in Multibank
Holding Companies as of December 29, 1972............ ... 4

1.3 Population, Personal Income, Deposits and Banks in
Florida, 1960-1972................................ ... 7

2.1 Registered Bank Holding Companies: Number, Number of-
Banks and Branches, and National Share of Offices
and Deposits, 1956, 1960, 1965............,........ ........ 22

2.2 Offices and Deposits of Banks Affiliated with
Registered Bank Holding Companies, 1965-1970. ........... 24

:2.3 Percentage of Banking Offices in Each State and
Percentage of Deposits in Each State Held by
Registered Bank Holding Companies, 1965 and 1970......... 25

6.1 Identification of Variables with Means and Standard
Deviations....................................... 82

6.2 ~Probit Model of Lead Bank Status Using Single Predictors... 83

6.3 Probit Model of Lead Bank Status Using Two Predictors...... 86

6.4 Final Equations for the Model of Lead Bank Status.....;..... 88

6.5 Correlations Between Variables in Model of Lead Bank
Status........................................... 90

6.6 The Probability of Lead Bank Status........................ 91

7. 1 Holding Company Proposals Denied or Abandoned, 1963-1972.. 95

7.2 Metropolitan Banking Markets in Florida. ................... 100

7.3 Multibank Holding Company Activity in Florida,
1960-1972........................................ 101

8.1~ Identification of Variables with Means and Standard
Deviations. ............. ..............................112

8.2 Tobit Model of Holding Company Activity Using
Single Predictors................................. .115


viii





Table Page

8.3 Tobit Model of Holding Company Activity Using Two
Predictors.............************************** 121

8.4 Tobit Model of Holding Company Activity Using
Three Predictors...********************************* 124

8.5 Final Equations for the Model of Holding Company
Activity*********************-........,....... ............ 125

8.6 Correlations Between Variables in Model of Holding
Company Activity....,................... ................ 127

8.7 The Probability of Activity and the Expected Value of
Activity. ..........*************************************** 129

A. 1 The Set of Potential Lead Banks..............,,..... ........ 135

A.2 Decisions Mhde by Potential Lead Bank Through
December 31, 1972...................................... 137








LIST OF FIGURES


Figure Page

5. 1 A Straight Line Fit with Regressian Model. ................ 71

5.2 .Homoskedasticity about the Regression Line................, 71

5.3 Normal Distribution of Dosage Thresholds....... ........... 74

5.4 Cumulative Distribution of Insects Killed.................- 74

5.5 The Probit Line for Lead Bank Status...................... 76

8. 1 The Tobit Line for Holding Company Activity. .............. 110








Abstract of Dissertation Presented to the
Graduate Council of the University of Florida in Partial
Fulfillment of the Requirements for the Degree of Doctor of Philosophy


UKE ANALYTICS OF MULTIBANK
HOLDING COMPANY BEHAVIOR


By


Gregory Edward Boczar


December, 1973


Chairman: Ralph H. Blodgett
Major Department: Economics


Multibank holding companies have developed rapidly since

1965. The purpose of the study is to explain the behavior of these

companies--their formation and expansion. The study is limited to

formations and acquisitions in Florida for the last decade. By

restricting the study to a single state, it is possible to control for

a number of significant differences among holding companies. Also,

Florida is an especially ideal state for study because it has experi-

enced more growth of holding companies, population, personal income

and bank deposits than any other major state.

By way of introduction to the recent upsurge of holding

contmpany growth, the historical and legal framework of multibank hold-

ing companies is described. A critical review of the scholarly

literature on holding companies is also undertaken. Both the

historical and legal review and the review of the literature provide

information on the motivation and behavior of multibank holding

companies. A profit maximization hypothesis is selected as the








appropriate analytical framework because the hypothesis is more

workable and somewhat more consistent with the facts than its chief

rival, the size maximization hypothesis.

In order to explain empirically the behavior of multibank

holding companies, a model of lead bank status and a model of holding

company activity are developed and tested. The probit statistical

technique rather than the more conventional regression approach was used

-to test the model of lead bank status because the dependent variable

only takes the value of zero or one. The explanatory .variables for

the model of lead bank status fall into three categories: regulatory

position, characteristics of the bank and environment of the bank.

It was not possible to measure the regulatory position of the Board

of Governors of the Federal Reserve System because of the lack of

overt actions by the Board. In terms of an individual bank's

characteristics, the amount of correspondent business done by the bank

and the absolute size of the bank proved to be significant determinants

of lead bank status. No evidence was found to support the idea that

banks with declining market shares tend to anchor holding companies..

Also, lead banks cannot be characterized as either wholesale- or

retail-oriented nor are they especially successful in gaining loans

in general. In terms of the environment of a bank, the relative

importance of a bank's market as a financial center had a favorable

influence on lead bank status. Finally, if relatively few banks in

the recent past have affiliated with holding companies, a bank is

more likely to become a lead bank than if: a substantial portion of the

banks in the market have come under holding company control.


xii








The model of holding company activity was tested using the

tobit statistical method. The tobit method is appropriate because the

dependent variable does not take on negative values, having a lower

limit of zero. The explanatory variables for the model of holding

company activity are divided into three categories: regulatory

environment, expected market profitability and a snowball of ect. The

regulatory attitude of the Federal Reserve Board toward holding company

activity in a market is measured by the level of market concentration.

As postulated, concentration had a dampening effect upon the amount of

activity in a market. In terms of expected market profitability, both

current market profits (income as a proportion of capital of asset) and

deposits per capital or per bank had a significant positive influence on

holding company activity. The level of per, capital personal income was

found to have a positive and highly significant effect on holding

company activity. Neither population per bank nor short term (two-year)

changes in population, personal income or deposits have a systematic

influence on holding company activity. Some evidence was found to

support the contention that holding company activity tends to snowball

over time.


xiii















CHAPTER 1

PURPOSE AND DESIGN OF THE STUDY


Multibank holding companies have developed rapidly in the

United States since 1965. This activity by bank holding companies has

been concentrated in a dozen unit-banking and limited-branching states.

The most active state has been Florida. In.1965 there were six multi-

bank systems in Florida which controlled 55 banks. By December 1972,

there were 27 bank holding companies which controlled 280 banks. In

terms of number of multibank systems and in terms of number of affiliated

banks, Florida currently ranks first in the nation.


PurPose of the Study

The purpose of the study is to explain the formations of, and

acquisitions by, multibank holding companies. In order~ to understand

the factors which influence holding company formations, a model of lead

bank status is developed and then empirically tested. In order to under-

stand the factors which cause holding companies to be more active in one

market than another, a model of holding company activity is developed

and then empirically tested. Both models are developed within the frame-

work of a profit maximization hypothesis.

Design of the Study: A Homogeneous Population

A recent study of multibank holding companies concluded:

It is almost impossible to make an unqualified statement
concerning the structure, organization, or operation of
holding companies which will fit all situations. Bank
holding companies, as they have developed in this country,
tend to reflect the special circumstances of their

-1 -





- 2-


organization and their location. The' factors which
brought them into existence have differed over time,
as have the financial requirements of the localities
or regions in which they operate. Given a banking system
with striking differences in bank size, operating powers,
management philosophies, supervisory status, and tradi-
tions,' it is not surprising that bank holding companies
are found to differ significantly one from the other.
...It [the above] makes difficult the task of the
analyst who seeks absolutes....1

Because of this wide diversity among bank holding companies, the usual

ceteris Dalibus assumptions are not satisfied in a general study of

these companies.

In order to control for the significant differences among

holding companies, the study is restricted to formations and acquisi-

tions in Florida for the last decade. By limiting the study to one

state, we control for differences in banking traditions and management

philosophies which appear to vary from one state to another and from

one region to another.2 Of course, all holding company banks in the

study will be subject to the same state banking laws.

In addition, there are no striking differences among Florida

holding companies in terms of total deposits controlled nor in terms of

number of banks controlled. .(See Table 1.1.) The size distribution

of affiliated banks is similar among holding companies except for the

billion dollar First :National Bank of Miami, the lead bank for Southeast

Banking Corporation. (See Table 1.2.)

In order to have a more homogeneous population, the study is

limited to the banking activities of multibank holding companies. Thus,



1' Carter H. Golembe and Associates, The Future of Registered Bank
Holding Companies (Washington, D. C.: The Association of Registered
Bank Holding Companies, 1971), pp. 35-36.
2 Ibid., p. 29.






-3-


Table 1.1

Multibank Holding Companies in Florida
as of December 29, 1972


Number Total
Rank Name and Location of Holding Companies of Banks Deposits
($millions)

1 Southeast Banking Corporation, Miami 18 1,517. 2
2 Barnett Banks, Jacksonville 36 .1,245.9
3 Florida N/B, Jacksonville 31 1,167.0
4 First at Orlando, Orlando 25 1,075. 3
5 Atlantic Bancorporation, Jacksonville' 20 855.9
6 First Financial Corp., Tampa 9 682. 3
7 First Florida Bancorp., Tampa 24 646. 8
8 E11is Banking Corp., Brandenton 15 529.4
9 Consolidated Bankshares, Ft. Lauderdale 8 505.0
10 Exchange Bancorporation, Tampa 10 478.9
11 United Bancshares, Miami 6 493.8
12 City National Bank Corp., Miami 3 441.5
13 Pan American Bancshares, Miami 9 436. 9
14 Charter Bankshares, Jacksonville 9 341. 9
15 Broward Bancshares, Ft. Lauderdale 4 304. 3
16 First State Banking Corp., Miami 5 247.7
17 Florida Commercial Banks, Miami 5 :243.8
18 First Bancshares, Boca Raton 6 224.6
19 American Bancshares, North Miami 6 191.3
20 .Palmer Bank Corp., Sarasota 4 167.3
21 First National Bankshares, Pompano 4 166.7
22 Combanks Corporation, Winter Park 5 131.5
23 Citizens Bancshares, Hollywood 5 113. 6
24 Central Bancorp., Miami 2 107. 1
25 Community Banks, Seminole 5 89. 4
26 Florida Bancorp, Pormpano 3 80.3
27 Jefferson Bancorporation, Miami 3 77. 3



Source: Florida Bankers Association; Board of Governors of the Federal
Reserve System.





a Rank is based on total deposits as of December 29, 1972.


Table 1.2

Deposit-Size Distribution of Banks in Multibank
Holding Companies as of December 29, 1972


Number of Banks in


Each Size Classb


Greater 100 25 .
than 500 to 500 to 100

1 -- 9


Less
than 25

8
19
19
10
9
3
18
9
4
7
2

4
7'
1
3
1
2
3
2.
1
3
3
1
4
2
2


Ranka Name of Holding Company



1 Southeast Banking Corp.
2 Barnett Banks
3 Florida National Banks
4 First at Orlando
5 Atlantic Bancorporation
6 First Financial Corp.
7 First Florida Bancorp.
8 Ellis Banking Corp.
9 Consolidated Bankshares
10 Exchange Bancorporation
11 United Bancshares
12 City National Bank Corp.
13 Pan American Bancshares
14 Charter Bancshares
15 Broward Bancshares
16 First State Banking Corp.
17 Florida Comrmercial Banks
18 First Bancshares
19 American Bancshares
20 Palmer Bank Corp.
21 First National Bankshares
22 Combanks Corporation .
23 Citizens Bancshares
24 Central Bancorp
25 Community Banks
26 Florida Bancorp
27 Jefferson Bancorporation


b Size

Source :
Reserve


classes are in terms of $millions of deposits.

Florida Bankers Association; Board of Governors of the Federal
System.





-5-


one bank holding companies and the nonbanking activities of multibank

holding companies are excluded from the study. Furthermore, no Florida

bank holding company appears to be a purely financial holding company;

that is, a company which acquires and holds bank stock solely for

investment purposes. Also, there are no Florida companies which have

grandfathered banking subsidiaries in other states.

By concentrating on the period from 1963 through- 1972, the

study will cover a period of sustained holding company activity by

newly-formed or newly-active holding companies. During this time

period the statewide bank holding companies were very active with the

exception of the Florida National Banks.1


Design of the Study: The Extreme Instance

As indicated above, Florida has recently experienced more

holding company activity than any other state. In terms of population

growth for the last decade, Florida was the fastest growing major state

in the U.S. and was second only to Nevada among all states. During the

past decade, Florida had the greatest percentage increase in personal

income of any major state and was second only to Nevada among all states.2!

For the period from 1960 to 1972, deposits of Florida banks increased by

406.'4 percent which is the highest percentage increase in total deposits

of any major state. More banks were opened in Florida from 1960 to 1972



1One researcher in studying holding company acquisitions nationwide
found a considerable difference in the results of his model by dividing
the population into growth-oriented and non-growth-oriented bank holding
companies. R. Charles Mayer, A Model of the Determinants of Registered
Bank Holding Company Acquisitions, Doctoral Dissertation, Graduate School
of Business, University of Pittsburgh, 1971, p. 100.

2 Florida Banking and Its Largest Banking Institutions, Allen C. Eving
& Co., Jacksonville, Florida, 1972, pp. 5-6.





- 6-


than in any other state. Finally, Florida is by far the most underbanked-

state in the country.1

What we have is a cluster of events which can be characterized

as an ~ext~reme instance. In an extreme instance, the variables of

interest are subject to major shifts. In this study, there are large

changes in holding company activity, population, personal income, bank

deposits and number of banks in a short period of time. (See -Table 1.3.)

The extreme instance is a way of approximating a controlled

experiment. Even if the cateris 2811MEs assumption cannot be strictly

enforced, nonetheless the cetera will be relatively small when compared

to the large changes in the variables of interest. Thus, the extreme

instance provides the researcher with a nearly perfect situation to

study.

In order to generalize from the extreme instance, the behavior

observed must accurately reflect the properties of the system and not

be pathological.

In general, as long as (1) the system responds in a
systematic and controlled fashion rather than in an
erratic and disorganized manner and (2) the actions
were taken as a result of deliberation rather than
spontaneously without calculated intent, the stress
created by the extreme instance can be assumed to be
producing a characteristic response



1 Florida's population per banking office was 12,738 in 1970. The
next least banked state was Illinois with a figure of 9,216.
2 The entire discussion of extreme instances is taken from Oliver E.
Williamson, The Economics of Discretionary Behavior: Managerial
Objectives in a Theory of the Firm (Englewood Cliffs, N.J.: Prentice-
Hall, Inc., 1964).
3 Ibid pp. 87-88.



































Source: Bureau of Economic and Business Research, University of.
Florida; Office of Business Economics, U.S. Department of Commerce;
Board of Governors of the Federal Reserve System; Florida Bankers
Association.


Design of the Study: Decisions and Independent Decision Makers

The study is designed to explain the decisions made by bank

holding companies. The decision makers may be top management and/or

the owners of a possible lead bank or of a holding company. It is

assumed that the decision makers are independent, i.e., that with a

population of k banking organizations there are k independent sources

of decision making.


Organization of the Study

Chapter 1 presents the purpose and design of the study. A

number of terms are defined.


- 7-


Table 1.3

Population, Personal Income, Deposits, and Banks
in Florida, 1960-1972


Number
of Banks


304
318
340
382
421
440
444
447
456
475
500
540
578


Year Population
(thousands)

1960 5,095
1961 5,270
1962 5,458
1963 5,621
1964 5,796
1965 5,974
1966 6,137
1967 6,289
1968 6,492
1969 6,699
1970 6,842
1971 7,025
1972 7,211


Personal Income
($ millions)

9,739
10,248
11,050
11,859
12,976
14,182
15,683
17,451
19,791
22,542
25,077
27,611


Total Deposite
($ millions)

4,866.8
.5,247.3
.5,534.9
6,011.4
6,802.4
7,686.3
8,302.9
9,681.8
11,513.0
12,326.6
13,971.2
16,249.2
19,780.1 .










Chapter 2 provides a brief history of the growth of multibank

holding companies. The legislation affecting bank holding companies is

reviewed.

Chapter 3 consists of a critical review of recent scholarly

work on bank holding companies.

Chapter 4 considers the profit maximization and size maximi-

sation hypotheses. The profit model is found to provide a useful

framework for analysis.

Chapter 5 presents a model o~f lead bank status. Both an

individual bank's characteristics and a bank's environment are thought

to influence a bank's decision regarding lead bank status. The profit

statistical model is used to. test the model.

Chapter 6 presents the empirical results for the model.of

lead .bank status.

Chapter 7 presents a model of holding company activity. The

regulatory environment, expected market profitability and the snowball

effect are thought to determine the level of holding company activity

in different metropolitan markets. The tobit statistical model is used

to test the model.

Chapter 8 presents the empirical results for the model of

holding company activity.

Chapter 9 summarizes the results of the study and discusses

the implications of the study for case analysis. Areas for future

research are suggested.


Definitions and Terminology

Terms used in the study are defined below.






- 9-


A bank holding company is defined as any company (or trust)

which owns or controls 25 percent or more of-the stock of each of two

or more banks.1 Such companies are regulated by the Federal Reserve

Board and must register with it.

In this paper, the term "bank holding company" always refers

to.a multibank holdiing company which is registered with the federall

Reserve Board.2! The terms "group banking" and "multibank system" are

used as synonyms for the term "bank holding company".

Chain banking refers to the ownership of controlling interest

in two or more banks by an individual or informal group of individuals.3

It is important to distinguish chain banking from bank holding

companies or group banking. Chains are subject to individual control

while groups are subject to corporate rule. A group customarily has

some form of central management while a chain typically does not. Also,

groups are frequently headed by a major metropolitan bank while chains

rarely have such an arrangement.



1 Ordinarily corporations have no inherent power to hold stock in
another corporation. Some state laws permit the acquisition of bank
stock by corporations; other states restrict or prohibit the acquisition
of bank stock. A useful summary of the state laws on this matter are
provided in Bank Holding Company Facts, Association of Registered Bank
Holding Companies, Washington, D.C., 1972.
2 The only exception is Florida National Banks (formerly the DuPont
Trust) which was not required to register with the Federal Reserve
Board until passage of the 1966 Amendments. In this study Florida
National Banks is treated as if it became a registered bank holding
company when the Bank Holding Company Act of 1956 was approved.
The term "controlling interest" is deliberately vague. For a dis-
cussion of the problems involved in defining chain banks see Jerome C.
Darnell, "Chain Banking," National Banking Review, vol. 3 (March,1966),
pp. 308-309.






* 10 -


The lead bank of a holding company is the bank which the

management of a holding company designates as the lead bank. The lead

bank is generally the largest affiliate and it is' usually located in

the same city as the headquarters of the bank holding company. Fre-

quently there are officers and directors common to the lead bank and

the holding company.

Holding company activity refers to decisions favoring holding

company formations and acquisitions. The formation of a bank holding

company requires the appr-oval of the Board of Governors of the Federal

Reserve System under Section 3(a)(1) of the Bank Holding Company Act,

as amended. The acquisition of a bank by a holding company must be approved

by the Board of Governors under Section 3(a)(3) of the Bank Holding Company

Act, as amended.














CHAPTER 2

A HISTORICAL AND LEGAL REVIEW OF BANK HOLDING COMPANIES


The Bank Holding Company Act became law in 1956. The statute

did not appear to have much immediate effect upon holding company

activity nationwide or in Florida. However, in 1965 there began a

significant upsurge in bank holding company activity and in 1966 the

Act was amended in a way which encouraged a further surge of interest

in these companies. By way of introduction to these recent events, a

brief historical review is undertaken.


Historical Review Up to 1956

The beginnings of the holding company framework of operation

are to be found in the early history of American banking. During the

first half of the nineteenth century many of the multiple office systems

closely resembled bank holding companies. Examples of institutions which

resembled holding companies would include the First and the Second Bank

of the United States and later several state branching systems in the

Midwest. At the turn: of the century, several bank holding companies were

formally incorporated in` the Northwest.1

Bank holding companies developed rapidly in the decade of the

twenties. Nearly all the major systems which registered as a result of

thie Bank Holding Company Act of 1956 were organized from 1928 to 1930. 2

Gerald C. Fisher, Bank Holding Companies (New York: Columbia University
Press, 1961), pp. 4-8.
W. Ralph Lamb, Group Banking (New Brunswick, N.J,: Rutgers University
Press, 1962), pp.. 82-83.

11 -







--12 -


The rapid expansion of group banking in this. period cannot be attributed

to any single factor but rather to diverse and complex factors.

...One would have to consider: the farm depression
and the failure of small rural banks; the general
consolidation movement in many other industries;
the "snowball effect" as one bank holding company was
formed and banks organized others to maintain their~
competitive position, correspondent business, and
prestige; the restrictions on branch banking and the
belief that the laws would soon change; and the
investor's eagerness tolpurchase stock, including
holding company shares.

Perhaps the most important of these diverse factors was the

restrictions on branch banking. As metropolitan areas grew, the larger

institutions, which usually are located downtown, became less convenient

sources of banking services as compared to suburban banks.

Congestion which existed in all major cities made
banking at a single office, usually located in the
business district, very difficult. In addition,
retail as opposed to wholesale banking was growing-
in importance and many bankers wished to adjust 2
their facilities to meet this new market pattern.

One way for a large unit bank to adjust to a new market pattern was to

organize or to join a holding company.

The economic, regulatory and psychological conditions cited

above had resulted in ninety-seven groups -by 1931. These groups con-

trolled 978 banks and 1,219 branches, and they held 22 percent of the

loans and investments of all commercial banks.3 However, the rapid




Gerald C. Fischer, American Banking Structure (New York: Columbia
University Press, 1968), p. 95.

Fiecher, Bank Holding Companies, pp. 23-24.

U.S., Congress, House, Committee on Banking and Currency, Control and
Regulation of Bank Holding Companies, 84th Cong., 1st sess., 1955, p. 33.






- 13-


growth of bank holding companies ended with the beginnings of the great

depression of the 1930's.

The 1930's constituted a period of retrenchment for bank

holding companies. The depression forced a drastic reorganization of

the banking structure. By the end of 1934 the number of commercial

banks had dropped to about 15,400 as compared to nearly 29,000 in 1920.1

Data released by the Federal Reserve Board for 1936 showed fifty-two

groups operating 479 banks, a decline of roughly 50 percent from the

number of groups and affiliated banks in operation in 1931.

The depressed state of the economy, the relaxation of legisla-

tion restricting branch banking and the general antagonism toward holding

companies in general severely curtailed interest in group banking well

into the next decade. As a result, the position of groups relative to

commercial banks in general showed little change between the mid-1930's

and 1948. In 1937 these firms controlled 7 percent of the banking offices

and 11 percent of the bank deposits of all commercial banks. These figures

were almost unchanged in 1948.2

During the early postwar- period, 1947-1953, the growth of bank

holding companies largely paralleled the expansion of comm~ercial banking

in general. The proportion of U.S. offices and deposits controlled by

bank holding companies remained essentially unchanged.3



1 Comptroller of the Currency, Annual Report (Washington, D.C.:
Government Printing Office, 1964), pp. 7-8.
2Fischer, Bank Holding Companies, pp. 35-37.
Thomas R. Piper, The Economics of Bank Acquisitions by Registered Bank
Holding& Companies (Boston: Federal Reserve Bank of Boston, 1971), pp. 47,49.






-14-


There was a marked acceleration in the expansion of bank

holding companies between January 1954 and June 1956. This increase in

the growth rate of group systems was dominated by the acquisitions of

Transamerica (later Western Bancorporation) and the Marine Midland

Company.1 Two major reasons for the spurt in growth for this 2-1/2

year period were: the fear of restrictive federal legislation and the

efforts to rebuild Transamerica following the sale of its interest in

Bank of America.2 .


Regulation of Bank Holding Companies

Commercial banking in the United Stat'es has been subject to

regulation and supervision by federal and state authorities since the

1780's. However., since the bank holding company itself is not a bank,

it did not come under state or federal banking laws. The rapid expan-

sion of bank holding.companies in the 1920's stirred congressional

interest in establishing some degree of federal regulation of group

banking. After extensive hearings by the House and Senate over a three

year period, the Banking Act of 1933 (Glass-Steagal Act) was passed.3


Banking Act of 1933

The Banking.Act of 1933 granted the Federal Reserve Board limited

powers to regulate holding company affiliates. A holding company, affiliate



1 A list of all 340 acquisitions made by holding companies between the
end of World War II and December 31, 1967, is provided in Piper, pp. 18-26.

Fischer, Bank Holding Companies, pp. 39-42.

The full text of the Banking Act of 1933 is contained in the June 1933
issue of the Federal Reserve Bulletin.






- 15


was defined as any organization which controls the majority of the stock

of a Federal Reserve System member bank.1 A firm which fell within this

definition must obtain a voting permit from the Board of Governors to

vote the stock of a subsidiary which is a member bank. Before granting

a voting permit, the Board of Governors considered the financial condi-

tion of the applicant and the general character of its management.2

There were several obvious weaknesses in this Act. First,

registration was not required of systems composed. of only nonmember

state banks and it was possible for some groups to control their banks

without obtaining a voting permit. Second, the statute did not attempt

to regulate the formation or expansion of bank holding companies. Third,

group systems were permitted to continue investing in nonbanking enter-

prises.3

The banking authorities sought the enactment of regulatory

legislation to supplement the bank holding company provisions of the

1933 Act. Recommendations of the regulatory agencies in the 1940's asked

for a curb on the further expansion or formation of groups. After 1950,

the position taken by the federal supervisory agencies moderated because

in many instances the groups were performing valuable service in upgrading

the management and condition of affiliated banks.4


1"Banking Act of 1933," Federal Reserve- Bu--lleti vol. 19 (June, 1933),

p. 385.
2 "Federal Reserve Act Digest," Encyclopedia of Banking Laws (Hartford,
Conn.: Lamont Cross & Company, 1964), pp. 8-9.
Fischer, American Banking Structure, pp. 103-104.
Lamb; ppt 177-193. Also, refer to Benjamin J. Klebaner, "The Bank
Holding Company Act of 1956," Southern Economic Journal, o.2 Jnay
1958), pp. 313-326.





- 16 -


The Bank Holding Company Act of 1956

The Bank Holding Company Act of 1956 was intended to accom-

plish several objectives: (1) to define bank holding companies; (2) to

prevent holding companies from acquiring banks across state lines; (3)

to control the formation and expansion of multibank systems within

their "home"' states; (4) to preserve the historical separation between

the suppliers of money and the users of money.l

A bank holding company was defined in the Act as any company

which owned or controlled 25 percent or more of the stock of two or

more banks or which controlled in any manner the election of a majority

of the directors of two or more banks. Companies falling within the

statutory definition were required to register with the Board.of Governors,

to disclose prescribed information in reports to the Board, and to submit

to examination by the Federal Reserve System.

The Board's consent was required before a bank holding company

could be formed, before a bank holding company could acquire over 5 per-

cent of the voting stock or substantially all the assets of any bank,

or before two bank holding companies could merge. In deciding whether

to.grant approval for these actions, the Board was required to consider

the following factors:

(1) the financial history and condition of company
or companies and the banks concerned;
(2) 'their prospects;
(3) the character of their management;
(4) the convenience, needs, and welfare of the
communities and areas concerned; and
(5) whether or not the effect of the acquisition
or merger or consolidation would be to expand
the size or extent of the bank holding



1 William F. Upshaw, "Antitrust and the New Bank Holding Company Act: Part
II," Monthly Review, Federal Reserve Bank of Richmond (March, 1971), p. 3.





- 17 -


system involved beyond limits consistent
with adequate and sound banking, the public
interest, and preservation of competition
in the field of banking

The first three factor pertained to the organization's solvency,

asset condition, capital and operations, thus continuing the underlying

concern for depositors reflected in the 1933 Act. The last two factors

represent a significant departure from earlier legislation. For the

first time, concern is shown for the competitive health of the banking

industry. It is recognized that the acquisition of a bank involves

important nonsafety-oriented considerations such as the expected effects

on banking competition, the possible introduction of new services at the

acquired bank, and changes in its lending behavior and pricing policies.2

Under the 1956 Act, a registered group was prohibited from

engaging in any business other than banking, managing banks, or providing

certain services to subsidiary banks. The holding company law also deals

with intra-system transactions, restricting lending and credit operations

of groups.3 The restrictions placed on intra-system dealings made it

difficult for bank holding companies to perform the function of facilitating

loan participation and other joint credit transactions. In fact, the

1956 legislation effectively made it easier for correspondent banks than

for holding company affiliates to take joint action.4


"Bank Holding Company Act of 1956," Federal Reserve Bulletin, vol. 42

(May, 1956),. p. 446.
2 nFederal Laws Regulating Bank Mergers and the Acquisition of Banks by
Registered Bank Holding Companies," Economic Review, Federal Reserve Bank
of Cleveland (January, 1971), pp. 20-21.
S"Bank Holding Company Act of 1956," Federal Reserve Bulletin, vol. 42
(May, 1956)-, p. 448.
SGeorge R. Hall, "Bank Holding Company Regulation," Southern Economic
Journal, vol. 31 (April, 1956), p. 344.






- 18 -


The Board of Governors in its annual report to Congress in

1958 discussed the regulation of holding companies. The Board had

encountered substantial difficulty in balancing "convenience and needs"

and "competitive impact" considerations, both of which the Board is

required to consider in passing upon applications. To compound the

problem, the terminology used in connection with the question (of

"competitive impact" was imprecise. Another problem of the 1956 Act

involved the severe restrictions placed on intra-system loan partici-

pations. In order to correct the problem, the Board recommended repeal

of the restrictions. The report also recommended that one bank holding

companies come under regulation and that several exemptions of organiza-

tions from the holding company law be repealed.1


The 1966 Amendments

The enactment of the 1966 Bank Holding Company Act Amendments

satisfied several of the Board's recommendations for changes in the law.

Several categories of exemption from the Act were repealed; nonbusiness

long-term trusts were brought within the coverage of the Act; the stan-

dards and antitrust procedures made applicable to bank mergers by the

Bank Merger Act of 1966 were made applicable also to bank holding company

cases; and the provisions of Section 6 of the Act, which restricted inter-

subsidiary loans and investments, were repealed.2

The Amendments of 1966 clarified Congressional intent with

respect to the relative importance of convenience and needs factors



1 "Report Under the Bank Holding Company Act," Federal Reserve Bulletin,
vol. 44 (July, 1958), pp. 776-796.
2 Amendments to Bank Holding Company Act," Federal Reserve- Buletin.,
vol. 52 (July, 1966), p. 966.






- 19 -


and considerations involving competitive impact. The Board was directed

not to approve:

(1) Any acquisition...which would result in monopoly,
or which would be in furtherance of any combina-
tion or conspiracy to monopolize or attempt to
monopolize the business of banking in any part of
the United States, or

(2) Any other proposed acquisition...whose effect in
any section of the country may be substantially
to lessen competition, or tend to create a monop-
oly, or which in any manner would be in restraint
of trade, unless it finds that the anticompetitive
effects of the proposed transaction are clearly
outweighted in the public interest by the probable
effect of the transaction in meeting the conven-
ience and needs of the community to be served.1

Contrary to the wishes of the Federal Reserve Board, the 1966

legislation continued to exempt one bank holding companies. Beginning

in 1968, a dramatic surge in the growth of one bank holding companies

stirred Congressional interest in bringing these companies within the

purview of the Act. By the end of 1968, seven of the ten largest com-

mercial banks in the United States had formed one bank holding companies.

A year later the list included 43 of the 100 largest banks.2

The most important factor motivating banks to adopt this

form of organization was the fact that one bank holding companies were

not subject to regulation and therefore were able to engage in any non-

banking activity. The increasing costs of funds in 1968 and 1969 and

the competitive disadvantage of the interest ceilings imposed by

Regulation Q encouraged banks -to organize these companies because a

bank holding company could issue commercial paper,.just as any ~other



1 Ii. pp. 967-9 68.
2 nThe 1970 Amendments to the Bank Holding Company Act: One Year Later,"
Business Conditions, Federal Reserve Bank of Chicago (December, 1971)
p. 3.





- 20 -


corporate borrower. Another factor stimulating change was the growing

investment by banks in data processing equipment and personnel. The

one bank holding company structure allows for the establishment of a

subsidiary which may adopt wage and personnel policies needed to

attract specialized personnel and the subsidiary is well suited to

marketing excess machine capacity. Also, the organization of a

holding company is one, and perhaps the only, method owners of banks

may use to insulate themselves against litigation attacking their

right to enter new areas not specifically authorized by statute.1


The 1970 Amendments

The Bank Rolding Company Act Amendments of 1970 expanded the

coverage of the Act to include a company that controls only one bank.

Other provisions of the 1970 legislation include an expansion of the

Board's authority to determine that a company controls a bank and a

prohibition against tie-in arrangements whereby a bank extends services

to a customer upon certain conditions.2

The 1970 Act revised Section 4(c)(8) of the Holding Company .

Act under which bank holding companies may acquire interests in non-

banking activities. The legislation provides that such activities may

be approved if they are determined by the Board to be ''...so closely

related to banking as to be a proper incident thereto....u3



William F. Upshaw, "Antitrust and the New Bank Holding Company Act:
Part III," Monthly Review, Federal Reserve Bank of Richmond (April,
1971) pp. 4-6.
2 'Bank Holding Company Act Amendments of 1970," Federal Reserve
Bulletin, vol. 57 (January, 1971), p. 29.

3 Ibid., p. 31.






- 21 -


The Board is also authorized to differentiate between an activity

comlmenced de novo and the acquisition of a going concern.1


Development of Bank Holding Companies After 1956

After passage of the 1956 Act, it was expected that holding

companies would expand substantially. The growth of multibank groups,

however, closely paralleled the general development of the banking

system in the period from 1956 to 1965. At the end of 1956 there were

49 separate bank groups registered with the Board pursuant to the~ Act.

For year end 1960, there were 42 distinct registered companies and

five years.1later there were 48 multibank systems. The share of U.S.

total deposits held by subsidiaries was little changed for the period;

7.5 percent ia 1956 compared to 8.3 percent in 1965. Similarly the

ratio of holding company offices to all commercial bank offices changed

by less than 1.0 percent, with groups possessing 6.7 percent of all

banking offices at the end of 1965. (See Table 2.1.)

Since the end of 1965, holding companies have expanded sub-

stantially. Many factors contributed to the increase in holding company

activity. The abolition in 1964 of the 2 percent federal tax penalty

imposed on companies filing a consolidated tax return was one contributing

factor. 2 Another possible consideration was that the holding company

idea had finally come of age. This was suggested by the Board's approval



1 For a good explanation of the 1970 Amendments see Donald L. Kohn and
John F. Zoellner, "The Amended Bank Holding Company Act," Monthly Review,
Federal Reserve Bank of Kansas City (May, 1971) pp. 11-20.
2 Ed Tyng, "New Moves to Form Holding Companies Seem Likely Now that
Tax Penalty is Dead," Journal of Commerce, April 7, 1964, pp. 1, 24.









in April 1966 of three holding company formations by New York

organizations.1


Table 2.1

Registered Bank Holding Companies: Number,
Number of Banks and Branches, and National Share
of Offices and Deposits, 1956i.1960, 1965



December 31 December 31 December 31
1956 1960 1965

Number of holding companies 49 42 48

Number of banks 428 426 468

Number of branches 783 1,037 1,486

Holding company offices as
a percentage of all banking
offices 5.8 6.2 -6. 7

Total deposits (millions of
dollars) 14,843 18,274 27,560

Total deposits of holding
companies as a percentage of
deposits of all banks 7.5 8.0. 8.3


- 2 -


a Numbers are for separate bank groups. There were
bank holding companies, in 1956, for example, but in
holding company controlled another.


53 registered
four cases one


Source: Unpublished Federal Reserve Board data; Federal Reserve
Bulletin, June -1961 and August 1966.





The three companies were BT New York Corporation, Charter New York
Corporation and -Security New York Corporation. The New York groups
had total deposits of over $8.2 billion which is approximately 30
percent of the total deposits figure for all registered companies
at the end of 1965.


C





- 23-


The 1966 Amendments (enacted in July of 1966) contributed to

the dramatic increase in holding company activity. Repeal of the

restrictions upon loan participation and the sale of loan paper

certainly encouraged the activity. The 1966 amendments also clari-

fied the applicability of the antitrust laws to banks and required

the Department of Justice to challenge the Board's decisions upon

holding company applications within thirty days of approval. The

formation or acquisition is thereafter permanently immune from anti-

trust attack except under the monopolization provisions of Section 2

of the Sherman Act.1 No longer were bank holding companies haunted by

the possibility of future challenges of their past actions by the

Department of Justice.

As shown in Table 2.2, there has been a steady increase in

the number of multibankc holding companies from 1965 to 1970. At the end

of 1970, holding companies controlled .11.8 percent of the banking

offices as compared to 6.7 percent of the offices in 1965. The per-

~centage of deposits controlled by holding companies nearly doubled

during this five-year period, with holding companies possessing 16.2

percent of U.S. deposits at the end of 1970.

Table 2.3 provides a state-by-state summary of the development

of multibank holding companies between 1965 and 1970.2 Unlike the


1"Amendments to Bank Holding Company Act," Federal---. Resere Buleti,

vol. 52 (July, 1966), p. 967.
2For a similar summary which uses the years 1956, 1962 and 1969,
see Golembe, pp. 7-10.















End of Number of Number of Banking Offices Deposits ($ millions)
Year Companiesa Banks Number % of U.S. Total % of U.S.

1965 48 468 1,954 6. 7 27,560 8. 3
1966 58 561 2,363 7.8 41,081 11.6
1967 65 603 2,688 8. 6 49,827 12..6
1968 71 629 2,891 8. 9 57,634 13. 2
1969 86 723 3,397 10. 1 62,574 14. 3
1970 111 895 4,155 11. 8 78,064 16. 2

a Separate bank groups only; if a subsidiary bank is also a registered
bank holding company only one is included in the total.
Source: Federal Reserve Bulletin, August issues from 1966 through 1971.


aggregate data of Table 2.2, which may suggest a broadly based advance

Ln holding company activity, the state data indicate that much of the

growth has taken place in a small number of states. Of the 34 states

and the District of Columbia in which registered bank holding companies

operated in 1970, seven states showed a decline' in holding company

offices as a percentage of total offices between 1965 and 1970, and

19 states noted an increase of less than 10 percent: The only areas

reporting an expansion of 10 percent or more of banking offices were

Alabana (13.7), Colorado (13.6), District of Columbia (12.8), Florida

(25. 7), Maine (42.7), New Jersey (12.2), New York (17.7), Ohio (10.6)

and Virginia (21.1). Florida and Colorado are unit-banking states while

Alabama, New Jersey, New York and Ohio are limited-branching states.

Although Maine and Virginia are classified as statewide-branching states



1 Benton E. Gup, A Study of Banking Structure (Occasional Report Series,
Vol. 2, No. 3; Tulsa, Oklahoma: University of Tulsa, 1970), pp. 38-54.


2A -


Table 2.2

Offices and Deposits of Banks Affiliated with
Registered Bank Holding Companies, 1965-1970







-25-


a 0

&$ duddooJdad oddR
Com M A NA
ggg + +++++. +rf+++++....++++v~m ~ ~ ~ m~~
"E I, rNOrQAmmd'M;~~i;'i~n~b;'l
a ll l n3 nh 1 \ r lr I
WH




a o am mooCI lomc e ws damoem eao
e ~ ~ od Mo io~o Oll dI



03


em M
C MA cllo M hcm 4@Ncv ~ m h~o~-umhlolON@ 0hNl
3"~ -l~ hi od 2M~O ; daoodo a;doo dd~; 0 \do:




S03
a am
rl MF
WO









uc 80 Udr lrmr ml r r3N(



oOn m 00 e & 0 0S00y O
r(a IP( COMM 0 ml Nr 9001 V1 Am*H
9 @ a u M & C O ad M
as G o u- o o me e ec
eaw e o a e m v m >
MO' 'N" ~kAMa g a m k
QM AC OO Samm c >> >
a MM O *MO M s a O w a m
ur: 0H HM M NE E BE E






-26.


am g
a mu 0~


** a, com we PO

M >0# e a LI a >

a a. 0 Cm omO a
r9b O o O 9
0 C a c d U
M CI 000m.L 1 M


a u ehMWOdO Odr
M O u D 0 I(mP3 cr
Oe a O ra ag

CCe1 h 00Mu m L 0egO .m B UO

mit a8 m ~ a LI7 '3 Oa
ae a ct u m we a
&MS O OlIP) a r Om


e. o a s ,b "F
><< CHCM hM amaC
.m O OIL O& 00m aar 0 usa
" "E 0 &# MO ctQo.~X
rc90l ErO 3 OM a P
U\I OOa M 0 A



O r Om w mO gme e Irl
NUM ## 0 a, GaI rM Ca
me d >CaM >oc a0e0a 1e



## MG Od I O rl Or
ruakO or
'CM rlr o C C 190 DO
NE 0~C 0 OM VI d &.


rd W O M Ma I OLIauO

hlM LI a 0 C X0
L C a, 0oc a a a r-1

uc V1 u U l 0 a m L a
ru OS L)m O \ I
0ono amd 49u0


UM / o
s 9
cu o










MO
CO
P 0


a~N
+i+++i++++







O\NCamrlm0mrD M
N44 44M4N













N
+.+++.++++






m~~mosom
04*^^ ~No ~











0r m u






QM c
a I 0010 0
Ocea, 90 .e
O MCI ern u mCM
dqeSQmFmmO
C rOUG a, ar 1lm~
00v18>>>>>


ol r
U

O v
DO




O O



PI r



a
o
F0





* 27 *


the statutes do not favor statevide-branching (although it is possible)

and the dominant structure is of the limited-branching variety.1 The

increase for the District of Columbia is due to the registration of

Financial General Corporation pursuant to the Bank Holding Company Act

Amendments of 1966.

In terms of deposits, 10 states showed holding company deposits

declining or remaining constant and in 16 states holding company control

of deposits rose by less than 10 percent from 1965 to 1970. The following

states had an increase of holding company deposits of 10 percent or more:

Alabama (16.0), Colorado (29.2), District of Columbia (11.4), Florida

(38.5), Maine (41.2), Missouri (23.7), New Jersey (16.8), New York (17.7),

Ohio (11.5) and Virginia (21.1). Except for the District of Columbia,

all of these states prohibit or make difficult statewide-branching.

Since the end of 1970, multibank holding companies have rapidly grown

in Texas (unit-banking) and Tennessee (limited-branching).



SFor example, in Virginia an organization can achieve- statewide
branching only by merger. The mergered bank loses its branching priv-
ileges. (Individual banks may branch countywide, more or less.) In
addition, only banks in existence for 5 years or more may be acquired
by merger. Because of these limitations on statewide-branching ,
~Virginia organizations have generally formed holding companies. Affi-
liates of holding companies do not lose- their branching privileges and
affiliates are not required to have been in existence for at least
five years.














CHAPTER 3


A CRITICAL REVIEW OF THE LITERATURE


In the preceding chapter the history of the bank holding

company movement was reviewed. In this chapter the scholarly litera-

ture on bank holding companies is reviewed.


A Selective and Critical Review: An Aid to Model Building

The review of literature will cover empirical studies of multi-

bankr holding companies since 1960.1 In the course of the review, we will

attempt to determine the characteristic operating policies of bank

holding companies. In order to determine if the results of the empirical

studies are reliable, the review of literature will be critical as well

as descriptive. The entire review of literature is designed to assist

in subsequent model building.


Fischer: Bank Holding Companies

An obvious starting point is Fischer's monograph on bank

holding companies. -Fischer- surveyed holding companies by means of

questionnaires and interviews to determine their organizational and

operational practices. .He found that most multibank systems employed

legal counsel and accounting and tax specialists. The headquarters of

the holding company did some, central purchasing and advertising;



SThe one exception is the Alhadeffs' study of bank mergers which was
published in 1955.


- Z8 -





- 29 -


operations specialists were made available to subsidiary banks; and

there was a considerable amount of committee activity.1

The lead bank of the holding company encouraged the less

active subsidiary banks to modernize by adopting the practices of the

lead bank. Almost without exception, the lead banks encouraged the

affiliated banks to emphasize consumer credit. (Consumer loans yield

a considerably higher rate of return than other types of loans and they

are subject to a number of cost savings if uniform systems and proce-

dures can be followed by all members of the bank holding company.)2

In the course of his interviews, Fischer found multibank

systems fully involved in the supervision of portfolios with special

assistance provided to affiliates in managing their cash positions.

The bank holding company actually participated in very little personnel

coordination or development despite claims to the .contrary.3

In unit-banking and limited-branching states, correspondent

banking is an important source of interbank service flows. Many of the

ways in which a depository bank aide its correspondents are very similar

to the services provided by a bank holding company. Because of the

similarity of services, many multibank holding companies have consolidated

correspondent accounts while having the lead bank or the holding company

itself provide subsidiary banks with specialized services.4



Fischer, Bank Holding Companies, pp. 86-87.
Ibid., pp. 87-92.
SIbid., pp. 94-98.
hid., p. 105.






- 30 -


Fischer observed that entry of a .bank holding company into a

rural community had .no effect on the general level of interest rates

although the new affiliate often stirred the competitive spirits in the

community as one bank remodeled or provided a drive-in facility and

other banks countered with innovations of their own.1

The research by Fischer is subject to several critidisms.

The review of operating policies of multibank systems is based upon

questionnaires and interviews. Information gathered by these means is

exposed to the problem of respondents giving good reasons as opposed to

the real reasons for operating policies. Also, there are difficulties

in classifying, and generalizing from, the information obtained from

questionnaires and interviews.

In Fischer's 1961 study, most of the holding companies surveyed

had been in existence since the late 1920's. One may wonder whether the

results of the study reflect the operating policies of the newly formed

or newly active multibank holding companies of the late 1960s.

As Fischer is well aware, he may be criticized for attempting

to generalize from a very diverse sample.2 His sample of holding com-

panies includes old and new groups, large and small .groups, groups which

operate in unit-banking, limited-branching, and unlimited-branching

states. The' sample includes states where the holding company movement

is in an advanced stage (Minnesota) and states where holding company

activity had yet (in 1961) to begin a sustained movement (Ohio).



Ibid., p. 130
2This criticism may be applied in varying degrees to all the works
surveyed in this chapter. Of course, there are multivariate statistical
methods (not used by Fischer) which would control for some of the
different characteristics of bank holding companies.






- 31 -


Because of these extreme differences, the behavior of the bank holding

companies sampled is very diverse. As a consequence, Fischer is unable

to make strong statements concerning holding company behavior because

of the wide differences in behavior observed.


Lawrence: The Performance of Bank holding Companies

The first rigorous study of bank holding company behavior was

done by Lawrence. In the study he employed the before-and-after method

of analysis in order to isolate the impact of holding company affilia-

tion upon certain performance variables. This method of analysis was

applied ~to the acquired bank and a comparable nonaffiliated bank. The

use of paired bank comparisons permitted Lawrence to hold constant all

relevant local market conditions.1

The acquired banks were essentially similar to the paired

independent--banks in terms of pre-affiliation performance variables with

the only exception being due to balances as a proportion of total deposits.

A possible explanation for this difference is that banks with relatively

large due to balances are reluctant to affiliate with a holding company

from fear of losing correspondent business upon affiliation.2 Additional

reasons for a reluctance to affiliate may be a lesser need on the part

of the bank for correspondent-type services, from a holding company and

the ability of the bank to possibly anchor a holding company itself.



1 For a good discussion of the paired bank methodology, see Piper,
pp. 166-168.
2 Robert J. Lawrence, The Performance of Bank Holding Companies
(Washington, D.C.: Board of Governors of the Federal Reserve System,
1967), pp. 15-16.





As for post-affiliation performance, acquired banks were much

more active than the paired independent banks in making loans. The

difference in the proportion of assets devoted to loans.is both statis-

tically significant and quantitatively significant. Lawrence reported

that "the result for the loans-to-assets ratio is one of the strongest in

the entire study--the subsidiaries of bank holding companies have on the

average a loans-to-assets ratio that is more than 5 percentage points

higher than the ratio for the independent banks."' Lawrence also found

that holding company banks had a higher percentage of their assets in

the obligations of states and political subdivisions and lower percentages

of their assets in U.S. Government securities and cash plus due from

balances.2 After affiliation, subsidiary banks were found to be much

more aggressive than the independent banks in seeking out consumer

instalment loans. The subsidiaries' ratio of instalment loans to assets

was 3.15 percentage points higher than the ratio of paired independent

banks .

In the Lawrence study, holding company banks do not appear..to

have any competitive advantage over nonaffiliated banks. The average

growth rates of affiliates were not superior to the growth rates of

the independent banks. The earnings of subsidiary banks were not signi-

ficantly different from the earnings of the independent banks. Also,

there was no evidence of operating efficiencies as measured by the

operating ratio (operating expenses divided by operating revenue).4



Ib d., p. 17.


SThis result is confirmed by other studies which employ less rigorous
measurement techniques. See, for example, Golembe, p. 49.
SLawrence, pp. 21-25.









No one to date has substantially improved upon the Lawrence

research although it is subject to some criticism. The paired bank

approach used in the study assumes that the actions of the paired bank

are independent of the operating changes of the acquired bank. One may

question this assumption since the banks are in the same market.

Both a strength and a weakness of the study is the sharp

differences among the banks in the study:

The selected banks include large and small banks,
banks in standard metropolitan statistical areas
and small towns, banks in all sections of the
nation, subsidiaries of large and small holding
companies, lead and on nead banks, and unit and
limited-branch banks.

From such a diverse sample, Lawrence may be justified in making general

statements about holding company behavior. However, it can be argued

that the univariate statistical method (t-test on difference of means)

employed by Lawrence did not adequately control for the sharp differences

in the banks sampled. It is possible that the extreme' differences among

the banks actually reduced the significance of some of the statistical

tests. Instead, if Lawrence had controlled for the differences in, say,

the size, location, length of affiliation and branching characteristics

of the banks sampled by using a multivariate statistical method, he may

have found that holding companies affiliates were, say, more efficient

or more profitable than independent banks,2



Ibjid., p. 12.
SLawrence did divide the sample into two groups on the bases of
location, size of holding company, size of bank, and length of .acquisi-
tion. These divisions did not yield more significant results. However,.
the sample was divided on the basis of one- characteristic at a time.
Also, the divisions were very rough; e. g., a bank was either a large
bank or a small one.





- 34 -


McLeary: Bank Holding Companies--Their Growth and Performance

Bank holding companies in the Sixth Federal Reserve District

(Atlanta) were studied by Mc~eary. His study was a replication of the

Lawrence research, with the sample consisting of holding company affil-

iates in Florida, Georgia, and eastern Tennessee. McLeary employed a

paired bank method in analyzing performance variables- after affiliation. 1

Four performance variables of affiliated banks were found to be signifi-

cantly different (in the statistical sense). Subsidiary banks .generally

charged lower interest rates on loans, carried fewer U.S. Government

securities and more state and local obligations relative to assets, and

had a higher percent of their deposits in demand accounts than did

independent banks.2

The McLeary study has several serious deficiencies. There

is no economic justification for excluding western Tennessee which is

not in the Sixth Federal Reserve District. McLeary failed to take

account of differences among the three states included in the study.

Florida is a unit-banking state while Georgia and.'Tennessee permit

limited-branching. In 1960 Georgia prohibited the formation of bank

holding companies or the further expansion of existing holding companies.

Florida and Tennessee place no restrictions on holding company activity.

Also, bankers in Florida and Tennessee have reacted differently to the

possibility of adopting the holding company structure. The beginning

of sustained holding company activity in Florida occurred in 1966. The



SMcLeary realized that a before-and-after study is needed to determine
rigorously whether operating differences observed are due to changes
following affiliation or are due to the fact that 'holding companies
acquired banks with certain performance characteristics.
2 Joe W. McLeary, "Bank Holding Companies: Their Growth and Performance,"
Monthly Review, Federal Reserve Bank of Atlanta .(October, 1968), p. 137.






- 35 -


take off for holding company activity in Tennessee occurred at least

three years later.1 The McLeary study covered holding company subsid-

iaries in existence at the end of 1966.


Weiss: Bank Holding Companies and Public Poliev

Weiss studied holding company structure and performance in

the New England states of Maine, Massachusetts and New Hampshire. He

found that the holding companies expected to be consulted by affiliated

banks on matters such as changes in dividend policy, establishment of

branches, or other factors that affect earnings. Most holding companies

in the study discouraged correspondent relationships with banks outside

the system in order to minimize the non-earning assets of subsidiaries. 2

In general, the acquired banks' loan to deposit ratio increased

in the post-acquisition period. Also substantial shifts occurred in the

composition of the acquired banks' non-loan assets in the post-acquisi-

tion period. The acquired banks generally reduced their holdings of

U.S. Government securities and their holdings of currency plus due from

balances -relative to total assets. The affiliated banks increased their

holdings of state and 1ocal government obligations. In the majority of

holding company acquisitions studied, Weiss found a decrease in consumer

instalment loans as a .proportion of total loans~. Earnings performance

and the growth of capital accounts for the acquired banks seemed unaffected




In Florida, holding companies' share of total deposits of the state
went from 22.4 percent in 1965 to 38.4 percent in 1969. The comparable
figures for Tennessee are 2.6 percent and 8.2 percent.
2Steven J. Weiss, "Bank Holding Companies and Public Policy," New
England Business Review, Federal Reserve Bank of Boston (January/February,
1969), pp. 18-19.






- 6 -


by holding company affiliation.1 None of Weiss' results were subjected

to statistical tests because the number of banks in the study was

extremely small.

Weiss discussed several factors which appeared to influence

holding company behavior. He contended that the holding company struc-

ture was particularly attractive to center city banks that lacked the

legal power to branch into growing suburban areas... Also, he suggested

that adoption of a holding company organization may be motivated in

part by a desire to increase the size of a single banking organization.

In the case of Maine, Weiss found that the aggressive activity of one

holding company had been an important factor contributing to an upsurge

of interest in holding companies by other Maine bankers.2


Talley: The Effects of Holding Company Acquisitions on Bank Performance

The Lawrence research was updated by Talley using a larger and

different sample of acquired banks for the period from 1966 through 1969.

In reviewing the portfolios of the banks in the study, Talley found that

the acquired banks tended to switch out of governments and into.state

and local obligations and loans, particularly consumer loads. *In com-

paring the mean difference of affiliated banks versus independent banks

before affiliation with the mean difference after affiliation, Talley



1 Ibid., pp. 21-22.
2 Ibid., pp. 9-15. Situations similar to the one in Maine have occurred
in other states. For example, in New Mlexico "the obvious success and
Lapressive growth record of Bank Securities has apparently triggered
action on the part of four or five bank groups that are reported in the
process of forming bank holding companies." "Bank Breakthrough in New
Mlexico," Burroughs Clearing House, vol. 54 (September, 1970), p. 25.








observed an increase in the subsidiaries' ratio of loans to assets of

3.85 percentage points.1 As measured by the mean change in the differ-

ence method, the subsidiaries' ratio of consumer loans to total assets

increased by 1.60 percentage points.2 Both results are statistically

significant at the .01 level.3

Talley suggested that the portfolio changes found may be due

to the holding company structure which permits affiliates to be located

in diverse locations thereby lowering risk to the system as a whole.

Also, acquired banks can expect assistance from other affiliates or the

holding company in the event of liquidity problems. In addition, the

portfolio changes may be due to headquarters replacing conservative

management with more aggressive~, profit-oriented management.4

Bank holding company affiliation did not result in statistically

significant changes in the capital, prices, expenses or profitability of

acquired banks with one exception. Other operating expenses as a per-

centage of total assets was significantly higher for acquired banks.

One plausible explanation for this higher expenses ratio is that the

acquired banks paid significant management fees to their holding companies.

The only major difference from the Lawrence study was that Talley did not



SThis result is observed in other research. For example, Piper, p. 129.
Also see Irving Schweiger, "Reply to Chicago Banking: A Critical Review,"
Journal of Finance, vol. 17 (October, 1962), p. 424.
2 Lawrence in his study excludes single payment loans from the consumer
instalment loan category. Talley includes single payment loans in his
consumer loan category.
Samuel H. Talley, The Effect of Holding Company Acquisitions on Bank
Performance (Washington, D.C.: Board of Governore of the Federal Reserve
System, 1972), pp. 8-9, 16.
SIbid., pp.' 9-10.





- 38


find that banks acquired by holding companies increased their service

charges.1

Talley's study is subject to the criticisms applied to the

Lawrence paper; namely, the assumption of independence implicit in the

paired bank method and inadequate controls for a very diverse sample.

It is possible that Talley's conclusion that holding company affiliation

did not have a broad impact on the performance of acquired banks is

invalid because his univariate method of analysis did not adequately

adjust for differences among the banks studied.


Piper: The Economics of Bank Acquisitions by Holding Companies

Recently, Piper has done extensive research on the profit-

ability of bank acquisitions by holding companies. He found that

affiliation with a holding company was not associated, on average, with

any important increase in bank profitability over and above that exper-

ienced by banks generally. Furthermore, there was little evidence of

superior growth by the acquired banks.2

Piper also considered the question of whether the acquisition

programs of holding companies are profitable.3 An acquisition was judged

profitable if earnings per share of the holding company were higher in

1967 as a result of the acquisition. Piper .omitted from the analysis of

profitability any favorable impact on the holding company's stock resulting



1Ibid., pp. 10-13.
Piper, p. 161.
3 It is quite possible that a bank may not be exceptionally profitable
either before or after acquisition and still be a profitable addition to
the holding company. Such would be the case if a bank were acquired by
a holding company at a bargain price.






- 39 -


from the acquisition activity itself. Also omitted were any increases

in the earnings of the parent company resulting from fees paid to it

by the acquired bank.1 Piper found that the 102 acquisitions studied

.were breakeven investments, on average, and they did not result in higher

earnings per share for the holding companies involved.2 There was a

definite tendency for acquisitions during the period 1946-1958 to be

less profitable than those completed during the period 1957-1967.. In

the second period from 1957 to 1967, holding companies acquired -banks

that had excess capital that either could support further bank expansion

or could be reallocated to other, more productive uses in the system.

Holding companies that were active in acquisitions during the 1957-1967

period were also more aggressive in theirluse of long term notes and

debentures.3

In considering the motivations for holding company behavior,

Piper stated:

Organizers and managements of bank holding companies
seemed motivated by two primary considerations: (1)
the prestige and increased lending capacity of a
large regional banking group and (2) beliefs, un-
substantiated by a substantial accumulation of
research, that scale economies and improved manage-
nent would offset premium prices paid for banks and
would permit profitable expansion by acquisitions



Piper, pp. 216-219. The omission of management fees paid to the
parent company is questionable. Both Lawrene and' Talley report signi-
ficant increases in other expenses for acquired banks. These other
expenses are very likely management fees.
2Piper has assumed that the growth prospects- of the bank and the
holding company were equal beyond 1967. He also assumed that the holding
company's policies with respect to the existing system were unaffected
by a change in its acquisition activity.
Piper, pp. 224-225.
Ibid., p. 252.






- 40 -


With regard to the considerations of prestige and large

lending limits, none of Piper's own research provides support for such

a contention.- The fact that the acquisitions studied by Piper were

at best a breakeven proposition for the holding companies involved

does not lead to the conclusion, which is implicit in the consideration

of prestige, that holding companies are not profit maximizers. One

may propose that for acquisitions made during the 1946-1956 period the

holding companies involved were overly optimistic; i.e., they incorrectly

estimated the future stream of earnings from their acquisitions. For

the 1957-1967 period, Piper actually found that holding company acquisi-

tions were, on balance, profitable.2

With respect to the unsubstantiated belief in scale economies,

Piper states that in almost all acquis-itions by holding companies, scale

economies were probably negligible since the size of the firm was in-

creased but plant size was not increased. Piper cites research by Bell

and Murphy which shows that the greater labor requirements of branching

offset the economies of acale which permit specialization in many

banking functions.3 Piper.is comparing a holding company system to a

branching system which seems inappropriate. A more appropriate basis




Piper appears to be strongly influenced by Reid's study of mergers
involving large banks. See Samuel R. Reid, Mergers, Managers and the
ESSESSI, @iew York: McGraw-Hill Book Co., 1968), Chapter 10.
2The sixty-seven acquisitions analyzed during the second eleven-year
period had a mean profitability index of 1.07. The breakeven point was
1.0. See Piper, p. 224.
3 Ibid., p. 256. The reference is to Frederick W. Bell and Neil B.
Murphy, Costs in Commercial Banking: A Quantitative Analysis of Bank
Behavior and Its Relation to Bank Regulation (Boston: Federal Reserve
Bank of Boston, 1968).





- 41-


(if one proposes to test the arguments for efficiencies as expressed by

advocates of the holding company structure) is to compare unit banks

operating independently with unit banks operating within a holding

company framework.


Possible Efficiencies of the Holding Company Structure

Conceptually, returns to scale refer to the percentage change

in physical output resulting from simultaneous and equal percentage

changes in the employment of each input. Thus, the advocates of

holding company efficiencies are not really claiming economies of

scale as this term is strictly defined. Rather, the emphasis is on

economies of vertical integration for the multibank holding company

(the multiplant firm).1 The Bell and Murphy research does not test

this claim. The use of the operating ratio by Lawrence and Talley as

a rough-and-ready measure of operating efficiencies is very rough.

It must be said that the case for economies of vertical

integration accruing to the multibank holding company is most convincing,

a 211211. Holding companies .have a greater ability to tap major money

markets than do individual banks. The credit of the parent organization

is usually stronger than that of individual units, so the parent can nego-

tiate better terms. A large organization can recruit better talent because

of the fringe benefits, salary structure and additional possibilities of

advancement. The holding company vehicle permits pension and profit

sharing plans, as well as other fringe benefits, to be uniformly applied

to all units in an efficient program. Coordinated advertising programs



1 The formation of a multibank holding company may be viewed as the
positioning of a layer of management on top of a group of semiautonomous
banks .





42 -


may be applied to all units. The bond portfolio of all units can be

administered more expertly by a central department. Centralized uniform

accounting can provide vital help in efficient operations and assist in

financial planning and budgeting. A centralized auditing program can

be beneficial to subsidiary banks both as a management tool and as a

means of supplementing director's examinations.1 Note that these argu-

ments for group efficiencies are based upon a fundamental economic

principle. Large markets promote the increased division of labor and

the increased division of labor makes possible a more efficient use of

resources.2


Schweitzer: Economies of Scale and Holding Company Affiliation

Schweitzer has conducted the only research to date which

directly examines the question of efficiencies in m~ultibank systems.

In the research, he specified and estimated a cost function for banks

which stressed their dual function as intermediaries and payments

clearinghouses. The results of the study suggest that there are economies

of holding company affiliation. Schweitzer concluded:-

On the holding company question, it appears that .-
there are cost savings to be achieved through
affiliation with a holding company group, and
particularly with one of the two large Ninth
District groups. But these savings accrue prin-
cipally to banks in the intermediate size ranges.



1 Eugene H. Adams, "Economy, System Changes Require States to Modernize
Bank Structure," American Banker, etme ,17,p 1
2 Although Adam Smith used the pin factory to explain this principle,
nonetheless the principle may be extended to the modern multiplant firm.
It should also be noted that the problems of a limited supply of manage-
ment services are less press-ing for registered bank holding companies
because federal statutes require relatively homogeneous components for
the company.






- 43 -


from $3.5 to $10.0 million and from $10.0 to $25.0
million total assets.1

These conclusions can only be termed suggestive. One may

question~ whether Schweitzer's use of only two independent production

functions adequately reflects the variety of output produced by banks.

Also, data on the factor ptices for labor and for demand deposits were

not available for estimating the cost function.2


Upson and Jessup: Returns from Bank Holding Companies

Research by Upson and Jessup bears upon the question of the

profit orientation of holding companies. (Recall that Piper stated

that holding companies have a prestige and size orientation.) They found

that the earnings of multibank holding companies and the valuation of

these earnings were very favorable. The researchers created a price

index of holding company stocks for the period from 1957 to 1971 for

companies with over $500 million in group deposits after -1960. Share

prices between 1957 and 1971 increased almost 200 percent ~(about 8 per-

cent, compounded annually). For the same period, Moody's common stock

price index for banks outside New York City increased about -117 percent

The behavior of the holding company indez is attributed to

the management policies of these companies which have resulted in divi-

dends growing more rapidly for holding companies than for individual

banks. This greater growth of dividends basically reflects real growth



1 Stuart A. Schweitzer, Cost and Production in Banking: The Case of the
Ninth Federal Reserve District, Doctoral Dissertation, Department of
Economics, University of Minnesota, 1970. Cited in Golembe, p. 59.
Stuart A. Schweitzer, "Economies of Scale and Holding Company Affilia-
tion in Banking," Southern. Economic Journal~, vol. 38 (October, 1972),
pp. 259-263.
SRoger B. Upson and Paul F. Jessup, "Returns from Bank Holding Companies,"
The Bankers Magazine, vol. 155 (Spring, 1972), pp. 60-61.






-4k-


in earnings per share because the holding companies have typically

not increased their dividend payout ratios over the time period.1

Upson and Jessup also developed a more comprehensive. measure

of performance than the price index of holding company stock by com-

bining price and dividend information into total rates of return. This

measure includes both dividend income and capital gain or loss and

allows for the timing of receipts and outlays. Using this measure of

performance, a complete matrix of returns for the shares of holding

companies was calculated. For the period from 1957 to 1971, the average

return for all holding periods was 11.2 percent, compounded annually.2

The results of the research by Upson and Jessup demonstrate that holding

company profits, on average, have been very favorable. The results also

imply that holding .companies are profit oriented.3


Alhadeffs: Recent Bank Mergers

Research by the Alhadeffs is relevant t'o the study of holding

company behavior. In analyzing bank mergers, which occurred in the

early 1950s, the researchers hypothesized that uneven rates of growth

among banks were a major reason for the observed increase in bank mergers.

The Alhadeffs wrote: "The very rapid growth of the past decade has created

stresses and maladjustments owing to uneven rates of growth of different



Ibid., pp. 61-62.
2 Ibid., p. 62. The returns are approximately symmetrical around the
overall mean of 11.2 percent. There are relatively few negative returns
and no negative, returns appear ~for holding periods of longer than two
years.
Note that Piper looked at holding company profitability in terms of
earnings per share for a selected set of acquisitions while Upson and
Jessup considered the returns to the entire holding company system in
terms of share prices and in terms of their comprehensive measure of
per formance.






-46-


banks."1 If a merger is not an acceptable solution to a lagging growth

rate--which is usually the case in unit-banking states and often the case

in limited-branching states--then a bank may decide to organize or to

join a holding company in order to remedy a record of poor growth.

In the same study, the Alhadeffs reported that mergers oc-

curred in clusters and gave two reasons for the merger clusters:

Banks which have not yet begun to acquire other
banks begin to do so in simple imitation of the
initiators. A pattern of acquisitions which is
desirable for the initiators can be equally
desirable for the imitators. The clustering is
also explained by the defensive reaction of non-
participating banks to the acquisition activities .
of the initiating banks. Once mergers have begun
in an area, the relative standing of the, other
banks is either actually or imminently threatened
and a retalitary merger is the surest way to
maintain a threatened position.2

The merger clusters cited by the Alhadeffs appear to have an analogue

Ln the snowballing of holding company; activityy which was observed in

several states.3


The Relative Standing of Banks .

In explaining the clustering of~nergers, the Alhadeffs

mentioned a' bank's concern for its relative standing in the market.

A bank's concern for its relative standing is quite understandable if

its relative position declines because of a movement of customers to.

other banks. The Alhadeffs pointed to an indirect effect of a decline

in market standing. A decline in market standing is accompanied by a



1 Charlotte P. Alhadeff and David A. Alhadeff, "Recent Bank Mergers,"
Quarterly Journal of Economics, vol. 55 (November, 1955),- p. 512.
2- Ibid., p. 518.
See Chapter 2, p.12 and this chapter, p. 36,





-46-


relative (and possibly an absolute) loss of deposits which often causes

an increase in the capital to deposit ratio which reduces the leverage

on the bank's earnings.1

Related to the question of relative standing is the. fact

that a bank possesses a supply of highly specialized resources, namely,

banking knowledge and experience. These specialized resources are

relatively fixed in the near term so these resources may be viewed as

involving a type .of fixed cost.2 It is advantageous for the bank to

attempt to spread these fixed costs (of specialized resources) over

additional units of output because, other things being equal, average

costs decline as fixed costs are stretched over added output. Of course,

the ability of a bank to spread out fixed costs is a function of its

ability .to expand. It should also be noted that a bank's knowledge and

experience increases over time, thereby augmenting the bank's capacity

for production. With the passage of time, lacklustre growth can result

in a serious underutilization of resources.3




The empirical studies consistently found that holdihg company

affiliates have altered their portfolios in a manner that should increase

returns. Cash, due from balances and' U.S. Governments have been reduced

and a larger part of the portfolio has been devoted to local and state



1Alhadeff, p. 516.
Note that if a bank forsakes its accumulated knowledge and experience
and seeks to acquire a completely different kind of knowledge and exper-
ience, the opportunity costs will be high.
3 The emphasis here is upon the human resources of the bank. Since the
provision of banking services is labor intensive, this emphasis seems
justified.






- 47 -


obligations and to loans, especially consumer loans. Many holding

companies provided correspondent-type services such as portfolio

analyses, accounting services and guidance on operational matters.1

The parent often charged the subsidiaries .fees for the services provided.

Also, banks were thought to be very sensitive to their relative position

in a market. Those banks whose relative positions were threatened--

for example, downtown banks in a unit-banking state--may be attracted

to the holding company structure as a way to improve their market

standing. Holding company growth has occurred in .spurts perhaps because

independent banks have frequently reacted to the aggressive actions of

holding companies by organizing or joining a holding company. The evi-

dence is inconclusive on the question of whether holding company affil-

Lates are more efficient than comparable independent banks. Holding

companies appear to have operated very profitably, especially in the

last decade or so.2




For a recent discussion of the range of services provided by holding
companies see Robert J. Lawrence, OperatinR Policies of Bank Holding
Companies--P-art 1 (Washington, D.C.: Board of Governors of the Federal
Reserve System, 1971).
2 A recent study of holding company performance was conducted by Golembe
and Associates. The study was not reviewed because the empirical work
is poorly done. A paired bank method is used on an after-acquisition
basis only in an attempt to analyze holding company behavior. The before-
acquisition analysis was never done using rigorous statistical tests..
There are only 16 and 22 group banks for the two post-acquisition dates.
The tests applied for the post-acquisition dates are biased since some
group-banks were paired with more than one independent bank. To confirm
the results of the paired bank approach, Golembe analyzed national data
for all insured banks and all affiliated banks as of December 31, 1968.
This use of aggregated data for a single point in time without the
benefit of any statistical tests is unacceptable. See Golembe, Chapter 3.













CHAPTER 4


BANK HOLDING COMPANIES:
PROFIT MAXIMIZERS OR SIZE MAXIMIZERS


Neither the history of holding companies covered in Chapter 2

nor the review of the literature presented in Chapter 3 have provided

clear evidence as to the underlying motivation behind the actions of

bank holding companies.

On the one hand, the changes observed in the portfolios of

affiliated banks suggest that holding companies are profit oriented.

Also, if one accepts the arguments in favor of efficiencies for the

holding company structure, then the adoption and use of this structure

by a banking organization may indicate that the organization is profit

conscious. Furthermore, holding companies have operated quite profit-

ably, at least in the last decade or so. Of course, profitable opera-

tions are supportive of a profit maximization hypothesis.

On the other hand, the very rapid external expansion of some

holding companies may imply that they are size maximizers. Also, the

very high prices paid for unaffiliated banks by holding companies suggest

that they are size oriented.l



1 Unpublished tabulations of the Federal Reserve Board show the premiums
paid by holding companies in Florida, Missouri, New Jersey and Ohio for
1970 and 1971. (The premium represents the difference between the market
value of the shares offered and the book value of the shares acquired,
expressed as a percent of book value.) The median premium ranged from
a low of 52.2 percent in Ohio to a high of 131.6 percent in Florida.


- 48 -






- 49 -


This chapter will discuss briefly profit maximization and

size maximization as alternative bases for .explaining holding company

behavior. Before model building can begin, it is necessary to decide

upon the underlying motivation of holding companies.


External Expansion and Profit Maximization

Most holding company affiliates are acquired by means of a

share exchange. In a share exchange, the holding companies with high

price-earnings multiples can pay a higher price for a bank than holding

companies with low multiples, other things being held equal.1 Also,

the highly valued company is able to pay the high acquisition prices

frequently demanded by independent banks without adversely affecting

per share earnings.

Accordingly, it seems necessary for the holding company that

dishes to undertake an aggressive program of external expansion to have

a relatively favorable price-earnings ratio. If a holding company with

a relatively unfavorable price-earnings ratio undertakes an active

program of expansion, the company will likely experience a dilution in

per share earnings and per share assets. The holding company will soon

be forced to curtail its expansion program since each additional

acquisition will increase the amount of dilution suffered by the



Southeast Banking .Corporation and United Bancshares, both Miami-
based holding companies, made a bid to acquire Lon Worth Crow, a
mortgage banking company with close ties to United Bancshares. South-
east Banking acquired the company because it was able to pay a higher
price than United. At the time, Southeast's P/E ratio was about
double the P/E ratio of United.






- 50 -


original stockholders.1

The primary determinants of the price-earnings multiple

appear to be the growth of earnings per share and the growth of the

associated dividends per share. There is a "growing consensus of

belief in a definite relationship (through the interaction between

the stock market, dividend policy and the reinvestment rate) between

the maximum obtainable growth rate of the supply of capital...and the

rate of profit earned on existing assets."2 If one accepts the argument

for a relationship between earnings growth, the P/E ratio and expansion,

then it follows that a holding company which intends to expand externally

nust be profit conscious.


The Alternative Hypothesis: Size Maximization

As noted above, a size maximization hypothesis and a profit

maximization hypothesis appear to provide a satisfactory framework for

explaining holding company behavior. As Lanzilotti has observed:



For example, compare First BancGroup and Huntington Bancshares.
Both Ohio companies have embarked on an aggressive expansion program.
First BancGroup with an especially high P/E multiple has been able to
nake acquisitions on terms that maintain or enhance its own per share
equity and earnings. tHuntington Bancshares with a much lower multiple
has diluted both its earnings and its book value in the course of
acquiring other banks.~ In 1972, Huntington Bancshares has acquired two
small banks whose deposits total $37.2 million, which is about one-fourth
of the deposits acquired in 1970 and in 1971. This reduced amount of
expansion is presumably due to the substantial dilution experienced by
Huntington Bancsharea. See Herbert F. Thomson and Jerry A. Meadows,
"The Expansion of Registered Bank Holding Companies in Ohio," Bulletin
of Business Research., Ohio State University, vol. 46 (July, 1971), p. 3.
2 R.L. Marris, "Review of E.T. Penrose: The Theory of the Growth of
the Firm," Economic Journal, vol. 71 (March, 1971), p. 147.
M Eultibank holding companies apparently believe in a relationship
between profits and multiples and expansion. See Paul S. Nadler,
"Craze for Raising Multiples Spreads with Multi-BHC Craze," American
Banker, October 31, 1972, p. 4.





- 51 -


"Given empirical findings are likely to be consistent with, or at

least partially reconcilable with, many hypotheses. The essential

question is: which hypotheses are likely to yield the most useful

and reliable short-run and long-run predictions."1

The key empirical finding favoring the size maximization

hypothesis is the fact that a number of bank holding companies have

expanded very rapidly. In addition, the fact that holding companies

frequently pay high acquisition premiums for banks is consistent with

this hypothesis. By paying especially high acquisition prices, the

holding company becomes larger at the expense oficarnings per share.

Another possible consideration in favor of this hypothesis

is that the decision-making power in a holding company usually rests

with management. The argument for management control of the holding

company is based upon the fact that the stock of a holding company is

usually widely held. It is often assumed that the management-dominated

firm will opt for some goal (such as size maximization) other than

profit maximization.

Vernon has tested this assumption for banking firms. He

concluded:

We uncover no evidence to suggest that control
status has exerted a significant influence on
profit. rates of large commercial banks during
recent' years. ...Owner-controlled banks did not
realize higher rates of return on invested capital
than management-controlled banks.2



1 Robert F. Lanzilotti, "Pricing Objectives in Large Companies: Reply,"
American Economic Review, vol. 49 (September, 1959), p. 685.
2Jack R. Vernon, "Separation of Ownership and Control and Profit Rates,
~the Evidence from Banking: Comment," Journal of Financial and Quantita-
tive Analysis, vol. 2 (January, 1971), p. 624.






- 52 -


If one is willing to extend Vernon's results to management-controlled

holding companies, then one cannot argue that holding companies are

not profit maximizers because these companies tend to be management

controlled.


Profit Maximization Versus Size Maximization

In practice, it is very difficult to distinguish between the

size maximizing firm and the profit maximizing firm. For instance, the

rapid expansion of holding companies and the payment of high acquisition'

premiums may be rationalized in terms of a longrun profit maximization

model.

Under certain conditions a size maximizing firm and a profit

maximizing may behave the same way. It was argued above, thiat a holding

company which wishes to expand rapidly must be profit conscious. Con-

sequently, some size maximizing companies may be forced to maximize

profits in order to pursue an aggressive program of external expansion.

Of course, a size-maximizing company subject to a profit maximization

constraint will behave in exactly the same way as a profit-maximising

company.

Although the evidence is mixed, it appears preferable to use

the traditional model, of profit maximization. The concept of size

maximization is difficult to apply empirically because the notion of a

satisfactory level of profits, which is associated with the size maximi-

zation hypothesis, is difficult to use in empirical work.1




Of course, the size maximizing firm cannot completely ignore its level
of profits. For a discussion of the concept of a satisfactory level of
proits se wilia J.BauolBusiness Behavior, Value and Grawth (e
York: Macmillan Company, 1959), p. 49. For a discussion of the problems
in applying this-concept, see Bevars D. Mabry and David L. Siders, "An
Empirical Test of the Sales Maximization Hypothesis," Southern Economic
Journal, vol. 28 (January, 1967), pp. 367-377.





- 53 -


'By way of contrast, the profit maximization model greatly

simplifies the analysis. Professor Machlup wrote that "the substitution

of~ money profits for a composite of pecuniary [satisfactory profits] and

non-pecuniary rewards [size and prestige] simplifies the analysis so

much that the gain in expediency far exceeds the loss in a~pplicability."1

A statement by Penrose indicates what the profit model must

accomplish in order to explain holding company behavior.

In an analysis of the expansion of individual firms
the profit-seeking assumption is useful so long as
it is possible to set forth in reasonably objective
economic terms the considerations that will determine
the probability that certain specified directions of
expansion will be more profitable than others.


Applications of the Profit Maximization Hypothesis

Models of lead bank status and of holding company activity

will be developed within the framework of the profit maximization

hypothes is. In the model of lead bank status, it is assumed that a

potential lead bank considers its characteristics and the characteristics

of its environment -in deciding whether to lead a holding company. The

profit maximization hypothesis predicts that the bank will lead a holding

company if this is the most profitable opportunity- available. For the

model of holding company activity, it is assumed that holding companies

as a group consider the expected profitability of different markets and

the regulatory position of the Federal Reserve Board in deciding where

to acquire affiliates.' The profit maximization hypothesis predicts



1 Fritz Machlup, "Theories of the Firm: Marginalist, Behavioral,
Managerial," American Economic Reviewvl57(ac 16)p..Th
size maximization model usually carries the suggestion that the firm
values highly the prestige associated with size.
2Edith Penrose, The Theory of the Growth of the Firm (New York: John
Wiley &r Sons, 1959), p. 185.






54-


that holding companies will be most active in the most attractive

markets (in terms of expected returns),- provided the regulatory

position is permissive.

It is important to note that the profit maximization hypothesis

is employed because the hypothesis provides us with a useful and con-

sistent framework for analysis. The historical evidence and scholarly

research suggest, but do not demonstrate, that holding companies are

profit maximizers. The reader should not be misled into thinking that

the study is designed to determine whether or not holding companies

are profit maximizers. Rather, the study is built upon the convenient

and well-known assumption of profit maximization.













CHAPTER 5

A MODEL OF LEAD BANK STATUS


Within the framework of the profit maximization hypothesis,

we will develop a model to analyze the factors which cause a bank to

become a lead bank of a'multibank holding company. Alternatively,

the model may be viewed as analyzing the factors which influence the

formation of a multibank holding company. For most cases, the decision

to lead a holding company and the decision to form a holding company

are different views of the same phenomenon.1 That is, a holding company

is generally formed by a bank which assumes lead bank status upon formation.


Definition of Lead Bank Status

In Chapter 1, the lead bank of a holding company is defined

as the bank which the management of a holding company designates as the

lead bank. The Federal Reserve -Bank of Atlanta asks each multibank

holding company to designate a bank as its lead bank in those instances

where it is not obvious which bank is the lead bank, so as a practical

matter it is not difficult to know which bank is the lead bank.




There are a few exceptions. For example, the McNulty Group.of Banks,
which consisted of eleven small banks in Central Florida, became a holding
company in 1967 with the title of First Florida Bancorporation. In 1969,
First Florida "acquired" Marine Bank and Trust Company, Tampa, which bank
became the lead bank of the holding company. See "First Florida Bancor-
poration, Haines City, Florida," Federal Reserve Bulletin, vol. 52 (November,
1966), pp. 1632-1635. "First Florida Bancorporation, Haines City, Florida,"
Federal Reserve Bulletin., vol. 55 (February, 1969), pp. 165-168.


* 55 -





- 56 -


Although it is the case that all multibank holding companies

have lead banks as defined above, nonetheless for a few holding com-

panies no bank has assumed lead bank status.'- The idea of lead bank

status is not easy to spell out, although it is usually evident

whether or not a bank has assumed lead bank status. A bank in taking

lead bank status will provide the top officers of the holding company.

The bank with lead bank status will set the tone for the entire system

in terms of management philosophy, operating procedures, policy guide-

lines and long range planning. The bank with lead bank status will

help the other bank subsidiaries solve problems of management and opera-

tions as well as providing these banks with specialized services.2


The Set of Potential Lead Banks

The set of potential lead banks is defined as the largest 75

banks (in terms of total deposits) in Florida as of December 31, 1971.

Although it is most unrealistic to consider all banks as potential lead

banks, the decision to use the top 75 banks may appear arbitrary. It

was reasoned that Dnly large banks have the opportunity of becoming a

lead bank because only large banks have the financial and human resources

needed to lead a multibank organization.3 A large bank by Florida



1 This appears to be the case with Ellis Banking Corporation of Bradenton,
Florida. Although the First National Bank of Bradenton was designated by
management as the lead bank (and it is used in the study), management of
Ellis Banking Corporation has indicated in discussions with the staff of
the Federal Reserve Board that the holding company "really" does not have
a lead bank.
2 As holding companies grow and diversify, the bank holding company
often takes on an existence distinct from that of the lead bank.
3 All lead banks of multibank holding companies are substantially larger
than the median size bank. The median size bank had total deposits of
$17.0 million as of December 31, 1971.






-57 -


standards was judged to be a bank with approximately $50 million in deposits

or more. The largest 75 banks satisfy this criterion.1

Three of the 75 largest banks were excluded from the set of

potential lead banks because they had been actual lead banks~ since

the 1920's.2. Fourteen of the 75 largest banks were excluded because

they were not independent decision makers. In every case bui one,3

the excluded bank was closely tied to a much larger bank which was

assumed to make the decisions. So, of the largest 75 banks, only 58

banks were judged to be both potential lead banks and independent

decision makers. (See. Table A.1 in the appendix.)~ Of the 58 potential

lead banks used in the study, 21 banks have become lead banks and of

the remaining 37 banks, ten banks have been acquired by holding companies

and 27 banks have made no decision regarding lead bank status. (See

Table A.2 in the appendix.)

It should be noted that three multibank holding companies are

led by banks which are not among the 75 largest banks. These three

lead banks are ranked 89th, 96th and 152nd among Florida's 540 banks on

the basis of total deposits as of December 31, 1971. Only one of the

three holding companies--American Bancshares (152nd)--has subsidiary

banks in more than one metropolitan market.4 The other two holding



1 The 75th largest bank as of year end 1971 had total deposits of
$48.9 million.
2The three banks are Atlantic National Bank, Florida National Bank
and Barnett First National Bank. All three banks are located in
Jacksonville. For a history of these groups see J.E. Dovell, Histor1
of Banking in Florida (Orlando: Florida Bankers Association, 1955),
pp. 183-189.
The .one~exception is Ellis Banking Corporation. The affiliated
Sarasota Bank and Trust Company is substantially larger than the
designated lead bank, First National Bank of Bradenton..
For, a discussion of metropolitan markets, see Chapter 7, pp. 97-99.






- 58 -


companies have subsidiary banks ini only one metropolitan market and are

among the smallest multibank holding companies in the state. The model

of lead bank status is not designed to explain the decisions ~by a few

medium size banks to anchor small local or regional holding companies.


Lead Banks and Chains

Before becoming the lead bank of a registered multibank

system, a bank may have anchored a chain.2 There are two main teasons

why a chain would form a holding company. First, the holding company

structure formalizes and makes permanent the relationships within the

chain.3 Second, the holding company structure allows a former chain

to rapidly expand its banking activities because the- holding company

can obtain outside funds to finance acquisitions.

Chain banking may be viewed as a substitute for the holding

company form of banking. Given this view, we should then hold constant

the influence of chain banking so that we can isolate the influence of

other factors upon lead bank status. Unfortunately, reliable data on

chain banks are not available for all years included in the study. The




SFor the distinction between chain banking and bank holding companies,
see Chapter 1, p. 9.
This was the case with the First National ~Bank at Orlando, Broward
National Bank and First National Bank of Fort Lauderdale. See "First
at Orlando Corporation, Orlando, Florida," Federal Reserve_ Bulletin,
vol. 53 (February, 1967), p. 236. "Broward Bancshares, Inc., Fort
Lauderdale, Florida," Federal Reserve Bulletin, vol. 56 (January, 1970),
p.. 85. "Consolidated Bancshares of Florida, Inc., Fort Lauderdale,"
Federal Reserve Bulletin, vol. 57 (February, 1971), p. 137.
3--- ---------
A chain may be dissolved. This did happen in t~he case of the First
National Group of Southeast Florida. See J.E. Dovell, History of Banking
in Florida: First Supplement (Or~lando: Florida Bankers Association, 1964),
p. 42.






- 59 -


omission of a variable to control for chain banking activity may not be

serious. Darnell, in studying chain banking, did not find a significant

relationship between chain banking and bank holding companies.1


The Time Framework of the Study

In deciding upon lead bank status the potential lead bank has

three options: (1) become a lead bank of a holding company, (2) become

a subsidiary bank of a holding company, or (3) remain unaffiliated.

There is a substantial time lag between the time a bank decides

to become a lead bank (or to be acquired by a holding company) and the

time. this decision is consummated. After a decision is made, time is

expended in finalizing negotiations with the proposed affiliates (or

with the acquiring holding company) and in preparing an application for

the Federal Reserve Board. After the application is filed with the

Federal Reserve Bank, it will take at least 105 calendar days for the

application to be processed by the Federal Reserve System and for the

Board of Governors to act on the proposal.2 An approval by the Board

may not be consummated for .thirty calendar days in order to permit the.

Department of Justice time to challenge the Board's decision. In all,

it appears that anywhere from six months to eighteen months will elapse

between the decision .date of the potential lead bank and the consummation

date.




Jerome C. Darnell, "Determinants of Chain Banking," National Bank
Review, Vol. 4 (June, 1967), pp. 459-468.
Current guidelines state that the Reserve Bank should attempt to
formally accept an application within ten business days after receipt.
The Federal Reserve System (i.e., the Reserve Bank and the Federal
Reserve Board) tries to bring the proposal before the Board of Governors
within ninety calendar days after formal acceptance.





- 60 -


All of the explanatory variables in the model are lagged one

year because of the considerable delay in consumrmating a decision.

The year a decision on lead bank status is consurmmated is determined

from Board documents. The previous year is assumed to be the decision

year and all explanatory variables are measured as of the ~decision year.

The variables are as of the decision year because it is necessary to

measure the characteristics of the bank at the time that the decision

was made. If a potential lead bank has not become a lead bank or been

acquired by a holding company by the end of 1972, which is the terminal

point of the study, then it is assumed that the bank has decided not

to become a lead bank for the time being.

From 1956, when the Bank Holding Company Act was passed, until

1964 no potential lead bank became a lead bank. In 1964, the First

National Bank of Tampa, a potential lead bank, became the lead bank of

a registered holding company. The decision to become a lead bank is

assumed to have been made in 1963. So the study of lead bank status

will cover the period from 1963 through 1971 in terms of decision years

or from 1964 through 1972 in terms of consummation years.


A Framework for the Empirical Model

The variables which affect lead bank status may be divided

into three groups: (1) variables which reflect the regulatory position

toward potential lead banks; (2) variables which measure the economic

characteristics of the potential lead bank; and.(3) variables which

reflect the environment in which a potential lead bank operates. The

relationship of the three types of variables to lead bank status is

summarized in equation (1).





- 61 -


(1) LBSt = f(FRBty IBCt-1, EEt-1)

where

LBSt = lead bank status in year t. LBS=1 if the
bank is a lead bank; LBS=0 otherwise.

FRBtpl = regulatory position of the Federal Reserve
Board in year t-1.

IBCt-1 = individual bank characteristics in year
t-1.

EEt-1 = environment of the bank in year t-1.


The Regulatory Position

TFor the period 1963-1972, the Federal Reserve Board has

approved all proposed formations.1 One cannot conclude from these

approvals that the Board has been completely permissive. The Federal

Reserve Board has certainly exerted an influence upon the decisions of

potential lead banks by means of its decisions on non-Florida cases and

by means of informal staff discussions with these banks. However, it

is an impossible task to measure this influence empirical-ly because of

the absence of any overt actions by the Board. Consequently, it has

been assumed that the Board's influence upon the decisions of potential

lead banks has not been significant and no variable reflecting Board

influence will be included in the model.



1 In 1962, the Board did disapprove the application by four of the eight
largest banks in Florida to form a holding company. Today the four banks
lead the first, second, fourth and tenth largest holding companies in
Florida. "First Bancorporation of Florida, Inc., Orlando, Florida,"
Federal Reserve Bulletin, vol. 48 (August, 1962), pp. 978-984. In 1971,
the Board disapproved a formation which involved a potential lead bank,
the Cormmercial Bank at Winter Park. The formation, however, was .contin-
gent upon the approval of an application by Southeast Banking-Corporation
to acquire all five banks of the proposed multibank holding companies.
"Combanks Corporation, Winter Park, Florida," Federal Reserve Bulletin,
vol. 58 (January, 1972), pp. 55-56.





- 62 -


Individual Bank Characteristics

In looking at individual bank characteristics, six character-

istics are thought to be important determinants of a bank's status and

they are shown in equation (2).

(2) IBCt = g (SZt, CBt, Lt/A C~L/Ag, CL /At, MSt)
where

IBCt = individual bank characteristics in year t.

SZt = size in terms of total deposits in year t.

CBt = correspondent business measured by due to
balances in year t.

Le/At = loan to asset ratio in year t.

CLt At = consumer loan to asset ratio in year t.

CLt/At = commercial and industrial loan to asset
ratio in year t.

MSt = change in market share for the period from
year t-5 to year t.

The most obvious characteristic of any bank is its size. It

is generally assumed that large banks are more likely to become lead

banks than small ones. The presumption is partly based on empirical

evidence--nine of the ten largest banks in Florida are lead banks.

Partly, it is assumed that the larger banks have the human-and financial

resources needed to lead a multibank system. .In terms of human resources,

the larger is the bank (in absolute terms) the more likely it is to have

specialized services such as investment guidance and data processing

services which can be offered to other banks. In terms of financial

resources, the larger is the lead bank (in absolute terms) the more





- 63 -


marketable will be the holding company stock and the more acceptable

will be the stock in ~share exchanges. In addition, the larger organ-

izations have easier access to financial markets when they publicly

offer additional stock or senior securities. This entire discussion

is intended to suggest that the larger is the bank the better will be

its ability to exhaust any efficiencies associated with the holding

company structure.

The size of a potential lead bank is measured by its total

deposits as of December 31 of the decision year. Total deposits are

deflated using the aggregate GNP deflator in order to remove price

changes in the deposit data for the time period of the study.

A second bank characteristic which is thought to affect the

decision on lead bank status is the size of the bank's correspondent

business. It is hypothesized that the greater is the bank's involve-

ment in correspondent banking the greater is the likelihood that the

bank will become a lead bank.

The hypothesis concerning correspondent business is based on

three considerations.2 First, the larger is the bank's correspondent

business, the more able is the bank to provide future subsidiary banks




Since the study is restricted to the state of Florida, we need not
control for inter-state differences in bank size. As an example of
state differences a $100 million bank, which is a very large bank in
New Mexico, may provide services which are inadequate by Texas standards.
2The three considerations have the same direction of influence and are
to some extent reenforcing. It is not necessary to specify which consid-
eration, if only one, is paramount to a particular bank.





- 64 -


with specialized services. In other words, the bank has expertise that

is useful to a holding company organization. Second, the larger the

correspondent business' the more likely the bank is to have a close

working relationship with other banks. The correspondent bank and its

family of closely-related client banks can easily adopt the holding

company structure and formalize the "family". Third, a bank with a

large amount of correspondent business may organize a holding company

in order to preserve its business from the encroachment of expanding

holding companies. .One sure way to suffer a decline in correspondent

business is for a bank to permit rival organizations to acquire respond-

ent banks.1

Due to balances are used as a measure of the correspondent

business done by a bank because such balances are the usual means of

payment for correspondent services. Due to balances are measured as of

year-end of the decision year and they are deflated with the aggregate

GNP deflator in order to adjust for inflation over time.

In isolating key characteristics of a potential lead-bank,

three portfolio ratios appear useful. These are the total loan to

asset ratio, the consumer loan to asset ratio, and the commercial and

industrial loan to asset ratio.

Both Lawrence and Talley, in studying holding company perform-

ance, found that affiliated banks had a substantially higher proportion

of their assets in loans, particularly consumer loans. In fact the

changes in the loan to asset ratio and the consumer loan to asset ratio



1. See Chapter 2, p. 12, and Chapter 3, pp. 29, 31. Also, Lawrence,
Operating Policies, p. 25.





- 65-


were among the most significant findings, both statistically and quanti-

tatively, in the two studies.1

The assumption is made that the changes in the operating

policies of subsidiary banks reflect the existing practices of the

lead bank. Accordingly, it is hypothesized that the higher the bank's

loan to asset ratio and the higher its consumer loan to asset ratio,

the more likely it is that the bank will become a lead bank. The loan

to asset ratio is measured by a bank's total loans and discounts

divided by its total assets.2 The consumer loan to asset ratio is

defined as total consumer loans divided by total assets.3 .

The third portfolio ratio considered in analyzing lead bank

status was the ratio of commercial and industrial loans to assets. This

ratio was used to indicate the extent to which a bank was wholesale-

or business-oriented. It is hypothesized that wholesale banks will be

likely to form holding companies in order to attract the business of,

and in order to better serve the banking needs of,; multi-establishment

firms. This portfolio ratio is defined as total commercial and industrial

loans divided by total assets.4

It should be noted that there is something of a contradiction

between the business loan ratio hypothesis and the consumer loan ratio

hypothesis. A bank cannot be wholesale-oriented and retail-oriented at



1See Chapter 3, pp. 32, 36-37.
2 In terms of the format of the Report of Condition, total loans and
discounts are item 8 of schedule A; total assets are item 14 of the
front page.
In terms of the Report of Condition format, total consumer loans
are the sum of items 6(a) through 6(f) of Schedule A.
Commercial and industrial loans are items 5 of Schedule A of the
Report of Condition.






- 66 -


the same time.1 The empirical results should determine which ratio,

if either, is a significant determinant .of lead bank status.

It is necessary to control for differences in market condi-

tions in using these portfolio ratios. Differences in market condi-

tions exist because potential lead banks are located in different

geographic markets and because potential lead banks in the same market

have different decision years. In order to control for market differ-

ences a bank's portfolio ratio is divided by the comparable ratio for

the entire market. Accordingly,.the market adjusted variable for, say,

the loan to asset ratio is equal to the loan to asset ratio of the bank

divided by the ratio of total loans to total assets for the market. If

this adjusted variable is above 1.00 then the bank's ratio exceeds the

market's ratio and conversely. The adjusted portfolio variables are used

in empirical tests of the model.

A bank's relative rate of growth is thought to influence a

bank's decision on lead bank status. It is proposed that the poorer a

bank's relative growth (i.e., the greater the decline in a bank's market

share) the more likely is the bank to become a lead bank in an attempt

to recover its relative position and to improve its profit performance.2

The relative rate of growth is measured by a bank's change in market



1 From an accounting point of view, a bank could conceivably have an
above-average business loan ratio and an above-average consumer loan
ratio. This might occur if the bank made relatively few real estate
loans, security loans, loans to financial institutions, and miscellaneous
loans. Such a loan portfolio does not seem probable, however.
2 The stresses caused by uneven rates of growth and the importance of
a bank's relative position were discussed in Chapter 3, pp. 44~46.





- 67 -


share.1 The change in market share is calculated for the period from

year t-5 to year t, where t is the decision year, using the formula in

equation (3).

t=0 t=-3
(3) MSt 1\ Bankr's Deposits)t 1\ Bank's Depositsy
3 Mrket Deposits)t 3 Mrket Deposits)t
t=- t=

It should be noted that a potential lead bank's market standing may

change drastically. This is due to Florida's unit-banking statutes

and its very rapid growth.


Environment

In considering the environment of a potential lead bank, three

variables are proposed as significant predictors of lead bank status.2

(4) EEt = h (ENC/BKt, ENC/TDt, MIt)

where

EEt = environment of bank's market in year t.

ENC/BKt = encroachment as measured by the change in
holding companies' share of banks in the
market from year t-2 to year t.

ENC/TDt = encroachment as measured by the change in
holding companies' share of market deposits
from year t-2 to year t.

MIt = market's importance as a financial center
in year t.



1 An attempt was made to measure market position over time by regressing
market share upon time for the set of potential lead banks. For about
one-quarter of the banks the coefficient of determination (R2), which is
a measure of the explanatory power of the regression equation, was rather
low so this method was not used. The low R2 was usually caused by market
share being very atypical for one year.
2 Other factors affecting the environment of a bank include market growth
and market concentration. However, these factors do not appear, a priori,
to influence a bank's decision regarding lead bank status. For example,
there is no clear reason to think that the likelihood of a bank selecting
lead bank status is.related to the rate of growth of its market.






- 68 -


The amount of recent holding company activity in a market is

thought to be an influence upon lead bank status. It is hypothesized that

encroachment by holding companies on a bank's local market, thereby

threatening its base of operations, will often result in a defensive

reaction by the bank.1 If the independent bank believes it is at a

disadvantage in competing with holding company affiliates, it can form

a holding company or join a holding company. However, which course of

action, if either, the bank will pursue cannot be decided a priori.

Despite the inability to predict the direction of influence,

it seems important to consider the encroachment effect. Much of the

snowballing of holding company activity observed in Florida and else-

where is probably due to the defensive response of banks to encroach-

ment by holding companies on their home markets.2 Given that the en-

croachment effect has a substantial impact upon the amount of holding

company activity, it is of interest to determine empirically whether or

not encroachment has a systematic effect upon a bank's decision regarding

lead bank status.

The encroachment effect is measured in two ways. The variable

ENC/BK measures encroachment in terms of the change in holding company

control of banks in a~ market. The variable ENC/TD measures encroachment

on the basis of the change in holding company control of market deposits.



For example, "the proposal of the Central National Bank of Cleveland
to acquire American Bank of Commerce in Akron was announced as a response
by Cleveland Banks to the encroachment on their traditional territory
by BancOhio Corporation, in its merger with Akron National Bank and
Trust Company." Thomson and Meadows, p. 3.
2 For a discussion of the snowballing of holding company activity, see
Chapter 2, pp. 24-27 and Chapter 3, p. 35.








































































___ __


- 69 -


Lead bank status is thought to be affected by a market's

importance as a financial center. It is proposed that, cateris

~agibus, the more important is the market as a financial center, the

more likely will be a bank in the market to become a lead bank. The

reason is that the more important a market is as a financial center,

the more likely it is that some banks in the market are informed about

the holding company structure and its possible benefits. The informed

banks are expected to have an influence upon the thinking of other

banks in the market. In other words, important financial centers

possess a stimulating environment which encourages banks to assume

leadership roles in multibank organizations.

A market's standing as a financial center is measured by the

market's share of total deposits in the state in the decision year. One

would expect some correlation between a market's share of deposits and

potential lead banks' deposit size and correspondent business. Nonethe-

less, a market's importance is considered to be a distinct determinant

of lead bank status.


The Statistical Modell

The model of lead bank status is intended to determine the

probability that a bank will opt for lead bank status. The probability

of lead bank status is thought to be a function of a bank's individual

characteristics (IBC) and its environment (EE). Clearly, the object of

prime interest is a probability.




T he discussion is intended to provide the reader with an intuitive
appreciation for why regression analysis is an inappropriate 'statistical
model for this study and why probit analysis is an appropriate statis-
tical model for this study.






- 70 -


The regression model is not a suitable statistical model for

determining the probability of lead bank status. In both ordinary and

generalized least squares regression models, the estimated value for

lead bank status may fall outside the interval from 0 to 1, because

regression models fit straight lines to the data points. .(See Figure

5.1.) An estimated value (probability) for lead bank status of less

than 0 or greater than 1 is inconsistent with interpreting the value

of the dependent variable as a probability.1

Another shortcoming of the regression model is the fact that

it is unable to handle a dichotomous dependent variable. It should be

realized that the actual value of the dependent variable lead bank

status is either, say, O or 1, this is to say, a bank either is or is

not a lead bank.2 Because the dependent variable is dichotomous, the

assumption of homoskedasticity, which is a basic assumption of the

regression model, is not satisfied The assumption of homoskedasticity

means that the expected variance of the error terms is equal for all

observations. In other words, the scatter of data points (the observed

values of the variables) about the estimated regression line is approxi-

mately uniform. (See Figure 5.2.)



1 It makes no sense to say that the probability of a bank selecting
lead bank status is -10 percent or +120 percent.
For example, the expected probability of getting heads from flipping
a coin may be 0.5. Nonetheless, for an actual flip of the coin, the
result is either heads (1) or tails (0).
SIn practice, generalized least squares regression, a technique for
adjusting for heterskedasticity, is unworkable because the calculated
values of the dependent variable are frequently very close to 0 or 1.
As a consequence, the inverse of the estimated disturbance covariance
matrix cannot be calculated (the inverse is singular or nearly singular).





- 71 -


1 agagagemano


Probability
of Lead
Bank Status


Figure 5.1


A Straight Line Fit with Regression Model


Figure 5.2


Homoskedasticity about the Regression Line





- 72 -


The variance of the error term is defined as the square of

the difference between the estimated value of the dependent variable

(the probability of lead bank status) and the actual value of the

dependent variable (either 0 or 1). The variance is weighted by the

probability of "success" (the bank is a lead bank) as "failure" (the

bank is not a lead bank).1 Applying the definition of the vari~ance of
the error term and letting "P" equal the probability of lead bank

status, we have:

(4) Variance of error term =(1-P)2(P)+(0-P)2(1-P)
=(1-2P+-P2)p~p2-p3
= P-P2
=P(1-P)

From (4), it is evident that the variance of the error term

depends on P, the probability of lead bank status;2 and P in turn

depends on the individual bank characteristics (IBC) and on the environ-

ment (EE). Consequently, the variance of the error term varies systema-

tically with the explanatory variables. Thus, the assumption of homo-

skredasticity, which is a basic assumption of the regression model, is

not satisfied.

In view of the problem with the regression model, an alterna-

tive statistical approach is needed to handle a dichotomous dependent



1 Probabilities are used for the theoretical distribution while frequen-
cies are used for the actual (sample) distribution.
2For example, if the expected value of lead bank status equals .9,
this means a probability of, 79 that the value will deviate from expectation
by +. 1 and a probability of .1 that the deviation will be -.9, while an
expected value of .4 means deviations of +.6 with probability .4 and
deviations of -.4 with probability .6. Using equation (4), the variance
equals .09 when P=.9, while the variance is .24 when P=.4.






- 73 -


variable and to restrict the value of the dependent variable to the

interval from 0 to 1. One alternative is the probit analysis model.1

Since probit analysis was developed with reference to the

field of biology, an example from this field may be most helpful in

explaining this statistical method. In biological assay, probit

analysis is used to determine the relationship between the pr ability

that organisms will be killed to the strength of the dose of poison

administered to them. The dependent variable for each organism in the

sample is dichotomous: killed or not killed. Each organism is assumed

to have a dosage threshold, such that a stronger dose will: kill that

organism and a weaker dose will not. Over the population of organisms

of a given kind, these dosage thresholds are assumed to be normally

distributed with a few organisms having very high or very low dosage,

thresholds.vhile most of the organism have intermediate-valued dosage

thresholds. (See Figure 5.3.) A graph of the percentage of insects

killed as the dose of poison increases will give a steadily rising

curve. The curve has an S-shape because the dosage thresholds are

assumed to be normally distributed. In other words, the rate of change

in the percentage killed per unit increase in dose is low in the region

of very small or very large doses of poison but is higher in the inter-

mediate region. (See Figure 5.4.)



SThe basic reference on probity analysis is D.J. Finney, Probit
Analysis (3rd ed.; Cambridge, England: The University Press, 1971).





U' -- -p
Dose of Poison

Figure 5.4

Cumulative Distribution of Insects Killed


- 74 *


Dosage


Dose of Poison

Figure 5.3

Normal Distribution of Dosage Thresholds


Percentage
of Insects
killed





- 75 -


The above ,example may be put into the following mnthe-

matical terms. Suppose the Ath organism has an actual value of the

dose T which is a linear function of the concentration of the

poison (CON) and the time .of; exposure to the poison (EXP) to which

the ith organism is subjected.

(5) Ti = Bo ICON +B2EXPi

Assume that Ti is the threshold value of the dose for the Ath

organism. If the actual value of the dose Ti is greater than or equal

to the threshold value of the dose Ti then the ith organism is

killed. If T is less than Ti then the ith organism is not

killed. The coefficients in equation (5)--i.e., Bo, B1 and B2--are

estimated within the maximum likelihood technique.1 Because the maxi-

man' likelihood I(rathetr- th'an :least squares) technique is used, it is

not necessary to assume homoskedasticity (equal variance in the error

terms).

The above discussion may not have convinced the reader that

the probit statistical model is appropriate for the study of lead bank

status. Figure 5.5 is designed to erase any lingering doubts. On the

horizontal axis of Figure 5.5 is measured bank size and on the verti-

cal axis is measured whether or not the bank has chosen lead bank



1 In order to find the maximum likelihood coefficients, it is neces-
sary to solve a set of rather complicated non-linear equations. The
solutions (the maximum likelihood estimates) are calculated using the
Newton-Raphson method of iteration. An exposition of the maximum
likelihood solution and a sample calculation of the solution are con-
tained in James Tobin, "The Application of ~Multivariate Probit Analysis
to Economic Survey Data," Cowles' Foundation Discussion Paper No. 1,
1955 (unpublished).















-~-s~-aaara~ara*


Size of Bank


- 76 -


Bank Is a
Lead Bank 1

















Bank Is Not
a Lead Bank 0


Line


Figure 5.5


The Probit Line for Lead Bank Status





77 -


status.` As bank size increases, it' is thought that the chances of a

bank choosing lead bank status increase. Some hypothetical data

points are shown as dots on the figure. Essentially what the probit

technique does is to fit an S-shaped curve to the data points as

shown on the figure.





CHAPTER 6

THE DETERMINANTS OF LEAD BANK STATUS


Using the probit statistical model, the model of lead bank

status was tested on the set of 58 potential lead banks listed in

Table A.2. The time period treated by the model is from 1963 through

1971 in terms of decision years or from 1964 through 1972 in terms of

decision years.


Interpreting the Probit Statistical Model

In order to assist one in interpreting the statistical

results, the probit model- is compared with the regression model.

Just as in regression analysis, the probit model assumes that the

dependent variable, lead bank status (LBS), may be expressed linearly

in terms of the explanatory variables, individual bank characteristics

(IBC) and environment (EE); i.e.,' LBS == BO+B IBC+B EE. In probit

analysis the parameters B B1 and B2 are estimated by the maximum

likelihood method while regression analysis uses the method of least

squares. There is no need to consider the maximum likelihood method

in detail. Suffice it to say that the method of maximum likelihood

gives estimators which have desirable properties and the estimators

are easy to find. For large samples, the maximum likelihood esti-

mates are approximately normally distributed. Because of this

approximate normality it is possible to estimate the standard error

for each coefficient and to calculate t-values for each


- 78 -





* 79 -


coefficient.1


The_ Likelihood Ratio Test

In addition to the t-test, the likelihood ratio test may

be used to test hypotheses about the independent variables both

singly (say, IBC = 0) and in combination (say, IBC = 0 and EE = 0).

The likelihood ratio test is closely related to maximum likelihood

estimation and it is easily applied to the probit model which employs

the maximum likelihood technique.

The likelihood ratio is equal to the maximum of the likeli-

hood function when a parameter is not allowed to vary freely

(say, EE = 0) divided by the maximum of the likelihood function when

a parameter freely varies (say, EE # 0).


(1) Likelihood Ratio = A = max L, (EE 0)
max L (EE #~ 0)

Note that max L(EE = 0) will be smaller than or at most equal to

max L(EE # 0) because there is less freedom in maximizing the likeli-

hood function when the hypothesis requires that EE equal zero.

The likelihood ratio test is based on the fact that -21o81

is approximately distributed like chi square with the number of



The probability that a bank will become a lead bank is equal to
the probability that the bank s value for LBS exceeds its threshold
value for LBS. A key portion of the probit model is the assumption
that the threshold values for LBS are normally distributed. In
symbols, Y
Pr(LBS) = 2 e~ _2d
-00
where Y = B0+B1B 2BEE.





- 80


degrees of freedom equal to the number of restrictions on the

parameters. If the likelihood ratio (i.e., -210gly is greater

than the tabled value of the chi-square distribution at some~chosen

level of significance, then the hypothesis of no relationship is

rejected.1 For example, if the likelihood ratio is greater than 2.71,

which is the value of chi .square at a .10 level of significance for

one degree of freedom, then one can reject the hypothesis that EE

equals zero and expect to be right nine out of ten times.


A Stepwise Testing Procedure

A kind of stepwise procedure was employed in testing the

model in order to avoid problems of multicallinearity2 and to con-

serve degrees of freedom. All of the independent variables were

tested one at a time. The more significant variables were then

tested using varying combinations of two variables and so on. In

addition, the initially significant variables were combined with all

other predictors in an attempt to detennine if initially unimportant

variables gained significance when combined with initially significant



In other words, the- critical region for -21ogX is the right-hand
tail of the chi-square distribution. For.a rigorous treatment of
this method, see Alexander Mood, Introduction to the Theory of
Statistics (New York: McGraw-Hill Book Co., 1950), pp. 257-259.
The problem of multicollinearity relates to the high covariance
between two estimators (say, Bl and B2). A principal cause of the
high covarianre is high correlation between the corresponding in-
dependent variables (say, X~ and X,). When multicollinearity exist
among the variables, it is difficult or even impossible to determine
the separate influences of the independent variables. As a result,
the estimated coefficients tend to be unstable and they are not overly
reliable.





- 81


predictors.1

Table 6.1 enumerates the variables used and reports the

mean and standard deviation of each variable. The relatively large

standard deviation for the correspondent business variable is due to

the rather pronounced bimodal distribution of the variable. In

addition, the values of the CB variable are positively skewed. It

should also be noted that consumer loans include single payment loans

while instalment loans exclude single payment loans.


The Results for Single Predictors

The results of the model when the independent variables

were tested one at~ a time are shown in Table 6.2. Not unexpectedly,

the SZ variable proved to ~be a significant predictors of lead bank

status. Another significant predictor of lead bank status was the

variable for correspondent business. The coefficient had a positive

sign as hypothesized.

None of the loan portfolio measures appeared to be good

predictors of lead bank status.3 The very low likelihood ratio values



As an example of this procedure, see Tong Itun Lee, "Alternative
Interest Rates and the Demand for Money: The Empirical Evidence,"
American Economic Review, vol. 57 (December, 1967), pp. 1168-1181.
2---- --- ---
Throughout the discussion, the term "significant" is used to denote
statistical significance. In Table 6.2, both the t-value and the
likelihood ratio test the same hypothesis; namely, that the estimated
coefficient is not significantly different from zero. It is gratify-
ing to note that the two statistical tests produce consistent results.
The ratio of instalment loans to assets was also tried in place of
the consumer loan to asset ratio. (Recall that Lawrence had used the
former ratio and Talley the latter in studying the performance of
holding company affiliates.) The instalment loan to asset ratio was
not significant and had a negative sign whereas a positive sign had
been hypothesized.






- 82 -


Table 6.1

Identification of Variables with Means and Standard Deviations

Standard
Variables Mean Deviation

LBS. Lead bank status. LBS = 1 if
the bank is a lead bank;
LBS = 0 otherwise. .362 .485

SZ. Size of bank in terms of deflated
total deposits ($100 millions). .670 .593

CB. Correspondent business measured
by deflated due to balances
($10 millions). .468 1.298

L/A. Loan to asset ratio of bank
relative to loan to asset fatio
of market. 1.046 .159

CL/A. Consumer loan to asset ratio of
bank relative to consumer loan
to asset ratio of market. .984 .475

IL/A. Instalment loan to asset ratio
of bank relative to instalment
loan to asset ratio of market. .933 .422

CI/A. Commercial and industrial loan
to asset ratio of bank relative
to this ratio for the market. 1.037 .463

MS. Change in market share ~of the
bank over a five-year period. -.0098 .0202

ACT/BK. Holding company activity as
unasured by the change in holding
company control of banks in the
market over a two-year period. .123 .114

ACT/TD. Holding company activity as measured
by the change in holding company
control of total deposits in the
market over a two-year period. .134 .151

MI. Market importance as a financial
center as measured by the market's
share of total deposits in the
state. .102 .073





- 83 -


Table 6.2

Probit Model of Lead Bank Status Using Single Predictors


Likelihood
Ratio Value'

13.87n**


19.84*k*


0.08


0.00


2.18


0.08


0.02


5.76k**-


1.35


5.39**


Dependent
Variable

LBS


LBS


LBS


LBS


LBS


LBS


LBS


LBS


LBS


LBS


Independent .
Variable

+ 2.122 SE
(3.05)***

+ 2.649 CB
(2.20)**

+ 0.297 L/A
(0.28)

+ 0.0086 CL/A
(0.02)

-0.605 IL/A
(1.46)

0.105 CI/A
(0.28)

+ 1.123 MS
(0.13)

-3.896 ACT/BK
(2.27)**

-1.362 ACT/TD
(1.14)

+ 5.463 MI
(2.28)**


Constant

=-1.668


=-0.88


=-0.665


=-0.361


== +0.200


=-0.244


=-0.42


=+0. 089


=-0.179


=-0.932


the hypothesis that the independent
effect upon holding company activ-


aThe likelihood ratio value tests


variable does not


have a systematic


ity.
** Significant at .05 level.
*** Significant at .01 level.

Note: The figures in parentheses are t-values for each estimated co-
efficient.






- 84 -


and t-values for the variables L/A, CL/A and CI/A indicates that

they have no systematic influence upon lead bank status. One may

conclude, at least for Florida banks, that lead banks are not espec-

ially aggressive in seeking loan business. Furthermore, one cannot

characterize lead banks as either wholesale-oriented or retail-

oriented .

The market share variable is not a significant predictor.

Given the very low likelihood ratio value, one may conclude that

changes in a bank's relative standing in its local market do not

influence the bank's decision regarding lead bank status.

The variable ACT/BK, which is used to test the encroachment

hypothesis, is significant with a negative sign.1 The variable

ACT/TD, which measures encroachment in terms of holding company

control of deposits, has a negative sign but it is not statistically

significant. These results imply that encroachment by holding

companies does not provoke a bank to form a holding company. It is

possible that encroachment influences banks to join holding companies

rather than form them.

Some care must be exercised in interpreting this result

because only ten out of the thirty-seven banks which were not lead

banks at the end of 1972 had made the decision to become subsidiaries

of holding companies. Of the twenty-seven banks that have not made



1Racall that it was not possible to determine a priori the sign for
the variable ACT/BK but it was felt that an encroachment variable
could help explain the snowballing of multibank holding company
af filiations .





- 85-


an overt decision it appears probable that most of these banks will

join existing holding companies. A few of these banks, especially the

one bank holding companies associated with large non-bank organiza-

tions, will likely remain independent.

The variable for market importance was statistically signifi-

cant at the .05 level of confidence. The variable has a positive

direction of influence as was originally postulated.2


The Results for Several Predictors

All of the significant predictors3 were paired with each

other and with each initially non-significant predictor. None of the

initially non-significant predictors gained significance when combined

with initially significant predictors.

The results from pairing the significant variables are pre-

sented in Table 6.3. Because of the high correlation between size and

correspondent business (0.93) the coefficients for the two variables

are not overly reliable. It is possible, however, to compare the

relative significance of the two variables. Given that the variable

SE is already present in the equation, the addition of the variable CB



1It may be that a potential lead bank has made the decision to remain
unaffiliated. However, there is no way to distinguish such a bank from
a bank that has made no decision one way or the other. Of course, the
decision to remain unaffiliated is revocable while the decision to be-
come affiliated, once consummated, is not revocable.
2 The reader should not suspect a spurious relationship because of the
fact that the largest banks come from the most important financial
centers. The most important financial centers also supplied the great-
est number of medium-large banks. For example, Dade County, which is
the most important financial center in Florida with 22.6 percent of
state deposits, supplied 13 of the potential lead banks, which is 22.4
percent of the total number of 58 banks.
SThe significant predictors from Table 6.2 are SZ, CB, ACT/BK and MI.














































The likelihood ratio value tests the hypothesis that the two
variables jointly do not have a systematic effect upon holding company
activity.
** Significant ~at .05 level.
*** Significant at .01 level.

Note: The figures in parentheses are t-values for each estimated
coefficient.


-' 86 -


Table 6.3

Probit Model of Lead Bank Status Using Two Predictors


Dependent
Variable


Likelihood
Ratio Valuea


Independent Variables


Constant


LBS = -1.224 + 0.697 SZ
(0.70)

LBS = -1.239 + -2.212 SZ
(2.99)***

LBS = -2.351 + 2.236 SZ
(3.10)***

LBS = -0.394 + 3.592 CB
(2.51)~**

LBS = -1.633 + 2.991 CB
'(2.27)+**


+ 2.362 CB
(1.87)**

-4.147 ACT/BK
(2.18)**

+ 5.806 MI
(2.19)**

-5.542 ACT/BK
(2.23)**

+ 6.442 MI
(2.39)**1R


20.34***


19.32***


18.92****


26.33***t


25.84***


LBS =-0.482


4.139 ACT/BK +
(2.34)**


5.750 MI 11.57***~k
(2.36)n*





- 87 -


increases the likelihood ratio by 6.47, which is significant at the

.05 level of confidence so one can reject the null hypothesis that

CB = 0, given that SZ is already included in the equation. When the

variable CB is entered first, the addition of the size variable in-

creases the likelihood ratio by only 1.02, which is not significant

at the .05 level of confidence. It is possible to conclude that both

variables supply much of the same information but that the correspond-

ent business variable provides information over and above the informa-

tion contained in the size variable while the reverse is not true.2

The two variables ACT/BK and MI appear to provide added

information. Whether combined with each other or when paired with the

variables SZ or CB, these variables increase the likelihood ratio by a

significant amount.

The final equations for the model of lead bank status are

listed in Table 6,4.3 The two equations are highly significant and

the signs of all the coefficients are correct. One may conclude that

the larger is a bank's size and correspondent business, the higher will

be the probability that the bank chooses lead bank status. If the

bank's market is an important financial center, this will increase the



1With the SZ variable in the equation, the likelihood ratio value is
13.87 (see Table 6.1). With the addition of the CB variable, the
ratio rises to 20.34, an increase of 6.47.
2The t-values for the two variables confirm the finding that the CB
variable is relatively more significant than the SZ variable.
The equations are final in the sense that the addition of a fourth
variable does not increase the likelihood ratio by a statistically
significant amount, with one exception. The sole exception is when the
variable CB is added to the first equation of Table 6.4. However, the
high collinearity between the SZ variable and the CB variable makes it
inappropriate to include both variables in the same final equation.




Full Text

PAGE 1

THE ANALYTICS OF MULTIBANK HOLDING COMPANY BEHAVIOR By GREGORY EDWARD BOCZAR A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1973

PAGE 2

flKS: J 1262 08552 8585

PAGE 3

Copyright by Gregory Edward Boczar 1973

PAGE 4

ACKNOWLEDGMENTS '" ' -'! ^ ' " There are many Individuals who have assisted in the completion of the dissertation. Special mention needs to be made of the assistance provided by Professor Ralph H. Blodgett, the chairman of my dissertation committee. He was a steady source of encouragement, advice and tolerance. His willingness to smooth over some of the obstacles to getting an education was essential to my getting this far. I am appreciative of the support in various forms provided by the Board of Governors of. the Federal System. Thanks go to Bob Creller and Joe Holmes for very competent programming assistance. Several of my colleagues were most generous with their time and advice i _ . including Steve Rhoades, Bud Talley, Art Fraas and Bob Lawrence. Steve Rhoades especially forced me time and again to clarify my thinking and to improve my writing. The Research Library at the Federal Reserve Board always provided professional service beyond the call of duty. Both Ann Marie Martin and Judy Back deserve special mention. Statistical and clerical assistance were competently provided by Anita Barley, Gloria Battle, Jackie Rupp, Ronnie McWilliams and Lucy Nelson among others. Typing of numerous drafts was done by Janet Breen, Ann Marie Cowan and Joan Albamonte. The person who deserves the most thanks is my wife Marilyn. She has helped in every conceivablie way assuming family responsibilities so that I could work, acting as a sounding board for my ideas, and encouraging me when I was discouraged. iv

PAGE 5

TABLE OF CONTENTS Page ACKNOWLEDGMENTS. iv LIST OF TABLES |. . . . .viii ! LIST OF FIGURES ..... x ABSTRACT j ^i Chapter 1. PURPOSE AND DESIGN OF THE STUDY. .'. 1 Purpose of the Study 1 Design of the Study: A Homogeneous Population. . , 1 Design of the Study: The Extreme Instance 5 Design of the Study: Decisions and Independent. . Decision Makers 7 Organization of the Study.................. 7 Definitions and Terminology. 8 2. A HISTORICAL AND LEGAL REVIEW OF BANK HOLDING COMPANIES..... 11 Historical Review Up to 1956 11 Regulation of Bank Holding Companies 14 Banking Act of 1933 14 The Bank Holding Company Act of 1956. 16 The 1966 Amendments. 18 The 1970 Amendments.. 20 Development of Bank Holding Companies After 1956 21 3. A CRITICAL REVIEW OF THE LITERATURE 28 A Selective and Critical Review: An Aid to Model Building , 28 Fischer: Bank Holding Companies 28 Lawrence: The Performance of Bank Holding Companies..... 31 McLeary: Bank Holding Companies — Their Growth and Performance.......... 34 Weiss: Bank Holding Companies and Public Policy......... 35 Talley: The Effects of Holding Company Acquisitions on Bank Performance. 36 Piper: The Economics of Bank Acquisitions by Holding Companies 38 Possible Efficiencies of the Holding Company Structure.. 41 Schweitzer: Economies of Scale and Holding Company Affiliation. 42 Upson and Jessup: Returns from Bank Holding Companies... 43 Alhadeffs: Recent Bank Mergers 44 The Relative Standing of Banks i ............... . ^5 Summary ^6

PAGE 6

Chapter Page 4. BANK HOLDING COMPANIES: PROFIT MAXIMZERS ORSIZE MAXIMIZERS. . . . . . . 48 External Expansion and Profit Maximization 49 The Alternative Hypothesis: Size Maximization. 50 Profit Maximization Versus Size Maximization 52 Application of the Profit Maximization Hypothesis..... 53 5. A MDDEL OF LEAD BANK STATUS 55 Definition of Lead Bank Status.. 55 The Set of Potential Lead Banks....... 56 Lead Banks and Chains 58 The Time Framework of the Study.......... 59 A Framework for the Empirical Model. ^ 60 The Regulatory Position. 61 Individual Bank Characteristics.. 62 Environment. 67 The Statistical Model.... 69 6. THE DETERMINANTS OF LEAD BANK STATUS ... 78 Interpreting the Probit Statistical Model... 78 The Likelihood Ratio Test 79 A Stepwise Testing Procedure 80 The Results for Single Predictors 81 The Results for Several Predictors 85 7. A MDDEL OF HOLDING COMPANY ACTIVITY, 93 The Profit Maximization Hypothesis and Holding Company Activity 93 Measures of Holding Company Activity. 94 The Dependent Variable and Tobit Analysis. ........... . 96 Metropolitan Markets ^ 97 The Time Period..... . 99 A Framework for the Explanatory Variables i ..102 The Regulatory Environment ; 102 Expected Profitability of the Market. 103 The Snowball Effect 106 Alternative Measures of Explanatory Variables. 108 The Model of Holding Company Activity. • • • • ^^^ 8. THE DETERMINANTS OF HOLDING COMPANY ACTIVITY... 109 Interpreting the Tobit Statistical Model ..109 The Results for Single Predictors HI The Results for Several Predictors .120 9. CONCLUSIONS FOR CASE ANALYSIS AND DIRECTIONS FOR FUTURE RESEARCH.... 130 A Summary of the Results. 130 Conclusions for Cases Analysis 131 vi

PAGE 7

Chapter Page Directions for Future Research 134 APPENDIX , 135 BIBLIOGRAPHY 139 vii

PAGE 8

LIST OF TABLES Table Page 1.1 Multibank Holding Companies In Florida as of December 29, 1972........ ..... 3 1.2 Deposit-Size Distribution of Banks in Multibank Holding Companies as of December 29, 1972................ 4 1.3 Population, Personal Income, Deposits and Banks in Florida, 1960-1972. i . . . 7 2.1 Registered Bank Holding Companies: Number, Number of Banks and Branches, and National Share of Offices and Deposits, 1956, 1960, 1965 , 22 2.2 Offices and Deposits of Banks Affiliated with Registered Bank Holding Companies, 1965-1970. 24 :2.3 Percentage of Banking Offices in Each State and Percentage of Deposits in Each State Held by Registered Bank Holding Companies, 1965 and 1970. 25 6.1 Identification of Variables with Means and Standard Deviations 82 6.2 Probit Model of Lead Bank Status Using Single Predictors... 83 6.3 Probit Model of Lead Bank Status Using Two Predictors 86 6.4 Final Equations for the Model of Lead Bank Status..... 88 6.5 Correlations Between Variables in Model of Lead Bank Status. 90 6.6 The Probability of Lead Bank Status. 91 7.1 Holding Company Proposals Denied or Abandoned, 1963-1972... 95 7.2 Metropolitan Banking Markets in Florida. 100 7.3 Multibank Holding Company Activity in Florida, 1960-1972 101 8.1 Identification of Variables with Means and Standard Deviations. 112 8.2 Tobit Model of Holding Company Activity Using Single Predictors 115 viii

PAGE 9

Tlble Page 8.3 Tobit Model of Holding Company Activity Using Two Predictors. 121 8.4 Tobit Model of Holding Company Activity Using Three Predictors 124 8.5 Final Equations for the Model of Holding Company Activity... 125 8.6 Correlations Between Variables in Model of Holding Company Activity. ;........ 127 8.7 The Probability of Activity and the Expected Value of Activity ^ . . ., 129 A. 1 The Set of Potential Lead Banks 135 A. 2 Decisions Made by Potential Lead Bank Through December 31, 1972......... 137 ix

PAGE 10

LIST OF FIGURES Figure Page 5.1 A Straight Line Fit with Regression Model................. 71 5.2 Homoskedasticity about the Regression Line 71 5.3 Normal Distribution of Dosage Thresholds^ 74 5.4 Cumulative Distribution of Insects Killed 74 5.5 The Probit Line for Lead Bank Status. 76 8. 1 The Tobit Line for Holding Company Activity. 110

PAGE 11

Abstract of Dissertation Presented to the Graduate Council of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE ANALYTICS OF MULTIBANK HOLDING COMPANY BEHAVIOR By Gregory Edward Boczar December, 1973 Chairman: Ralph H. Blodgett Maj,or Department: Economics Multibank holding companies have developed rapidly since 1965. The purpose of the study is to explain the behavior of these companies — their formation and expansion. The study is limited to formations and acquisitions in Florida for the last decade. By restricting the study to a single state, it is possible to control for a number of significant differences among holding companies. Also, Florida is an especially ideal state for study because it has experienced more growth of holding companies, population, personal income and bank deposits than any other major state. By way of introduction to the recent upsurge of holding cOTijpany growth, the historical and legal framework of multibank holding companies is described. A critical review of the scholarly literature on holding companies is also undertaken. Both the historical and legal review and the review of the literature provide information on the motivation and behavior of multibank holding companies. A profit maximization hypothesis is selected as the " . xi

PAGE 12

appropriate analytical framework because the hypothesis is more workable and somewhat more consistent with the facts than its chief rival, the size maximization hypothesis. In order to explain empirically the behavior of multibank holding companies, a model of lead bank status and a model of holding company activity are developed and tested. The probit statistical technique rather than the more conventional regression approach was used to test the model of lead bank status because the dependent variable only takes the value of zero or one. The explanatory variables for the model of lead bank status fall into three categories: regulatory position, characteristics of the bank and environment of the bank. It was not possible to measure the regulatory position of the Board of Governors of the Federal Reserve System because of the lack of overt actions by the Board. In terms of an individual bank's characteristics, the amount of correspondent business done by the bank and the absolute size of the bank proved to be significant determinants of lead bank status. No evidence was found to support the idea that banks with declining market shares tend to anchor holding companies. Also, lead banks cannot be characterized as either wholesaleor retail-oriented nor are they especially successful in gaining loans in general. In terms of the environment of a bank, the relative Importance of a bank's market as a financial center had a favorable influence on lead bank status. Finally, if relatively few banks in the recent past have affiliated with holding companies, a bank is more likely to become a lead bank than if a substantial portion of the banks in the market have come under holding company control. xli

PAGE 13

The model of holding company activity was tested using the toblt statistical method. The tobit method is appropriate because the dependent variable does not take on negative values, having a lower limit of zero. The explanatory variables for the model of holding company activity are divided into three categories: regulatory environment, expected market profitability and a snowball effect. The regulatory attitude of the Federal Reserve Board toward holding company activity in a market is measured by the level of market concentration. As postulated, concentration had a dampening effect upon the amount of activity in a market. In terms of expected market profitability, both current market profits (income as a proportion of capital of asset) and deposits per capita or per bank had a significant positive influence on holding company activity. The level of per. capita personal income was found to have a positive and highly significant effect on holding company activity. Neither population per bank nor short term (two-year) changes in population, personal income or deposits have a systematic Influence on holding company activity. Some evidence was found to support the contention that holding company activity tends to snowball over time. xiii

PAGE 14

CHAPTER 1 PURPOSE AND DESIGN OF THE STUDY Multibank holding companies have developed rapidly in the United States since 1965. This activity by bank holding companies has been concentrated in a dozen unit-banking and limited-branching states. The most active state has been Florida. In. 1965 there were six multibank systems in Florida which controlled 55 banks. By December 1972, there were 27 bank holding companies which controlled 280 banks. In terms of number of multibank systems and in terms of number of affiliated banks, Florida currently ranks first in the nation. Purpose of the Study The purpose of the study is ta explain the formations of, and acquisitions by, multibank holding companies. In order to understand the factors which influence holding company formations, a model of lead bank status is developed and then empirically tested. In order to understand the factors which cause holding companies to be more active in one market than another, a model of holding company activity is developed and then empirically tested. Both models are developed within the framework of a profit maximization hypothesis. Design of the Study; A Homogeneous Population A recent study of multibank holding companies concluded: It is almost impossible to make an unqualified statenent concerning the structure, organization, or operation of holding companies which will fit all situations. Bank holding companies, as they have developed in this country, tend to reflect the special circumstances of their . 1 -

PAGE 15

2 organization and their location. The factors which brought them into existence have differed over time, as have the financial requirements of the localities or regions in which they operate. Given a banking system with striking differences in bank size, operating powers, management philosophies, supervisory status, and traditions, it is not surprising that bank holding companies are found to differ significantly one from the other. ...It [the above] makes difficult the task of the analyst who seeks absolutes...;^ Because of this wide diversity among bank holding companies, the usual ceteris paribus assumptions are not satisfied in a general study of these companies. In order to control for the significant differences among holding companies, the study is restricted to formations and acquisitions in Florida for the last decade. By limiting the study to one state, we control for differences in banking traditions and management philosophies which appear to vary from one state to another and from one region to another.' Of course, all holding company banks in the study will be subject to the same state banking laws. In addition, there are no striking differences among Florida holding companies in terms of total deposits controlled nor in terms of number of banks controlled. (See Table 1.1.) The size distribution of affiliated banks is similar among holding companies except for the billion dollar First National Bank of Miami, the lead bank for Southeast Banking Corporation. (See Table 1.2.) In order to have a more homogeneous population, the study is limited to the banking activities of multibank holding companies. Thus, ^" Carter H. Golembe and Associates , The Future of Registered Bank Holding Companies (Washington, D. C. : The Association of Registered Bank Holding Companies, 1971), pp. 35-36. 2 Ibid . . p. 29.

PAGE 16

-3 Table 1. 1 Multlbank Holding Companies in Florida as of December 29, 1972 Rank Name and Location of Holding Companies 1 Southeast Banking Corporation, Miami 2 Barnett Banks, Jacksonville 3 Florida N/B, Jacksonville 4 First at Orlando, Orlando 5 Atlantic Bancorporation, Jacksonville 6 First Financial Corp., Tampa 7 First Florida Bancorp., Tampa 8 Ellis Banking Corp. , Brandenton 9 Consolidated Bankshares, Ft. Lauderdale 10 Exchange Bancorporation, Tampa 11 United Bancshares, Miami 12 City National Bank Corp. , Miami 13 Pan American Bancshares, Miami 14 Charter Bankshares, Jacksonville 15 Broward Bancshares, Ft. Lauderdale 16 First State Banking Corp., Miami 17 Florida Commercial Banks, Miami 18 First Bancshares, Boca Raton 19 American Bancshares , North Miami 20 Palmer Bank Corp., Sarasota 21 First National Bankshares, Pompano 22 Combanks Corporation, Winter Park 23 Citizens Bancshares, Hollywood 24 Central Bancorp. , Miami 25 Community Banks, Seminole 26 Florida Bancorp, Pompano 27 Jefferson Bancorporation, Miami Number

PAGE 17

Table 1.2 Deposit-Size Distribution of Banks in Multibank Holding Companies as of December 29, 1972 Rank^ Name of Holding Company 1 Southeast Banking Corp. 2 Barnett Banks 3 Florida National Banks 4 First at Orlando 5 Atlantic Bancorporation 6 First Financial Corp. 7 First Florida Bancorp. 8 Ellis Banking Corp. 9 Consolidated Bankshares 10 Exchange Bancorporation 11 United Bancshares 12 City National Bank Corp. 13 Pan American Bancshares 14 Charter Bancshares 15 Broward Bancshares 16 First State Banking Corp. 17 Florida Commercial Banks 18 First Bancshares 19 American Bancshares 20 Palmer Bank Corp. 21 First National Bankshares 22 Combanks Corporation 23 Citizens Bancshares 24 Central Bancorp 25 Community Banks 26 Florida Bancorp 27 Jefferson Bancorporation Number of Banks in Each Size Class^ Greater 100 25 Less than 500 to 500 to 100 than 25 9 15 9 14 10 5 5 5 2 2 2 1 3 1 2 1 4 4 3 1 3 2 2 1 1 1 8 19 19 10 9 3 18 9 4 7 2 4 7 1 3 1 2 3 2 1 3 3 1 4 2 2 Rank is based on total deposits as of December 29, 1972. Size classes are in terms of $millions of deposits. Source: Florida Bankers Association; Board of Governors of the Federal Reserve System.

PAGE 18

one bank holding companies and the nonbahking activities of multlbank holding companies are excluded from the study. Furthermore, no Florida bank holding company appears to be a purely financial holding company; that Is, a company which acquires and holds bank stock solely for investment purposes. Also, there are no Florida companies which have grandfathered banking subsidiaries In other states. By concentrating on the period from 1963 through 1972, the study will cover a period of sustained holding company activity by newlyformed or newly-active holding companies. During this time period the statewide bank holding companies were very active with the exception of the Florida National Banks. Design of the Study; The Extreme Instance As indicated above, Florida has recently experienced more holding company activity than any other state. In terms of population growth for the last decade, Florida was the fastest growing major state in the U.S. and was second only to Nevada among all states. During the past decade, Florida had the greatest percentage Increase in personal income of any major state and was second only to Nevada among all states. For the period from 1960 to 1972, deposits of Florida banks Increased by 406.4 percent which is the highest percentage increase in total deposits of any major state. More banks were opened in Florida from 1960 to 1972 2 One researcher In studying holding company acquisitions nationwide found a considerable difference in the results of his model by dividing the population into growth-oriented and nongrowth-oriented bank holding companies. R. Charles Moyer, A Model of the Determinants of Registered Bank Holding Company Acquisitions . Doctoral Dissertation, Graduate School of Business, University of Pittsburgh, 1971, p. 100. 2 ' Florida Banking and Its Largest Banking Institutions . Allen C. Ewlng & Co. , Jacksonville,, Florida, 1972, pp. 5-6.

PAGE 19

6 than in any other state. Finally, Florida is by far the most underbanked state in the country. ^ What we have is a cluster of events which can be characterized 2 as an extreme instance. In an extreme instance, the variables of interest are subject to major shifts. In this study, there are large changes in holding company activity, population, personal income, bank deposits and number of banks in a short period of time. (See Table 1.3.) The extreme instance is a way of approximating a controlled experiment. Even if the ceteris paribus assumption cannot be strictly enforced, nonetheless the cetera will be relatively small when compared to the large changes in the variables of interest. Thus, the extreme instance provides the researcher with a nearly perfect situation to study. In order to generalize from the extreme instance, the behavior observed must accurately reflect the properties of the system and not be pathological. In general, as long as (1) the system responds in a systematic and controlled fashion rather than in an erratic and disorganized manner and (2) the actions were taken as a result of deliberation rather than spontaneously without calculated intent, the stress created by the extreme instance can be assumed to be producing a characteristic response.^ Florida's population per banking office was 12,738 in 1970. The next least banked state was Illinois with a figure of 9,216. The entire discussion of extreme instances is taken from Oliver E. Williamson, The Economics of Discretionary Behavior: Managerial Objectives in a Theory of the Firm (Englewood Cliffs, N.J. : PrenticeHall, Inc. , 1964). 3 Ibid. , pp. 87-88.

PAGE 20

Table 1.3 Population, Personal Income, Deposits, and Banks in Florida, 1960-1972

PAGE 21

-;8 Chapter 2 provides a brief history of the growth of multibank holding companies. The legislation affecting bank holding companies is reviewed. Chapter 3 consists of a critical review of recent scholarly work on bank holding companies. Chapter 4 considers the profit maximization and size maximization hypotheses. The profit model is found to provide a useful framework for analysis. Chapter 5 presents a model of lead bank status. Both an individual bank's characteristics and a bank's environment are thought to influence a bank's decision regarding lead bank status. The probit statistical model is used to, test the model. Chapter 6 presents the empirical results for the model of lead bank status. Chapter 7 presents a model of holding company activity. The regulatory environment, expected market profitability and the snowball effect are thought to determine the level of holding company activity in different metropolitan markets. The tobit statistical model is used to test the model. Chapter 8 presents the empirical results for the model of holding company activity. Chapter 9 summarizes the results of the study and discusses the implications of the study for case analysis. Areas for future research are suggested. Definitions and Terminology Terms used in the study are defined below.

PAGE 22

A bank holding company is defined as any company (or trust) which owns or controls 25 percent or more of the stock of each of two or more banks. Such companies are regulated by the Federal Reserve Board and must register with it. ! In this paper, the term "bank holding company" always refers to a multibank holding company which is registered with the Federal . i 2 Reserve Board. The terms "group banking" and "multibank system" are used as synonyms for the term "bank holding company". Chain banking refers to the ownership of controlling interest 3 in two or more banks by an individual or informal group of individuals. It is important to distinguish chain banking from bank holding companies or group banking. Chains are subject to individual control while groups are subject to corporate rule. A group customarily has some form of central management while a chain typically does not. Also, groups are frequently headed by a major metropolitan bank while chains rarely have such an arrangement. Ordinarily corporations have no inherent power to hold stock in another corporation. Some state laws permit the acquisition of bank stock by corporations; other states restrict or prohibit the acquisition of bank stock. A useful summary of the state laws on this matter are provided in Bank Holding Company Facts . Association of Registered Bank Holding Companies, Washington, D.C. , 1972. 2 The only exception is Florida National Banks (formerly the DuPont Trust) which was not required to register with the Federal Reserve Board until passage of the 1966 Amendments. In this study Florida National Banks is treated as if it became a registered bank holding company when the Bank Holding Company Act of 1956 was approved. 3 The term "controlling interest" is deliberately vague. For a discussion of the problems involved in defining chain banks see Jerome C. Darnell, "Chain Banking," National Banking Review , vol. 3 (March, 1966) , pp. 308-309.

PAGE 23

10 The lead bank of a holding company is the bank which the management of a holding company designates as the lead bank. The lead bank is generally the largest affiliate and it is usually located in the same city as the headquarters of the bank holding company. Frequently there are officers and directors common to the lead bank and the holding company. Holding company activity refers to decisions favoring holding company formations and acquisitions. The formation of a bank holding company requires the approval of the Board of Governors of the Federal Reserve System under Section 3(a)(1) of the Bank Holding Company Act, as amended. The acquisition of a bank by a holding company must be approved by the Board of Governors under Section 3(a)(3) of the Bank Holding Company Act, as amended.

PAGE 24

CHAPTER 2 A HISTORICAL AND LEGAL REVIEW OF BANK HOLDING COMPANIES The Bank Holding Company Act became law in 1956. The statute did not appear to have much immediate effect upon holding company activity nationwide or in Florida. However, in 1965 there began a significant upsurge in bank holding company activity and in 1966 the Act was amended in a way which encouraged a further surge of interest in these companies. By way of introduction to these recent events, a brief historical review is undertaken. Historical Review Up to 1956 The beginnings of the holding company framework of operation are to be found in the early history of American banking. During the first half of the nineteenth century many of the multiple office systems closely resembled bank holding companies. Examples of institutions which resembled holding companies would include the First and the Second Bank of the United States and later several state branching systems in the Midwest. At the turn, of the century, several bank holding companies were formally incorporated in the Northwest. Bank holding companies developed rapidly in the decade of the twenties. Nearly all the major systems which registered as a result of the Bank Holding Company Act of 1956 were organized from 1928 to 1930.^ Gerald C. Fisher, Bank Holding Companies (New York: Columbia University Press, 1961), pp. 4-8. W. Ralph Lamb, Group Banking (New Brunswick, N.J.: Rutgers University Press, 1962), pp.. 82-83. 11:-

PAGE 25

12 The rapid expansion of group banking in this period cannot be attributed to any single factor but rather to diverse and complex factors. ...One would have to consider: the farm depression and the failure of small rural banks; the general consolidation movement in many other industries; the "snowball effect" as one bank holding company was formed and banks organized others to maintain their; competitive position, correspondent business, and ! prestige; the restrictions on branch banking and the belief that the laws would soon change; and the investor's eagerness to purchase stock, including holding company shares. Perhaps the most important of these diverse factors was the restrictions on branch banking. As metropolitan areas grew, the larger institutions, which usually are located downtown, became less convenient sources of banking services as compared to suburban banks. Congestion which existed in all major cities made banking at a single office, usually located in the business district, very difficult. In addition, retail as opposed to wholesale banking was growing in importance and many bankers wished to adjust ^ their facilities to meet this new market pattern. One way for a large unit bank to adjust to a new market pattern was to organize or to join a holding company. The economic, regulatory and psychological conditions cited above had resulted in ninetyseven groups by 1931. These groups controlled 978 banks and 1,219 branches, and they held 22 percent of the 3 loans and investments of all commercial banks. However, the rapid Gerald C. Fischer, American Banking Structure (New York: Columbia University Press, 1968), p. 95. 2 '' Fischer, Bank Holding Companies , pp. 23-24. 3 U.S., Congress, House, Committee on Banking and Currency, Control and Regulation of Bank Holding Companies . 84th Cong. . 1st sess. , 1955, p. 33.

PAGE 26

13;.growth of bank holding companies ended with the beginnings of the great depression of the 1930' s. The 1930's constituted a period of retrenchment for bank holding companies. The depression forced a drastic reorganization of the banking structure. By the end of 1934 the number of commercial banks had dropped to about 15,400 as compared to nearly 29,000 in 1920. Data released by the Federal Reserve Board for 1936 showed fifty-two groups operating 479 banks, a decline of roughly 50 percent from the number of groups and affiliated banks in operation in 1931. The depressed state of the economy, the relaxation of legislation restricting branch banking and the general antagonism toward holding companies in general severely curtailed interest in group banking well into the next decade. As a result, the position of groups relative to commercial banks in general showed little change between the mid-1930's and 1948. In 1937 these firms controlled 7 percent of the banking offices and 11 percent of the bank deposits of all commercial banks. These figures were almost unchanged in 1948.^ During the early postwar period, 1947-1953, the growth of bank holding companies largely paralleled the expansion of commercial banking in general. The proportion of U.S. offices and deposits controlled by bank holding companies remained essentially unchanged.^ Comptroller of the Currency, Annual Report (Washington, D.C. : Government Printing Office, 1964) , pp. 7-8. 2 Fischer, Bank Holding Companies , pp. 35-37. 3 Thomas R. Piper, The Economics of Bank Acquisitions by Registered Bank Holding Companies (Boston: Federal Reserve Bank of Boston, 1971), pp. 47,49.

PAGE 27

14 . There was a marked acceleration In the expansion of bank holding companies between January 1954 and June 1956. This increase in the growth rate of group systems was dominated by the acquisitions of Transamerica (later Western Bancorporation) and the Marine Midland Company. Two major reasons for the spurt in growth for this 2-1/2 year period were: the fear of restrictive federal legislation and the efforts to rebuild Transamerica following the sale of its interest in 2 Bank of America. Regulation of Bank Holding Companies Commercial banking in the United States has been subject to regulation and supervision by federal and state authorities since the 1780*s. However, since the bank holding company itself is not a bank, it did not come under state or federal banking laws. The rapid expansion of bank holding companies in the 1920' s stirred congressional interest in establishing some degree of federal regulation of group banking. After extensive hearings by the House and Senate over a three ' 3 year period, the Banking Act of 1933 (Glass-Steagal Act) was. passed. Banking Act of 1933 The Banking. Act of 1933 granted the Federal Reserve Board limited powers to regulate holding company affiliates. A holding company affiliate ^ A list of all 340 acquisitions made by holding companies between the end of World War II arid December 31, 1967, is provided in Piper, pp. 18-26. 2 Fischer, Bank Holding Companies , pp. 39-42. 3 The full text of the Banking Act of 1933 is contained in the June 1933 issue of the Federal Reserve Bulletin.

PAGE 28

15 was defined as any organization which controls the majority of the stock of a Federal Reserve System member bank. A firm which fell within this definition must obtain a voting permit from the Board of Governors to vote the stock of a subsidiary which is a member bank. Before granting a voting permit, the Board of Governors considered the financial condi, 2 tion of the applicant and the general character of its manaigement. There were several obvious wealoiesses in this Act. First, registration was not required of systems composed, of only nonmember state banks and it was possible for some groups to control their banks without obtaining a voting permit. Second, the statute did not attempt to regulate the formation or expansion of bank holding companies. Third, group systems were permitted to continue investing in nonbanking enter" 3 ' prises. The banking authorities sought the enactment of regulatory legislation to supplement the bank holding company provisions of the 1933 Act. Recommendations of the regulatory agencies in the 1940' s asked for a curb on the further expansion or formation of groups. After 1950, the position taken by the federal supervisory agencies moderated because in many instances the groups were performing valuable service in upgrading 4 ' " the management and condition of affiliated banks. ^ "Banking Act of 1933," Federal Reserve Bulletin , vol. 19 (June, 1933), p. 385. ^ "Federal Reserve Act Digest," Encyclopedia of Banking Laws (Hartford, Conn.: Lamont Cross 6e Company, 1964), pp. 8-9. '3 Fischer. American Banking Structure , pp. 103-104. Lambv pp, 177-193. Also, refer to Benjamin J. Klebaner, "The Bank Holding Company Act of 1956," Southern Economic Journal , vol. 24 (January, 1958), pp. 313-326.

PAGE 29

16 The Bank Holding Company Act of 1956 The Bank Holding Company Act of 1956 was intended to accomplish several objectives: (1) to define bank holding companies; (2) to prevent holding companies from acquiring banks across state lines; (3) to control the formation and expansion of multibank systems within their "home" states; (4) to preserve the historical separation between the suppliers of money and the users of money. A bank holding company was defined in the Act as any company which owned or controlled 25 percent or more of the stock of two or more banks or which controlled In any manner the election of a majority of the directors of two or more banks. Companies falling within the statutory definition were required to register with the Board of Governors, to disclose prescribed information in reports to the Board, and to submit to examination by the Federal Reserve System. The Board's consent was required before a bank holding company could be formed, before a bank holding company could acquire over 5 percent of the voting stock or substantially all the assets of any bank, or before two bank holding companies could merge. In deciding whether to. grant approval for these actions, the Board was required to consider the following factors: (1) the financial history and condition of company or companies and the banks concerned; (2) their prospects; (3) the character of their management; (4) the convenience, needs, and welfare of the communities and areas concerned; and (5) whether or not the effect of the acquisition or merger or consolidation would be to expand the size or extent of the bank holding 1 William F. Upshaw, "Antitrust and the New Bank Holding Company Act: Part II," Monthly Review . Federal Reserve Bank of Richmond (March , 1971), p. 3.

PAGE 30

17 system involved beyond limits consistent with adequate and sound banking, the public interest, and preservation of competition in thefield of banking.! The first three factors pertained to the organization's solvency, asset condition, capital and operations, thus continuing the underlying concern for depositors reflected in the 1933 Act. The last two factors represent a significant departure from earlier legislation. For the first time, concern is shown for the competitive health of the banking industry. It is recognized that the acquisition of a bank involves important nonsafety-oriented considerations such as the expected effects on banking competition, the possible introduction of new services at the acquired bank, and changes in its lending behavior and pricing policies. Under the 1956 Act, a registered group was prohibited from engaging in any business other than banking, managing banks, or providing certain services to subsidiary banks. The holding company law also deals with intra-system transactions, restricting lending and credit operations .• 3 of groups. The restrictions placed on intra-system dealings made it difficult for bank holding companies to perform the function of facilitating loan participations and other joint credit transactions. In fact, the 1956 legislation effectively made it easier for correspondent banks than for holding company affiliates to take joint action.^ "Bank Holding Company Act of 1956," Federal Reserve Bulletin , vol. 42 (May, 1956), p. 446. o "Federal Laws Regulating Bank Mergers and the Acquisition of Banks by Registered Bank Holding Companies," Economic Review . Federal Reserve Bank of Cleveland (January, 1971), pp. 20-21. ^ "Bank Holding Company Act of 1956," Federal Reserve Bulletin , vol. 42 (May, 1956), p. 448. 4 .George R. Hall, "Bank Holding Company Regulation," Southern Economic Journal , vol. 31 (April, 1956), p. 344.

PAGE 31

18 The Board of Governors in its annual report to Congress in 1958 discussed the regulation of holding companies. The Board had encountered substantial difficulty in balancing "convenience and needs" and "competitive impact" considerations, both of which the Board is required to consider in passing upon applications. To compound the problem, the terminology used in connection with the question iof "competitive impact" was imprecise. Another problem of the 1956 Act involved the severe restrictions placed on intra-system loan participations. In order to correct the problem, the Board recommended repeal of the restrictions. The report also recommended that one bank holding companies come under regulation and that several exemptions of organizations from the holding company law be repealed. The 1966 Amendments The enactment of the 1966 Bank Holding Company Act Amendments satisfied several of the Board's recommendations for changes in the law. Several categories of exemption from the Act were repealed; nonbusiness long-term trusts were brought within the coverage of the Act; the standards and antitrust procedures made applicable to bank mergers by the Bank Merger Act of 1966 were made applicable also to bank holding company cases; and the provisions of Section 6 of the Act, which restricted inter2 subsidiary loans and investments, were repealed. The Amendments of 1966 clarified Congressional intent with respect to the relative importance of convenience and needs factors "Report Under the Bank Holding Company Act," Federal Reserve Bulletin , vol. 44 (July, 1958), pp. 776-796. ^ "Amendments to Bank Holding Company Act." Federal Reserve Bulletin . vol. 52 (July, 1966), p. 966.

PAGE 32

19 and considerations involving competitive impact. The Board was directed not to approve: (1) Any acquisition. . .which would result in monopoly, or which would be in furtherance of any combination or conspiracy to monopolize or attempt to monopolize the business of banking in any part of the United. States, or (2) Any other proposed acquisition. . .whose effect in any section of the country may be substantially to lessen competition, or tend to create a monopoly, or which in any manner would be in restraint of trade, unless it finds that the anticompetitive effects of the proposed transaction are clearly outweighted in the public interest by the probable effect of the transaction in meeting the convenience and needs of the community to be served. 1 Contrary to the wishes of the Federal Reserve Board, the 1966 legislation continued to exempt one bank holding companies. Beginning in 1968, a dramatic surge in the growth of one bank holding companies stirred Congressional interest in bringing these companies within the purview of the Act. By the end of 1968, seven of the ten largest commercial banks in the United States had formed one bank holding companies. A year later the list included 43 of the 100 largest banks. ^ The most important factor motivating banks to adopt this form of organization was the fact that one bank holding companies were not subject to regulation and therefore were able to engage in any nonbanking activity. The increasing costs of funds in 1968 and 1969 and the competitive disadvantage of the interest ceilings imposed by Regulation Q encouraged banks to organize these companies because a ^ bank holding company could issue commercial paper, just as any other ^ Ibid. , pp. 967-968. -~ — o "The 1970 Amendments to the Bank Holding Company Act: One Year Later," Business Conditions . Federal Reserve Bank of Chicago (December, 1971) p. 3.

PAGE 33

20 corporate borrower. Another factor stimulating change was the growing Investment by banks In data processing equipment and personnel. The one bank holding company structure allows for the establishment of a subsidiary which may adopt wage and personnel policies needed to attract specialized personnel and the subsidiary is well suited to marketing excess machine capacity. Also, the organization of a holding company is one, and perhaps the only, method owners of banks may use to Insulate themselves against litigation attacking their right to enter new areas not specifically authorized by statute.^ The 1970 Amendments The Bank Holding Company Act Amendments of 1970 expanded the coverage of the Act to Include a company that controls only one bank. Other provisions of the 1970 legislation Include an expansion of the Board's authority to determine that a company controls a bank and a prohibition against tle-lh arrangements whereby a bank extends services to a customer upon certain conditions. The 1970 Act revised Section 4(c)(8) of the Holding Company Act under which bank holding companies may acquire interests in nonbanking activities. The legislation provides that such activities may bie approved if they are determined by the Board to be '.'..iso closely related to banking as to be a proper incident thereto...."^ William F. Upshaw, "Antitrust and the New Bank Holding Company Act; Part III," Monthly Review . Federal Reserve Bank of Richmond (April, 1971) pp. 4-6. 'bank Holding Company Act Amendments of 1970," Federal Reserve Bulletin , vol. 57 (January, 1971), p. 29. 3 Ibid . . p. 31.

PAGE 34

21 The Board is also authorized to differentiate between an activity commenced de novo and the acquisition of a going concern. Development of Bank Holding Companies After 1956 After passage of the 1956 Act, it was expected that holding companies would expand substantially. The growth of multibank groups, however, closely paralleled the general development of the banking system in the period from 1956 to 1965. At the end of 1956 there were 49 separate bank groups registered with the Board pursuant to the Act. For year end 1960, there were 42 distinct registered companies and five years later there were 48 multibank systems. The share of U.S. total deposits held by subsidiaries was little changed for the period; 7.5 percent in 1956 compared to 8.3 percent in 1965. Similarly the ratio of holding company offices to all commercial bank offices changed by less than 1.0 percent, with groups possessing 6. 7 percent of all banking offices at the end of 1965. (See Table 2.1.) Since the end of 1965, holding companies have expanded substantially. Many factors contributed to the increase in holding company activity. The abolition in 1964 of the 2 percent federal tax penalty imposed on companies filing a consolidated tax return was one contributing 2 factor. Another possible consideration was that the holding company idea had finally come of age. This was suggested by the Board's approval For a good explanation of the 1970 Amendments see Donald L. Kohn and John F. Zoellner, "The Amended Bank Holding Company Act," Monthly Review . Federal Reserve Bank of Kansas City (May, 1971) pp. 11-20. 2 Ed Tyng, "New Moves to Form Holding Companies Seem Likely Now that Tax Penalty is Dead." Journal of Commerce . April 7, 1964, pp. 1, 24.

PAGE 35

-22 in April 1966 of three holding company formations by New York organizations. Table 2.1 Registered Bank Holding Companies: Number, Number of Banks and Branches , and National Share of Offices and Deposits, 1956-, 1960, 1965 December 31 December 31 December 31 1956 1960 1965 49

PAGE 36

.-23 The 1966 Amendments (enacted in July of 1966) contributed to the dramatic increase in holding company activity. Repeal of the restrictions upon loan participations and the sale of loan paper certainly encouraged the activity. The 1966 amendments also clarified the applicability of the antitrust laws to banks and required the Department of Justice to challenge the Board's decisions upon holding company applications within thirty days of approval. The formation or acquisition is thereafter permanently immune from antitrust attack except under the monopolization provisions of Section 2 of the Sherman Act. No longer were bank holding companies haunted by the possibility of future challenges of their past actions by the Department of Justice. As shown in Table 2. 2, there has been a steady increase in the number of multibank holding companies from 1965 to 1970. At the end of 1970, holding companies controlled 11. 8 percent of the banking offices as compared to 6. 7 percent of the offices in 1965. The percentage of deposits controlled by holding companies nearly doubled during this five-year period, with holding companies possessing 16. 2 percent of U.S. deposits at the end of 1970. Table 2.3 provides a state-by-state summary of the development of multibank holding companies between 1965 and 1970.^ Unlike the "Amendments to Bank Holding Company Act," Federal Reserve Bulletin vol. 52 (July, 1966), p. 967. "" " ~~' 2 For a similar summary which uses the years 1956, 1962 and 1969, see Golembe, pp. 7-10.

PAGE 37

24 Table 2.2 Officesand Deposits of Banks Affiliated with Registered' Bank Holding Companies, 1965-1970 End of Number of Number of Banking Offices Deposits ($ millions) Year Companies^ Banks Number °L of U.S. Total 'L of U.S. 1965

PAGE 38

-25 09

PAGE 39

26 -

PAGE 40

.'27 . the statutes do not favor statewide-branching (although it is possible) and the dominant structure is of the limited-branching variety.^ The increase for the District of Columbia is due to the registration of Financial General Corporation pursuant to the Bank Holding Company Act Amendments of 1966. In terms of deposits, 10 states showed holding company deposits declining or remaining constant and in 16 states holding company control of deposits rose by less than 10 percent from 1965 to 1970. The following states had an increase of holding company deposits of 10 percent or more: Alabama (16.0), Colorado (29.2), District of Columbia (11.4), Florida (38.5), Maine (41.2), Missouri (23. 7) , New Jersey (16.8), New York (17.7), Ohio (11.5) and Virginia (21.1). Except for the District of Columbia, all of these states prohibit or make difficult statewide-branching. Since the end of 1970, multibank holding companies have rapidly grown in Texas (unit-banking) and Tennessee (limited-branching). For example, in Virginia an organization can achieve statewide branching only by merger. The mergered bank loses its branching privileges. (Individual banks may branch countywide, more or less.) In addition, only banks in existence for 5 years or more may be acquired by merger. Because of these limitations on statewide-branching, Virginia organizations have generally formed holding companies. Affiliates of holding companies do not lose their branching privileges and affiliates are not required to have been in existence for at least five years.

PAGE 41

CHAPTER 3 A CRITICAL REVIEW OF THE LITERATURE In the preceding chapter the history of the bank holding company movement was reviewed. In this chapter the scholarly literature on bank holding companies is reviewed. A Selective and Critical Review; An Aid to Model Building The review of literature will cover empirical studies of multibank holding companies since 1960. In the course of the review, we will attempt to determine the characteristic operating policies of bank holding companies. In order to determine if the results of the empirical studies are reliable, the review of literature will be critical as well as descriptive. The entire review of literature is designed to assist in subsequent model building. Fischer; Bank Holding Companies An obvious starting point is Fischer's monograph on bank holding companies. Fischer surveyed holding companies by means of questionnaires and interviews to determine their organizational and operational practices^ He found that most multibank systems employed legal counsel and accounting and tax specialists. The headquarters of the holding company did some central purchasing and advertising; The one exception is the Alhadeffs ' study of bank mergers which was published in 1955. -28 -

PAGE 42

29 operations specialists were made available to subsidiary banks; and there was a considerable amount of committee activity. The lead bank of the holding company encouraged the less active subsidiary banks to modernize by adopting the practices of the lead bank. Almost without exception, the lead banks encouraged the affiliated banks to emphasize consumer credit. (Consumer loans yield a considerably higher rate of return than other types of loans and they are subject to a number of cost savings if uniform systems and proce2 dures can be followed by all members of the bank holding company. ) In the course of his interviews, Fischer found multibank systems fully involved in the supervision of portfolios with special assistance provided to affiliates in managing their cash positions. The bank holding company actually participated in very little personnel -' " 3 ' ' ' coordination or development despite claims to the contrary. In unitbanking and limited-branching states , correspondent banking is an important source of interbank service flows. Many of the ways in which a depository bank aids its correspondents are very similar to the services provided by a bank holding company. Because of the similarity of services, many multibank holding companies have consolidated correspondent accounts while having the lead bank or the holding company itself provide subsidiary banks with specialized services. Fischer. Bank Holding Companies , pp. 86-87. ^ Ibid . . pp. 87-92. 3 Ibid . . pp. 94-98. * Ihid. , p. 105.

PAGE 43

30 Fischer observed that entry of a bank holding company into a rural conununity had no effect on the general level of interest rates although the new affiliate often stirred the competitive spirits in the community as one bank remodeled or provided a drive-in facility and other banks countered with innovations of their own. The research by Fischer is subject to several criticisms. The review of operating policies of multibank systems is based upon questionnaires and interviews. Information gathered by these means is exposed to the problem of respondents giving good reasons as opposed to the real reasons for operating policies. Also, there are difficulties in classifying, and generalizing from, the information obtained from questionnaires and interviews. In Fischer's 1961 study, most of the holding companies surveyed had been in existence since the late 1920's. One may wonder whether the results of the study reflect the operating policies of the newly formed or newly active multibank holding companies of the late 1960s. As Fischer is well aware, he may be criticized for attempting 2 to generalize from a very diverse sample. His sample of holding companies includes old and new groups, large and small groups, groups which operate in unitbanking, limitedbranching, and unlimitedbranching states. The sample includes states where the holding company movement is in an advanced stage (Minnesota) and states where holding company activity had yet (in 1961) to begin a sustained movement (Ohio). ^ Ibid., p. 130 2 This criticism may be applied in varying degrees to all the works surveyed in this chapter. Of course, there are multivariate statistical methods (not used by Fischer) which would control for some of the different characteristics of bank holding companies.

PAGE 44

31 Because of these extreme differences, the behavior of the bank holding companies sampled is very diverse. As a consequence, Fischer is unable to make strong statements concerning holding company behavior because of the wide differences in behavior observed. Lawrence; The Performance of Bank Holding Companies The first rigorous study of bank holding company behavior was done by Lawrence. In the study he employed the before -and-after method of ianalysls in order to isolate the impact of holding company affiliation upon certain performance variables. This method of analysis was applied to the acquired bank and a comparable nonaffiliated bank. The use of paired bank comparisons permitted Lawrence to hold constant all relevant local market conditions. The acquired banks were essentially similar to the paired Independent -banks in terms of pre-af filiation performance variables with the only exception being due to balances as a proportion of total deposits. A possible explanation for this difference is that banks with relatively large due to balances are reluctant to affiliate with a holding company 2 from fear of losing correspondent business upon affiliation. Additional reasons for a reluctance to affiliate may be a lesser need on the part of the bank for correspondent-type services from a holding company and the ability of the bank to possibly anchor a holding company itself. * For a good discussion of the paired bank methodology, see Piper, pp. 166-168. * Robert J. Lawrence, The Performance of Bank Holding Companies (Washington, D.C. : Board of Governors of the Federal Reserve System, 1967), pp. 15-16.

PAGE 45

32 . As for post-affiliation performance, acquired banks were much more active than the paired independent banks in making loans. The difference in the proportion of assets devoted to loans^is both statistically significant and quantitatively significant. Lawrence reported that "the result for the loans-to-assets ratio is one of the strongest in the entire study — the subsidiaries of bank holding companies have on the average a loans-to-assets ratio that is more than 5 percentage points higher than the ratio for the independent banks." Lawrence also found that holding company banks had a higher percentage of their assets in the obligations of states and political subdivisions and lower percentages of their assets in U.S. Government securities and cash plus due from 2 balances. After affiliation, subsidiary banks were found to be much more aggressive than the independent banks in seeking out consumer instalment loans. The subsidiaries' ratio of instalment loans to assets was 3.15 percentage points higher than the ratio of paired independent banks. In the Lawrence study, holding company banks do not appear;, to have any competitive advantage over nonaffiliated banks. The average growth rates of affiliates were not superior to the growth rates of the independent banks. The earnings of subsidiary banks were not significantly different from the earnings of the independent banks. Also, there was no evidence of operating efficiencies as measured by the operating ratio (operating expenses divided by operating revenue). ^ Ibid., p: 17. ^ Ibid. 3 This result is confirmed by other studies which employ less rigorous measurement techniques. See, for example, Golembe, p. 49. ^ Lawrence, pp. 21-25.

PAGE 46

33 No one to date has substantially improved upon the Lawrence research although it is subject to some criticism. The paired bank approach used in the study assumes that the actions of the paired bank are independent of the operating changes of the acquired bank. One may question this assumption since the banks are in the same market. Both a strength and a weakness of the study is the sharp differences among the banks in the study: The selected banks include large and small banks, banks in standard metropolitan statistical areas and small towns, banks in all sections of the nation, subsidiaries of large and small holding companies, lead and nonlead banks, and unit and limited-branch banks. From such a diverse sample, Lawrence may be justified in making general statements about holding company behavior. However, it can be argued that the univariate statistical method (t-test on difference of means) employed by Lawrence did not adequately control for the sharp differences in the banks sampled. It is possible that the extreme differences among the banks actually reduced the significance of some of the statistical tests. Instead, if Lawrence had controlled for the differences in, say, the size, location, length of affiliation and branching characteristics of the banks sampled by using a multivariate statistical method, he may have found that holding companies affiliates were, say, more efficient 2 or more profitable than independent banks. ^ Ibid ., p. 12. 2 Lawrence did divide the sample into two groups on the bases of location, size of holding company, size of bank, and length of acquisition. These divisions did not yield more significant results. However, the sample was divided on the basis of one characteristic at a time. Also, the divisions were very rough; e.g., a bank was either a large bank or a small one.

PAGE 47

34 McLeary; Bank Holding Companies — Their Growth and Performance Bank holding companies in the Sixth Federal Reserve District (Atlanta) were studied by McLeary. His study was a replication of the Lawrence research, with the sample consisting of holding company affiliates in Florida, Georgia, and eastern Tennessee. McLeary employed a paired bank method in analyzing performance variables after affiliation. Four performance variables of affiliated banks were found to be significantly different (in the statistical sense). Subsidiary banks generally charged lower interest rates on loans, carried fewer U.S. Government securities and more state and local obligations relative to assets, and had a higher percent of their deposits in demand accounts than did o independent banks. The McLeary study has several serious deficiencies. There is no economic justification for excluding western Tennessee which is not in the Sixth Federal Reserve District. McLeary failed to take account of differences among the three states included in the study. Florida is a unitbanking state while Georgia and Tennessee permit limited-branching. In 1960 Georgia prohibited the formation of bank holding companies or the further expansion of existing holding companies. Florida and Tennessee place no restrictions on holding company activity. Also, bankers in Florida and Tennessee have reacted differently to the possibility of adopting the holding company structure. The beginning of sustained holding company activity in Florida occurred in 1966. The McLeary realized that a be fore -and -after study is needed to determine rigorously whether operating differences observed are due to changes following affiliation or are due to the fact that holding, companies acquired banks with certain performance characteristics. Joe W. McLeary, "Bank Holding Companies: Their Growth and Performance," Monthly Review, Federal Reserve Bank of Atlanta (October ,1968) , p. 137.

PAGE 48

35 take off for holding company activity in Tennessee occurred at least three years later. The McLeary study coviered holding company subsidiaries in existence at the end of 1966. Weiss; Bank Holding Companies and Public Policy Weiss studied holding company structure and performance in the New England states of Maine, Massachusetts and New Hampshire. He found that the holding companies expected to be consulted by affiliated banks on matters such as changes in dividend policy, establishment of branches, or other factors that affect earnings. Most holding companies in the study discouraged correspondent relationships with banks outside 2 the system in order to minimize the non-earning assets of subsidiaries. In general, the acquired banks' loan to deposit ratio increased in the post-acquisition period. Also substantial shifts occurred in the composition of the acquired banks' non-loan assets in the post-acquisition period. The acquired banks generally reduced their holdings of U.S. Government securities and their holdings of currency plus due from balances relative to total assets. The affiliated banks increased their holdings of state and local government obligations. In the majority of holding company acquisitions studied, Weiss found a decrease in consumer instalment loans as a proportion of total loans. Earnings performance and the growth of capital accounts for the acquired banks seemed unaffected In Florida, holding companies' share of total deposits of the state went from 22.4 percent in 1965 to 38.4 percent in 1969. The comparable figures for Tennessee are 2.6 percent and 8. 2 percent. 2 Steven J. Weiss, "Bank Holding Companies and Public Policy," New England Business Review . Federal Reserve Bank of Boston (January/ February, 1969), pp. 18-19.

PAGE 49

36 by holding company affiliation. None of Weiss' results were subjected to statistical tests because the number of banks in the study was extremely small. ! Weiss discussed several factors which appeared to influence holding company behavior. He contended that the holding company structure was particularly attractive to center city banks that lac|ced the legal power to branch into growing suburban areas. Also, he suggested that adoption of a holding company organization may be motivated in part by a desire to increase the size of a single banking organization. In the case of Maine, Weiss found that the aggressive activity of one holding company had been an important factor contributing to ain upsurge of interest in holding companies by other Maine bankers. Talley; The Effects of Holding Company Acquisitions on Bank Performance The Lawrence research was updated by Talley using a larger and different sample of acquired banks for the period from 1966 through 1969. In reviewing the portfolios of the banks in the study, Talley found that the acquired banks tended to switch out of governments and into. state and local obligations and loans, particularly consumer loanS. in comparing the mean difference of affiliated banks versus independent banks before affiliation with the mean difference after affiliation, Talley ^ Ibid . , pp. 21-22. , ^ Ibid . , pp. 9-15. Situations similar to the one in Maine have occurred in other states. For example, in New Mexico "the obvious success and impressive growth record of Bank Securities has apparently triggered action on the part of four or five bank groups that are reported in the process of forming bank holding companies." "Bank Breakthrough in New Mexico," Burroughs Clearing House , vol. 54 (September, 1970) , p. 25.

PAGE 50

37 observed an increase in the subsidiaries' ratio of loans to assets of 3. 85 percentage points. As measured by the mean change in the difference method, the subsidiaries' ratio of consumer loans to total assets 9 increased by 1.60 percentage points. Both results are statistically significant at the . 01 level. ^ Talley suggested that the portfolio changes found may be due to the holding company structure which permits affiliates to be located in diverse locations thereby lowering risk to the system as a whole. Also, acquired banks can expect assistance from other affiliates or the holding company in the event of liquidity problems. In addition, the portfolio changes may be due to headquarters replacing conservative 4 management with more aggressive, profit-oriented management. . Bank holding company affiliation did not result in statistically significant changes in the capital, prices, expenses or profitability of acquired banks with one exception. Other operating expenses as a percentage of total assets was significantly higher for acquired banks. One plausible explanation for this higher expenses ratio is that the acquired banks paid significant management fees to their holding companies. The only major difference from the Lawrence study was thiat Talley did not This result is observed in other research. For example, Piper, p. 129. Also see Irving Schweiger, "Reply to Chicago Banking: A Critical Review," Journal of Finance , vol. 17 (October, 1962), p. 424. 2 Lawrence in his study excludes single payment loans from the consumer Instalment loan category. Talley includes single payment loans in his consumer loan category. 3 Samuel H. Talley, The Effect of Holding Company Acquisitions on Bank Performance (Washington, D.C. : Board of Governors of the Federal Reserve System, 1972), pp. 8-9, 16. ^ Ibid ., pp. 9-10.

PAGE 51

38 find that banks acquired by holding companies increased their service charges. Talley's study is subject to the criticisms applied to the Lawrence paper; namely, the assumption of independence implicit in the paired bank method and inadequate controls for a very diverse sample. It is possible that Talley's conclusion that holding company affiliation did not have a broad impact on the performance of acquired banks is invalid because his univariate method of analysis did not adequately adjust for differences among the banks studied. ! Piper; The Economics of Bank Acquisitions by Holding Companies Recently, Piper has done extensive research on the profitability of bank acquisitions by holding companies. He found that affiliation with a holding company was not associated, on average, with any important increase in bank profitability over and above that experienced by banks generally. Furthermore, there was little evidence of 2 superior growth by the acquired banks. Piper also considered the question of whether the acquisition programs of holding companies are profitable. An acquisition was judged profitable if earnings per share of the holding company were higher in 1967 as a result of the acquisition. Piper omitted from the analysis of profitability any favorable impact on the holding company's stock resulting ^ Ibid . . pp. 10-13. ^ Piper, p. 161. 3 " It is quite possible that a bank may not be exceptionally profitable either before or after acquisition and still be a profitable addition to the holding company. Such would be the case if a bank were acquired by a holding company at a bargain price.

PAGE 52

..-39 -. from the acquisition activity itself. Also emitted were any increases in the earnings of the parent company resulting from fees paid to it by the acquired bank. Piper found, that the 102 acquisitions studied were breakeven investments, on average, and they did not result in higher earnings per share for the holding companies involved.^ There was a definite tendency for acquisitions during the period 1946-195^ to be less profitable than those completed during the period 1957-1967. In the second period from 1957 to 1967, holding companies acquired banks that had excess capital that either could support further bank expansion or could be reallocated to other, more productive uses in the system. Holding companies that were active in acquisitions during the 1957-1967 period were also more aggressive in theiriuse of long term notes and 3 debentures. In considering the motivations for holding company behavior. Piper stated: Organizers and managements of bank holding companies seemed motivated by two primary considerations: (1) the prestige and increased lending capacity of a large regional banking group and (2) beliefs, unsubstantiated by a substantial accumulation of ' ^ research, that scale economies and improved manage• ment would offset premium prices paid for banks ^nd would permit profitable expansion by acquisition.^ Piper, pp. 216-219. The ommission of management fees paid to the parent company is questionable. Both Lawrence and Talley report significant increases in other expenses for acquired banks. These other expenses are very likely management fees. 2 Piper has assumed that the growth prospects of the bank and the holding company were equal beyond 1967. He also assumed that the holding company's policies with respect to the existing system were unaffected by a change in its acquisition activity. ^ Piper, pp. 224-225. ^ Ibid. , p. 252.

PAGE 53

40 With regard to the considerations of prestige and large lending limits, none of Piper's own research provides support for such a contention. The fact that the acquisitions studied by Piper were at best a breakeven proposition for the holding companies involved does not lead to the conclusion, which is implicit in the consideration of prestige, that holding companies are not profit maximizers. One may propose that for acquisitions made during the 1946-1956 period the holding companies involved were overly optimistic; i.e., they incorrectly estimated the future stream of earnings from their acquisitions. For the 1957-1967 period. Piper actually found that holding company acquisi2 tions were, on balance, profitable. With respect to the unsubstantiated belief in scale economies. Piper states that in almost all acquisitions by holding companies, scale economies were probably negligible since the size of the firm was increased but plant size was not increased. Piper cites research by Bell and Murphy which shows that the greater labor requirements of branching offset the economies of scale which permit specialization in many 3 banking functions. Piper is comparing a holding company system to a branching system which seems inappropriate. A more appropriate basis Piper appears to be strongly influenced by Reid's study of mergers involving large banks. See Samuel R. Re id. Mergers, Managers and the Economy . (New York: McGraw-Hill Book Co. , 1968), Chapter 10. The sixty-seven acquisitions analyzed during the second eleven-year period had a mean profitability index of 1.07. The breakeven point was 1.0. See Piper, p. 224. ^ Ibid . . p. 256. The reference is to Frederick W. Bell and Neil B. Murphy, Costs in Commercial Banking; A Quantitative Analysis of Bank Behavior and Its Relation to Bank Regulation (Boston: Federal Reserve Bank of Boston, 1968).

PAGE 54

-41(if one proposes to test the arguments for efficiencies as expressed by advocates of the holding company structure) is to compare unit banks operating independently with unit banks operating within a holding company framework. Possible Efficiencies of the Holding Company Structure Conceptually, returns to scale refer to the percentage change in physical output resulting from simultaneous and equal percentage changes in the employment of each input. Thus, the advocates of holding company efficiencies are not really claiming economies of scale as this term is strictly defined. Rather, the emphasis is on economies of vertical integration for the multibank holding company (the multiplant firm)» The Bell and Murphy research does not test this claim. The use of the operating ratio by Lawrence and Talley as a rough-and-ready measure of operating efficiencies is very rough. It must be said that the case for economies of vertical integration accruing to the multibank holding company is most convincing, a priori . Holding companies have a greater ability to tap major money markets than do individual banks. The credit of the parent organization is usually stronger than that of individual units, so the parent can negotiate better terms. A large organization can recruit better talent because of the fringe benefits, salary structure and additional possibilities of advancement. The holding company vehicle permits pension and profit sharing plans, as well as other fringe benefits, to be uniformly applied to all units in an efficient program. Coordinated advertising programs The formation of a multibank holding company may be viewed as the positioning of a layer of management on top of a group of semiautonomous banks.

PAGE 55

42 may be applied to all units. The bond portfolio of all units can be administered more expertly by a central department. Centralized uniform accounting can provide vital help in efficient operations and assist in financial planning and budgeting, A centralized auditing program can be beneficial to subsidiary banks both as a management tool and as a means of supplementing director's examinations. Note that these arguments for group efficiencies are based upon a fundamental economic principle. Large markets promote the increased division of labor and the increased division of labor makes possible a more efficient use of resources. Schweitzer; Economies of Scale and Holding Company Affiliation Schweitzer has conducted the only research to date which directly examines the question of efficiencies in mult ibank systems. In the research, he specified and estimated a cost function for banks which stressed their dual function as intermediaries and payments clearinghouses. The results of the study suggest that there are economies of holding company affiliation. Schweitzer concluded: On the holding company question, it appears that •• " . there are cost savings to be achieved through . affiliation with a holding company group, and particularly with one of the two large Ninth District groups. But these savings accrue principally to banks in the intermediate size ranges; Eugene H. Adams, "Economy, System Changes Require States to Modernize Bank Structure." American Banker . September 6. 1972, p. 11. ^Although Adam Smith used the pin factory to explain this principle, nonetheless the principle may be extended to the modern multiplant firm. It should also be noted that the problems of a limited supply of management services are less pressing for registered bank holding companies because federal statutes require relatively homogeneous components for the company.

PAGE 56

43 from $3.5 to $10. million and from $10.0 to $25.0 million total assets. ' These conclusions can only be termed suggestive. One may question whether Schweitzer's use of only two independent production functions adequately reflects the variety of output produced by banks. Also, data on the factor prices for labor and for demand deposits were not available for estimating the cost function. . Upson and Jessup; Returns from Bank Holding Companies Research by Upson and Jessup bears upon the question of the profit orientation of holding companies. (Recall that Piper stated that holding companies have a prestige and size orientation. ) They found that the earnings of multibank holding companies and the valuation of these earnings were very favorable. The researchers created a price index of holding company stocks for the period from 1957 to 1971 for companies with over $500 million in group deposits after i960. Share prices between 1957 and 1971 increased almost 200 percent (about 8 percent, compounded annually). For the same period, Moody's common stock 3 price index for banks outside New York City increased about 117 percent. The behavior of the holding company index is attributed to the management policies of these companies which have resulted in dividends growing more rapidly for holding companies than for individual banks. This greater growth of dividends basically reflects real growth /• Stuart A. Schweitzer, Cost and Production in Banking; The Case of the Ninth Federal Reserve District . Doctoral Dissertation, Department of Economics, University of Minnesota, 1970. Cited in Golembe, p. 59. 2 Stuart A. Schweitzer, "Economies of Scale and Holding Company Affiliation in Banking." Southern Economic Journal , vol. 38 (October, 1972), pp. 259-263. ^ Roger B. Upson and Paul F. Jessup, "Returns from Bank Holding Companies," The Bankers Magazine , vol. 155 (Spring. 1972), pp. 60-61.

PAGE 57

44 in earnings per share because the holding companies have typically not increased their dividend payout ratios over the time period. Upson and Jessup also developed a more comprehensive measure of performance than the price index of holding company stock by combining price and dividend information into total rates of return. This measure includes both dividend income and capital gain or loss and allows for the timing of receipts and outlays. Using this measure of performance, a complete matrix of returns for the shares of holding companies was calculated. For the period from 1957 to 1971, the average return for all holding periods was 11.2 percent, compounded annually.'' The results of the research by Upson and Jessup demonstrate that holding company profits, on average, have been very favorable. The results also 3 ' imply that holding companies are profit oriented. Alhadeffs; Recent Bank Mergers Research by the Alhadeffs is relevant to the study of holding company behavior. In analyzing bank mergers , which occurred in the early 1950s, the researchers hypothesized that uneven rates of growth among banks were a major reason for the observed increase in bank mergers. The Alhadeffs wrote: "The very rapid growth of the past decade has created stresses and maladjustments owing to uneven rates of growth of different ^ Ibid . . pp. 61-62. . n_.-'. ^ Ibid . .p. 62. The returns are approximately symmetrical around the overall mean of 11. 2 percent. There are relatively few negative returns and no negative returns appear for holding periods of longer than two years. ' 3Note that Piper looked at holding company profitability in terms of earnings per share for a selected set of acquisitions while Upson and Jessup considered the returns to the entire holding company system in terms of share prices and in terms of their comprehensive measure of performance.

PAGE 58

45 banks." If a merger is not an acceptable solution to a lagging growth rate — which is usually the case in unit -banking states and often the case in limited -branching states--then a bank may decide to organize or to join ia holding company in order to remedy a record of poor growth. In the same study, the Alhadeffs reported that mergers occurred in clusters and gave two reasons for the merger clusters: Banks which have not yet begun to acquire other banks begin to do so In simple imitation of the initiators. A pattern of acquisitions which is desirable for the initiators can be equally desirable for the imitators. The clustering is also explained by the defensive reaction of nonparticipating banks to the acquisition activities of the initiating banks. Once mergers have begun in an area j the relative standing of the other banks is either actually or imminently threatened and a retalitory merger is the surest way to maintain a threatened position.^ The merger clusters cited by the Alhadeffs appear to have an analogue in the snowballing of holding comgany. activity which was observed in " 3 " ' . ' -, several states. . The Relative Standing of Banks , In explaining the clustering of mergers, the Alhadeffs mentioned a bank's concern for its relative standing in the market. A bank's concern for its relative standing is quite understandable if its relative position declines because of a movement of customers to other banks. The Alhadeffs pointed to an indirect effect of a decline in market standing. A decline in market standing is accompanied by a ^ Charlotte P. Alhadeff and David A. Alhadeff , "Recent Bank Mergers," Quarterly Journal of Economics , vol. 55 (November, 1955),. p. 512. 2 Ibid. , p. 518. 3 See Chapter 2, p. 12 and this chapter, p. 36.

PAGE 59

46 ,''•• relative (and possibly an absolute) loss of deposits which often causes an increase in the capital to deposit ratio which reduces the leverage on the bank's earnings. Related to the question of relative standing is the i fact that a bank possesses a supply of highly specialized resources, namely, banking knowledge and experience. These specialized resource^ are relatively fixed in the near term so these resources may be viewed as 2 involving a type of fixed cost. It is advantageous for the bank to attempt to spread these fixed costs (of specialized resources) over additional units of output because, other things being equal, average costs decline as fixed costs are stretched over added output. Of course, the ability of a bank to spread out fixed costs is a function of its ability to expand. It should also be noted that a bank's knowledge and experience increases over time, thereby augmenting the bank's capacity for production. With the passage of time, lacklustre growth can result in a serious underutilization of resources. Summary The empirical studies consistently found that 'hold ihg company affiliates have altered their portfolios in a manner that should increase returns.. Cash, due from balances and U.S. Governments have been reduced and a larger part of the portfolio has been devoted to local and state ^ Alhadeff, p. 516. 2 Note that if a bank forsakes its accumulated knowledge and experience and seeks to acquire a completely different kind of knowledge and experience, the opportunity costs will be high. The emphasis here is upon the human resources of the bank. Since the provision of banking services is labor intensive, this emphasis seems justified.

PAGE 60

. -47 -obligations and to loans, especially consumer loans. Many holding companies provided correspondent-type services such as portfolio analyses, accounting services and guidance on operational matters. The parent often charged the subsidiaries fees for the services provided. Also, banks were thought to be very sensitive to their relative position in a market. Those banks whose relative positions were threatened — for example, downtown banks in a unitbanking state — may be attracted to the holding company structure as a way to improve their market standing. Holding company growth has occurred in spurts perhaps because independent banks have frequently reacted to the aggressive actions of holding companies by organizing or joining a holding company. The evidence is inconclusive on the question of whether holding company affiliates are more efficient than comparable independent banks. Holding companies appear to have operated very jprofitably, especially in the last decade or so. ." For a recent discussion of the range of services provided by holding companies see Robert J. Lawrence, Operating Policies of Bank Holding Companies-P-art 1 (Washington, D.C. : Board of Governors of the Federal Reserve System, 1971). A recent study of holding company performance was conducted by Golembe and Associates. The study was not reviewed because the empirical work is poorly done. A paired bank method is used on an after-acquisition basis only in an attempt to analyze holding company behavior. The beforeacquisition analysis was never done using rigorous statistical tests. There are only 16 and 22 group banks for the two post-acquisition dates. The tests applied for the post-acquisition date^ are biased since some groups banks were paired with more than one independent bank. To confirm the results of the paired bank approach, Golembe analyzed national data for all insured banks and all affiliated banks as of December 31, 1968. This use of aggregated data for a single point in time without the benefit of any statistical tests is unacceptable. See Golembe, Chapter 3.

PAGE 61

CHAPTER 4 BANK HOLDING COMPANIES: PROFIT MAXIMIZERS OR SIZE MAXIMIZERS Neither the history of holding companies covered in Chapter 2 nor the review of the literature presented in Chapter 3 have provided clear evidence as to the underlying motivation behind the actions of bank holding companies. On the one hand, the changes observed in the portfolios of affiliated banks suggest that holding companies are profit oriented. Also, if one accepts the arguments in favor of efficiencies for the holding company structure, then the adoption and use of this structure by a banking organization may indicate that the organization is profit conscious. Furthermore, holding companies have operated quite profitably, at least in the last decade or so. Of course, profitable operations are supportive of a profit maximization hypothesis. On the other hand, the very rapid external expansion of some holding companies may imply that they are size maximizers. Also, the very high prices paid for unaffiliated banks by holding companies suggest that they are size oriented. ^ Unpublished tabulations of the Federal Reserve Board show the premiums paid by holding companies in Florida, Missouri, New Jersey and Ohio for 1970 and 1971. (The premium represents the difference between the market value of the shares offered and the book value of the shares acquired, expressed as a percent of book value.) The median premium ranged from a low of 52.2 percent in Ohio to a high of 131.6 percent in Florida. -48

PAGE 62

49 This chapter will discuss briefly profit maximization and size maximization as alternative bases for explaining holding company behavior. Before model building can begin, it is necessary to decide upon the underlying motivation of holding companies. External Expansion and Profit Maximization Most holding company affiliates are acquired by means of a share exchange. In a share exchange, the holding companies with high price-earnings multiples can pay a higher price for a bank than holding compianies with low multiples, other things being held equal. Also, the highly valued company is able to pay the high acquisition prices frequently demanded by independent banks without adversely affecting per share earnings. Accordingly, it seems necessary for the holding company that wishes to undertake an aggressive program of external expansion to have a relatively favorable price-earnings ratio. If a holding company with a relatively unfavorable price-earnings ratio undertakes an active program of expansion, the company will likely experience a dilution in per share earnings and per share assets. The holding company will soon be forced to curtail its expansion program since each additional acquisition will increase the amount of dilution suffered by the Southeast Banking Corporation and United Bancshares, both Miamibased holding companies, made a bid to acquire Lon Worth Crow, a mortgage banking company with close ties to United Bancshares. Southeast Banking acquired the company because it was able to pay a higher price than United. At the time, Southeast's P/E ratio was about double the P/E ratio of United.

PAGE 63

50 original stockholders. The primary determinants of the price-earnings multiple appear to be the growth of earnings per share and the growth of the associated dividends per share. There is a "growing consensus of belief in a definite relationship (through the interaction between the stock market, dividend policy and the reinvestment rate) between the maximum obtainable growth rate of the supply of capital. . .and the rate of profit earned on existing assets."^ if one accepts the argument for a relationship between earnings growth, the P/E ratio and expansion, then it follows that a holding company which intends to expand externally 3 must be profit conscious. The Alte rnative Hypothesis; Size Maximization As noted above, a size maximization hypothesis and a profit maximization hypothesis appear to provide a satisfactory framework for explaining holding company behavior. As Lanzilotti has observed: For example, compare First BancGroup and Huntington Bancshares. Both Ohio companies have embarked on an aggressive expansion program. First BancGroup with an especially high P/E multiple has been able to make acquisitions on terms that maintain or enhance its own per share equity and earnings. Huntington Bancshares with a much lower multiple has diluted both its earnings and its book value in the course of acquiring other banks. In 1972, Huntington Bancshares has acquired two small banks whose deposits total $37.2 million, which is about one -fourth of the deposits acquired in 1970 and in 1971. This reduced amount of expansion is presumably due to the substantial dilution experienced by Huntington Bancshares. See Herbert F. Thomson and Jerry A. Meadows, "The Expansion of Registered Bank Holding Companies in Ohio," Bulletin of Business Research. Ohio State University, vol. 46 (July, 1971), p. 3. R.L. Marris, "Review of E.T. Penrose: The Theory of the Growth of the Firm." Economic Journal , vol. 71 (March, 1971), p. 147. Multibank holding companies apparently believe in a relationship between profits and multiples and expansion. See Paul S. Nadler, "Craze for Raising Multiples Spreads with Multi-BHC Craze," American Banker, October 31, 1972, p. 4. "

PAGE 64

51 "Given empirical findings are likely to be consistent with, or at least partially reconcilable with, many hypotheses. The essential question is: which hypotheses are likely to yield the most useful and reliable short-run and long-run predictions." The key empirical finding favoring the size maximization hypothesis is the fact that a number of bank holding companies have expanded very rapidly. In addition, the fact that holding companies frequently pay high acquisition premiums for banks is consistent with this hypothesis. By paying especially high acquisition prices, the holding company becomes larger at the expense of earnings per share. Another possible consideration in favor of this hypothesis is that the decision-making power in a holding company usually rests with management. The argument for management control of the holding company is based upon the fact that the stock of a holding company is usually widely held. It is often assumed that the management -dominated firm will opt for some goal (such as size maximization) other than profit maximization. Vernon has tested this assumption for banking firmis. He concluded: We uncover no evidence to suggest that control status has exerted a significant influence on profit, rates of large commercial banks during recent years. ...Owner-controlled banks did not realize higher rates of return on invested capital than management-controlled banks.Robert F. Lanzilotti, "Pricing Objectives in Large Companies: Reply," American Economic Review , vol; 49 (September, 1959), p. 685. 2 Jack R. Vernon, "Separation of Ownership and Control and Profit Rates, the Evidence from Banking: Comment," Journal of Financial and Quantitative Analysis , vol. 2 (January, 1971), p. 624.

PAGE 65

52 If one is willing to extend Vernon's results to management-controlled holding companies, then one cannot argue that holding companies are not profit maximizers because these companies tend to be management controlled. i Profit Maximization Versus Size Maximization In practice, it is very difficult to distinguish between the size maximizing firm and the profit maximizing firm. For Instance, the rapid expansion of holding companies and the payment of high acquisition premiums may be rationalized in terms of a long run profit maximization model. Under certain conditions a size maximizing firm and a profit maximizing may behave the same way. It was argued above, that a holding company which wishes to expand rapidly must be profit conscious. Consequently, some size maximizing companies may be forced to maximize profits in order to pursue an aggressive program of external expansion. Of course, a size -maximizing company subject to a profit maximization constraint will behave in exactly the same way as a profit maximizing company. Although the evidence is mixed, it appears preferable to use the traditional model of profit maximization. The concept of size maximization Is difficult to apply empirically because the notion of a satisfactory level of profits, which is associated with the size maximization hypothesis, is difficult to use in empirical work. Of course, the size maximizing firm cannot completely Ignore its level of profits. For a discussion of the concept of a satisfactory level of profits, see William J. Baumol, Business Behavior, Value and Growth (New York: Macmillan Company, 1959), p. 49. For a discussion of the problems in applying this concept, see Bevars D. Mabryand David L. Slders, "An Empirical Test of the Sales Maximization Hypothesis," Southern Economic Journal , vol. 28 (January, 1967), pp. 367-377.

PAGE 66

53 By way of contrast, the profit maximization model greatly simplifies the analysis. Professor Machlup wrote that "the substitution of money profits for a composite of pecuniary [satisfactory profits] and non-pecuniary rewards [size and prestige] simplifies the analysis so much that the gain in expediency far exceeds the loss in applicability. "1 A statement by Penrose indicates what the profit model must accomplish in order to explain holding company behavior. In an analysis of the expansion of individual firms the profit-seeking assumption is useful so long as it is possible to set forth in reasonably objective economic terms the considerations that will determine the probability that certain specified directions of expansion will be more profitable than others. ^ Applicat ions of the Profit Maximization Hypothesis Models of lead bank status and of holding company activity will be developed within the framework of the profit maximization hypothesis. In the model of lead bank status, it is assumed that a potential lead bank considers its characteristics and the characteristics of its environment in deciding whether to lead a holding company. The profit maximization hypothesis predicts that the bank will lead a holding company if this is the most profitable opportunity available. For the model of holding company activity, it is assumed that holding companies as a group consider the expected profitability of different markets and the regulatory position of the Federal Reserve Board in deciding where to acquire affiliates. The profit maximization hypothesis predicts Fritz Machlup, "Theories of the Firm: Marginalist, Behavioral, Managerial." American Econo mic Review , vol. 57 (March, 1967), p. '5. The size maximization model usually carries the suggestion that the* firm values highly the prestige associated with size. 2 Edith Penrose, The Theory of the Growth of the Firm (New York: John Wiley 6e Sons, 1959), p. 185.

PAGE 67

54 that holding companies will be most active in the most attractive markets (in terms of expected returns), provided the regulatory position is permissive. i . , It is important to note that the profit maximization hypothesis is employed because the hypothesis provides us with a useful and consistent framework for analysis. The historical evidence and Scholarly research suggest, but do not demonstrate, that holding companies are profit maximizers. The reader should not be misled into thinking that the study is designed to determine whether ov not holding companies are profit maximizers. Rather, the study is built upon the convenient and well-known assumption of profit maximization.

PAGE 68

CHAPTER 5 A MODEL OF LEAD BANK STATUS Within the framework of the profit maximization hypothesis, we will develop a model to analyze the factors which cause a bank to become a lead bank of a'multibank holding company. Alternatively, the model may be viewed as analyzing the factors which influence the formation of a multibank holding company. For most cases, the decision to lead a holding company and the decision to form a holding company are different views of the same phenomenon. That is, a holding company is generally formed by a bank which assumes lead bank status upon formation. Definition of Lead Bank Status In Chapter 1, the lead bank of a holding company is defined as the bank which the management of a holding company designates as the lead bank. The Federal Reserve Bank of Atlanta asks each multibank holding company to designate a bank as its lead bank in those instances where it is not obvious which bank is the lead bank, so as a practical matter it is not difficult to know which bank is the lead bank. There are a few exceptions. For example, the McNulty Group of Banks, which consisted of eleven small banks in Central Florida, became a holding company in 1967 with the title of First Florida Bancorporation. In 1969, First Florida "acquired" Marine Bank and Trust Company, Tampa, which bank became the lead bank of the holding company. See "First Florida Bancorporation, Haines City, Florida," Federal Reserve Bulletin , vol. 52 (November, 1966), pp. 1632-1635. "First Florida Bancorporation, Haines City, Florida," Federal Reserve Bulletin , vol. 55 (February, 1969), pp. 165-168. 55 -

PAGE 69

, . . . 56 Although it Is the case that all mult ibank holding companies have lead banks as defined above, nonetheless for a few holding companies no bank has assumed lead bank status. The idea of lead bank status is not easy to spell out, although it is usually evident whether or not a bank has assumed lead bank status. A bank in taking lead bank status will provide the top officers of the holding company. The bank with lead bank status will set the tone for the entire system in terms of management philosophy, operating procedures, policy guidelines and long range planning. The bank with lead bank status will help the other bank subsidiaries solve problems of management and opera2 tions as well as providing these banks with specialized services. The Set of Potential Lead Banks The set of potential lead banks is defined as the largest 75 banks (in terms of total deposits) in Florida as of December 31, 1971. Although it is most unrealistic to consider all banks as potential lead banks, the decision to use the top 75 banks may appear arbitrary. It was reasoned that only large banks have the opportunity of becoming a lead bank because only large banks have the financial and human resources needed to lead a multibank organization."^ A large bank by Florida ^ This appears to be the case with Ellis Banking Corporation of Bradenton, Florida. Although the First National Bank of Bradenton was designated by management as the lead bank (and it is used in the study), management of Ellis Banking Corporation has indicated in discussions with the staff of the Federal Reserve Board that the holding company "really" does not have a lead bank. ^ As holding companies grow and diversify, the bank holding company often takes on an existence distinct from that of the lead bank. ^ All lead banks of multibank holding companies are substantially larger than the median size bank. The median size bank had total deposits of $17.0 million as of December 31, 1971.

PAGE 70

57 standards was judged to be a bank with approximately $50 million in deposits or more. The largest 75 banks satisfy this criterion.^ Three of the 75 largest banks were excluded from the set of potential lead banks because they had been actual lead banks since 2 the 1920 's. . Fourteen of the 75 largest banks were excluded because they were not independent decision makers. In every case but one,^ the excluded bank was closely tied to a much larger bank which was assumed to make the decisions. So, of the largest 75 banks, only 58 banks were judged to be both potential lead banks and independent decision makers. (See Table A. 1 in the appendix.) Of the 58 potential lead banks used in the study, 21 banks have become lead banks and of the remaining 37 banks, ten banks have been acquired by holding companies and 27 banks have made no decision regarding lead bank status. (See Table A. 2 in the appendix.) It should be noted that three multibank holding companies are led by banks which are not among the 75 largest banks. These three lead banks are ranked 89th, 96th and 152nd among Florida's 540 banks on the basis of total deposits as of December 31, 1971. Only one of the three holding companies — American Bancshares (152nd)--has subsidiary banks in more than one metropolitan market.^ The other two holding The 75th largest bank as of year end 1971 had total deposits of $48.9 million. 2 The three banks are Atlantic National Bank, Florida National Bank and Barnett First National Bank. All three banks are located in Jacksonville. For a history of these groups see J.E. Dovell, History of Banking in Florida (Orlando: Florida Bankers Association, 1955), pp. 183-189. 3 The one exception is Ellis Banking Corporaition. The affiliated Sarasota Bank and Trust Company is substantially larger than the designated lead bank. First National Bank of Bradenton. . For a discussion of metropolitan markets, see Chapter 7, pp. 97-99.

PAGE 71

58 companies have subsidiary banks in only one metropolitan market and are among the smallest multibank holding companies in the state. The model of lead bank status is not designed to explain the decisions by a few medium size banks to anchor small local or regional holding companies. i Lead Banks and Chains \ j. Before becoming the lead bank of a registered multibank 2 system, a bank may have anchored a chain. There are two main teasons why a chain would form a holding company. First, the holding company structure formalizes and makes permanent the relationships within the 3 chain. Second, the holding company structure allows a former chain to rapidly expand its banking activities because the holding company can obtain outside funds to finance acquisitions. Chain banking may be viewed as a substitute for the holding company form of banking. Given this view, we should then hold constant the influence of chain banking so that we can isolate the influence of other factors upon lead bank status. Unfortunately, reliable data on chain banks are not available for all years included in the study. The For the distinction between chain banking and bank holding companies, see Chapter 1, p. 9. 2 This was the case with the First National Bank at Orlando, Broward National Bank and First National Bank of Fort Lauderdale. See "First at Orlando Corporation, Orlando, Florida," Federal Reserve Bulletin . vol. 53 (February, 1967) i p. 236. "Broward Bancshares, Inc., Fort Lauderdale, Florida," Federal Reserve Bulletin , vol. 56 (January, 1970), p. 85. "Consolidated Bancshares of Florida, Inc., Fort Lauderdale," Federal Reserve Bulletin , vol. 57 (February, 1971), p. 137. 3 A chain may be dissolved. This did happen in the case of the First National Group of Southeast Florida. See J.E. Dovell^ History of Banking in Florida; First Supplement (Orlando: Florida Bankers Association, 1964), p. 42.

PAGE 72

59 omission of a variable to control for chain banking activity may not be serious. Darnell, in studying chain banking, did not find a significant relationship between chain banking and bank holding companies. • The Time Framework of the Study In deciding upon lead bank status the potential lead bank has three options: (1) become a lead bank of a holding company, (2) become a subsidiary bank of a holding company, or (3) remain unaffiliated. There is a substantial time lag between the time a bank decides to become a lead bank (or to be acquired by a holding company) and the time this decision is consummated. After a decision is made, time is expended in finalizing negotiations with the proposed affiliates (or with the acquiring holding company) and in preparing an application for the Federal Reserve Board. After the application is filed with the Federal Reserve Bank, it will take at least 105 calendar days for the application to be processed by the Federal Reserve System and for the Board of Governors to act on the proposal.^ An approval by the Board may not be consummated for thirty calendar days in order to permit the Department of Justice time to challenge the Board's decision. In all, it appears that anywhere from six months to eighteen months will elapse between the decision .date of the potential lead bank and the consummation date. Jerome C. Darnell, "Determinants of Chain Banking." National Bank Review, Vol. 4 (June, 1967), pp. 459-468. [ ' Current guidelines state that the Reserve Bank should attempt to formally accept an application within ten business days after receipt. The Federal Reserve System (i.e., the Reserve Bank and the Federal Reserve Board) tries to bring the proposal before the Board of Governors within ninety calendar days after formal acceptance.

PAGE 73

60 All of the explanatory variables in the model are lagged one year because of the considerable delay in consunmating a decision. The year a decision on lead bank status is consummated is determined from Board documents. The previous year is assumed to be the decision year and all explanatory variables are measured as of the decision year. The variables are as of the decision year because it is necessary to measure the characteristics of the bank at the time that the decision was made. If a potential lead bank has not become a lead bank or been acquired by a holding company by the end of 1972, which is the terminal point of the study, then it is assumed that the bank has decided not to become a lead bank for the time being. From 1956, when the Bank Holding Company Act was passed, until 1964 no potential lead bank became a lead bank. In 1964, the First National Bank of Tampa, a potential lead bank, became the lead bank of a registered holding company. The decision to become a lead bank is assumed to have been made in 1963. So the study of lead bank status will cover the period from 1963 through 1971 in terms of decision years or from 1964 through 1972 in terms of consunmation years. A Framework for the Empirical Model The variables which affect lead bank status may be divided into three groups: (1) variables which reflect the regulatory position toward potential lead banks; (2) variables which measure the economic characteristics of the potential lead bank; and (3) variables which reflect the environment in which a potential lead bank operates. The relationship of the three types of variables to lead bank status is summarized in equation (1).

PAGE 74

61 (1) LBSt = f (FRBt.i, IBCt.i, EEj-.i) where LBSj. = lead bank status in year t. LBS=1 if the bank is a lead bank; LBS-0 otherwise. FRB^_j = regulatory position of the Federal Reserve Board in year t-1. IBCj._j^ = individual bank characteristics in year . t-1. ,, ! EE^_i = environment of the bank in year t-1. The Regulatory Position For the period 1963-1972, the Federal Reserve Board has approved all proposed formations. One cannot conclude from these approvals that the Board has been completely permissive. The Federal Reserve Board has certainly exerted an influence upon the decisions of potential lead banks by means of its decisions on non-Florida cases and by means of informal staff discussions with these banks. However, it is an impossible task to measure this influence empirically because of the absence of any overt actions by the Board. Consequently, it has been assumed that the Board's influence upon the decisions of potential lead banks has not been significant and no variable reflecting Board influence will be included in the model. 1 in 1962, the Board did disapprove the application by four of the eight largest banks in Florida to form a holding company. Today the four banks lead the first, second, fourth and tenth largest holding companies in Florida "First Bancorporation of Florida, Inc., Orlando, Florida," Federal Reserve Bulletin , vol. 48 (August, 1962), pp. 978-984. In 1971, the Board disapproved a formation which involved a potential lead bank, the Commercial Bank at Winter Park. The formation, however, was contingent upon the approval of an application by Southeast Banking Corporation to acquire all five banks of the proposed multibank holding companies. "Combanks Corporation, Winter Park, Florida," Federal Reserve Bulletin, vol. 58 (January, 1972), pp. 55-56.

PAGE 75

62 Individual Bank Characteristics In looking at individual bank characteristics, six characteristics are thought to be important determinants of a bank's staitus and they are shown in equation (2). (2) IBCt = g (SZt, CB^, L^/At, CL^/A^, CL^/k^, MS^) where IBCt = individual bank characteristics in year t. SZt = size in terms of total deposits in year t. CBt = correspondent business measured by due to balances in year t. LtMt loan to asset ratio in year t. CLj./A^ = consumer loan to asset ratio in year t. CL^Mt ~ commercial and industrial loan to asset ratio in year t. MSj = change in market share for the period from year t-5 to year t. The most obvious characteristic of any bank is its size. It is generally assumed that large banks are more likely to become lead banks than small ones. The presumption is partly based on empirical evidence --nine of the ten largest banks in Florida are lead banks. Partly, it is assumed that the larger banks have the human and financial resources needed to lead a multibank system. In terms of human resources, the larger is the bank (in absolute terms) the more likely it is to have specialized services such as investment guidance and data processing services which can be offered to other banks. In terms of financial resources, the larger is the lead bank (in absolute terms) the more

PAGE 76

63 marketable will be the holding company stock and the more acceptable will be the stock in share exchanges. In addition, the larger organizations have easier access to financial markets when they publicly offer additional stock or senior securities. This entire discussion is intended to suggest that the larger is the bank the better will be its ability to exhaust any efficiencies associated with the holding company structure. The &ize of a potential lead bank is measured by its total deposits as of December 31 of the decision year. Total deposits are deflated using the aggregate GNP deflator in order to remove price changes in the deposit data for the time period of the study. A second bank characteristic which is thought to affect the decision on lead bank status is the size of the bank's correspondent business. It is hypothesized that the greater is the bank's involvement in correspondent banking the greater is the likelihood that the bank will become a lead bank. The hypothesis concerning correspondent business is based on 2 three considerations. First, the larger is the bank's correspondent business, the more able is the bank to provide future subsidiary banks Since the study is restricted to the state of Florida, we need not control for inter-state differences in bank size. As an example of state differences a $100 million bank, which is a very large bank in New Mexico, may provide services which are inadequate by Texas standards. 2 The three considerations have the same direction of influence and are to some extent reenforcing. It is not necessary to specify which consideration, if only one, is paramount to a particular bank.

PAGE 77

64 with specialized services. In other words, the bank has expertise that is useful to a holding company organization. Second, the larger the correspondent business the more likely the bank is to have a close working relationship with other banks. The correspondent bank and its family of closely-related client banks can easily adopt the holding company structure and formalize the "family". Third, a bank with a large amount of correspondent business may organize a holding company in order to preserve its business from the encroachment of expanding holding companies. One sure way to suffer a decline in correspondent business is for a bank to permit rival organizations to acquire respondent banks. Due to balances are used as a measure of the correspondent business done by a bank because such balances are the usual means of payment for correspondent services. Due to balances are measured as of year-end of the decision year and they are deflated with the aggregate GNP deflator in order to adjust for inflation over time. In isolating key characteristics of a potential lead bank, three portfolio ratios appear useful. These are the total loan to asset ratio, the consumer loan to asset ratio, and the commercial and industrial loan to asset ratio. Both Lawrence and Talley, in studying holding company performance, found that affiliated banks had a substantially higher proportion of their assets in loans, particularly consumer loans. In fact the changes in the loan to asset ratio and the consumer loan to asset ratio ^ See Chapter 2, p. 12, and Chapter 3, pp. 29, 31Also, Lawrence, Operating Policies , p. 25.

PAGE 78

65 were among the most significant findings, both statistically and quantitatively, in the two studies. The assumption is made that the changes in the operating policies of subsidiary banks reflect the existing practices of the lead bank. Accordingly, it is hypothesized that the higher the bank's loan to asset ratio and the higher its consumer loan to asset ratio, the more likely it is that the bank will become a lead bank. The loan to asset ratio is measured by a bank's total loans and discounts 2 divided by its total assets. The consumer loan to asset ratio is o defined as total consumer loans divided by total assets. The third portfolio ratio considered in analyzing lead bank status was the ratio of commercial and industrial loans to assets. This ratio was used to indicate the extent to which a bank was wholesaleor business-oriented. It is hypothesized that wholesale banks will be likely to form holding companies in order to attract the business of, and in order to better serve the banking needs of,, multi-establishment firms. This portfolio ratio is defined as total commercial and industrial loans divided by total assets. It should be noted that there is something of a contradiction between the business loan ratio hypothesis and the consumer loan ratio hypothesis. A bank cannot be wholesale-oriented and retail-oriented at ^ See Chapter 3, pp. 32, 36-37. In terms of the format of the Report of Condition, total loans and discounts are item 8 of schedule A; total assets are item 14 of the front page. 3 In terms of the Report of Condition format, total consumer loans are the sum of items 6(a) through 6(f) of Schedule A. Commercial and industrial loans are items 5 of Schedule A of the Report of Condition.

PAGE 79

66 the same time. The empirical results should determine which ratio, if either, is a significant determinant .of lead bank status. It is necessary to control for differences in market conditions in using these portfolio ratios. Differences in market conditions exist because potential lead banks are located in different geographic markets and because potential lead banks in the same market have different decision _years. In order to control for market differences a bank's portfolio ratio is divided by the comparable ratio for the entire market. Accordingly, the market adjusted variable for, say, the loan to asset ratio is equal to the loan to asset ratio of the bank divided by the ratio of total loans to total assets for the market. If this adjusted variable is above 1.00 then the bank's ratio exceeds the market's ratio and conversely. The adjusted portfolio variables are used in empirical tests of the model. A bank's relative rate of growth is thought to influence a bank's decision on lead bank status. It is proposed that the poorer a bank's relative growth (i.e., the greater the decline in a bank's market share) the more likely is the bank to become a lead bank in an attempt to recover its relative position and to improve its profit performance. ^ The relative rate of growth is measured by a bank's change in market From an accounting point of view, a bank could conceivably have an above-average business loan ratio and an above -average consumer loan ratio. This might occur if the bank made relatively few real estate loans, security loans, loans to financial institutions, and miscellaneous loans. Such a loan portfolio does not seem probable, however. The stresses caused by uneven rates of growth and the importance of a bank s relative position were discussed in Chapter 3, pp. 44-46.

PAGE 80

67 share. The change in market share is calculated for the period from year t-5 to year t, where t is the decision year, using the formula in equation (3). t=0 t=-3 ! (3) MS = -i-^ "(Bank's Deposits) t_ 1^''^C "(Bank's Deposits)t ^ 3.r
PAGE 81

68 The amount of recent holding company activity in a market is thought to be an influence upon lead bank status. It is hypothesized that encroachment by holding companies on a bank's local market, thereby threatening its base of operations, will often result in a defensive reaction by the bank. If the independent bank believes it is at a disadvantage in competing with holding company affiliates, it can form a holding company or join a holding company. However, which course of action, if either, the bank will pursue cannot be decided a priori . Despite the inability to predict the direction of influence, it seems important to consider the encroachment effect. Much of the snowballing of holding company activity observed in Florida and elsewhere is probably due to the defensive response of banks to encroach2 ment by holding companies on their home markets. Given that the encroachment effect has a substantial impact upon the amount of holding company activity, it is of interest to determine empirically whether or not encroachment has a systematic effect upon a bank's decision regarding lead bank status. The encroachment effect is measured in two ways. The variable ENC/BK measures encroachment in terms of the change in holding company control of banks in a market. The variable ENC/TD measures encroachment on the basis of the change in holding company control of market deposits. For example, "the proposal of the Central National Bank of Cleveland = to acquire American Bank of Commerce in Akron was announced as a response by Cleveland Banks to the encroachment on their traditional territory by BancOhio Corporation, in its merger with Akron National Bank and Trust Company." Thomson and Meadows, p. 3. 2 For a discussion of the snowballing of holding company activity, see Chapter 2, pp. 24-27 and Chapter 3, p. 35.

PAGE 82

69 Lead bank status is thought to be affected by a market's Importance as a financial center. It is proposed that, ceteris paribus . the more important is the market as a financial center, the more likely will be a bank in the market to become a lead bank. The reason is that the more important a market is as a financial center, the more likely it is that some banks in the market are informed about the holding company structure and its possible benefits. The informed banks are expected to have an Influence upon the thinking of other banks in the market. In other words, important financial centers possess a stimulating environment which encourages banks to assume leadership roles in multibank organizations. A market's standing as a financial center is measured by the market's share of total deposits in the state in the decision year. One would expect some correlation between a market's share of deposits and potential lead banks' deposit size and correspondent business. Nonetheless, a market's importance is considered to be a distinct determinant of lead bank status. The Statistical Model The model of lead bank status is intended to determine the probability that a bank will opt for lead bank status. The probability of lead bank status is thought to be a function of a bank's individual characteristics (IBC) and its environment (EE). Clearly, the object of prime Interest is a probability. The discussion is Intended to provide the reader with an intuitive appreciation for why regression analysis is an inappropriate statistical model for this study and why probit analysis is an appropriate statistical model for this study.

PAGE 83

70 The regression model is not a suitable statistical model for determining the probability of lead bank status. In both ordinary and generalized least squares regression models, the estimated value for lead bank status may fall outside the interval from to 1, because regression models fit straight lines to the data points. (See Figure 5.1.) An estimated value (probability) for lead bank status of less than or greater than 1 is inconsistent with interpreting the value of the dependent variable as a probability. ^ Another shortcoming of the regression model is the fact that it is unable to handle a dichotomous dependent variable. It should be realized that the actual value of the dependent variable lead bank status is either, say, or 1; this is to say, a bank either is or is 2 not a lead bank. Because the dependent variable is dichotomous, the assumption of homoskedasticity, which is a basic assumption of the regression model, is not satisfied. The assumption of homoskedasticity means that the expected variance of the error terms is equal for all observations. In other words, the scatter of data points (the observed values of the variables) about the estimated regression line is approximately uniform. (See Figure 5.2.) It makes no sense to say that the probability of a bank selecting lead bank status is -10 percent or +120 percent. 2 For example, the expected probability of getting heads from flipping a coin may be 0.5. Nonetheless, for an actual flip of the coin, the result is either heads (1) or tails (0). o In practice, generalized least squares regression, a technique for adjusting for heterskedasticity, is unworkable because the calculated values of the dependent variable are frequently very close to or 1. As a consequence, the inverse of the estimated disturbance covariance matrix cannot be calculated (the inverse is singular or nearly singular).

PAGE 84

-71 Probability of Lead Bank Status Figure 5.1 A Straight Line Fit with Regression Model Observed Data ** Regression Line Figure 5,2 Homoskedasticity about the Regression Line

PAGE 85

72 " The variance of the error term is defined as the square of the difference between the estimated value of the dependent variable (the probability of lead bank status) and the actual value of the dependent variable (either or 1). The variance is weighted by the ' " . " ! probability of "success" (the bank is a lead bank) as "failure" (the bank is not a lead bank).-'Applying the definition of the variiance of the error term and letting "P" equal the probability of lead bank status, we have: (4) Variance of error term = (l-P)^(P)+(0-P)2(l-P) = (l-2P+p2)P+p2-p3 = P-p2 = P(l-P) From (4), it is evident that the variance of the error term 2 depends on P, the probability of lead bank status; and P in turn depends on the individual bank characteristics (IBC) and on the environment (EE). Consequently, the variance of the error term varies systematically with the explanatory variables. Thus, the assumption of homoskedasticity, which is a basic assumption of the regression model, is not satisfied. In view of the problem with the regression model, an alternative statistical approach is needed to handle a dichotomous dependent ^ Probabilities are used for the theoretical distribution while frequencies are used for the actual (sample) distribution. 2 For example, if the expected value of lead bank status equals .9, this means a probability of ,. 9 that the value will deviate from expectation by +.1 and a probability of .1 that the deviation will be -.9, while an expected value of .4 means deviations of +.6 with probability .4 and deviations of -.4 with probability .6. Using equation (4), the variance equals .09 when P=. 9, while the variance is .24 when P=.4.

PAGE 86

73 variable and to restrict the value of the dependent variable to the interval from to 1. One alternative is the probit analysis model. Since probit analysis was developed with reference to the field of biology, an example from this field may be most helpful in explaining this statistical method. In biological assay, probit analysis is used to determine the relationship between the probability that organisms will be killed to the strength of the dose of poison administered to them. The dependent variable for each organism in the sample is dichotomous: killed or not killed. Each organism is assumed to have a dosage threshold, such that a stronger dose will; kill that organism and a weaker dose will not. Over the population of organisms of a given kind, these dosage thresholds are assumed to be normally distributed with a few organisms having very high or very low dosage thresholds. while most of the organism have intermediate-valued dosage thresholds. (See Figure 5.3.) A graph of the percentage of insects killed as the dose of poison increases will give a steadily rising curve. The curve has an S-shape because the dosage thresholds are assumed to be normally distributed. In other words, the rate of change in the percentage killed per unit increase in dose is low in the region of very small or very large doses of poison but is higher in the intermediate region. (See Figure 5.4.) The basic reference on proiait analysis Is D.J. Finney, Probit Analysis (3rd ed.; Cambridge, England: The University Press, 1971).

PAGE 87

74 Dosage Thresholds Dose of Poison Figure 5.3 Normal Distribution of Dosage Thresholds 100 Percentage of Insects killed 50 B^ Dose of Poison Figure 5.4 Cumulative Distribution of Insects Killed

PAGE 88

75 The above example may be put into the following mathematical terms. Suppose the _ith organism has an actual value of the dose T^, which is a linear function of the concentration of the poison (CON) and the time of exposure to the poison (EXP) to which the ith organism is subjected. (5) T^ = BQ+BiC0Ni+B2EXPi * Assume that T^ is the threshold value of the dose for the ith organism. If the actual value of the dose T^ is greater than or equal to the threshold value of the dose T^^ then the ith organism is killed. If T^ is less than Tj^, then the ith organism is not killed. The coefficients in equation (5) — i.e., Bq, Bj^ and B2 — are estimated with the maximum likelihood technique. Because the maxi"momlikelihood (rather thanleast squares) technique is used, it is not necessary to assume homoskedasticity (equal variance in the error terms). The above discussion may not have convinced the reader that the probit statistical model is appropriate for the study of lead bank status. Figure 5.5 is designed to erase any lingering doubts. On the horizontal axis of Figure 5.5 is measured bank size and on the vertical axis is measured whether or not the bank has chosen lead bank In order to find the maximum likelihood coefficients, it is necessary to solve a set of rather complicated nonlinear equations. The solutions (the maximum likelihood estimates) are calculated using the Newton-Raphson method of iteration. An exposition of the maximum likelihood solution and a sample calculation of the solution are contained in James Tobin, "The Application of Multivariate Probit Analysis to Economic Survey Data," Cowles Foundation Discussion Paper No. 1 . 1955 (unpublished).

PAGE 89

76 Bank Is a Lead Bank i Bank Is Not a Lead Bank Size of Bank Figure 5.5 The Probit Line for Lead Bank Status

PAGE 90

77 status. As bank size increases, it is thought that the chances of a bank choosing lead bank status increase. Some hypothetical data points are shown as dots on the figure. Essentially what the probit technique does is to fit an S-shaped curve to the data points as shown on the figure.

PAGE 91

CHAPTER 6 THE DETERMINANTS OF LEAD BANK STATUS Using the probit statistical model, the model of lead bank status was tested on the set of 58 potential lead banks listed in Table A. 2. The time period treated by the model is from 1963 through 1971 in terms of decision years or from 1964 through 1972 in terms of decision years. Interpreting the Probit Statistical Model In order to assist one in interpreting the statistical results, the probit model is compared with the regression model. Just as in regression analysis, the probit model assumes that the dependent variable, lead bank status (LBS), may be expressed linearly in terms of the explanatory variables, individual bank characteristics (IBC) and environment (EE); i.e., LBS = B +B.IBC+B-EE. In probit analysis the parameters B^, B, and B_ are estimated by the maximum likelihood method while regression analysis uses the method of least squares. There is no need to consider the maximum likelihood method in detail. Suffice it to say that the method of maximum likelihood gives estimators which have desirable properties and the estimators are easy to find. For large samples, the maximum likelihood estimates are approximately normally distributed. Because of this approximate normality it is possible to estimate the standard error for each coefficient and to calculate t-values for each 78 -

PAGE 92

coefficient 79 1 The Likelihood Ratio Test In addition to the t-test, the likelihood ratio test may be used to test hypotheses about the independent variables both singly (say, IBC = 0) and in combination (say, IBC = and EE = 0). The likelihood ratio test is closely related to maximum likelihood estimation and it is easily applied to the probit model which employs the maximum likelihood technique. The likelihood ratio is equal to the maximum of the likelihood function when a parameter is not allowed to vary freely (say, EE = 0) divided by the maximum of the likelihood function when a parameter freely varies (say, EE ^* 0). (1) Likelihood Ratio = X = 22^LiiJM_Z_gI max L (EE 4 0) Note that max L(EE = 0) will be smaller than or at most equal to max L(EE 4 0) because there is less freedom in maximizing the likelihood function when the hypothesis requires that EE equal zero. The likelihood ratio test is based on the fact that -21bg\ is approximately distributed like chi square with the number of The probability that a bank will become a lead bank is equal to the probability that the bank's value for LBS exceeds its threshold value for LBS. A key portion of the probit model is the assumption that the threshold values for LBS are normally distributed. In symbols, Y _ 2 Pr(LBS) i— 'e ^ du -00 2TT _)e 2 -00 where Y = B0+B1IBC+B2EE.

PAGE 93

80 degrees of freedom equal to the number of restrictions on the parameters. If the likelihood ratio (i.e., -21og\)'is greater than the tabled value of the chi-square distribution at some chosen level of significance, then the hypothesis of no relationship is rejected. For example, if the likelihood ratio is greater than 2.71, which is the value of chi square at a .10 level of significance for one degree of freedom, then one can reject the hjrpothesis that EE equals zero and expect to be right nine out of ten times. A Stepwise Testing Procedure A kind of stepwise procedure was employed in testing the 2 model in order to avoid problems of multicollinearity and to conserve degrees of freedom. All of the independent variables were tested one at a time. The more significant variables were then tested using varying combinations of two variables and so on. In addition, the initially significant variables were combined with all other predictors in an attempt to determine if initially unimportant variables gained significance when combined with initially significant In other words, the critical region for -21ogX is the right-hand tail of the chi-square distribution. For a rigorous treatment of this method, see Alexander Mood, Introduction to the Theory of Statistics (New York: McGraw-Hill Book Co., 1950), pp. 257-259. 2 The problem of multicollinearity relates to the high covariance between two estimators (say, B, and B-). A principal cause of the high covariance is high correlation between the corresponding independent variables (say, X and X_). When multicollinearity exist among the variables, it is aifficult or even impossible to determine the separate influences of the independent variables. As a result, the estimated coefficients tend to be unstable and they are not overly reliable.

PAGE 94

81 predictors. Table 6.1 enumerates the variables used and reports the mean and standard deviation of each variable. The relatively large standard deviation for the correspondent business variable is due to the rather pronounced bimodal distribution of the variable. In addition, the values of the CB variable are positively skewed. It should also be noted that consumer loans include single payment loans while instalment loans e^cclude single payment loans. The Results for Single Predictors The results of the model when the independent variables were tested one at a time are shown in Table 6.2. Not unexpectedly, 2 the SZ variable proved to be a significant predictor of lead bank status. Another significant predictor of lead bank status was the variable for correspondent business. The coefficient had a positive. sign as hypothesized. None of the loan portfolio measures appeared to be good 3 predictors of lead bank status. The very low likelihood ratio values As an example of this procedure, see Tong Hun Lee, "Alternative Interest Rates and the Demand for Money: The Empirical Evidence," American Economic Review , vol. 57 (December, 1967), pp. 1168-1181. 2 Throughout the discussion, the term "significant" is used to denote statistical significance. In Table 6.2, both the t-value and the likelihood ratio test the same hypothesis; namely, that the estimated coefficient is not significantly different from zero. It is gratifying to note that the two statistical tests produce consistent results. 3 The ratio of instalment loans to assets was also tried in place of the consumer loan to asset ratio. (Recall that Lawrence had used the former ratio and Talley the latter in studying the performance of holding company affiliates.) The instalment loan to asset ratio was not significant and had a negative sign whereas a positive sign had been hypothesized.

PAGE 95

-82 Table 6.1 Identification o£ Variables with Means and Standard Deviations Variables Standard Mean Deviation LBS. Lead bank status. LBS = 1 if the bank is a lead bank; LBS = otherwise. .362 .485 SZ. Size of bank in terms of deflated total deposits ($100 millions). .670 .593 CB. Correspondent business measured by deflated due to balances ($10 millions). .468 1.298 L/A. Loan to asset ratio of bank relative to loan to asset ratio of market. 1.046 .159 CL/A. Consumer loan to asset ratio of bank relative to consumer loan to asset ratio of market. .984 .475 IL/A; Instalment loan to asset ratio ~of bank relative to instalment loan to asset ratio of market. .933 .422 CI/A. Commercial and industrial loan to asset ratio of bank relative to this ratio for the market. 1.037 .463 MS. Change in market share of the bank over a five-year period. -.0098 .0202 ACT/BK. Holding company activity as measured by the change in holding company control of banks in the market over a two-year period. .123 .114 ACT/TD. Holding company activity as measured by the change in holding company control of total deposits in the market over a two-year period. .134 .151 MI. Market importance as a financial center as measured by the market's share of total deposits in the state. .102 .073

PAGE 96

83 Table 6.2 Probit Model of Lead Bank Status Using Single Predictors Dependent Variable

PAGE 97

84 and tvalues for the variables L/A, CL/A and CI/A indicates that they have no systematic influence upon lead bank status. One may conclude, at least for Florida banks, that lead banks are not especially aggressive in seeking loan business. Furthermore, one cannot characterize lead banks as either wholesale-oriented or retailoriented. The market share variable is not a significant predictor. Given the very low likelihood ratio value, one may conclude that changes in a bank's relative standing in its local market do not Influence the bank's decision regarding lead bank status. The variable ACT/BK, which is used to test the encroachment hypothesis, is significant with a negative sign. The variable ACT/TD, which measures encroachment in terms of holding company control of deposits, has a negative sign but it is not statistically significant. These results imply that encroachment by holding companies does not provoke a bank to form a holding company. It is possible that encroachment influences banks to join holding companies rather than form them. Some care must be exercised in interpreting this result because only ten out of the thirty-seven banks which were not lead banks at the end of 1972 had made the decision to become subsidiaries of holding companies. Of the twenty-seven banks that have not made Recall that it was not possible to determine £ priori the sign for the variable ACT/BK but it was felt that an encroachment variable could help explain the snowballing of multibank holding company affiliations.

PAGE 98

85 an overt decision, it appears probable that most of these banks will join existing holding companies. A few of these banks, especially the one bank holding companies associated with large non-bank organizations, will likely remain independent. The variable for market importance was statistically significant at the .05 level of confidence. The variable has a positive 2 direction of influence as was originally postulated. The Results for Several Predictors 3 All of the significant predictors were paired with each other and with each initially non-significant predictor. None of the initially non-significant predictors gained significance when combined with initially significant predictors. The results from pairing the significant variables are presented in Table 6.3. Because of the high correlation between size and correspondent business (0.93) the coefficients for the two variables are not overly reliable. It is possible, however, to compare the relative significance of the two variables. Given that the variable SZ is already present in the equation, the addition of the variable CB It may be that a potential lead bank has made the decision to remain unaffiliated. However, there is no way to distinguish such a bank from a bank that has made no decision one way or the other. Of course, the decision to remain unaffiliated is revocable while the decision to become affiliated, once consummated, is not revocable. 2 The reader should not suspect a spurious relationship because of the fact that the largest banks come from the most important financial centers. The most important financial centers also supplied the greatest number of mediumlarge banks. For example, Dade County, which is the most important financial center in Florida with 22.6 percent of state deposits, supplied 13 of the potential lead banks, which is 22.4 percent of the total number of 58 banks. 3 The significant predictors from Table 6.2 are SZ, CB, ACT/BK and MI.

PAGE 99

-86 Table 6.3 Probit Model of Lead Bank Status Using Two Predictors Dependent Variable

PAGE 100

87 increases the likelihood ratio by 6.47, which is significant at the .05 level of confidence so one can reject the null hypothesis that CB = 0, given that SZ is already included in the equation. When the variable CB is entered first, the addition of the size variable increases the likelihood ratio by only 1.02, which is not significant at the .05 level of confidence, it is possible to conclude that both variables supply much of the same information but that the correspondent business variable provides information over and above the informa2 tion contained in the size variable while the reverse is not true. The two variables ACT/BK and MI appear to provide added information. Whether combined with each other or when paired with the variables SZ or CB, these variables increase the likelihood ratio by a significant amount. The final equations for the model of lead bank status are 3 • . ' ' , , listed in Table 6.4. The two equations are highly significant and the signs of all the coefficients are correct. One may conclude that the larger is a bank's size and correspondent business, the higher will be the probability that the bank chooses lead bank status. If the bank's market is an important financial center, this will increase the With the SZ variable in the equation, the likelihood ratio value is 13.87 (see Table 6.1). With the addition of the CB variable, the ratio rises to 20.34, an increase of 6.47. 2 The t-values for the two variables confirm the finding that the CB variable is relatively more significant than the SZ variable. 3 The equations are final in the sense that the addition of a fourth variable does not increase the likelihood ratio by a statistically significant amount, with one exception. The sole exception is when the variable CB is added to the first equation of Table 6.4. However, the high collinearity between the SZ variable and the CB variable makes it inappropriate to include both variables in the same final equation.

PAGE 101

88 likelihood of the bank's selecting lead bank status. If holding companies have gained control of a substantial percentage of banks in the market, this reduces the chances of a bank heading a holding company. There are probably two reasons for the negative impact of holding company encroachment. First, the most effective way to defend against substantial encroachment may be to join an established holding company. Second, the opportunities for forming a holding company are reduced because substantial encroachment has reduced the number of attractive unaffiliated banks. Table 6.4 Final Equations for the Model of Lead Bank Status ^ Likelihood Dependent 3 Variable Constant Independent Variable Ratio Value LBS = -1.875 + 2.288 SZ 4.258 ACT/BK + 5.852 MI 24.37*** (2.99)*** (2.18)** (2.19)** LBS = -1.126+ 4.049 CB 6.107 ACT/BK + 6.745 MI 32.67*** (2.71)*** (2.28)** (2.45)*** ^ The likelihood ratio value tests the hypothesis that the three variables jointly do not have a systematic effect upon holding company activity. ** Significant at the .05 level. *** Significant at the .01 level. Note: The figures in .parentheses are t-values for each estimated coefficient. The first equation of Table 6.4 properly classifies 48 out of the 58 potential lead banks, using an 0.5 decision rule. The In using a 0.5 decision rule, the model is judged to classify properly if the probability of lead bank status is above 0.5 and the bank does become a lead bank or the probability is below 0.5 and the bank does not become a lead bank.

PAGE 102

89 second equation properly classifies 47 out of the 58 potential lead banks when an 0.5 decision rule is empljoyed. The simple correlations between variables are displayed in Table 6.5. The four instances of rather high correlations between independent variables are circled. In general, the correlation matrix does not suggest any serious problems of multicollinearity. In order to confirm the suggestion of no serious multicollinearity, one may review Tables 6.2 through 6.4. The reasonably stable maximum likelihood coefficients of these tables (except for the pairing of the SZ variable and the CB variable) indicate that the results are not strongly distorted by statistical multicollinearity. Table 6.6 is intended to provide the reader with the quantitative impact of each significant predictor upon the probability of lead bank status. The table presents the probability of lead bank 2 status for selected values of the independent variables. In using the table, emphasis should be placed upon the change of probability rather than the level of probability associated with different values of the independent variables . For an example where the maximum likelihood estimates are quite sensitive to the model specification, see Thomas Johnson, "Qualitative and Limited Dependent Variables in Economic Relationships," Econo metrica, vol. 40 (May, 1972), pp. 460-461. 2 In calculating the probability of lead bank status, one first finds the value of the dependent variable LBS for the selected values of the independent variables. (The independent variables whose values are not changing are held constant at their mean values.) The probability of lead bank status equals the value of the unit-normal distribution function corresponding to the calculated value of LBS.

PAGE 103

90 <:Ih HI <:|m t)|< Ml< oK iJ|<: o en NO O CM r-l O 0> O i-H O -• eg P>J o / «n A f-i o -* o I ON I eg eg t»j i-J \-^ I* I* i' O 00 i-l
PAGE 104

91 Table 6.6 The Probability of Lead Bank Status' Variable SZ = 0.670* SZ = 1.340 (1.13 S.D.) SZ = 2.010 (2.26 S.D.) Probability of Lead Bank Status .394 .897 .997 Change of Probability +.503 +.100 CB = 0.0 (0.36 S.D.) CB = 0.468* CB = 0.936 (0.36 S.D.) .100 .731 .994 +.631 +.263 ACT/BK = 0.123* ACT/BK = 0.246 (1.08 S.D.) ACT/BK = 0.369 (2.16 S.D.) .731 .446 .187 -.285 -.259 MI = 0.0 (1.40 S.D.) MI = 0.102* MI = 0.204 (1.40 S.D.) .507 .731 .887 +.224 +.156 The first equation of Table 6.4 is used for calculations involving the variable SZ. The second equation of Table 6.4 is used for the other three variables. Intervals for each variable are equal to 100 percent or 200 percent deviations from the variable's mean. The mean values of each variable are marked with an *. Deviations from the mean in terms of standard deviations of the variables are given in parentheses. Note that the intervals are especially large in order to provide a clear indication of the quantitative impact of each predictor.

PAGE 105

92 Correspondent business appears to have the largest impact upon the probability of lead bank status. If a bank changes from an unimportant to an important source of correspondent services, the probability of its becoming a leiad bank increases by over 60 percentage points. Both the SZ variable and the ACT/BK variable cause large changes in the probability measure. The market importance variable has the smallest impact upon lead bank status relative to the other significant variables. Nonetheless, the quantitative impact of the variable MI from an absolute, rather than a relative, point of view Is substantial. For instance, if a bank is located in a market which is twice as important as a financial center as the average metropolitan market, then the likelihood of that bank becoming a lead bank is about 15 percentage points higher than the likelihood of a comparable bank i deated i n an average-size metropolitan; market. The mean value for the variable CB is somewhat misleading because the actual values for the variable are positively skewed. It turns out that a bank with correspondent balances equal to the mean of $.468 million is one of the leading sources of correspondent services.

PAGE 106

CHAPTER 7 I A MODEL OF HOLDING COMPANY ACTIVITY It has been hypothesized that bank holding companies are profit maximizers. Using this hypothesis, we will develop a model .of holding company activity that is intended to explain the amount of activity by holding companies in various markets for several time periods. By focusing upon the activity of holding companies in the aggregate rather than upon the activity of individual companies, we hope to obtain a picture of typical holding company behavior. The Profit Maximization Hypothesis and Holding Company Activity With a profit maximization hypothesis, one would expect holding company activity to occur in the markets where the expected return from acquiring bank affiliates is highest. As the more profitable opportunities are exhausted, one would expect activity to occur in the markets which were at first less attractive. Holding company activity would continue in the initially more attractive markets at a reduced level. After several time periods, it is expected that activity will differ among markets on the basis of slight differences in market attractiveness. It has been implicitly assumed in the above discussion that the markets are not subject to exogeneous changes during the several time periods. This assumption does not hold in the case of markets It has also been implicitly assumed that market participants are well informed and that market relations are generally impersonal. 93 -

PAGE 107

94 in Florida which have experienced sharp changes in population, personal income, bank deposits and number of banks in a short period of time. Consequently, very attractive markets may experience activity for several time periods because of rapid changes which maintain the market's attractiveness. . _ Measures of Holding Company Activity Holding company activity consists of decisions in favor of the affiliation of banks with multibank holding companies. The affiliation decision may be associated with the proposed formation of a holding company or with the proposed acquisition of a bank by an existing holding company. Holding company activity is measured in four different ways. The two more comprehensive measures are: (1) the proportion of banks in a market that become affiliated with holding companies during a time period; (2) the proportion of deposits in a market which become controlled by holding companies as a result of affiliations during a time period. The holding company structure may be viewed as a means by which lead banks (i.e., large unit banks) expand. Given this viewpoint See Chapter 1^ pp. 6-7. 2 Note that the nonbanking activities of holding companies are excluded from the definition. Also, some proposals favoring affiliation are denied by the Board or are abandoned by the applicant after Board approval. Since the model is intended to explain the decisions of holding companies, these denials and abandonments are included in the measures of holding company activity for the time period in which they occur. See Table 7.1 for a complete list of denials and abandonments. All of these are 3(a)(3) proposals, i.e., affiliations proposed by existing holding companies.

PAGE 108

95 Table 7.1 i Holding Company Proposals Denied or Abandoned, 1963-1972 Proposa l Barnett Banks to acquire First National Beach Bank, Jacksonville Barnett Banks to acquire Union Trust National Bank, St. Petersburg Charter Bancshares to acquire First National Beach Bank, Jacksonville Barnett Banks to acquire Bank of Osceola, Kissimmee Exchange Bancorporation to acquire Peninsula State Bank, Tampa Southeast Banking to acquire First N/B of Eau Gallie and Indiatlantic Beach Bank First Financial to acquire Bank of Clearwater Southeast Banking to acquire Hollywood Bank and Trust Pan American Bancshares to acquire First Bank of Plantation First at Orlando to acquire National Bank of Sarasota First at Orlando to acquire National Bank ' Gulf Gate; Sarasota Florida National Banks to acquire Ormond Beach First National Bank Southeast Banking to acquire South Seminole Bank, North Orlando Bank, Commercial Bank at Apopka, Commercial Bank at Pine Castle, Camnercial Bank at Winter Park Eirst -Financial to acquire Union Trust National Bank, St. Petersburg First at Orlando to acquire Seminole Bank of Tampa Denied o^ Abandoned^ Denied 12/27/65 Denied 6/26/69 Approved 3/12/70 Later Abandoned Approved 3/17/70 Later Abandoned Denied 3/26/70 Denied 7/14/70 Denied 8/ 6/70 Denied 11/3/70 Approved 1/11/71 Later 'Abandoned Denied 11/9/71 Approved 11/9/71 Later Abandoned Denied 12/3/71 Denied 12/3/71 Denied 4/11/72 Denied 8/17/72 Three proposals were abandoned because insufficient shares were tendered to make an effective exchange. One proposal was abandoned because of the close affiliation of the bank with a denied bank. Source: Board of Governors of the Federal Reserve System. _;

PAGE 109

96 one should exclude lead banks in measuring holding company activity. The appropriate measures are: (3) the proportion of nonlead banks in a market which become affiliated with holding companies during a time period; (4) the proportion of nonlead bank deposits in a market which become controlled by holding companies as a result of affiliations during a time period. The Dependent Variable and Tobit Analysis All four measures of the dependent variable, which is holding company activity, will range in value from zero to +1.0. The measures conceivably could take on negative values if holding companies sold affiliated banks. In practice, however, holding companies do not sell 2 affiliated banks. The dependent variable has a lower limit of zero^ and there is a concentration of observations at this limit. ^ Since there cannot be negative deviations from the expected value of the dependent variable at its limiting value of zero, the assumption of homoskedasticity does not hold. As a consequence, multiple regression analysis cannot be used to test the model of holding company activity. If only the probability of limit and non-limit observations were to be explained — i.e., there In other words, if large lead banks are expanding by means of the holding company device, it is inappropriate to include in the measure of expansion the lead banks which are the bases of the expansion. 2 For a discussion of this phenomenon, see Paul F. Jessup, "Portfolio Strategies for Bank Holding Companies," The Bankers Magazine , vol. 52 (Spring, 1969), pp. 78-85. 3 The upper limit of +1.0 is not a binding constraint. The dependent variable has the limiting value of zero for 26.7 percent of the observations for all four measures of holding company activity.

PAGE 110

97 either is or is not activity in a market — then probit analysis provides a suitable statistical model. However, it is inefficient to throw away information on the value of the dependent variable for non-limit observations. Tobit analysis, a hybrid of probit analysis and multiple regression, is designed to handle a dependent variable which has a limit and which takes on the limit for a substantial number of observations. The tobit model allows the explanatory variables to influence both the probability of limit observations and the size of non-limit observations. Accordingly, tobit analysis is a suitable statistical model for testing the model of holding company activity since we are interested in both the probability of holding company activity in a market and the expected amount of activity in a market. Metropolitan Markets The study is limited to holding company activity in metropolitan markets. Most of the activity has taken place in the metropolitan markets 2 of Florida. Furthermore, the exclusion of rural markets will increase 3 homogeneity among the markets in the study. The use of metropolitan markets assures that there are a reasonable number of banks which may affiliate with holding companies. Because there are a reasonable number ^ For a presentation of the tobit model, see James Tobin, "Estimation of Relationships for Limited Dependent Variables," Econometrica, vol. 26 (January, 1958) , pp. 24-36. 2 As of December 31, 1972, 227 banks of the 280 banks affiliated with holding companies were located in metropolitan markets. These 227 banks had deposits of $11,221 million, which is 89. 3 percent of the deposits controlled by multibank systems. ' The need for homogeneity was treated in Chapter 1, pp. 1-5.

PAGE 111

98 of potential affiliates, holding company activity in metropolitan markets usually reflects activity on the part of several holding companies. In this study, a metropolitan market is a market classified as a Standard Metropolitan Statistical Area (SMSA) or a part of such an area. In establishing an SMSA, the Bureau of the Budget states that "the criteria of metropolitan character relates primarily to the attributes of the county as a place of work or as a home for a concentration of non-agricultural workers. "2 There are 14 SMSA's in Florida, three of which are composed of two counties and eleven which are one-county SMSAs. The one-county SMSAs appear to be reasonable approximations to the local banking 3 market.-* The two-county metropolitan area are the Orlando SMSA, the Pensacola SMSA and the Tampa-St. Petersburg SMSA. The Orlando and Pensacola SMSA are dominated by one large city which is in the center of the two-county, area so the two-county area is a useful measure of the banking market. In the case of the Tampa-St. Petersburg SMSA there are two large cities, neither one is dominant and the two cities are separated by Tampa Bay. Furthermore, Tampa-Hillsborough County has If rural markets were included in the study, affiliation of a single bank in a rural market would be shown by a large change in the measure of holding company activity. In addition, holding company activity in rural markets often reflects the actions of a single holding company. Of course, if only one holding company is active in a'market the chance that the measure of activity will not reflect typical holding company behavior is increased. 2 U.S. Bureau of Budget, Standard Metropolitan Statistical Areas (Washington, D.C. : U.S. Government Printing Office, 1967), p. 1. For the specific criteria of metropolitan character, see ibid , p. 2. 3The local banking market is defined as that geographic area in which all banks exert and react to essentially the same set of competitive forces which influence the price and quality of services to local customers.

PAGE 112

99 economic characteristics which are very different from the characteristics of St. Petersburg-Pinellas County. Because the SMSA in this instance is a poor approximation to the banking market, Tampa-Hillsborough County and St. Petersburg-Pinellas County are each considered banking markets. The 15 metropolitan markets used are listed in Table 7.2. • The Time Period As ishown in Table 7.3, the period from 1967 to 1972 was one of substantial holding company activity, the only exception being 1968. The model of holding company activity will be applied to two-year intervals--1967-1968, 1969-1970 and 1971-1972. Because the model is not believed to be sensitive enough to explain year-to-year variations in activity, two-year time periods are used. In addition, fortuitous circumstances in a single year are "standardized" by being combined with normal circumstances in another year. There is a considerable delay from the time a decision favoring a formation or an acquisition is made to the time a decision 3 is consummated.-^ Because of the delay, the decision favoring activity is assumed to have been made one year before the date of consummation. The explanatory variables are measured for the time when decisions are made, not for the time when decisions are consummated, because the model is Intended to explain the decisions of holding companies. Tampa-Hillsborough County is a diversified agricultural, manufacturing and trade center while St. Petersburg-Pinellas County is basically a tourist and retiree center. 2 The very low level of holding company activity in 1968 may be considered a fortuitous circumstance. 3 ^ See Chapter 5, pp. 59-60, for an explanation of the long delay in consummating decisions.

PAGE 113

100 Table 7.2 Metropolitan Banking Markets in Florida Name of Market Gainesville SMSA Me Ibourne-Titusvi lie -Cocoa SMSA Fort Lauderdale -Hollywood SMSA Miami SMSA Jacksonville SMSA Pensacola SMSA Tampa Market Fort Myers SMSA Tallahassee SMSA Orlando SMSA West Palm Beach-Boca Raton SMSA St. Petersburg Market Lake land -Winter Haven SMSA Sarasota SMSA Daytoria Beach SMSA Geographic Area Alachua County Brevard County Broward County Dade County Duval County Escambia & Santa Rosa Counties Hillsborough County Lee County Leon County Orange & Seminole Counties Palm Beach County Pinellas County Polk County Sarasota County Volusia County

PAGE 114

101 u to U CO (0 o .c a. o n fe <« c O -H o) S 1-1 -H <8 OJ 1-H 4J O .-I O Ou-43 O ^ CO OONONr-IC>l>tVOO\r^N*vOiriU-| en.-ioocMCN^ncsoooocofn c\ieMcMcMCMCMCMncnn..-Hr^ •H . T-* l-^O^O^00.-^O^I-l^^Cl^.-^ P^^m C^ 1-1 pvivo 1^ t-ii— ir-(CN)u-ir>.OcNp^r^r^vOin i-(i-if-ii-ii-ii-icMnc»i>*vooocM •CO1-1 f-i a pq .2d 5>s a ^ 4J to i-( ^5 i-tinoon«Hioiri
PAGE 115

102 Given the assumption of a year interval from decision date to consummation date, the variables related to consummations in, say, 1967 should be measured for the appropriate dates in 1966. The same is true for consummations in 1968. If consummations are randomly distributed through the two-year interval 1967-1968, then the average date for the measurement of variables is the end of 1966. Accordingly, for the two-year interval the explanatory variables are measured as of the end of the year preceding the interval. A Framework for the Explanatory Variables The explanatory variables for the model of holding company activity may be divided into three groups: (1) variables which reflect the regulatory environment; (2) variables which affect the expected profitability of the market; and (3) variables which measure the snowball effect. (1) HCAt = f(FRBt_i,E(n)^_^,SB^._p where HCA|. = holding company activity in period t. FRB^_, = regulatory environment in period t-1. E(tt)^_j^ = expected profitability of the market as of period t-1, SBj^_j^ = snowball effect variable. SB,= 1 if there was prior activity; SB «= otherwise. The Regulatory Environment The regulatory environment is essentially a function of the '• Data on population and per capita personal income are not available on a year-end basis. Consequently, population and personal income are as of mid-year of the year preceding the interval.

PAGE 116

103 policy followed by the Board of Governors of the Federal Reserve System. In measuring the regulatory environment, one needs to consider Board policy which will affect the level of holding company activity in the different markets. The amount of concentration in a market is believed to influence the Board's regulation of holding company activity. If a market is highly concentrated, the Board of Governors is more likely to restrict holding company activity than if the market is not concentrated. It is suggested that the Board believes holding company activity in a concentrated market will either entrench the existing amount of concen2 tration or increase the amount of concentration. Concentration is measured by the proportion of total deposits in a market which are controlled by the top two banking organizations. Expected Profitability of the Market It is hypothesized that the greater is the expected profitability of the market, the higher will be the probability of activity 3 and the higher will be the expected amount of activity. The expected profitability of a market may be viewed as a function of the current Although denials by the Board are included in the measures of activity, it is thought that the Board's attitude toward activity in concentrated markets will discourage holding companies from even submitting proposals for affiliates in these markets. 2 For example, see "First at Orlando Corporation, Orlando, Florida: Order Denying Acquisition of a Bank." Federal Reserve Bulletin , vol. 58 (September, 1972), pp. 818-819. ^ We are implicitly assuming that expected profits are not fully reflected in the value of the stock of banks to be acquired. This assumption is valid if, say, the stockholders of banks to be acquired have higher discount rates or shorter time horizons or lower profit expectations than holding companies.

PAGE 117

104 profitability of the market (iTt), market condition variables (MC) , market growth variables (GR) , and the level of per capita personal income (PI) . (2) E(rT)t = g(rTj., MC^, GR^, Pit) 1 Bank holding companies are thought to consider the current profitability of a market in estimating the market's future profitability. Ceteris paribus , the higher are current market profits, the higher will be the estimate of future market profits. Two measures of market profitability are used: (1) net income as a proportion of average total capital; and (2) net income as a proportion of average total assets. Because net income represents the earnings of a bank for a full year, the average of a bank's capital or assets over the year is used. The year-end measure of capital or assets for year t-1, the midyear measure for year t, and the year-end measure for year t are used to get the average figure. The net income and the average capital or assets of all banks in each market are aggregated separately. In this way, the earnings of large banks are given more weight than the earnings of small banks in calculating current market profitability. Current market conditions are thought to be an important consideration in the estimation of a market's future profitability. Current market conditions are measured in three ways: (1) deposits per bank, (2) population per bank and (3) deposits per capita. Deposits per bank may be interpreted as a measure of the market's ability to support banks. Population per bank is thought of as a measure of the Population per bank is a better measure of market condition than the more customary population per banking office in a strict unit-banking state like Florida. This is because some facilities of unit banks such as detached drive-in tellers are classified as offices for technical reasons. For example, as of June 30, 1972, there were 52 branches in Florida even though branches are prohibited by Florida law.

PAGE 118

105 market's need for banks. Deposits per capita is a mixture of the other two measures. The higher are deposits per bank or population per bank or deposits per capita, the higher, it is hypothesized, will be the anticipated profitability of the market. For the market condition variables, deposits per. bank and deposits per capita, total deposits are deflated with the impljicit GNP deflator. Since the study covers three two-year periods, it is necessary to control for price changes . Holding companies' profit expectations are believed to be influenced by market growth. Market growth is measured in three ways: (1) growth of deposits; (2) population growth; and (3) per capita personal income growth. The greater is the growth in deposits, population or personal income, the higher will be the expected profitability of the market . The market growth variables are measured for the two-year Interval preceding the time period under study. The growth variables are intended to reflect short run changes in the market. The long run changes in these variables will be reflected in the market condition variable. In addition, if the growth variable were measured for time periods longer than two years, the danger of high serial correlation in the error terms would exist. In calculating the growth rate of deposits, the deposit data are deflated with the implicit GNP deflator. In determining the percentage growth of per capita personal income, the income date are deflated with the consumer price index. Holding companies are assumed to consider the level of per capita personal income in estimating a market's profitability. The higher is the level of per capita income, the higher will be the hold-

PAGE 119

-losing companies' estimate of profits in the market. The personal income data are deflated with the consumer price index. There is some overlap between the level of personal income and the three market condition variables. Nonetheless, the personal income variable is believed to capture distinct market forces. High per capita personal income suggests a good supply of investable funds which may or may not have been placed with banks (and thus reflected in deposits per bank). Also, the demand for banking services appears to be elastic with respect to personal income. The greater is personal Income per head, the even greater will be the actual (or latent) demand for banking services. The Snowball Effect It is hypothesized that holding company activity in one period is influenced by whether or not there was activity in the preceding period. This is the snowball effect which reflects the fact that there are substantial frictions present in the early stages of activity in a market. There are several possible reasons for these frictions. First, the adoption and aggressive use of the holding company structure represent a drastic change in the customary way banks behave. It is natural to expect banks to hestitate and reflect before changing traditional behavior patterns. Furthermore, these traditional ways of acting have often evolved into a gentlemen's code of conduct regarding competition. Banks are expected to be reluctant to change the "rules of the game". ^ See Chapter 5, pp. 67-68, for a discussion of the snowball effect with reference to the model of lead bank status.

PAGE 120

107 The snowball effect may be a type of demonstration effect. The initial actions of a few bank holding companies in a market show that the holding company structure presents the opportunity for profitable expansion in the market. In other words, the initial activity may alter holding companies' perception of anticipated profits in a market. The initial stage of activity in a market is also likkly to produce a defensive reaction among competing banks . In an attempt to preserve their market position, competing banks are likely to form or join a holding company. A dummy variable is used to capture the impact of the snowball effect, the idea being that the ice is broken or it is not broken. If there were no holding company activity in the two-year interval preceding the period under study, the snowball effect variable is zero. If there were activity, the variable is one. For purposes of this variable, no distinction is made between formations or 3(a)(1) activity and acquisitions or 3(a)(3) activity. A positive relationship is posited for the snowball effect variable. If the variable is zero, little or no activity is predicted. If the variable is one, a substantial amount of 2 activity is expected. Y . For a discussion of the importance of market position, see Chapter 3, pp. 44-46. 2 It is impossible to determine precisely a_ priori what impact the snowball effect has after the ice has been broken for several periods, although it is clear that high levels of activity cannot be maintained indefinitely. We do not have to answer this question, however, since the study only covers three time periods. The snowball effect is in operation for all three periods in only three of the fifteen markets.

PAGE 121

108 Alternative Measures of Explanatory Variables Current market profitability, market condition and market growth are measured in two or more ways. It is not hypothesized that the three measures of market condition or the two measures of market profitability or the three measures of market growth will gll be satistically significant variables. Rather, an attempt will be mad^ to determine which measure of the explanatory variable is most significant. In order to make this determination, the alternative measures of a variable will be tried in the tobit statistical model one at a time This procedure avoids the problem of multicollinearlty which is likely to occur if several of the measures are tested at the same time. This procedure has the_added-advantage-of-conserving degrees of freedom. The Model of Holding Company Activity After substituting equation (2) into equation (1), we have the model of holding company activity which will be empirically tested. (3) HCAt = f(FRBt>i, TT^_^, MC^.i, GR^_i, ^l^^i, SB^-i)

PAGE 122

CHAPTER 8 THE DETERMINANTS OF HOLDING COMPANY ACTIVITY The toblt statistical model will be used to test empirically the model of holding company activity which was developed in Chapter 7. The results from testing each independent variable alone will first be reported. Then the significant results from using two or more variables will be presented. Interpreting the Tobit Statistical Model The tobit model is similar to the probit model which was described and used in Chapter 6. Just as in probit analysis, the tobit technique assumes that differences in behavior (among holding companies) are normally distributed and maximum likelihood methods are employed to estimate the parameters. Both the t-ratio and the likelihood ratio may be used to test hypotheses about parameters which are estimated from large samples by the tobit approach. Unlike probit analysis, the tobit model assumes that there is a limit on the dependent variable. (In our case, holding company' activity (IK3A) has a lower limit of zero.) Another difference is the fact that the actual value of the dependent variable can take values other than or 1. (In our case, HCA can take on values equal to or greater than zero.) Figure 8.1 is intended to provide the reader with an intuitive picture of what the tobit model does. Market profitability is measured on the horizontal axis and holding company activity on the vertical axis. 109 -

PAGE 123

110 Holding Company Activity Market Profitability Figure 8.1 The Tobit Line for Holding Company Activity

PAGE 124

Ill Zero activity is the general case at low levels of profitability with activity increasing rapidly beyond some critical level of profitability. Of course, there are a few holding companies with very low critical levels of profitability which are active in some markets despite the markets' low profitability. In Figure 4, the broken line QAB is, let us say, the maximum likelihood estimate of the relationship of iactivity to profitability. The expected value of activity implied by this relationship is shown by the curve OB.^ The Results for Single Predictors The model of holding company activity was tested on 15 metropolitan markets for three time periods, so there are 45 observations in all. Table 8.1 enumerates the variables used and reports the mean and standard deviation of each variable. Table 8.2 presents the results of the model when the independent variables were tested one at a time.As shown in Table 8.2, the level of concentration in a market For an actual example see Tobin, "Estimation of Relationships for Limited Dependent Variables," pp. 31-36. The expected value of activity is calculated using the following formula: E(HCA) HCA.F(HCA)+Sigma.f(HCA) where VCA Bo+Bi(Market Profitability). F( ) = the cumulative distribution function of a standardized normal random variable. f ( ) a the probability density function of a standardized normal random variable. Sigma « the standard deviation of the error term, i.e., the standard error of the estimate.

PAGE 125

112 Table 8.1 Identification of Variables with Means and Standard Deviations^ Standard Dependent Variables Mean Deviation i ,' KIA-BK. Proportion of banks which became I affiliated with holding companies during a time period. .149 ' .145 ! HCA-TD. Proportion of total deposits which | came under control of holding companies during a time period .146 .168 HCA-NLBK. Proportion of non-lead banks which became affiliated with holding companies during a time period. .145 .141 HCA-NLTD. Proportion of deposits of non-lead banks which came under control of holding companies during a time period. The time periods are 1967-68, 1969-70, 1971-72. .135 .148 Explanatory Variables CON. Proportion of total deposits controlled by the largest two organizations as of year-end 1966, 1968, 1970. .435 .140 I/CAP. Net income to total capital for the years 1966, 1968, 1970. .112 .030 I/AST. Net income to total assets for the years 1966, 1968, 1970. .0077 .0019 PP/BK. Population per bank as of year-end 1966, 1968, 1970 (100 thousands). .145 .034 TD/BK. Total deposits deflated per bank as of year-end 1966, 1968, 1970 ($100 million). .199 .081 TD/PP. Total deposits deflated per capita as of year-end 1966, 1968, 1970 ($10 thousands). .137 .043 GR-TD Growth of total deposits deflated for the two-year periods 1965-66, 1967-68, 1969-70. .194 .145

PAGE 126

113 Table 8.1 Continued Explanatory Variables Mean GR-PP. Growth of population for the twoyear periods 1965-66, 1967-68, 1969-70. .066 Standard Deviation .043 GR-PI. Growth in per capita personal income i deflated for two-year periods 19651 66, 1967-68, 1969-70. .087 .061 PI. Level of per capita personal income deflated for mid-year 1966, 1968, 1970 ($10 thousands). .292 .045 SNW. Prior holding company activity variable. No activity during prior two-year period = 0. Activity during prior two-year period = 1. .511 .506 ^ All variables are measured in terms of metropolitan banking markets .

PAGE 127

114 has a statistically significant effect upon holding company activity. Also, the direction of influence is negative as hypothesized. Recall that the level of concentration was employed as an indicator of Federal Reserve Board policy toward holding company activity. The thinking was that the Board would frown upon holding company activity in the more concentrated markets. This hypothesis appears to be confirmed by the empirical results. The possibility of a spurious relationship does exist. Holding company activity may be relatively low in concentrated markets because there are few targets for affiliation in concentrated markets. In order to examine this possibility the percentage of deposits and the percentage of banks which are controlled by holding companies were correlated with market concentration. The simple correlation between concentration and holding company control of total deposits was 0.13, while the correlation between concentration and holding company, control of banks was 0.02. Since the two measures of correlation show very weak relationships, the possibility of a spurious relationship between 2 concentration and holding company activity may be rejected. ^ Throughout the discussion, the term "significant" is used to denote statistical significance. In Table 8.2, both the t-value and the likelihood ratio test the significance of each individual coefficient. The two statistical tests produce compatible results. In this chapter , the t-value attributes slightly greater significant to the results than the likelihood ratio. This is in contrast to Chapter 6, where the likelihood ratio gave slightly higher significance to the results than the t-value. 2 It should be reported that there is no apparent connection between market concentration and market profitability. The correlation between the two is very low, being .02 between CON and I/CAP and -.03 between CON and I/AST.

PAGE 128

115 Table 8.2 Tobit Model of

PAGE 129

ii6

PAGE 130

117 Table 8.2 Continued Dependent

PAGE 131

118 between the two measures of profitability cannot be attributed to differences across markets in the use of long term debt. A likely explanation for the slightly better performance of the capital measure is the fact that banks in rapidly growing markets tend to be undercapitalized. As a result income to capital tends to be higher in the faster growing markets. There is no compelling reason to expect income to assets to be higher in the faster growing markets. The suggestion being made is that holding company activity is more likely in markets which are profitable and growing rapidly. In terms of t-values, deposits per bank and deposits per capita have a significant influence on holding company activity. Deposits per capita appears to be slightly better predictor of activity than deposits per bank. Both measures of market conditions have positive coefficients as originally hypothesized. The very low t-values and likelihood ratio values for the third measure of market conditions, population per bank, indicate that this measure of market conditions has no systematic effect upon activity. It is reasonable to conclude that holding companies in general are more interested in deposits thian in population. All three measures of short term market growth are statistically non-significant. The t-values and likelihood ratios are Banks began in 1969 to report their assets on a gross rather than a net basis, and to include on the liability side of their balance sheets an entry for reserves on loans and securities. In addition, changes were made in the treatment of charges to account for loan losses. Therefore, the measures of profitability in 1970 are not strictly comparable with the measure for 1966 and for 1968. Because the profit measures are for markets rather than for individual banks* the differences in profit figures due solely to reporting changes may not be substantial. For a detailed discussion of the reporting changes, see Caroline H. Cagle, "Member Bank Income, 1969." Federal Reserve Bulletin, vol. 56 (July, 1970), pp. 571-572.

PAGE 132

119 consistently small. In addition, several of the coefficients have negative signs whereas a positive direction of influence was hypothesized.. Measures of future market growth were also tested. They too proved to be unsatisfactory predictors of holding company, activity. It must be concluded that holding company activity is not sensitive to short run changes in deposits, population or personal income. The most significant single determinant of activity is the level of per capita personal income. The coefficient carries a positive sign as postulated. There is some empirical support for the snowball hypothesis when the dependent variable, holding company activity, is measured in terms of banks. When activity is measured in terms of deposits, the snowball variable is not significant although it does have a positive sign. One possible explanation for these results may be the fact that the activity which first breaks the ice in a market involves large banks in terms of deposits. Because of these early affiliations by the largest banks in a market, subsequent activity may not substantially exceed the initial activity when one uses deposit measures of activity. Subsequent activity in terms of banks could easily exceed the initial For example, instead of assuming activity during 1969 and 1970 was related to growth over the period 1967-68, it was assumed that the activity was related to growth for the period 1969-70.

PAGE 133

120 activity which frequently involves only a few banks. The Results for Several Predictors All of the significant predictors were paired with each 2 other and with each initially non-significant predictor. None of the initially non-significant predictors gained significance when combined with initially significant predictors. Table 8.3 is designed to show the relative importance of various pairs of significant predictors other than personal income. No pairings involving personal income are shown in the table because this highly significant variable overshadows the other predictors, thereby making it difficult to determine the relative importance of the other less significant predictors. Also only pairings involving income to capital and deposits per capita are shown in the table because these variables were slightly better predictors of activity than income to assets and deposits per bank. The results from pairing significant predictors with each other are largely additive. In other words, if two variables which were found to be significant in terms of t-values are combined the A more retined snowball variable was tested. The refinement consisted of setting the variable equal to 2 if there had been activity during each of the previous two 2-year periods. The results were not as good as the original definition. It had been noted earlier (Chapter 7, p. 107) that the precise direction and impact of the snowball effect after several time periods was unclear. 2 The significant predictors from Table 8.2 are CON, l/CAP, l/AST, TD/BK, TD/PP and SNW. The reader should remember that a stepwise procedure is being used in order to avoid problems of multicollinearity and to conserve degrees of freedom.

PAGE 134

121

PAGE 135

122 Table b.3 Continued Dependent

PAGE 136

123 previously, the snowball variable performs poorly when activity is measured in terms of deposits. As a consequence, pairings with the snowball variable are generally not significant in terms of the likelihood ratio test when the dependent variable is in terms of deposits. The results from combining three significant variables, personal income again being excluded, are shown in Table 8.4. The set of variables I/CAP, TD/PP and SNW do not perform well in explaining the dependent variable. The poor performance is due to the overlapping influence of l/CAP and TD/PP and the non-significance of the variable SNW when deposit measures of activity are used. The other sets of variables perform more or less equally well in terms of likelihood ratio values. The final equations for the model of holding company activity -1 . are listed in Table 8.5. All of the equations are highly significant and the signs of all the coefficients are as originally hypothesized. The CON variable, which performs well in combination with I /CAP and TD/PP, does not work very well with the personal income variable. The poor performance is apparently due to the collinearity between the CON variable and the PI variable. The simple correlation between the two variables is somewhat high (-0.42), which suggests that the variables provide some similar information on holding company activity. Income to capital and deposits per capita perform well when combined with personal income, although the rather high correlation between the two variables leads to some instability in the coefficients when both are included in The addition of a fourth variable caused only very small increases in the likelihood ratio value.

PAGE 137

124 Table 8.4 Tobit Model of Holding Company Activity Using Three Predictors

PAGE 138

125

PAGE 139

126 the same equation. The snowball variable (not shown) performed poorly ^en personal income also appeared in the same equation. This poor performance is likely due to the fact that the personal income variaole is much more significant than the snowball variable, and to the fact that the two variables have a rather high measure of correlation (0.46). It should be noted that given that the variable PI is already included in the equations of Table 8.5, one cannot reject the hypothesis that the other two variables in the equations have coefficient of zero at the .10 level of significance. This result is partly due to the importance of the personal income variable in its own right. This result also appears to be caused by the ability of the PI variable to mirror the influence of other variables. In addition to its high correlation with the snowball variable, the correlation of personal income with concentration is -0.42, with income to capital it is 0.17, and with total deposits per capita it is 0.18. Numerous combinations of initially insignificant predictors with significant predictors were tested. At no time, did an initially insignificant predictor approach a ,10 level of significance. The correlations between variables are displayed in Table 8.6. The three instances of rather high correlations between independent variables are circled.: In general, the correlation matrix does not suggest any serious problems of multicollinearity. In addition, the reasonably stable maximum likelihood coefficients reported in Tables 8.2 through 8.5 indicate that the results are not strongly distorted by At the .25 level of significance, one would be able to reject the null hypothesis tor nine of the twelve equations in the table.

PAGE 140

127 U P4 QIP4 hIpq o vO c3q 1^

PAGE 141

128 statistical multicollinearity. The probability of activity and the expected value of activity are presented in Table 8.7 for selected values of the independent variables. The level of personal income deflated has the largest impact on holding company activity. A twenty percent increase in per capita personal income from a mean level of $2,920 to a level of $3,500 will increase the likelihood of activity by nearly 15 percentage points. The expected amount of activity will increase by over 8 percentage points. The variable with the second largest quantitative influence is l/CAP followed by TD/PP. The concentration variable has the smallest quantitative influence on activity with a 40 percent increase in concentration from a mean of .435 resulting in a decrease of only 4 percentage points in the likelihood of activity and a decrease of just 1.5 percentage points in the expected value of activity. The reader should not attempt a linear extrapolation of the results in the table since the tobit technique is non-linear. For instance, an 80 percent change, in a variable does not have twice the quantitative impact of a 40 percent change .

PAGE 142

129 r^ M o > G U *a u .S (U u a X H 0) 3*0 > -H u o 9 o<>t-i X o r-» 30 r^ n oo o CMlOr^OCM>OONU^r»»iACMi--l>-l O "H 0) 00 43 C «> ^ O u H 1-1 > •H cooo «*>*r^ cMvooo a\ m cnop«.CTviocnr-iO'noo o^r«>iocnr-.cMooc*ir^oco « Q O

PAGE 143

CHAPTER 9 CONCLUSIONS FOR CASE ANALYSIS AND DIRECTIONS FOR FUTURE RESEARCH The purpose of this dissertation has. been to explain empirically the behavior of multibank holding companies. In order to accomplish this purpose, a model of lead bank status and a model of holding company activity were developed and tested. In an effort to approximate a controlled experiment, both models were applied exclusively to Florida, a state with a number of striking characteristics. A Summary of the Results The probit statistical technique rather than the more conventional regression approach was used to test the model of lead bank status because the dependent variable only takes the value of zero or one. It was found that the most important factor causing a bank to select lead bank status is the amount of correspondent business done by the bank. Another very important determinant of lead bank status is the absolute size of the bank. The other important variables influencing lead bank status are prior encroachment by holding companies, which tends to discourage banks from becoming lead banks, and the When correspondent business and size are measured at different points in time, it is appropriate to deflate both variables in order to control for pure price changes. 130

PAGE 144

131 importance of a market as a financial center, which has a positive effect on lead bank status. No evidence was found to support the idea that banks with declining market shares tend to anchor holding companies. Finally, lead banks cannot be characterized as either wholesaleor retail-oriented nor are they especially successful in gaining loans in general. 1 The model of holding company activity was tested using the tobit statistical method. The tobit method is appropriate because the dependent variable does not take on negative values, having a lower limit of zero. Market profitability (income as a proportion of capital or assets), market condition (deposits per bank or per capita), and especially the level of per capita personal income were found to have a positive and significant effect on holding company activity. Highly concientrated markets experience somewhat less activity than unconcentrated markets. Neither population per bank nor short term (two-year) changes in population, deposits or personal income have a systematic influence on holding company activity. There is some empirical support for the contention that holding company activity tends to snowball over several time periods. Conclusions for Case Analysis Perhaps the best way to describe the case implications of the lead bank model is to evaluate the probability of a hypothetical bank 1 The author worked as an economist in the Banking Markets Section at the Federal Reserve Board while completing the dissertation. A major part of his duties consisted of reviewing bank holding company applications, participating in the economic analyses of these applications, and assisting in the preparation of case memoranda to the Board. It was his intention that the dissertation would assist in the economic analyses of holding company cases.

PAGE 145

132 becoming a lead bank. One almost automatically looks at the size of the bank and size was found to be an important determinant of lead bank status. Assuming the bank is one of the largest banks in its state, it is most necessary to determine if it is an important source of correspondent services. If it is, this increases substantially the probability of the bank becoming a lead bank. Other things being equal, the more important the bank's local market as a financial center, the more likely the bank to opt for lead bank status. Also, if relatively few banks in the recent past have affiliated with holding companies, the bank is more likely to become a lead bank than if a substantial portion of the banks in the market have come under holding company control. The fact that the bank has experienced a decline in market share has no impact on the likelihood of its choosing lead bank status. Even though the bank is heavily involved in business lending, this does not affect the likelihood of its becoming a lead bank. If the bank were especially aggressive in seeking loans in general, or consumer loan in particular, these facts would shed no light on the probability of its choosing lead bank status. The results from the model of holding company activity will be useful in analyzing the relative attractiveness of different markets from the viewpoint of holding companies. One way to describe the implications of this model for case analysis is to analyze a hypothetical application which involves the market-extension acquisition of a sizeable bank by one of the largest holding companies in a state. The

PAGE 146

133 key issue centers around potential competition. If the application is denied, what is the probability that this holding company will 2 acquire a smaller bank or open a new bank in the local market. A most important factor to consider in analyzing this hypothetical case is the level of per capita personal income. If personal income is very high in the market, this increase substantially the probability of the holding company entering the market in a less anticompetitive fashion. The higher are the profits of banks in the market, the better are the chances of the applicant entering the market by an alternate method. Also, a favorable level of deposits 3 per capita increases the chances of the bank entering the market de novo or by means of a toehold acquisition, assuming the application is denied. The ratio of population to bank does not have a systematic influence on the probability of the applicant entering the market in a more competitive fashion. It appears that holding companies are more interested in deposits that in population. Short term (two-year) For a discussion of the concept of potential competition as applied by the regulatory authorities, see S,A. Rhoades and A, J. Yeats, "A Clarification of the Potential Competition Doctrine in Bank Merger Analysis," Board of Governors; Staff Economic Comment , vol. 52 (April, 1972), pp. 25-33. 2 The term "entry" is deliberately avoided. All of the entry studies in banking have used a narrow definition of entry which does not allow for re-entry or multiple entry which are quite usual in states were holding companies are active. In addition, these studies focus upon de novo entry despite the fact that entry via toehold acquisitions and de novo entry seems to be close substitutes for holding companies.^., The model of holding company activity avoids both these shortcomings. 3 ' An alternate measure Is deposits per bank. The correlation between deposits per capita and deposits per bank is very high (0.77) and both measures of market condition were significant predictors of activity.

PAGE 147

134 rates of growth in deposits, population and personal income do not affect a holding company's evaluation of the attractiveness of a market. Directions for Future Research a' logical next step would he to apply the two models to other states where holding companies are active. There would be serious problems, however, with using the model of lead bank status in other states. This model worked well in Florida because Florida has by far the greatest number of multibank holding companies. In other states, the number of "successes" — banks that become lead banks— would be rather small, say a dozen or less per state. As a consequence, statistical tests would be hampered. The model of holding company activity can be. rather easily applied to other states. The data and the statistical program are readily available. Care would have to be exercised to assure that different states were at the same stage of holding company development if the model were applied to several states at once.

PAGE 148

APPENDIX Table A.l The Set of Potential Lead Banks Rank Name of Bank 1 First N/B of Miami 2 First N/B of Tampa 3 Atlantic N/B of Jacksonville 4 First N/B at Orlando 5 Florida N/B of Jacksonville 6 Exchange N/B of Tampa 7 City N/B of Miami 8 Barnett Bank of Jacksonville 9 First N/B in Fort Lauderdale 10 Miami Beach First N/B 11 First N/B in St. Petersburg 12 Union Trust N/B, St. Petersburg 13 Florida N/B and Trust, Miami 14 Broward N/B of Fort Lauderdale 15 Marine Bank and Trust, Tampa 16 Bank of Clearwater 17 First State Bank of Miami 18 Pan American Bank of Miami 19 Coral Gables First N/B 20First N/B in Palm Beach 21 Peoples Bank of Lakeland 22 Sarasota Bank and Trust 23 First N/B in Fort Myers 24 Palmer First N/B at Sarasota 25 Florida N/B at St. Petersburg 26 Barnett First N/B of Winter Park 27 First N/B of Hollywood 28 Citizens N/B of Orlando 29 American N/B of Jacksonville 30 City N/B of Miami Beach 31 Miami N/B 32 Lee County Bank, Fort Myers 33 First N/B of Pompano Beach 34 First N/B of Clearwater 35 American N/B, Fort Lauderdale 36 First Bank and Trust, Boca Raton Lead Bank Status Potential Lead Bank Potential Lead Bank Lead Bank Since 1920s Potential Lead Bank Lead Bank Since 1920s Potential Lead Bank Potential Lead Bank Lead Bank Since 1920s Potential Lead Bank Potential Lead Bank Potential Lead Bank Potential Lead Bank Decisions Made by 5 Potential Lead Bank Potential Lead Bank Potential Lead Bank Potential Lead Bank Potential Lead Bank Decisions Made by 10 Potential Lead Bank Potential Lead Bank Decisions Made by 47 Potential Lead Bank Potential Lead Bank Decision Made by 5 Potential Lead Bank Potential Lead Bank Potential Lead Bank Potential Lead Bank Decisions Made by 7 Potential Lead Bank Potential Lead Bank Potential Lead Bank Potential Lead Bank Potential Lead Bank Potential Lead Bank -135

PAGE 149

136 Table A.l Continued Rank Name of Bank 37 Barnett Bank of Holljwood 38 Hileah-Miami Springs First State 39 Dania Bank 40 Jacksonville N/B 41 Central Bank and Trust, Miami 42 State Bank of Jacksonville 43 Fort Lauderdale N/B 44 First N/B, Lake Worth 45 St. Petersburg Bank and Trust 46 Boca Raton N/B 47 First N/B of Bradenton 48 Bank of Palm Beach and Trust 49 Coral Ridge N/B of Fort Lauderdale 50 Mercantile N/B of Miami Beach 51 Florida N/B of Orlando 52 Atlantic N/B of West Palm Beach 53 Exchange N/B of Winter Haven 54 Riverside Bank, Miami 55 Capital City First National, Tallahassee 56 First N/B, New Port Richey 57 First N/B and Trust, Dunedin 58 Bank of Naples 59 Manatee N/B 60 Commercial Bank at Winter Park 61 Lewis State Bank, Tallahassee 62 Commercial Bank and Trust, Miami 63 Springfield Atlantic Bank, Jacksonville 64 First N/B of South Miami 65 First N/B of Delray Beach 66 First N/B of Gainesville 67 First N/B and Trust, Stuart 68 Jefferson N/B of Miami Beach 69 Bank of Hallandale and Trust 70 First N/B and Trust, Naples 71 First Marine Bank and Trust, Riviera Beach 72 Southeast Bank of Deerfield 73 Citizens N/B of West Hollywood 74 First N/B, Hialeah 75 Central Plaza Bank, St. Petersburg Lead Bank Status Potential

PAGE 150

137 Table A. 2 Decision made by Potential Lead Banks Through December 31, 1972 Rank Name of Bank 1 First N/B of Miami 2 First N/B of Tampa 3 First N/B at Orlando 4 Exchange N/B of Tampa 5 City N/B of Miami 6 First N/B in Fort Lauderdale 7 Miami Beach First N/B 8 First N/B in St. Petersburg 9 Union Trust N/B, St. Petersburg 10 Broward N/B of Fort Lauderdale 11 Marine Bank and Trust, Tampa 12 Bank of Clearwater 13 . First State Bank of Miami 14 Pan American Bank of Miami 15 First N/B in Palm Beach 16 Peoples Bank of Lakeland 17 First N/B in Fort Myers 18 Palmer First N/B at Sarasota 19 Barnett First N/B of Winter Park 20 First N/B of Hollywood 21 Citizens N/B of Orlando 22 American N/B of Jacksonville 23 Miami N/B 24 Lee County Bank, Fort Myers 25 First N/B of Pompano Beach 26 First N/B of Clearwater 27 American N/B, Fort Lauderdale 28 First Bank and Trust, Boca Raton 29 Barnett Bank of Hollywood 30' Dania Bank 31 Jacksonville N/B 32 Central Bank and Trust, Miami 33 State Bank of Jacksonville 34 First N/B, Lake Worth 35 St. Petersburg Bank and Trust 36 Boca Raton N/B 37 First N/B of Bradenton 38 Bank of Palm Beach and Trust 39 Mercantile N/B of Miami Beach 40 Riverside Bank, Miami 41 Capital City First National, Tallahassee 42 First N/B and Trust, Dunedln 43 Bank of Naples 44 Manatee N/B 45 Commercial Bank at Winter Park Decision on

PAGE 151

138 Table A. 2 Continued Rank Name of Bank • 46 Lewis State Bank, Tallahassee 47 Commercial Bank and Trust, Miami 48 First N/B of South Miami 49 First N/B of Delray Beach 50 First N/B and Trust, Stuart 51 Jefferson N/B of Miami Beach 52 Bank of Hallandale and Trust 53 First N/B and Trust, Naples 54 First Marine Bank and Trust, Riviera Beach 55 Southeast Bank of Deerfield 56 Citizens N/B of West Hollywood 57 First N/B, Hialeah 58 Central Plaza Bank, St. Petersburg Decision on

PAGE 152

SELECTED BIBLIOGRAPHY Adams, Eugene H. "Economy, System Changes Require States to Modernize Bank Structure," American Banker . September 6, 1972, pp. 5, 10-11. 1 Alhadeff, Charlotte P. and Alhadeff, David A. "Recent Bank Mergers," Quarterly Journal of Economics , vol. 55 (November, 1955), pp. 503-532. Allen C. Ewing & Company. Florida Banking and Its Largest Banking Institutions . Jacksonville, Florida: Allen C. Ewing 6e Company, 1972. "Amendments to Bank Holding Company Act," Federal Reserve Bulletin , vol. 52 (July, 1966), pp. 966-971. "The Amendments to the Bank Holding Company Act: One Year Later/' Business Conditions . Federal Reserve Bank of Chicago (December, 1971), pp. 2-11. "Bank Breakthrough in New Mexico," Burroughs Clearing House , vol. 54 (September, 1970), pp. 24-25, 58-60. "Bank Holding Company Act Amendments of 1970," Federal Reserve Bulletin . vol. 57 (January, 1971), pp. 29-33. "Bank Holding Company Act of 1956," Federal Reserve Bulletin , vol. 42 (May, 1956), pp. 444-453. "Banking Act of 1933," Federal Reserve Bulletin , vol. 19 (June, 1933), pp. 385-401. Baumol, William J. Business Behavior. Value and Growth . New York: Macmillan Company, 1959. Bell, Fredericks, and Murphy, Neil B. Costs in Commercial Banking : A Quantitative Analysis of Bank Behavior and Its Relation to Bank Regulation . Boston: Federal Reserve Bank of Boston, 1968. Cagle, Caroline H. "Member Bank Income, 1969," Federal Reserve Bulletin . vol. 56 (July, 1970), pp. 564-572. Carter H. Golembe and Associates. The Future of Registered Bank Holding Companies . Washington, D.C. : The Association of Registered Bank Holding Companies, 1971. 139 -

PAGE 153

140 Comptroller of the Currency. Annual Report . Washington, D.C.: Government Printing Office, 1964. Cornfield, Jerome and Mantel, Nathan. ".Some New Aspects of the Application of Maximum Likelihood to the Calculation of the Dosage Response Curve," Journal of the American Statistical Association , vol. 45 (June, 1950), pp. 181-210. Darnell, Jerome C. "Chain Banking," National Banking Review , vol. 3 (March, 1966), pp. 307-319. Darnell, Jerome C. "Determinants of Chain Banking," National Banking Review , vol. 4 (June, 1967) pp. 459-468. Dovell, J.E. History of Banking in Florida . Orlando, Florida: Florida Bankers Association, 1955. "Federal Reserve Act Digest," in Encyclopedia of Banking Law . Hartford, Conn.: Lamont Cross & Company, 1964 Finney, D.J. "Estimation from Individual Records of the Relationship Between Dose and Quantal Response," Biometrika, vol. 34 (June, 1947), pp. 320-334. Finney, D.J. Probit Analysis . 3rd ed. Cambridge, England: The University Press, 1971. Fischer, Gerald C. American Banking Structure . New York: Columbia UMversity Press, 1968. Fischer, Gerald C. Bank Holding Companies . New York: Columbia University Press, 1961. Gup, Benton E. A Study of Bank Structures . Occasional Report Series, vol. 2, no. 3; Tulsa, Oklahoma: University of Tulsa, 1970. Hall, George R. "Bank Holding Company Regulation," Southern Economic Journal , vol. 31 (April, 1965), pp. 342-355. Jessup, Paul. F. "Portfolio Strategies for Bank Holding Companies," The Bankers Magazine , vol. 52 (Spring, 1969), pp. 78-85. Johnson, Thomas. "Qualitative and Limited Dependent Variables in Economic Relationships." Econometrica , vol. 40 (May, 1972), pp. 455-462. Kohn, Donald L. and Zoellner, John F. "The Amended Bank Holding Company Act," Monthly Review . Federal Reserve Bank of Kansas City (May, 1971), pp. 11-20. Lamb, W. Ralph. Group Banking . New Brunswick, N.J. : Rutgers University Press, 1962.

PAGE 154

141 Lanzilottl, Robert F. "Pricing Objectives in Large Companies: Reply." American Economic Review , vol. 49 (September, 1959), pp. 679-687. Lawrence, Robert J. Operating Policies of Bank Holding Companies . Washington, D.C. : Board of Governors of the Federal Reserve System, 1971. Lawrence, Robert J. The Performance of Bank Holding Companies . Washington, D.C. : Board of Governors of the Federal Reserve System, 1967. Lee, Tong Hun. "Alternate Interest Rates and the Demand for Money: The Empirical Evidence," American Economic Review , vol. 57 (December, 1967), pp. 1168-1181. Mabry, Bevars D. and Siders, David L. "An Empirical Test of the Sales Maximization Hypothesis," Southern Economic Journal , vol. 28 (January, 1967), pp. 367-377. MachluELjJ!ritz.._; "Theories of the Firm: Marginalist, Behavioral, Managerial." American Economic Review , vol. 57 (March, 1967), pp. 1-33. Harris, "R.L. "Review of E.T. Penrose: The Theory of the Firm ." Economic Journal , vol. 71 (March, 1961), pp. 144-148. Harris, Robin. The Economic Theory of Managerial Capitalism . London: Ma cmil Ian Co., 1964 HcLeary, Joe W. "Bank Holding Companies: Their Growth and Performance," Monthly Review , Federal Reserve Bank of Atlanta (October, 1968), pp. 131-138. Hood, Alexander. Introduction to the Theory of Statistics . New York: HcGraw-Hill Book Company, 1950. Hpyer, R. Charles. A Model of the Determinants of Registered Bank -^^ Holding Company Acquisitions . Doctoral Dissertation, Graduate School of Business, University of Pittsburg, 1971. Nodler, Paul S. "Craze for Raising Multiples Spreads with Multi-BHC Craze," American Banker . October 31, 1972, p. 4. Penrose, Edith. The Theory of the Growth of the Firm . New York: John Wiley & Sons, 1959. Piper, Thomas R. The Economics of Bank Acquisitions by Registered Bank Holding Companies . Boston: Federal Reserve Bank of Boston, 1971. Reid, Samuel R. Mergers. Managers and the Economy . New York: McGraw-Hill Book Company, 1968.

PAGE 155

142 Rhoades, S.A. and Yeats, A.J. "A Clarification of the Potential Competition Doctrine in Bank Merger Analysis," Board of Governors; Staff Economic Comment , vol. 52 (April, 1972), pp. 25-33. "Report Under the Bank Holding Company Act," Federal Reserve Bulletin . vol. 44 (July, 1958), pp. 776-796. Schwelger, Irving. "Reply to Chicago Banking: A Critical Review," Journal of Finance , vol. 17 (October, 1962), pp. 424-428. Schweitzer, Stuart A. "Economies of Scale and Holding Company Affiliation in Banking," Southern Economic Journal , vol. 38 (October, 1972), pp. 255-263. Suits, Daniel B. "Use of Dummy Variables in Regression Equations," Journal of the American Statistical Association , vol. 52 (December, 1957), pp. 548-551. Talley, Samuel H. The Effects of Holding Company Acquisitions on Bank Performance . Washington, D.C.: Board of Governors of the Federal Reserve System, 1972. Thomson, Herbert F. and Meadows, Jerry A. "The Expansion of Registered Bank Holding Companies in Ohio," Bulletin of Business Research , vol. 46 (July, 1971), pp. 1-8. Tobln, James. "The Application of Multivariate Problt Analysis to Economic Survey Data," in Cowles Foundation Discussion Paper 1 . 1955 (mimeographed). Tobln, James. "The Estimation of Relationships for Limited Dependent Variables," Econometrica . vol. 26 (January, 1958), pp. 24-36. Tyng, Ed. "New Moves to Form Holding Companies Seem Likely Now that Tax Penalty Is Dead," Journal of Commerce . April 7, 1964, pp. 1, 24. Upshaw, William F. "Antitrust and the New Bank Holding Company Act: Part II," Monthly Review , Federal Reserve Bank of Richmond (March, 1971), pp. 3-10. Upson, Roger B. and Jessup, Paul F. "Returns from Bank Holding Companies," The Bankers Magazine , vol. 155 (Spring, 1972), pp. 59-62. U.S. Bureau of the Budget. Standard Metropolitan Statistical Areas . Washington, D.C. : Government Printing Office, 1967. U.S. Congress. House. Committee on Banking and Currency. Control and Regulation of Bank Holding Companies . 84th Cong. , 1st Sess. , 1955.

PAGE 156

143 Vernon, Jack R. "Separation of Ownership and Control and Profit Rates, the Evidence from Banking: Comment." Journal of Financial and Quantitative Analysis , vol. 2 (January, 1971), pp. 615-625. Warner, Stanley Leon. Stochastic Choice of Mode in Urban Travel; A Study in Binary Choice . Chicago: Northwestern University Press, 1962. Williamson, Oliver E. The Economics of Discretionary Behavior; Managerial Objectives in a Theory of the Firm . Englewood Cliffs, N.J. ; Prentice-Hall, Inc., 1964. Weiss, Steven J. "Bank Holding Companies and Public Policy," New England Business Review . Federal Reserve Bank of Boston (January /February, 1969), pp. 3-29.

PAGE 157

. BIOGRAPHICAL SKETCH Gregory Edward Boczar was born in Panama City, Republic of Panama on March 16 ^ 1943. He grew up in the southern. part of Florida. I He graduated magna cum laude from Xavier University, Cincinnati, Ohio in June of 1965. He received University Scholarships during his four years of undergraduate study. He was a member of student government his junior and senior years, and he was active on the student newspaper. He spent the summers of 1963 and 1964 in Mexico where he studied Spanish. He was awarded a Woodrow Wilson Fellowship and a National Science Foundation Fellowship for graduate study. He entered the doctoral program in economics at Stanford University in September 1965. In September 1967 he took an extended leave of absence from Stanford University for reasons of health. He was an instructor of economics at Florida Atlantic University, Boca Raton, Florida from June 1968 to June 1969. He resumed his graduate studies at the University of Florida in September 1969. He began employment as an economist with the Board of Governors of the Federal Reserve System in October 1970. He received the Ph. D. degree from the University of Florida in December 1973. His major field was economics and his minor field was mathematics. In 1965 he married Marilyn Trauth. They have two children.

PAGE 158

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Ralph IH. Blodgett, Chairm^Ji Professor of Economics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. W^^./. obert G. Blake Associate Professor of Mathematics I certify that I have read this study and that in my opinion it confoirms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. I certify that I have repd this study and that in my opinion it conforms to acceptable stan'dards of scholarly presentation and is. fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. O'l l^^'C'^rU*^ Vernon iate Profiessor of Economics

PAGE 159

This dissertation was submitted to the Department of Economics In the College of Business Administration and. to the Graduate Council, and was accepted as a partial fulfillment of the requirements for the degree of Doctor of Philosophy. Dean, Graduate School