Title: Long-term incentive compensation plan adoption and the capital spending decisions of managers /
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00099326/00001
 Material Information
Title: Long-term incentive compensation plan adoption and the capital spending decisions of managers /
Physical Description: vii, 133 leaves : ; 28 cm.
Language: English
Creator: Baril, Charles Purdom, 1948-
Publication Date: 1986
Copyright Date: 1986
Subject: Management science   ( lcsh )
Capital budget -- United States   ( lcsh )
Corporate profits -- United States   ( lcsh )
Incentives in industry   ( lcsh )
Accounting thesis Ph. D
Dissertations, Academic -- Accounting -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Thesis: Thesis (Ph. D.)--University of Florida, 1986.
Bibliography: Bibliography: leaves 126-132.
Statement of Responsibility: by Charles Purdom Baril.
General Note: Typescript.
General Note: Vita.
 Record Information
Bibliographic ID: UF00099326
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000905636
notis - AEL4679
oclc - 015682862


This item has the following downloads:

longtermincentiv00bari ( PDF )

Full Text






Copyright 1986


Charles Purdom Baril


I wish to thank my dissertation committee, Professors Bipin

Ajinkya, E. Dan Smith, and Stephen Cosslett, for their inspiration,

suggestions, and constructive criticism throughout the development of

this dissertation. A special note of thanks is due to my chairman,

Bipin Ajinkya, whose understanding and guidance contributed immensely to

my academic and personal development. I also wish to thank Professors

John Simmons and Gerry Salamon, who read earlier drafts and provided

many valuable comments and much encouragement, and also Professors

Sanjai Bhagat, James Brickley, and Ronald Lease, who permitted me to use

their list of firms which had voted on incentive plans. Furthermore,

the generous financial support of the Deloitte, Haskins & Sells

Foundation is gratefully acknowledged.

Finally, I want to express my appreciation to those people whose

emotional and spiritual support made this all possible. To my mother

who many years ago suggested that my calling might be in the academic

field, to my father whose patience with and confidence in me will always

be remembered, to my wife and best friend, Candy, whose love is the

single most important thing in my life, to my daughter, Amanda, whose

creative and imaginative play helped me to maintain a more wholesome

perspective on my life, and to my Virginia and Florida friends who

helped me to laugh when I needed it most, I give my heartfelt thanks.



ACKNOWLEDGEMENTS . . . . . . . . . 1iii

ABSTRACT . . . . . . vi


I. INTRODUCTION . . . . . . . . .. . . 1

RELATED LITERATURE . . . . . . . . ... . 8

2.1 Long-term Compensation Plans . . . . . . . 8
2.2 Literature Review . . . . . . . . . .. 13

III. DEVELOPMENT OF HYPOTHESES . . . . . . . . ... 27

3.1 The Adoption Hypotheses . . . . . . ... 27
3.2 Past Performance Interaction Hypothesis . . . ... 32
3.3 Ownership Control Interaction Hypothesis . . . ... 37
3.4 Notes . . . . . . . . . . . . . 43

IV. RESEARCH METHODOLOGY .. . . . . . . . ... 45

4.1 Sample Selection . . . . . . .. .... . 45
4.2 Research Design . . . . . . . . . .. 46
4.3 Dependent Variables . . . . .. . . . .. 47
4.4 Independent Variables . . . . . . . ... 52
4.5 Notes . . . . . . . . . . . . . 63

V. EMPIRICAL ANALYSIS . . . . . . .. . . . 64

5.1 Dependent Variables -- Descriptive Statistics . . .. .64
5.2 Tests of Hypotheses . . . . . .. . . 70
5.3 Notes . . . . . . . .. .... . .. .... . 101

VI. SUMMARY AND CONCLUSIONS . . . . . . . .... . .102

6.1 Discussion of Results . . . . . . . ... 103
6.2 Conclusions and Limitations . . . . . . .. .109



REFERENCES . . . . . . . . . . . . . 126

BIOGRAPHICAL SKETCH .. ............ ........ 133

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



Charles Purdom Baril

December 1986

Chairman: Bipin B. Ajinkya
Major Department: Accounting

The motivation and incentives theory of compensation posits that

performance-contingent compensation plans are designed and implemented

to reconcile the incentives of managers with those of stockholders.

Kaplan and Rappaport among others have noted that resourceful managers

can enhance short-run profits by rejecting profitable investment and

discretionary expenditure projects. This dissertation examines whether

the introduction of long-term incentive compensation contracts may

induce myopic managers to extend their decision horizons, making capital

investment and research and development projects more acceptable.

It is hypothesized that the adoption of long-term compensation

plans is positively associated with unexpected capital spending level

increases. Three separate models are used to generate firm-specific

expectations of capital spending levels: (i) Value Line forecasts,

(ii) econometric model predictions of investment, and (iii) industry

average. Furthermore, in an attempt to enhance our understanding of the

conditions under which incentive compensation schemes are more

effective, the dependence of unexpected capital spending level changes

on the interaction between plan adoption and (i) a firm's ownership

control status and (ii) its prior earnings growth (both proxies for

agency costs) is investigated. These hypotheses are tested

cross-sectionally using a multivariate regression model and a sample of

216 adopting and 139 non-adopting firms.

The evidence shows the hypothesized positive relationship between

long-term plan adoption and unexpected capital spending level increases

to be significant across both performance and restricted stock plans for

the year of adoption and the two-year period commencing with the

adoption year. Furthermore, the results exhibit a significant

association between unexpected capital spending increases and the

relative magnitude of the economic award under the adopted plan.

Finally, it is shown that the degree of separation between a firm's

ownership and its control is significant in its interaction with plan

adoption in explaining the magnitude of the capital spending increase.

These findings are interpreted as supporting the incentives theory of

compensation, and as being inconsistent with either the competing

screening-sorting or the signalling theories operating exclusively.

However, this finding does not negate the applicability of these

alternative theories in the settings examined.


Concern over American corporation management's preoccupation with

short-term profits at the expense of long-term firm profitability has

recently resurfaced. Kaplan (1984, p. 409) notes that during difficult

times, resourceful managers can "earn" profits "not just by selling

more or producing for less, but by engaging in a variety of

nonproductive and typically nonvalue-creating activities." Examples of

such behavior include managing accruals, changing accounting methods,

and generating earnings through financial transactions like mergers and

debt swaps. While such actions may ineffectively utilize a firm's

resources, the most damaging dysfunctional behavior induced by this

preoccupation with short-run performance may be the incentive created

for managers to make investment and discretionary expenditure decisions

which undermine firm value [Kaplan (1984)].

Sunder (1980) finds that capital investment expenditures

significantly reduce reported earnings in the year following the out-

lay. As possible reasons for this effect he points to the practice of

expensing (i) individual items (otherwise capitalizable) smaller than a

specified amount and (ii) difficult-to-separate administrative and

factory start-up costs. For discretionary outlays which must be

charged to expense when incurred, the short-run negative earnings

consequence is clear. Research and development (R & D) costs exemplify

this latter category.

It is apparent then that a manager who is attending to short-term

concerns can immediately and positively impact reported profits by

rejecting capital investment and R & D proposals. (Future references to

capital spending imply both capital investment and R & D expenditures.)

However, if these projects promise to generate added firm value, their

rebuff may sacrifice shareholder worth and the long-term economic

health of the firm. In light of this potentiality, this research study

examines whether long-term incentive compensation contracts may be

effective in redirecting the decision focuses of managers.

Rappaport (1978) explains why managers may ignore or discount the

expected long-term consequences of their decisions.

What is economically rational from the corporate or
social viewpoint may, however, be an irrational course
of action for the executives charged with the decision
making. If their incentives conflict with the longer-
term view of economic efficiency, we can expect their
resource allocation decisions to be 'rational' but
inefficient. [p. 82]

One of the strongest incentives for managers to disregard or underweigh

the long-range effects of their decisions is the evaluation of and

reward for their performance based on reported annual earnings. Promo-

tion decisions both within and outside of the firm generally employ

some earnings measure. While trends in earnings may be considered, it

is not unusual for the most recent earnings report to be weighed most

heavily. Furthermore, executive compensation is tied directly to

earnings whenever income-based bonus plans are in effect. Over ninety

percent of the eleven hundred largest U.S. manufacturing concerns used

bonus schemes to compensate managers in 1984, with the 1983 median

award representing fifty percent of salary [Fox (1985)]. Evaluating

performance on the basis of annual earnings may compel managers to

concentrate on short-run results, thereby adopting policies

discouraging growth.

The property rights literature posits that an individual's

behavior within an organization is dependent upon the nature of his

compensation contract [Coase (1937), Alchian and Demsetz (1972), Jensen

and Meckling (1976), and Monsen and Downs (1965)]. If contractually

specified incentive schemes can influence managerial behavior, the

introduction of long-term performance-contingent arrangements may help

redirect management's focus to longer-termed performance goals.

Examples of long-term incentive arrangements include stock options,

stock appreciation rights, restricted stock plans, and phantom stock

plans. These plans reward managers on the basis of stock price perfor-

mance which has an inherently long-term focus. Another long-term

scheme which has recently come into vogue is performance plans. These

arrangements measure and reward performance on the basis of growth in

various accounting measures (typically earnings per share or return on

equity) over a three- to five-year performance period. This study

concentrates on the motivational effects of market-based restricted

stock and phantom stock plans and accounting-based performance plans.

Only scant empirical evidence exists on the effectiveness of such

plans in extending the decision horizons of managers. Preliminary

evidence provided by Larcker (1983) suggests that the introduction of

performance plans may induce an increase in capital investment

expenditures. Furthermore, studies by Brickley, Bhagat, and Lease

(1985) and Larcker (1983) provide evidence supporting the hypothesis

that the introduction of long-term incentive compensation schemes can

increase shareholder wealth. While this positive market reaction has

been attributed by some to the tax implications of the plans [Miller

and Scholes (1982)], the analysis of Smith and Watts (1982) suggests

that each of these long-term compensation plans may also have an

incentive/signalling effect.

The primary objective of this research effort is to examine

whether the adoption of long-term incentive compensation plans leads to

changes in the capital spending decisions of managers, in a manner

contributing to an extended decision-making horizon. Unexpected

increases in capital spending levels systematically associated with

plan adoption will provide further evidence that compensation contracts

can affect the production/investment decisions of managers.

Studies of this type must seriously consider the problem of self-

selection. Because the firms have self-selected into the experimental

and control groups, randomization cannot be relied upon to control for

pre-existing differential conditions between firms. Previous research

in this area [Larcker (1983)] utilized a matched pairs design in an

attempt to control for potentially confounding factors influencing

investment levels. However, the use of such a design presupposes that

cross-sectional differences in investment expenditure levels between

paired firms remain stable over time. Larcker (1983) does not

explicitly model the expected capital spending behavior of the firm.

To partially alleviate the problem of confounding influences this study

utilizes three distinct models of expected capital spending:

(i) Value Line forecasts, (ii) firm-specific econometric model

predictions of the investment process, and (iii) industry average (a

naive model). In explicitly or implicitly considering macroeconomic,

industry, and firm-level factors influencing capital spending levels,

the use of these models accords greater assurances that changes in

managerial capital spending decisions are due to the adoption of

long-term compensation plans, rather than the presence of other

differential investment opportunities.

In an attempt to further enhance controls on and provide

additional evidence concerning the hypothesized relationship between

plan adoption and ensuing changes in capital spending decisions, two

additional independent variables are introduced; the past performance

of the firm and the ownership control status of the firm. The adoption

of long-term incentive compensation plans should be most effective in

motivating capital spending increases in firms where managers have been

behaving in a self-serving manner. To the extent that these variables

are associated with excessive agency costs, the significance of their

interaction with the adoption of such plans will help in validating

the theory of compensation by providing support for the motivation and

incentives hypothesis.

To summarize, this study examines whether the adoption of

long-term compensation arrangements may effectively induce unexpected

increases in capital spending levels. The results indicate that the

adoption of both performance plans and restricted stock plans is

associated with significant unexpected increases in capital spending

for the one- and two-year periods commencing with the year of adoption.

Moreover, a firm's ownership control status is found to be significant

in its interaction with plan adoption in explaining the magnitude of

the capital spending level increase. This result is interpreted as

supporting the motivation and incentives theory of compensation. While

previous research has shown incentive compensation plan adoption to be

associated with changes in managerial decisions [Larcker (1983) and

Waegelein (1985)] and with positive abnormal stock returns [Larcker

(1983), Brickley, Bhagat and Lease (1985), and Tehranian and Waegelein

(1985)], this study is the first to obtain results which distinguish

among the alternative compensation theories.

Further contributions made by this study to the theory of

compensation involve improvements in Larcker's (1983) methodology.

Sample size is increased substantially. Instead of a matched-pairs

design, three expectation models are utilized to provide additional

controls for potentially confounding or otherwise omitted variables

that may influence capital spending levels. In addition to performance

plans, this study considers the relationship between restricted stock

plan adoptions and capital spending changes. Finally, the results of

this study are substantially stronger than those of Larcker (1983).

In the next chapter a variety of long-range compensation plans are

characterized and the related literature is reviewed. The research

hypotheses are developed in Chapter III, and in Chapter IV the research

methodology and sample selection techniques are described. Chapter V

provides a description of the results. The final chapter of this


dissertation, Chapter VI, discusses these findings and their



This chapter characterizes long-term compensation plans and

briefly describes those considered in this study. The empirical

literature which examines the incentive effects of executive

compensation contracts is then reviewed. Finally, long-term

compensation arrangements are analyzed in terms of their ability to

control the conflict of interests between stockholders and managers.

2.1 Long-term Compensation Plans

Long-term compensation plans are characterized by the measurement

of performance over a period of time exceeding one year. This distin-

guishes these plans from bonus plans which reward managers on the basis

of annual performance, generally measured as a function of accounting

earnings. In addition to the more commonly used stock option and stock

appreciation rights (SAR) plans, three other major compensation schemes

are classified as long-term: restricted stock plans, phantom stock

plans and performance plans.

Restricted stock plans make outright awards of shares to

executives subject to certain restrictions, the most important being

that the shares are subject to forfeiture until they are "earned out"

on the basis of continued employment over a specified period. The

executive is prevented from assigning, transferring or selling the

stock during the vesting period, which generally runs between three and

ten years. However, during the restriction period the manager receives

the dividends and usually votes the shares. More common in smaller

firms, about 15 percent of the Conference Board sample had such plans

in 1984 (Fox (1985)].

Related to restricted stock plans, phantom stock plans award cash

to the executive equivalent to share value, rather than the actual

shares. By elimination of the need to sell the shares to obtain cash,

transactions costs for the recipient are reduced. Under both

restricted stock and phantom stock plans, managers are compensated on

the basis of their continuing to remain in the firms' employ.

Additionally, the awards are made in shares or based upon share price.

In making a manager's compensation dependent upon the future price of

shares, his performance and related remuneration are determined by

measures which presumably reflect the long-run effects of his

decisions. In light of these similarities, these plans are aggregated

and referred to simply as restricted stock plans throughout this study.

Performance plans on the other hand explicitly state performance

goals in terms of accounting-based measures (earnings per share, growth

in earnings per share, accounting rate of return on assets, etc.) at

the beginning of a three- to five-year award period. Resulting

compensation is paid at the end of the award period and is generally

forfeited if the executive leaves the firm during that period.

Performance plans are comprised of two types: performance units

and performance shares. Under both types the executive is allocated a

fixed number of units or shares at the beginning of the award period,

with the number actually earned depending upon the extent to which the

goals are met. The primary difference between the two classes of

performance plans lies in the variability in value of the award. While

the dollar value of each performance unit is assigned at the beginning

and remains fixed throughout the award period, the dollar value of each

performance share depends upon the market price per share at the end of

the performance period. The use of performance plans has grown rapidly

since their inception in the early 1970s, particularly among the larger

firms. While fifty-four of the Fortune 100 firms had performance plans

in 1983 [Towers, Perrin, Forester, and Crosby (1983)], only 25 percent

of the Conference Board sample of 1053 firms had such plans in 1984

[Fox (1985)1.

Stock option plans provide managers with rights to purchase a

specified number of the firms' shares during a given exercise period at

a prescribed price. Options have a legal time limit of ten years for

their exercise, but a shorter limit may be specified by the plan. If

the manager leaves the company prior to exercising his options, the

options are generally terminated. Option plans are the most popular

long-term incentive scheme with 59 percent of the Conference Board

sample having such plans in 1984 [Fox (1985)]. Related to stock

options, SAR's give the executive the right to relinquish his option

and receive cash equal to the difference between the stock price and

the exercise price, thereby reducing transactions costs. Of the firms

with stock options in 1984, more than half had SAR's attached [Fox

(1985)]. Like restricted stock plans, both stock options and SAR's

compensate managers on the basis of share price performance.

This study limits its examination to the incentive effects arising

from the adoption of performance and restricted stock plans. While

stock options and SAR's are also inherently long-term, they are omitted

from consideration due to their immense popularity. Their exclusion

makes it more likely that a representative and adequately sized control

sample of firms (those which do not have in place any of the long-term

plans considered during the test period) is obtained. Furthermore,

stock option plans may have lost some incentive effects if general

market conditions limit the managers' controllability of future

rewards. In fact, this contention may explain the more recent

emergence of performance plans and restricted stock plans, which makes

their study more interesting. Nevertheless, disregarding the stock

option plans which the sample firms have in place does impose certain

limitations. Of greatest concern is the bias their omission creates

against demonstrating the hypothesized incentive effects for the

compensation plans considered. However, their omission also limits the

inferences which may be drawn from the results of this study, as it may

contribute to the problem of self-selection.

A growing number of firms are utilizing long-term compensation

plans each year. When presented for stockholder approval, the

incentive effects of these arrangements are generally underscored, as

evidenced by the following Proxy Statement excerpts:

The purpose of the 1980 Performance Plan is to
establish an additional incentive program for certain
key managers which will (i) stimulate, recognize and
reward the contributions of key executives to
achievement of long-range corporate goals...

Meredith Corporation's 1980 Proxy Statement, p. 4

Moreover, the Board feels that it is very desirable
that the personal interests of the Company's key
executives be closely associated with the interests of
the stockholders, and that this can be accomplished in
part by relating the executives' potential for capital
accumulation with increases in the shareholders'
return. The Board therefore recommends that the
shareholders authorize the adoption of a Long-Term
Incentive Plan.

Scott Paper Company's 11 March 1982 Proxy
Statement, p. 14

If the introduction of long-term plans can in fact provoke such an

incentive effect, the divergence of interests between stockholders and

managers may be reduced, thereby enhancing the probability of

successfully competing with firms whose managers act in the

shareholders' interests.

On the other hand, some argue that management compensation plans do

not encourage management to act in the interests of shareholders. As

evidence, Baumol (1967), Marris (1963), and Loomis (1982) point to the

lack of close association between management compensation and any

measure of shareholder wealth. They argue that boards provide little

control over top management and often authorize the establishment of

compensation plans on the basis of management recommendations and

persuasion. If these plans merely provide a means of transferring

wealth from stockholders to managers, rather than promoting more

congruent (with stockholders) behavior on the part of managers, the

chances of a firm competing successfully are reduced. Given these

disparate scenarios, it seems appropriate that the alleged incentive

effects of long-term incentive plans be examined further.

2.2 Literature Review

Recent economics research has modeled the firm as a set of

contracts among individuals who are factors in a team production

setting [Alchian and Demsetz (1972), Jensen and Meckling (1976), and

Fama (1980)]. Because the rewards of production to these individuals

are determined in part by the performance of the entire team or firm,

conflicts of interest may arise. It is hypothesized that various

contractual agreements between team members arise in order to control

these potential conflicts. Of interest to this research effort is the

conflict of interest between the managers and the firm's shareholders,

and the effectiveness of long-term incentive compensation contracts in

controlling this conflict by aligning the decisions of managers with

the desires of shareholders. More specifically, this study examines

whether the introduction of certain compensation arrangements may lead

to changes in the production/investment decisions of managers. In the

next section the empirical literature examining whether and how

incentive contracts can influence managerial behavior is reviewed.

This is followed by an analysis of the incentive effects of

compensation contracts, focusing specifically on the plans considered

in this study: performance plans and restricted stock plans.

2.2.1 Incentive effects of managerial compensation contracts

The incentive effects conjecture arose from the property rights

stream of research, pioneered by Coase and extended by Alchian, Demsetz

and others [Coase (1937, 1959), Alchian (1965, 1968), Alchian and

Kessel (1962), Demsetz (1967), Alchian and Demsetz (1972), Jensen and

Meckling (1976), Monsen and Downs (1965), Silver and Auster (1969), and

McManus (1975)]. Because contractually specified individual rights

determine how costs and rewards will be allocated within a firm,

individual behavior in organizations will be shaped by the nature of

these contracts.

Accounting researchers typically invoke these agency arguments in

maintaining that an incentive effect does exist [Watts and Zimmerman

(1978), Hagerman and Zmijewski (1979)]. They fundamentally assume that

managers are rational economic agents, thereby acting in a self-

interested manner. Furthermore, they assume that the negotiation and

monitoring of these contracts is costly. (With costless negotiation

and monitoring, the compensation agreements can precisely control

desired managerial behavior, thereby eliminating the "moral hazard"

problem.) Additionally, the labor market for managers is assumed to be

characterized by less than full "ex post" settling up. Rational

managers may anticipate the extent to which the value of their human

capital falls if they opportunistically take actions to increase their

own wealth, while diminishing the value of the firm. In order that

compensation contracts affect managerial decisions, this adjustment in

the value of.the managers' human capital must be less than perfectly

offsetting. Finally, it is assumed that the market for corporate

control is characterized by costly takeovers [Smiley (1976)], causing

this mechanism to be less than perfectly effective in controlling the

self-serving actions of managers.

While researchers generally agree at a theoretical level that

incentive compensation contracts should affect managerial decisions,

the empirical evidence is mixed. Managerial decisions which have been

examined include those which are accounting measurement oriented, as

well as those associated with allocating the firm's real resources.

In considering accounting method selection and accrual decisions,

researchers have concentrated on the implications of the most common

accounting-based incentive arrangement, the bonus plan. They

hypothesize that managers so compensated make accounting method and

accrual decisions which enhance reported earnings. Hagerman and

Zmijewski (1979), in an attempt to explain the cross-sectional

variation in corporate accounting procedures, investigate the effect of

an earnings-based incentive compensation plan on managerial choices

concerning inventory methods, depreciation methods, investment tax

credit treatment and amortization periods for past service pension

costs. In classifying specific choices as either income increasing or

income decreasing, they estimate the effects of size, risk, capital

intensity, concentration, and the existence of a bonus plan, on the

manager's choice. They find the compensation plan variable to be

significant in only two of the four managerial choices (they conjecture

that the inventory model was misspecified due to its real tax

consequences and offer no explanation for their insignificant findings

concerning the investment tax credit choice). Their results are

further tainted because their analysis did not include any variable

directly representing debt contracts whose existence has been shown to

be highly correlated with that of bonus plans [Watts and Zimmerman

(1986)]. Therefore, the significant association found between bonus

compensation schemes and accounting methods selection may be driven by

bond covenant considerations, rather than by compensation concerns.

Aware of this criticism, Zmijewski and Hagerman (1981) responded

by introducing a total debt/total assets variable to represent the

effect of debt contracts. Furthermore, instead of looking at each

accounting procedure choice separately, they consider the portfolio of

accounting method choices. They contend that a manager may choose a

mixed portfolio (some income increasing and some income decreasing

procedures) due to his counter-balancing incentives associated with

management compensation and political costs. Utilizing the same sample

firms as their earlier study, they find the coefficient of the

management compensation variable to be significant in predicting a

manager's choice of a portfolio of accounting methods. However, as

noted by Ball and Foster (1982), while their models predict

significantly better than a random model, they do not predict

significantly better than a strategy of predicting that all firms

utilize the most commonly used accounting method in their sample.

In a related study Bowen, Noreen and Lacey (1981) examine the

manager's decision concerning the voluntary capitalization of interest

costs associated with capital projects. However, contrary to the

spirit of Zmijewski and Hagerman (1981), they find no association

between the presence of a bonus plan and the manager's choice. In fact

the management compensation dummy variable, while insignificant, has

the wrong sign. A possible explanation for these conflicting results

is provided by Watts and Zimmerman (1986) who suggest that Bowen,

Noreen and Lacey's tests of management compensation lack power because

many bonus plans add interest back in defining earnings.

Healy (1985), in another study of the effect of accounting bonus

plans on accounting procedures selection, considers the net accruals of

the firm. He recognizes that many bonus plans not only do not pay

bonuses until an earnings target is met, but many also have upper

bounds on the size of the bonus. Based upon this more complete

characterization of bonuses, Healy (1985) finds that if a firm's

earnings are outside the bounds, managers will reduce earnings, and

therefore minimize accruals, to the extent possible. By so doing the

firm's earnings in the following year will naturally be enhanced.

Healy (1985) also examines the incentive effect of bonus schemes

on voluntary changes in accounting methods. While net accruals reflect

both discretionary and nondiscretionary accounting procedures,

voluntary changes in accounting methods reflect purely discretionary

accounting procedure decisions. Interestingly, the results of this

test fail to support his hypothesis, and are inconsistent with his net

accrual results.

The effect of incentive compensation contracts on the resource

allocation decisions of managers has been considered only recently.

Waegelein (1985), employing a matched-pairs design, examines the

association between bonus plan adoption and changes in a firm's R & D,

advertising, and capital expenditure decisions. He finds that the

introduction of a bonus plan is associated with an increase in

advertising expenditures and a decrease in R & D expenditures, while

capital expenditures are not significantly affected.

Larcker (1983) is the first to consider the incentive effects of

compensation arrangements other than bonus plans. Utilizing a matched

pairs design, Larcker (1983) examines the association between

performance plan adoption and corporate capital investment. He finds

that relative to similar control firms, his twenty-five sample firms

which adopt performance plans experience a weak but statistically

significant (.10) increase in investment levels in the year following

plan adoption.

While Larcker (1983) is to be commended for his path-breaking

efforts, several problems constrain the inferences which may be drawn

from his results. While his low significance levels and small sample

size are troublesome, the most problematic aspect of his study is his

inability to attribute his results to the incentive effects of the

adopted plans. Not only are his findings consistent with the

motivation and incentives hypothesis which he embraces, but they are

also consistent with the alternative signalling and screening-sorting

hypotheses (see Section 3.2 for definitions and discussion). Each of

these hypotheses provides an alternative explanation for establishing

performance-contingent compensation arrangements. As a result Larcker

(1983) is unable to infer a causal relationship between performance

plan adoption and increased investment levels. His matched pairs

design contributes substantively to this problem, as it provides

inadequate controls for differential investment opportunities which may

exist between adopting and non-adopting firms. The introduction of an

explicit model of firm-specific expected investment behavior would

mitigate this limitation.

Utilizing a different approach, Walking and Long (1984) consider

the manager's decision to resist takeover bids. According to their

"managerial welfare hypothesis," target managements will react to

tender offers based upon the bid-induced changes in their own utility.

They measure changes in managerial utility through increments in share

and option wealth and through adjustments to the flow of future

compensation from salary. The authors find support, based on their

analysis of 95 cash tender offers, for the hypothesis that the

resistance decision is significantly conditioned on the potential

wealth change to officers and directors.

In a related vein, Larcker (1983) and Brickley, Bhagat, and Lease

(1985) examine the stock price reaction upon announcing the adoption of

long-term plans. Larcker (1983) finds a statistically significant

two-day positive abnormal return of 0.8 percent associated with the

announcement of performance plan adoption. Brickley, Bhagat, and Lease

(1985) find statistically significant positive abnormal returns of

approximately 3.5 percent over the period between the board of

directors, approval date and the stockholder meeting date for New York

Stock Exchange firms announcing the adoption of long-term compensation

plans. While Miller and Scholes (1982) argue that all of these plans

have tax implications which may explain the positive market reaction,

the analysis of Smith and Watts (1982) suggests that these plans may

also have incentive or signalling effects. Moreover, Bhagat, Brickley

and Lease (1984) find two-day positive abnormal returns associated with

the announcement of stock purchase plans of 3.4 percent. Since Miller

and Scholes (1982) argue that stock purchase plans are tax neutral,

this unexplained stock price increase may be attributed to productivity

increases either associated with the incentive effect of these plans,

or being signalled by the managers.

In summary, the empirical evidence linking managerial decisions

("accounting" or "real") to various incentive compensation arrangements

is inconsistent. This has been attributed in part to compensation plan

misspecification [Ball and Foster (1982)]. Given the complexity of

compensation arrangements, the use of a dichotomous measurement of

incentive compensation, focusing solely on the existence of an

executive profit-sharing plan, is clearly simplistic. Furthermore,

many of these studies ignore or discount completely the multi-year

impact of managerial decisions on earnings and incentive compensation.

If the extended effects of these decisions were considered, it is

possible that the results of the accounting procedure selection studies

in particular might be strengthened.

Solely on the basis of the evidence available, one cannot yet

infer that incentive compensation arrangements affect managerial

decisions. Further refinements in empirical measurement and testing

must be undertaken before one can start questioning the basic theory.

That is the aim of this research effort. In the following section the

incentive effects of long-term compensation plans are analyzed from the

perspective of agency theory.

2.2.2 Long-term compensation contracts and the stockholder-manager
conflict of interest

Viewed from the agency theory perspective, there are three

potential sources of conflict between stockholders and managers: 1)

differences in time horizons, 2) differences in risk attitudes, and 3)

the managers' disutility for effort. The introduction of long-term

incentive compensation contracts may help to control each of these

sources of disagreement, thereby lowering agency costs. Certainly the

evidence relating a positive stock price reaction to the initiation of

these compensation schemes is consistent with such an effect [Larcker

(1983), Brickley, Bhagat, and Lease (1985), Bhagat, Brickley, and Lease


The differential decision horizon problem arises because a

manager's decisions may be evaluated on the basis of short-term

measurements of performance and his claim on the firm is generally

limited to his tenure with the firm. Stockholder claims on the other

hand are tradeable claims on the entire future stream of residual cash

flows. To maximize the market value of the stockholders' claims, a

manager must consider the effect of his decision alternatives on the

firm's cash flows for the remaining life of the firm. A self-serving

manager, on the other hand, will consider only those cash flows or

other measures which are expected to positively affect his own utility.

Market-based compensation arrangements including restricted stock

plans may be particularly helpful in controlling this problem. If

share market value reflects the present value of the entire future

stream of expected cash flows, any action taken which disregards

long-run opportunity costs may adversely affect the market price of the

firm's stock. If a manager's compensation is made dependent upon share

price, his wealth may also be adversely affected by his myopic deci-


With regard to performance plans, an additional aspect of

compensation warrants attention. As noted previously, most firms

sponsor some form of short-term bonus plan which ties the manager's

compensation in part to yearly accounting measures. Consistent with a

shortened decision-making horizon, Waegelein's (1985) findings suggest

that the adoption of such plans is associated with reduced expenditures

for research and development. The introduction of accounting-based

long-term performance plans may help mitigate this dysfunctional effect

by delaying a large component of compensation until better information

on the manager's decision quality is available. In this way the

manager's reward can be tied more closely to his actual performance.

Delaying compensation provides a means of "bonding" managers to their

firms [Eaton and Rosen (1983)].

The manager's horizon can also be extended through the deferred

payment and forfeiture provisions of all the long-term compensation

plans under discussion. Both performance and restricted stock plans

typically defer payment for three to five years. Consequently, if a

manager is seriously considering leaving his current position or retir-

ing, the payment deferral and forfeiture clauses make this alternative

more costly, and therefore less appealing. The manager's expected

tenure with the firm may be extended upon the introduction of a

long-term compensation plan, thereby extending his decision horizon and

reducing agency costs.

The second major source of potential conflict between stockholders

and managers is differences in the risk attitudes of the two parties.

It is generally assumed that stockholders are essentially risk neutral

as a result of their ability to diversify their investment portfolio.

A manager, however, whose human capital investment in the firm is

generally non-marketable, is unable to diversify this risk away.

Hence, the manager has the incentive to invest in projects with lower

levels of risk, and correspondingly lower returns, than are acceptable

to the risk-neutral stockholders. Because long-run expected cash flows

are oftentimes less certain than those in the short-run, a difference

in risk attitudes may lead the manager to ignore or underweigh the

projected long-range effects of various alternative projects.

It has been suggested that the adoption of plans with rewards

dependent upon market performance may encourage managers to be less

risk-averse in their decision-making, thereby helping to alleviate this

second source of conflict. Smith and Watts (1982) argue that the

expected payoff for such plans is an increasing function of the

variance of the firm's stock. If so, a manager whose compensation

package includes a market-based plan can increase its value by

selecting risky actions which increase the variance of the firm's


However, it is not clear that this argument is applicable to the

manager's valuation of performance share and restricted stock awards.

Not only is the transfer of these awards constrained both contractually

and institutionally, but an element of "downside risk" exists. Both

performance share and restricted stock awards have value, regardless of

market performance. There may be no incentive for a risk-averse

manager to take risky action, when the potential for associated losses


Another component of "downside risk" may be present for managers

compensated via performance plans because their awards depend upon the

firm's long-term earnings performance. David Kraus, a director of

McKinsey & Co., observed that the earnings targets of performance plans

are often set below the inflation rate, making a fail safe system of

lush rewards [Meadows (1981)]. Once again there may be no incentive

for the risk-averse manager to take added risks; inordinately risky

behavior could lead to the reduction or loss of an otherwise guaranteed

reward. Therefore, it is not clear that the introduction of either

performance or restricted stock plans will mitigate the potential

shareholder/manager conflict arising from differences in risk exposure

attitudes. Any action taken which increases a firms's earnings and

stock price variance will, ceteris paribus, also impose additional

"downside risk" on the manager. Their introduction may in some cases

actually discourage executive risk-taking. On the other hand, to the

extent that a manager considers long-run expected cash flows riskier

than those of the short-run, any extension of a manager's

decision-making focus will also encourage managers to be less


While the initiation of long-term plans may help extend managerial

decision horizons, thereby encouraging a more thorough consideration of

long-run projected cash flows, it is likely that additional managerial

effort (the third source of conflict) will also be necessary if

investment/production decisions are to be altered, particularly if

capital spending levels need to be increased. Whenever a manager's

compensation derives solely from a fixed salary, he has little

incentive to exert decision-making effort above that required to insure

that the firm's cash flows are sufficient to cover his fixed claims.

In such a setting the introduction of any incentive arrangement may

encourage the manager to exert additional effort in order to attain the

performance goal and the related reward. However, most firms already

have bonus plans in place. Therefore, upon adoption of a long-term

incentive plan, the manager's total compensation is a function of both

a short-term accounting-based measure of performance, and a market-

based and/or long-term accounting-based measure of performance.

Holmstrom (1979) shows that if no one measure alone is sufficient

to assess managerial effort, the use of additional measures will enable

the owner to more precisely determine effort level. Lambert (1983), in

extending this work to a multi-period setting, demonstrates that the

dependence of second-period compensation on a first-period measure of

performance allows the stockholders to diversify away some of the

uncertainty surrounding the manager's actions. As managerial effort is

better monitored, a greater level of effort can be encouraged through

appropriate contractual arrangements. Therefore, greater effort levels

can be prompted through the introduction of any of these long-term

incentive compensation schemes.

In summary, while the empirical literature provides little basis

for inferring that performance-contingent contracts lead to changes in

managerial decisions, the agency theory analysis suggests that the

adoption of long-term incentive compensation plans may help reduce the

divergence of interests between stockholders and managers by

lengthening the manager's decision-making focus. Furthermore, the


installation of these plans can encourage effort-averse managers to

provide the additional effort necessary to pursue additional endeavors.

However, it is not clear that the introduction of these plans (as

currently structured) will alter the effect of risk on the managers'



This chapter develops the hypotheses which are empirically

examined in this study. To provide some insight into the effectiveness

of establishing long-term incentive compensation contracts to control

the stockholder/manager conflict of interest, the firms' long-term

capital spending decisions are investigated. Because both capital

investment [Sunder (1980)] and R & D projects negatively impact early

period reported profits, a manager with a short-term decision focus may

reject proposals which shareholders would find acceptable. Therefore,

capital spending decisions provide a promising arena for studying the

incentive effects of long-term performance-contingent contracts.

3.1 The Adoption Hypotheses (Hypotheses 1 and 2)

To analyze the relationship between the introduction of a

performance-contingent compensation contract and changes in a manager's

capital spending decisions, several assumptions concerning the decision

setting are made. First, the manager is assumed to be a rational

economic person. Second, both the internal and external managerial

labor markets are assumed to be characterized by less than full "ex

post" settling up. Third, the manager's compensation stream is assumed

to depend in part on the firm's annual earnings. This dependence may

arise from an existing contractual arrangement (bonus), or from the

labor market's implicit use of annual earnings for managerial

performance evaluation purposes. As a result, the manager may

overemphasize the short-term earnings effects of his alternatives when

making decisions.

Given these assumptions, the introduction of a long-term incentive

compensation plan may extend the manager's decision horizon. This is

accomplished through (i) the payment deferral and forfeiture provisions

of these plans, (ii) the direct effect of changes in share price on the

value of the award (restricted stock and performance share plans), and

(iii) for accounting measurement based arrangements (performance unit

and performance share plans), the use of an award period which extends

significantly beyond the annual basis often used by bonus plans and the

managerial labor market.2

In order to predict the effect of long-term compensation plan

adoption on capital spending levels, three additional factors must be

considered. First, all projects under consideration by the manager

must have relatively low or negative earnings and cash flows in early

years and relatively large and positive earnings and cash flows in

later years. Then, upon implementing a long-term incentive

arrangement, the appeal of those projects with later payoffs will be

enhanced for managers whose decision-making horizon has been

lengthened. Moreover, the exclusion of projects with negative earnings

and cash flows in later years implies that any project deemed

acceptable by the manager prior to the adoption of a long-term

compensation scheme remains acceptable. Consistent with this

characterization of the available project set, Sunder (1980) finds that

capital expenditures significantly reduce earnings in the year

following outlay for his sample of 755 firms.

Secondly, development and selection of additional projects

undoubtedly require the exertion of additional managerial effort. As

indicated previously, when stockholders are unable to observe

managerial effort perfectly, the use of multiple period performance

measures [Lambert (1983)], or the use of multiple measures related to a

single period [Holmstrom (1979)], allows for better monitoring of

managerial effort by stockholders. The introduction of either

market-based or accounting-based long-term compensation plans makes

managerial compensation dependent upon more "informative" measures of

performance. Therefore, their adoption can encourage the added effort

necessary for increased capital spending that is in the shareholders'

long-term interests.

Finally, the option-like characteristics of these plans must be

considered. If the adoption of performance or restricted stock plans

makes riskier projects more acceptable to the manager [Smith and Watts

(1982)], it follows then that low-risk projects may become less desir-

able. However, because an element of "downside risk" is likely to be

present in these plans, it is not clear that their introduction will

alter the effect of risk on the manager's decisions. Therefore,

because its existence is indefinite and direction unclear, the

hypothesized option-like characteristics of these plans are assumed to

have a negligible effect on the level of capital spending.

Under the circumstances described, the introduction of the

long-term incentive compensation plans under consideration may result

in an increase in the level of capital spending, relative to the level

expected prior to knowledge of the adoption. This follows from the

ability of the plan to extend the managers' decision-making horizons

and induce the managers to provide the additional effort needed.

Hence, the following research hypothesis is advanced and empirically


Hypothesis 1

The introduction of long-term compensation plans leads to
higher than expected capital spending levels.

The effect of adopting long-term compensation plans on capital

spending levels is studied in the aggregate, as well as for the

following plan categories:

1. performance plans (accounting-based plans), and

2. restricted stock plans (market-based).

While the analysis in this section suggests that the direction of the

incentive effects of these plan classifications will be similar, it is

nonetheless also interesting to consider whether the magnitude of the

incentive effects may differ between market-based and accounting

earnings-based plans. At this point in its development, the theory of

compensation is unable to specify when or why compensation contracts

are changed or what contractual arrangements should be introduced in

specific decision settings. As Smith and Watts (1983) argue, firms

often face a variety of conflicts of interest between managers and the

firms' other claimholders. In order to help control these conflicts

and minimize the resulting agency costs, situation-specific

compensation contracts are designed and introduced. Given the

diversity of the settings, it follows that the appropriate

(equilibrium) compensation package can vary between firms at any point

in time, and change over time for a given firm.

While the magnitude of the incentive effects may be dominated by

the magnitude of the long-term award relative to total compensation,

another factor which may influence the capital spending decisions of

managers may be the extent to which decision horizons are lengthened by

these plans. Market-based restricted stock plans make managerial

compensation dependent upon share price, which may reflect the entire

future stream of expected cash flows. As such, market-based plans may

extend the decision horizons of managers indefinitely. Performance

plans, on the other hand, make managerial compensation dependent upon

firm earnings over the award period, which is generally three to six

years. This suggests that the decision horizons of managers

compensated by performance plans may be limited by the length of the

award period. Thus, managers may make capital spending decisions which

maximize earnings over the award period, in conflict with the longer

decision horizons of stockholders. However, this divergence of

interests may be mitigated by the staging of performance plan

offerings. Oftentimes under performance plans, executives are offered

additional units or shares each year, with a new award period and new

earnings targets. If a performance plan compensated manager feels that

such offerings will continue, his decision horizon may be extended


Additionally, the impact of controllability on managerial

motivation must be considered. Executives may be more motivated to

achieve superior performance, and therefore make capital spending

decisions which are more consistent with stockholder interests, when

they feel that their actions directly affect their incentive award.

Because executives have greater control over company profit performance

than over stock market performance, accounting earnings-based

performance plans may provide the greater incentive to behave in the

stockholders' interests with regard to capital spending decisions.

The effect of compensation plan type on capital spending levels is

therefore ambiguous. While market-based plans may encourage longer

managerial decision horizons, accounting-based plans may induce greater

effort given the added controllability of accounting earnings. As

a result, there is no reason to expect a priori that the introduction

of market-based and accounting-based plans will differ in their effect

on capital spending levels. The following hypothesis is empirically


Hypothesis 2

The capital spending levels following from the
introduction of market-based compensation plans will
differ from those following from the introduction of
accounting-based plans.

3.2 Past Performance Interaction Hypothesis (Hypothesis 3)

The adoption hypothesis hinges on the contention that the managers

of firms which introduce long-term compensation plans have been, or

would otherwise be, making self-serving capital spending decisions.

Because these managers' compensation streams depend in part on the

firms' annual earnings, they may overemphasize the short-term negative

earnings impact of capital spending projects. As a result these

managers may reject "late payoff" proposals, which the stockholders

would find acceptable. In such settings stockholders, disgruntled with

the managers' performance, may adopt long-term compensation plans to

encourage the managers to act in a manner which is more consistent with

stockholder welfare.

While the direction of the hypothesized adoption effect is clear,

the magnitude is dependent upon (i) the degree of existing divergence

between manager and shareholder preferences with regard to capital

spending decisions and (ii) the magnitude of the adopted incentive plan

relative to other components of total compensation. The effect of plan

magnitude is considered in conjunction with the adoption hypothesis

(Hypothesis 1). With regard to managerial self-interest, it is clear

that the greater the agency costs associated with the divergence

between expected managerial capital spending actions and those actions

which maximize shareholder welfare, the greater the room available for

improved decision making by the manager. Therefore, if the long-term

compensation plans adopted by stockholders promise incentive awards

which are sufficient to bring about more congruent capital spending

decisions by executives, then the magnitude of the adoption effect will

be positively associated with the existing degree of managerial

digression from the interests of stockholders regarding these

decisions. Alternatively, the adoption of a long-term compensation

plan will have little impact on capital spending levels if the manager

is already making decisions which are consistent with owner


The examination of the interaction between the adoption of

long-term compensation contracts and the level of agency costs is

particularly interesting, as it may provide some insight into why

compensation schemes are established. The theory of compensation in

its current state of development is unable to explain (i) the

characteristics of the compensation arrangements observed and (ii) the

managerial decisions which might be affected by these contracts [Raviv

(1985)]. As Raviv (1985, p. 242) notes,

[tJhere are at least three different theories capable of
explaining the observed contracts...I will argue that each
one of these hypotheses is consistent with the observed
contracts and almost all of the papers in this conference
[Conference on Management Compensation and the Managerial
Labor Market in honor of William H. Meckling, University of
Rochester, April 27-28, 1984] can be interpreted in 'any
one' of the the three frameworks. Thus, we have a serious
problem of identifying the causality behind the contracts

Raviv refers to these theories as the screening-sorting

hypothesis, the signalling hypothesis, and the motivation and

incentives hypothesis. The screening-sorting hypothesis is based on

the inability of claimholders to observe managerial ability.

Performance-contingent contracts are therefore introduced to

distinguish among managers of different ability levels, by providing

incentives to the less capable to abstain from applying to or remaining

with the firm. This hypothesis does not resort to moral hazard

problems in explaining the existence of incentive contracts.

The signalling hypothesis [Ross (1977, 1979)] assumes not only

that managers act in the best interests of stockholders (no moral

hazard problem), but also that their ability is known. Under this

view, incentive contracts provide a means for managers to signal the

existence of good news to the market. The manager's act of accepting

the added risk associated with the introduction of a performance-

contingent contract provides the basis for the market's belief in the


The motivation and incentives hypothesis assumes that neither the

discipline of the managerial labor market nor the threat of takeovers

is sufficient to control the stockholder/manager conflict of interests.

According to this hypothesis, performance-contingent contracts are

introduced as another mechanism for controlling this conflict (the

moral hazard problem), thereby reducing agency costs. (The adoption

hypotheses, Hypotheses 1 and 2, were developed on the basis of the

motivation and incentives proposition.) One testable implication of

this proposition is that the degree of change in managerial behavior

induced by the introduction of incentive contracts is dependent upon

the magnitude of pre-adoption agency costs. Neither the screening-

sorting nor the signalling hypothesis bases the establishment of

incentive contracts on the degree of prevailing agency costs.

Therefore, a finding of positive association between the magnitude of

the adoption effect and the degree of agency costs existing prior to

plan adoption can be interpreted as supporting the motivation and

incentives hypothesis, and as being inconsistent with either of the two

alternative compensation hypotheses operating exclusively. However,

such a finding does not negate the possible relevance of the screening-

sorting and signalling hypotheses. These alternative hypotheses may

also apply in the environments tested.

One indicator of the degree of stockholder-manager divergence of

interests may be earnings performance in the recent prior period.

Because sustained earnings depend upon the quality of a manager's

decisions, poor earnings performance prior to adoption may evidence

poor capital spending decisions. This linkage, however, requires an

intermediate assumption. The premise of the adoption effect is that

poor capital spending decisions may arise from a manager's inordinate

concern with enhancing short-run earnings. Therefore, in order for

poor capital spending decisions to be positively associated with poor

earnings, the tenure of the manager's short-run orientation must be

assumed to be of sufficient duration to result in the inevitable

earnings decline. To the extent that poor past earnings performance

connotes poor capital spending decisions, the magnitude of the adoption

effect will be greater for poor past earnings performance firms than

for good past earnings performance firms. Hence, the following

research hypothesis is posited for empirical examination.

Hypothesis 3

Following the introduction of a long-term compensation
plan, the unexpected capital spending levels associated
with poor prior earnings performance firms will be
higher than those associated with good prior earnings
performance firms.

No hypothesis is entertained with regard to a direct (main) effect

of prior earnings performance on the dependent variable (unexpected

capital spending level). While the various capital spending

expectation models may not explicitly consider a firm's past earnings

performance, implicitly each does to some degree. Therefore, no

association between unexpected capital spending levels and the prior

period earnings performance of firms is anticipated. It is only to the

extent that this variable is under or overweighed by the expectation

models that one would envision such a relationship.

3.3 Ownership Control Interaction Hypothesis (Hypothesis 4)

While poor past performance may signal that a manager is making

capital spending decisions which are inconsistent with shareholder

interests, it is also interesting to consider the characteristics of an

organization's structure which may contribute to this divergence. As

previously noted, managers can enhance earnings by engaging in a

variety of nonproductive and sometimes dysfunctional activities

(including managing accruals, changing accounting methods, and reducing

discretionary expenditures). Therefore, organizational characteristics

which may lead to this divergence of interests may actually be a better

surrogate for excessive agency costs than past earnings performance.

Various authors have suggested that the separation of ownership

from control within an organization may lead to a conflict of interest

between these parties [Berle and Means (1932), Williamson (1964), and

Monsen and Downs (1965)]. According to Smith (1976, p. 709) "control

refers to the power to direct the affairs of the corporation or to

determine the broad policies guiding the corporation." The board of

directors is generally charged with providing such guidance.

Therefore, control is generally considered to be separated from

management when, due to the wide dispersion of a firm's stock among

many small holdings and the ability of management to solicit proxies

for its slate of candidates, management is able to dominate the

selection of board members [Larner (1970), p. 3]. Upon the separation

of ownership from the control function, the ability of the principal

(owner) to monitor (observe) the actions of the agent ("controlling"

manager) is abridged. In such settings the additional discretionary

power which ensues to the agent may contribute to acts of moral hazard.

The manager may act in his own interests, rather than the interests of

the owners.

The empirical evidence concerning the opportunistic use of

managerial discretion is mixed, however. Smith (1976), Salamon and

Smith (1979), and Dhaliwal, Salamon and Smith (1982) present evidence

indicating that the accounting choices made by a firm are associated

with the firm's ownership control status. Smith (1976) finds that

manager-controlled firms (those in which stockholders may lack adequate

information or power to exact profit-maximizing behavior from

management) make more smoothing changes than owner-controlled firms

(those in which the owners are able to control the selection of the

board of directors). Salamon and Smith (1979) find that manager-

controlled firms tend to make accounting changes that increase earnings

in years when their stock price decreases. Moreover, Dhaliwal, Salamon

and Smith (1982) find that there is a significant difference in the

depreciation methods adopted by manager-controlled and owner-controlled

firms for financial reporting purposes, with manager-controlled firms

using the income-enhancing straight-line method more extensively.

Each of the above studies finds that managers with discretionary

power derived from the separation of ownership and control make

accounting decisions which enhance artificially the reported earnings

performance of their firms. Such income-bolstering decisions may be

driven by the need to offset the impact of non-optimal resource

allocation decisions [Hindley (1970) and Williamson (1967)], thereby

maximizing their own lifetime incomes [Monsen and Downs (1965)].

However, when the resource allocation decisions of managers are

empirically examined, the results are less consistent.

Grabowski and Mueller (1972) consider the association between R&D

levels and the percentage of insiders on the board of directors, their

measure of the degree of separation of ownership and control. From the

premises that (i) managerial compensation is more closely tied to firm

size than profitability, and (ii) R & D expenditures positively impact

long-run growth in firm size, they argue that R & D expenditures will

be greater for manager-controlled (those firms with the higher

percentage of insiders on the board) than owner-controlled firms.

While their findings support this contention, they fail to consider

differences in compensation plan characteristics, assuming instead that

firm size determines compensation levels. Moreover, their measure of

ownership-control separation, the percentage of insiders on the board,

may not provide an accurate portrayal of this condition. In those firms

where inside board members comprise a large percentage of the board and

control a large portion of the voting stock, it is not clear that

ownership is separated from control. Hence, interpreting their results

in these terms may be misleading.

In a related test, Dukes, Dyckman, and Elliott (1980) examine the

relationship between R & D expenditures and a firm's ownership control

status in connection with their investigation of the effects of issuing

the FASB Statement No. 2. Contrary to Grabowski and Mueller (1972),

they classify firms as owner-controlled if one party controls ten

percent or more of the voting stock; otherwise, firms are classified as

manager-controlled. No arguments are provided concerning the direction

of the hypothesized owner-manager effect on R & D expenditures, nor are

they able to discern any significant relationship.

Alternatively, capital investment decisions are the focus of

Sunder's (1980) study. From the evidence of Smith (1976) and Salamon

and Smith (1979), he argues that managers in manager-controlled firms

may be expected to select investment projects for which the negative

short-run effect of investment on earnings is smaller than for managers

in owner-controlled firms. He finds however, that the degree of

ownership control does not seem to affect the earnings profile of the

projects accepted. As did Dukes, Dyckman, and Elliott (1980), Sunder

(1980) classifies firms as manager-controlled if no single party owns

more than ten percent of the voting stock. However, he does point out

that this definition of ownership control may be too strict, as a five

percent ownership in a large publicly held corporation with widely held

stock may give an owner substantial control of the affairs of the firm

[Sunder (1980), p. 562].

In summary the empirical evidence suggests that differences may

exist in the managerial decisions made by owner-controlled and

manager-controlled firms. However, more research is needed in order to

ascertain whether the managers of manager-controlled firms use their

discretionary power to make self-interested decisions which conflict

with the welfare of stockholders. This conclusion is particularly

applicable when specific resource allocation decisions of managers are


If the managers of manager-controlled firms use their

discretionary power to make self-interested decisions, then their

capital spending levels may be affected as follows. It may be in the

best interests of such managers to reject some projects which impact

early future period earnings negatively. By doing so they can enhance

the short-run earnings performance of their firms, and potentially

their own lifetime compensation. Therefore, given the additional

discretionary power available, the capital spending levels of

manager-controlled firms may be lower than those of owner-controlled

firms. However, with regard to this contention, no hypothesis is

introduced for testing. It is only to the extent that the ownership

control status of firms is disregarded by the expectation models that

one would anticipate the unexpected capital spending levels of owner-

controlled firms to be greater than those of manager-controlled firms.

Both the Value Line and firm-specific econometric models are assumed to

implicitly consider ownership status.

Of greater interest to this study is the interaction between a

firm's ownership control status and the hypothesized adoption effect.

As suggested, if considerable discretionary power is in fact available

to the managers of manager-controlled firms, they may reject profitable

"late payoff" investment projects in order to enhance short-run

earnings and their self-interest. Conversely, relatively little

discretionary power (that goes unnoticed by the controlling owners) is

available to the managers of owner-controlled firms. Hence, in these

firms, the managers' capital spending decisions and the associated

expenditure levels may be more aligned with stockholder preferences.

Under the latter scenario, the introduction of long-term compensation

arrangements may have little effect. It then appears that the

magnitude of the adoption effect may be more pronounced in firms that

are controlled by managers, given the greater divergence between

stockholder interests and managerial investment decisions.

However, one additional associated factor which may

counterbalance this effect in part should be considered. Executive

compensation plans must be authorized by the boards of directors and

must receive a majority vote of the stockholders in order to be

ratified. However, because stockholder rejections are extremely rare,

the effectiveness of control lies largely in the hands of the

directors. In manager-controlled firms the selection of these

directors may be controlled by the manager, thereby increasing the

probability of a compliant board. This suggests that these managers

may be able to design and implement with little opposition compensation

schemes which may be effective only in shifting wealth from

stockholders to managers in the form of additional incentive

compensation. This may be accomplished by incorporating easily

reachable performance goals. Moreover, the outside owners of

owner-controlled firms, aware of the managers' self-interests, may

successfully urge the adoption of more effective (stringent) long-term

incentive compensation contracts to further encourage managerial

actions which are compatible with stockholder interests. Under this

line of reasoning, the hypothesized adoption effect may be greater for

owner-controlled than for manager-controlled firms.

Because the two arguments above are in opposition to each other

and their relative strengths are unknown, the influence of the

ownership control status variable on the hypothesized adoption effect

is ambiguous. However, if the long-term compensation arrangements

adopted by owner and manager-controlled firms are comparable in terms

of reasonableness of goals and size of awards (assuming the existence

of sufficient external discipline in the form of comparability with

relevant competitors), then the adoption effect may be more pronounced

in settings where managers have the greatest discretionary power in

decision making, manager-controlled firms. Hence, the following

research hypothesis is posited.

Hypothesis 4

Following the introduction of a long-term compensation
plan, the unexpected capital spending levels associated
with manager-controlled firms will be higher than those
associated with owner-controlled firms.

3.4 Notes

1. The existence of other long-term compensation plans may depress or
eliminate a manager's incentive to overemphasize annual earnings.
Stock options, the most popular long-term incentive plan with 59
percent of the Conference Board sample of 1053 firms having such
plans in 1984 [Fox (1985)], may adequately control the
differential decision horizon problem in some firms. However, for
such firms there is little reason for stockholders to ratify an
additional long-term compensation scheme.

2. This assumes that the managerial incentive created by the
implementation of a long-term incentive plan is large enough to
overcome any pre-existing short-term incentive associated with
bonuses and labor market evaluation based on annual earnings. For
the experimental firms in this study which provided such data,
awards associated with the adopted long-term plans averaged 23
percent of total salary plus bonus in the two year period
commencing with the adoption year and ranged from 0 percent to 207

3. This contention relies upon the existence of inefficient markets
for corporate control and managers. As previously discussed in
this chapter, the discipline provided by the threat of takeover is
hindered by substantial transactions costs [Smiley (1976)]. With


regard to the market for managers, little empirical evidence is

4. See Hunt (1986) for a more complete review of the effect of
separating ownership and control.


This chapter describes the procedures used in selecting the sample

and delineates the research design of this study. Moreover, the

dependent and independent variables are operationally defined.

Finally, the descriptive statistics for the independent variables are

presented. Descriptive statistics for the alternative dependent

variable measures are provided in Chapter V.

4.1 Sample Selection

A listing of firms which adopted performance and restricted stock

plans was obtained through a review of the 1980 and 1981 proxy

statements available in the University of Florida library. To augment

this listing of 162 adopting firms, Brickley, Bhagat, and Lease (1985)

generously provided a list of approximately 350 firms which had voted

on incentive plans during the period from 1978 through 1982.1 The

proxy statements related to these adoptions were reviewed to confirm

that the plan was introduced to the stockholders for ratification, and

to ascertain the type of plan. Only those firms which (i) could be

confirmed in this manner, (ii) were plan types of concern to this

study, and (iii) were included on COMPUSTAT'S list of firms, were

included in the sample group. This procedure yielded 54 additional

firms, making a total sample of 216 adopting firms. While these

procedures merely identify plans presented for ratification, it is rare

that such plans do not receive the simple majority vote needed to pass.

Each firm was categorized as adopting restricted stock plans (including

phantom stock), performance plans, or mixed plans which contained both

restricted stock and performance plan components.

A control sample of 139 non-adopting firms was generated through

the following procedures. From COMPUSTAT's listing of firms, a group

was randomly selected. The proxies of these firms were reviewed where

available to insure that no restricted stock or performance plan was

adopted during the seven year test period (three years prior to the

"adoption" year, the year of "adoption", and three years subsequent to

the "adoption" year). "Adoption" years, 1978 through 1982, were

randomly assigned to these non-adopting firms in the same proportion as

in the sample of adopting firms. This procedure yielded 116 control

firms. Appended to these was a group of 23 firms selected on the basis

of industry affiliation to insure proportional representation in the

experimental and control samples. To be included in the control sample

a firm needed a proxy statement available for the assigned adoption

year and sufficient others to provide reasonable assurance that no

restricted stock or performance plan adoption occurred during the test


4.2 Research Design

The hypotheses developed in the previous section are tested using

a multivariate model of the following form. As explained previously no

hypotheses (of significance or direction) are introduced for the main

effects of prior earnings performance and ownership control status due

to the likelihood that the expectation models implicitly consider these

variables. The multivariate tests employ ordinary least squares

estimation of the following linear model:

Actual Expected bo + bl ADOPT + b2PERF + b CONTROL
Capital Capital
Spending Spending + b4 ADOPT*PERF + b5 ADOPT*CONTROL
Level Level

where ADOPT = 1 if long-term compensation plan adopted
0 if no long-term compensation plan adopted

PERF = firm earnings performance
prior to adoption (continuous variable)

CONTROL = 1 if owner-controlled firm
0 if manager-controlled firm.

The hypothesized signs of the coefficients are as follows: ADOPT is

expected to be positive, while both the ADOPT*PERF and ADOPT*CONTROL

interactions are expected to be negative. No a priori directional

expectations are entertained for the "main effect" coefficients of PERF


4.3 Dependent Variables

4.3.1 Actual capital spending level

The capital spending level of the sample firms is operation-

alized as the following ratio(s):

Capital = Capital investment expenditures or
Spending Sales (or Property, Plant and Equipment)

S (Capital investment + R&D) expenditures
Sales (or Property, Plant and Equipment)

Capital spending may include outlays for R & D as well as capital

investment. There is no difference conceptually in managerial

decisions concerning these two categories. Each impacts early future

period earnings negatively, an effect which may result in project

rejection by a self-interested manager. Moreover, firm-specific

budgeting, marketing, technological, and accounting differences

undoubtedly exist. Therefore, in order to increase the likelihood that

unexpected changes in capital spending levels are captured, both R & D

and investment expenditures together are used as an alternative

comprehensive measure for testing purposes.

To normalize capital spending across firms and reduce the impact

of inflation across years, this measure is divided alternatively by

sales and property, plant and equipment (PP & E). The association of

capital spending and sales is based on the relationship between the

capital turnover desired and managements' projections of demand for

their product. To the extent that actual post-adoption sales reflect

these projections, basing capital spending on sales provides a relative

measure across firms. The use of PP&E as a deflator is supported by

the evidence that replacement investment may predominate total

investment expenditures, at least at the aggregate level [Kuznets

(1961)]. To the extent that replacement investment is proportional to

capital stock and represents a major segment of capital spending, one

would expect that PP&E-deflated investment expenditures may also

provide a reasonable cross-sectional measure. Capital investment,

R & D, PP & E and sales data are obtained from COMPUSTAT's Industrial

data base.

In order to minimize the potential for misspecification associated

with annualized expenditure measures, the capital spending levels of

firms are examined not only for the year of adoption, but also for the

two-year periods commencing with the year of adoption, and with the

year following adoption. Mayer (1960) provides survey evidence

indicating that the weighted average lag time from drawing plans to the

completion of construction of new industrial plants and additions is

twenty-one months. Therefore, in order to better determine whether

long-term compensation plan adoption is associated with unexpectedly

high capital spending levels, it is desirable to allow for this lag by

considering the longer accumulation periods.

4.3.2 Expectation models of capital spending

The following models are used to generate firm-specific expected

levels of capital spending:

1. Value Line prediction model, and

2. Almon polynomial distributed lag investment model.2

A description of each of these expectation models follows. Value Line prediction model

The Value Line Investment Survey provides firm-specific forecasts

of capital investment expenditures for the current year, the upcoming

year, and three to five years hence. In predicting post adoption

capital spending levels, the Value Line forecast made four quarters

prior to the adoption date is utilized. This early forecast is used to

increase assurances that it was made without knowledge of the

prospective long-term compensation plan adoption. While Value Line

analysts were unwilling to disclose the variables considered in

generating their predictions, they noted that managerial estimates

oftentimes are used. It follows then that this expectation model must

control to some extent for possible macroeconomic, industry, and

firm-specific factors influencing investment expenditures. Almon polynomial distributed lag investment model

As an alternative estimate of the major factors (other than plan

adoption) that determine capital spending, an Almon polynomial

distributed lag model is used to predict firm-specific investment

levels. This model, based on the flexible accelerator model of

investment originated by Chenery (1952) and Koyck (1954), focuses on

the time structure of the investment process. Therefore, the firm is

taken to have a desired level of capital, determined by long-run

considerations. General agreement exists among the alternative

theories of investment behavior on the validity of the flexible

accelerator mechanism for translating changes in desired capital into

actual investment expenditures [Jorgenson and Siebert (1968a)]. Under

this mechanism capital is adjusted each period toward its desired level

by a proportion of the difference between desired and actual capital.

As noted by Jorgenson and Siebert (1968a), the flexible

accelerator mechanism can be transformed into a more complete model of

investment behavior by specifying the desired capital level.3 It is on

this dimension that the alternative theories diverge. Oftentimes the

level of desired capital is assumed to be proportional to the level of

output. This use of output is consistent with the capital utilization

theory which argues that high levels of investment expenditure are

associated with high ratios of output to capital, and low levels of

investment with low ratios of output to capital. While output (a

demand factor) has performed well in predicting investment expenditure

levels [Jorgenson (1971)], others citing Tinbergen (1939) argue that

the rate of investment may be constrained by the supply of funds. This

follows from Tinbergen's theory of investment which posits that desired

capital is proportional to expected profits, the cost of capital being

clearly related to profit expectations. Given the divergence in

theories regarding the determinants of desired capital, both variables

are included in the capital spending models used in this study.

The distributed lag investment models used for predicting post

adoption capital spending levels are of the following form,

Iit = bo + bi,t- + b2 i,t-2 + b3it-3 + b4i,t + b5Qi,t-l

+ b6 i,t-2 b7Ki,t + b Ki,t- + b9Ki,t-2 + eit
where the subscripts refer to firm "i" and year "t" respectively, and

the variables are measured as

it = Capital spending level,

Qit = Cost of goods sold, and

Kit = Interest expense/long-term debt.

The capital spending, cost of goods sold, interest expense and

long-term debt data were obtained from COMPUSTAT's Industrial data base

for the period 1962 through 1984. In a manner consistent with this

study's operational definitions of capital spending levels, four

distinct sets of distributed lag model predictions are obtained. The

predictions of the two sets which include R & D as part of capital

spending are based solely on the lagged capital spending levels. This

change was prompted by a drastic reduction in the number of firms for

which time-series predictions could be generated due to the limited

availability of R & D data from COMPUSTAT, particularly in the early

years. By deleting the output (Q) and cost of capital (K) variables

from the investment expenditure model, predictions are obtained for an

adequate number of firms to run the tests.

The measurement of output (Q) as cost of goods sold is once again

affected by data availability considerations. Output (Q) may be more

appropriately defined as cost of goods sold plus any change in finished

goods inventory levels. However, due to the limited availability of

data on finished goods inventories prior to the early 1970's, it was

decided to use cost of goods sold in order to maintain a reasonable

sample size. Moreover, output is included in this model because of its

theoretical relationship with desired capital. It may be that desired

capital is more strongly associated with cost of goods sold alone, as

changes in finished goods inventories may include management judgment

errors regarding expected sales.

With regard to the cost of capital measure, "K", Elliott's (1980)

empirical test of its influence on aggregate investment finds the

effect to be significant, regardless of the cost of capital measure

used. Furthermore, he finds that a debt interest measure of capital

cost provides results which are quite close to the superior weighted

average-permanent income measure. The cost of capital measure selected

for this study is based on these empirical findings and the ready

availability of the necessary data.5

4.4 Independent Variables

The operational definitions of the independent variables are

provided below. Additionally, the descriptive statistics (Table 5.1)

of these variables and their correlation matrix (Table 5.2) are

discussed. The descriptive statistics for the dependent variable

measures are reported in Chapter VI, along with the results.

4.4.1 Adoption of long-term compensation plan (ADOPT)

The classification of firms for the dichotomous adoption variable

(ADOPT) is based on whether they adopted one of the long-term

compensation plans being considered. Consequently, firms which were

found to have adopted restricted stock or performance plans were

assigned a value of "I", while the control (non-adopting) sample firms

were assigned a value of "O". The adopting firms were further

classified according to type of plan adopted; performance plan,

restricted stock plan, or some combination of performance and

restricted stock plans. Additionally, the adopting firms were

partitioned on the basis of whether the adoption was an initial

introduction of the long-term plan, or whether the change was an

extention of a plan which had been previously initiated. A review of

the proxy statements for the three years preceding the adoption year

helped make this determination. Of the 216 adoptions, 149 represent an

initial introduction of a long-term compensation plan.

In order to more completely specify the magnitude of the adopted

plans, the proxy statements in the year of adoption and the subsequent

year were reviewed for data indicating the value of the restricted

stock awards granted and the potential value of the performance units

or shares received. This measure was used because restricted stock and

performance plan compensation expense was able to be discerned in only

a few cases. Generally these expenses were aggregated with retirement,

employee savings plan, and pension plan expenses under the contingent

forms of renumeration category, or with salaries, fees, and bonuses.

The magnitude of the adopted plans' incentive effects is expected to be

dependent upon the size of their performance awards relative to other

management compensation. Findings supporting this contention will

enhance the validity of this study's results. For 176 of the 216

adopting firms, the proxy statements provided sufficiently detailed

textual data to estimate MADOPT, the magnitude of the restricted stock

and performance awards relative to salary and bonus compensation for

the management group.

4.4.2 Past performance of firm (PERF)

To operationalize the measure of past earnings performance of a

firm prior to adoption, PERF, the relative growth in earnings per share

is utilized. More specifically, the measure is defined as follows:


where t = year of adopting long-term compensation plan, or
assigned adoption year,

EPSt = earnings per share for the year preceding adoption
adjusted for stock splits and stock dividends, and

Sales = net sales in millions of dollars.

The earnings per share growth over the five years preceding the year of

adoption (assigned adoption year for control sample firms) is divided

by net sales to normalize this measure of earnings performance. In

addition to the clarity of the sales-earnings relationship and the need

to utilize a relative measure of growth in any cross-sectional study,

the use of net sales eliminates the source of two other measurement

problems. First, utilizing net sales does away with the inconvenience

associated with negative denominators typically encountered when

employing some measure of earnings per share. Secondly, it obviates

the statistical difficulties attached to the use of extremely small

denominators in measuring relative growth.

4.4.3 Ownership control status (CONTROL)

Ownership control status (CONTROL) is used as a dichotomous

variable with manager-controlled firms represented as "0" and owner-

controlled firms represented as "1". The classification of firms on

the basis of this variable is accomplished in the following manner.

From a review of the proxy statements for the adoption year, firms are

categorized as manager-controlled if no single block of stock greater

than five percent is controlled by any party, and if the aggregate

ownership of the board of directors is not greater than five percent.

To be classified as owner-controlled, one party must own ten percent or

more of the voting stock of the firm, or the aggregate ownership of the

board of directors must be twenty percent or greater.

To provide further assurances that the behavior of these managers

are closely alligned with the long-run interests of stockholders,

another criterion for classification as an owner-controlled firm was

implemented. A firm is deleted from the owner-controlled group if at

least two large owners (five percent or greater stock ownership) exist,

and one is a member of management (an insider) while the other is not.

This competing owners condition may signal an excessive concern on the

part of the inside owner for short-term profits, in order to forestall

an internal fight for control. Because owner-control is used in this

study as a proxy for a low or reasonable level of agency costs, these

firms are eliminated from this classification. The earlier criteria

are generally consistent with definitions used in previous empirical

studies [Monsen and Downs (1965), Smith (1976), Dhaliwal, Salamon, and

Smith (1982)].

4.4.4 Descriptive statistics

A summary of the distributional properties of the independent

variables is provided in Tables 4.1 and 4.2. For this purpose the data

are partitioned on the ADOPT variable, thereby yielding descriptive

statistics for both the experimental and control sample groups. Table

4.3 details the correlations among the independent variables.

Generally, the statistics in Table 4.1 and 4.2 conform with

expectations. The distribution of firms over the five adopting years,

1978 through 1982, is the result of the sampling technique utilized.

The majority of adopting firms was discovered by reviewing the 1980 and

1981 proxy statements available in the University of Florida library.

This procedure generated most of the 1979 adoptions, all of the 1980

adoptions, and a substantial portion of the 1981 adoptions. As

discussed previously, the non-adopting firms were randomly assigned

"adoption" years in the same proportion as the experimental group.

Categorization by plan type was accomplished by reviewing the

description provided in the proxy statements announcing the upcoming

vote on the plan. The adopted plan is composed of both restricted

stock and performance plan elements for 34 of the firms. A review of

the actual awards made in the adopting and subsequent years provides a

measure of the relative size of each of these elements. For seven

firms the performance plan element comprised greater than eighty

percent of the total long-term award value. In these cases the firm is

classified as adopting a performance plan. Of the 34 firms with

composite plans, another three have large (greater than 80 percent)

restricted stock components and were so classified. Subsequent to this

reclassification, 24 firms remain in the mixed component plan category.

With regard to the magnitude of the adopted plan (MADOPT), the

range from 0 percent to 207 percent reflects the disparity in the

potential value of awards relative to salary and bonus. This is

consistent with Ball and Foster's (1982) criticism that the use of a

dichotomous variable to represent compensation plans is overly

simplistic. Of the 176 adopting firms which provided sufficient data

to estimate the relative award size, 47 firms made no awards during the

adoption or subsequent year. The large size of a few of the awards

represents firms which made large restricted stock or performance

awards during the period whose potential value may or may not be

reached. Because the composition of compensation expense is generally

not publicly available, the stated potential value at the time of the

grant is utilized.

A review of the ownership control status (CONTROL) distribution

(Table 4.2) points out two factors of interest. First, the percentage

of manager-controlled firms in the adopting group is substantially

larger than that of the non-adopting group. This relationship is

consistent with the ownership control interaction hypothesis

(Hypothesis 4) arguments previously developed. If control by

management is a good surrogate for excessive agency costs, then it is

more likely that such firms will adopt performance-contingent

compensation plans in an attempt to reduce agency costs and enhance the

benefits of management. This relationship is also indicative of the

negative correlation between those variables presented in Table 4.3.

Moreover, the negative correlation between MADOPT and CONTROL further

supports the Hypothesis 4 arguments. Again, it is more likely that

manager-controlled firms will adopt the larger incentive schemes

(relative to total compensation) in order to substantially cut agency

costs and augment the monetary rewards of management. Secondly, the

distribution of the non-adopting group merits further comment, given

the large percentage of owner-controlled firms relative to previous

studies using similar definitions [Hunt (1986) As noted by Hunt

(1986), as the size of corporations increases, their ownership

structures generally change from one involving a dominant stockholder

to one with widespread ownership among a diverse group. Consistent

with this observation, relatively small firms (less than $100 million

of sales) comprise 29.5 percent of the non-adopting group.

The partitioning method used in defining ownership control status

resulted in a subset of 131 firms whose status could not be clearly

defined. Generally for these firms the largest party controls between

five and ten percent of the voting stock. Ownership is too

concentrated to be included in the manager-controlled group, but not

concentrated enough to be included in the owner-controlled group. For

these firms it is not clear that either party controls the selection of

the board of directors.

The distribution of earnings performance (PERF) provides no

surprises. Early period earnings per share are nct available on

COMPUSTAT for 19 firms, resulting in the reduced sample size of 336.

As may be noted, the range in PERF is greater within the control group

of firms, and the mean is somewhat larger. This suggestion that firms

with superior earnings performance are less likely to be adopting

long-term compensation plans is also consistent with the arguments

presented in Chapter III concerning the past performance interaction

hypothesis (Hypothesis 3). If past earnings performance is a good

surrogate for the degree of existing agency costs, then poorly

performing firms have more to gain from adopting a long-term

compensation plan to help control the divergence of interests between

owners and managers. The negative correlation between ADOPT and PERF

displayed in Table 4.3 further validates this contention.

Table 4.3 provides the correlation coefficients for the three

independent variables for the full sample, as well as for the

experimental and control groups separately. While in some cases the

correlations are statistically significant, fortunately they are not

sufficiently large to indicate a serious multicollinearity problem in

the multivariate regressions. Furthermore, this suggests that the

independent variables might interact with each other in explaining

unexpected capital spending levels.


All Adopting Non-adopting
Firms (Experimental) (Control)
Firms Firms

Adoption year No. (%) No. (%) No. (%)
1978 21 ( 5.9) 14 ( 6.5) 7 ( 5.0)
1979 59 ( 16.6) 35 ( 16.2) 24 ( 17.3)
1980 91 ( 25.6) 57 ( 26.4) 34 ( 24.5)
1981 132 ( 37.2) 78 ( 36.1) 54 ( 38.8)
1982 52 ( 14.7) 32 ( 14.8) 20 ( 14.4)
355 (100.0) 216 (100.0) 139 (100.0)

Plan type
Performance plan 111 ( 51.4)
Restricted stock plan 81 ( 37.5)
Mixed plan 24 ( 11.1)
216 (100.0)

Plan origination
Initial adoptions 149 ( 69.0)
Extensions 67 ( 31.0)
216 (100.0)

Plan magnitude (MADOPT)2
Mean .225
Minimum 0
Maximum 2.070
Standard deviation .278
No. of firms (for which magnitude 176
data available)

1. ADOPT = 1 if long-term compensation plan adopted
0 if no long-term compensation plan adopted

2. MADOPT = magnitude of the performance and restricted stock
awards made to the management group relative to salary and
bonus compensation.


All Adopting Non-adopting
Firms (Experimental) (Control)
Firms Firms

Past performance of firm (PERF)1
Mean .008 .006 .012
Minimum -.138 -.023 -.138
Maximum .237 .074 .237
Standard deviation .028 .012 .042
N 336 209 127

Ownership control status (CONTROL)2
No. (%) No. (%) No. (%)
Manager-controlled 78 ( 22.0) 61 ( 28.2) 17 ( 12.2)
Owner-controlled 146 ( 41.1) 71 (32.9) 75 ( 54.0)
Other 131 ( 36.9) 84 ( 38.9) 47 ( 33.8)
355 (100.0) 216 (100.0) 139 (100.0)

1. PERF = EPSt-l EPSt-6

where t = year of adoption

EPSt_1 = earnings per share for the year preceding adoption,
Sales = net sales in millions of dollars.

2. CONTROL = 1 if owner-controlled firm
0 if manager-controlled firm.


All Adopting Non-adopting
Firms (Experimental) (Control)
Correlations between Firms Firms

ADOPT and PERF -.11 b b

ADOPT and CONTROL -.29 b b

MADOPT and PERF c .05 c

MADOPT and CONTROL c -.10 c

ADOPT (Initial) and PERF -.09 b b

ADOPT (Initial) and CONTROL -.25 b b

PERF and CONTROL .15 .19 .13
(.0280) (.0313) (.2375)

apearson correlation coefficient and significance level (in

bBecause ADOPT is a dichotomous variable distinguishing adopting and
non-adopting firms, these partitioned correlations are irrelevant.

CThe magnitude of the adopted plan, MADOPT, pertains to the adopting
firms only.

4.5 Notes

1. Their list was procurred from Stephen Walsh, a NYSE proxy
specialist. In addition to the plan adoptions of interest, this
list included short-term bonus plan, employee savings plan and
stock option adoptions. While the list is of firms voting on such
plans, Mr. Walsh believes that all of the plans were adopted.

2. An industry average model is also used to predict expected capital
spending levels on the basis of actual capital spending levels of
all COMPUSTAT firms with the same two-digit SIC Code. A further
description of the model and associated results are provided in
Appendix A.

3. According to Jorgenson and Siebert (1968a), to finalize the theory
a model of replacement investment should be added. In line with
the internal funds investment model of Dhrymes and Kurz (1967),
depreciation is oftentimes used to proxy for investment
replacement. However, Fama (1974, p. 308) found that depreciation
did not seem to be systematically important in explaining changes
in capital stock when using an investment regression model similar
to this study's models. Hence, depreciation is not included in
the models.

4. The mean squared prediction errors over the three year
pre-adoption period of these reduced models are comparable to
those of the full model which define investment expenditures as
capital spending only.

5. Consistent with the theory of Tinbergen (1939) and the findings of
Elliott (1980), the mean squared prediction errors of the
investment models improved with the inclusion of the cost of
capital variable.

6. An attempt was made to scale earnings per share growth using both
EPS6 and its absolute value. However, apparently due to the
problems noted, stronger results were obtained with the sales
scaling variable, than with either of the EPS variable


This chapter presents the empirical results for the hypotheses

developed in Chapter III. Descriptive statistics are also provided for

the various dependent variable measures. The descriptive statistics

for the independent variables were presented in Chapter IV.

5.1 Dependent Variable -- Descriptive Statistics

The summary statistics for the dependent variable measures are

provided in Tables 5.1 through 5.4. The definitions of these measures

are presented in Table 5.0. Tables 5.1 and 5.2 furnish the statistics

for the measures of unexpected capital spending level based on Value

Line predictions. For the statistics of the comparable measures based

on the econometric models of investment, refer to Tables 5.3 and 5.4.

Statistics for the adopting and non-adopting sets of firms are

summarized in the first table of these sets of two, with the second

table presenting the simple correlations between the independent and

the dependent variables.

5.1.1 Value Line model expectations

Table 5.1 summarizes the statistics for all the sample firms for

which Value Line capital spending predictions are available. Of the

224 sample firms classified according to ownership control status

(CONTROL), 131 or 58 percent were included in the Value Line Investment

Survey. The forecasted capital spending levels and the resulting

dependent variable measures (unexpected capital spending levels) were

checked for reasonableness. The primary purpose of this procedure was

to avoid introducing errors due to inaccurate transcription of the

Value Line predictions. In two instances the dependent variable

measures were found to be clearly unreasonable. These cases were

eliminated in the subsequent analysis, reducing to 129 the number of

sample firms available for analysis based on Value Line expectations.

Under no circumstances were observations deleted due to a perceived

anomalous relationship between the dependent and independent variables.

Therefore, while the population for which the inferences and

conclusions apply excludes outliers, there should be no biases in the

statistical analysis. A further review of Table 5.1 indicates that

Value Line provided capital spending forecasts for approximately 67

percent of the adopting firms, but only 43 percent of the non-adopting

group. This is readily explained by the disparity in firm size between

the groups as previously discussed, taken in conjunction with Value

Line's policy of reporting on the larger companies.

A review of the descriptive statistics for the partitioned samples

(Table 5.1) reveals that the mean unexpected capital spending levels of

the adopting firms are positive and larger than those of the

non-adopting firms for all of the sales deflated measures, as expected.

This expectation follows from the hypothesized adoption effect

(Hypothesis 1) and the presumption that Value Line's capital spending

forecasts (dated four quarters prior to the proxy statement announcing

the adoption) are made without knowledge of the upcoming adoption.

With regard to the PP & E deflated dependent variable measures however,

the non-adopting firms have the larger means in five of the six cases

(VLCRPI is the exception), although the means for the adopting firm

group remain positive. This unexpected relationship is further

evidenced in Table 5.2 by the negative correlations between these

measures and the ADOPT variable. Subsequent analysis of the regression

results will refer to this unexplained condition.

Further examination of the Value Line expectation model tables

reveals nothing else of an unforeseen nature. The mean values of the

unexpected capital spending levels which consider both capital

investment and R & D expenditures are consistently greater than those

which consider capital investment only. This is a direct result of

using the same Value Line forecast for each dependent variable measure.

Value Line provides a single capital spending forecast for the upcoming

two to four years. Because it is not clear whether this forecast

includes R & D expenditures for plant and equipment, unexpected capital

spending levels are measured in both ways. Moreover, as discussed

previously there may exist differences cross-sectionally in the

accounting classification of these R & D expenditures.

With regard to the sales deflated versus PP & E deflated measures

of the dependent variable, the smaller unexpected investment levels

based on sales demonstrates that the magnitude of sales generally

outweighs that of PP & E. Furthermore, it may be noted that the

standard deviations of the PP & E based measures are uniformly larger.

Finally, the relationship between the one- and two-year measures

of the dependent variable deserves comment. As expected, the means of

the variable measures for the adopting firms generally diminish as

post-adoption years are added to the operational definitions. However,

this relationship is not maintained by the non-adopting firm sample.

In fact, their means tend to increase in the periods which include

post-adoption years. This divergence in sample statistics emphasizes

the need to evaluate unexpected capital spending levels over more than

one year.

5.1.2 Econometric model expectations

The descriptive statistics for the dependent variable measures

which base unexpected capital spending levels on the predictions of the

investment econometric models are presented in Tables 5.3 and 5.4.

Statistics for the partitioned sample are summarized in Table 5.3, and

the correlation matrix is exhibited in Table 5.4.

In reference to Table 5.3, the procedures used to obtain

econometric model predictions and the related issue of sample sizes

will first be discussed. The polynomial distributed lag investment

models used to generate predictions require an uninterrupted flow of

data for a minimum of four years prior to the year of prediction, given

the lags specified. This, taken in conjunction with the reasonableness

checks described later, requires that the distributed lag model

variables be available for a minimum of seven years prior to adoption.

Because of this data requirement, econometric model predictions were

unable to be generated for a number of the firms. For the distributed

lag models which defined capital spending as capital investment

deflated by sales and by PP & E, predictions were generated for 130 and

141 firms respectively, of the 224 sample firms which were classified

according to their ownership control status (CONTROL). However, when

R & D was added to the operational definition of capital spending,

model generated predictions for both sales and PP & E deflated measures

fell to 89 sample firms. This difference, as reflected in the

observations of Table 5.3, is the result of differential reporting

requirements. While capital investment data is readily available from

COMPUSTAT, R & D data is generally not available prior to 1971, as

reporting was voluntary. Subsequent to that time R & D expenditures

are available on a limited basis (approximately 37 percent of COMPUSTAT

firms), as reporting was required only when such expenditures exceeded

one percent of sales.

The predictions generated by the econometric models were examined

for reasonableness at two separate stages. These reasonableness checks

were particularly critical given the nature of the prediction

generating process. While annual COMPUSTAT data was generally

available for the period 1962 through 1984, data specific to an

individual firm may have been limited to a shorter time series. This

situation, in combination with the ability of the distributed lag

models to generate predictions on the basis of extremely limited time

series data (six years), suggests that outliers may be common.

The initial examination for reasonableness considered the

prediction error during each of the three years preceding adoption.

Outliers were identified and firms were deleted from subsequent

analysis if the squared prediction error exceeded 1.0.1 A second

review for reasonableness considered the prediction errors in the

adoption and following two years. Although these predictions were made

on the basis of predicted rather than actual capital spending levels,

only a limited number of outliers (less than two percent) were evident.

The extent of deletions on the basis of the two reviews for

reasonableness varied for the four investment models, ranging from two

to eight percent of the generated predictions. The sales based models

exhibited a greater number of outliers than did the PP & E based

models. Surprisingly, the models which define capital spending as

capital investment only generated a larger percentage of outliers than

those which also include R & D expenditures. Given that capital

investment data was generally available for ten more years than was

R & D data, the opposite result was expected. This relationship

suggests that any advantage associated with additional observations

was, for some firms, more than offset by changes over time

(non-stationarity) in the process which determines capital spending

levels. As was the case with the Value Line based dependent variable

measures, none of the deletions was the result of an exceptional

relationship between dependent and independent variables. Therefore,

while the population for which inferences apply excludes outliers, the

statistical analysis remains unbiased.

A further review of Table 5.3 reveals that consistent with the

hypothesized adoption effect (Hypothesis 1), the means of the adopting

firm sample are generally positive and larger than those of the control

(non-adopting form) sample. (There is only one case, EMCRS1, in which

the mean of the adopting firm group is not larger than the mean of the

control group.) This condition is reiterated in the correlations

matrix of Table 5.4.


Furthermore, as was the case with the Value Line measures, the

means and standard deviations of the dependent variable measures

deflated by PP & E are larger than those deflated by sales. However,

when comparing the statistics of the measures which include R & D as a

capital spending expenditure to those which do not, no consistent

patterns emerge. The same can be stated for the relationship between

the time periods measured. Since such relationships are more evident

for the Value Line dependent variables measures, the differences in

forecasting methods need to be considered. The same Value Line

forecast is used for each of the twelve dependent variable measures.

As a result, any differences between the Value Line dependent variable

measures are due entirely to alternative measures of actual capital

spending levels. On the other hand, four distinct but related

econometric models were used to generate forecasts for the four

alternative definitions of capital spending. Moreover, separate

forecasts were made for the year of adoption and for each of the

following two years. Given that the forecasts are peculiar to capital

spending definitions, bases (sales versus PP & E), and years, it is not

surprising that clear-cut relationships do not exist. A further review

of the statistics in Tables 5.3 and 5.4 indicates nothing else out of

the ordinary.

5.2 Tests of Hypotheses

Tables 5.5 through 5.16 present the empirical results for the four

hypotheses developed in the preceding chapter. Results based on the

Value Line expectations are presented first, followed by those based on

the econometric model expectations. As noted previously the results

based on industry average expectations are provided in Appendix A. A

brief review of each of the hypotheses follows.

From the motivation and incentives theory of compensation, the

adoption hypothesis, Hypothesis 1, posits that the establishment of a

long-term compensation plan will lead to higher than expected capital

spending levels. Hypothesis 2 is introduced to examine whether the

hypothesized adoption effect might differ between performance plans and

restricted stock plans. A priori, no difference is expected.

Hypotheses 3 and 4 are presented in an attempt to determine the

circumstances under which long-term incentive contracts are more

effective, and to provide evidence of the applicability of the

motivation and incentives hypothesis. The motivation and incentives

theory argues that performance-contingent compensation plans are

designed to control the stockholder/manager conflict of interests,

thereby reducing agency costs. It follows then that plan adoptions

will be most effective when agency costs are greatest. Based on the

assumption that past earnings performance is a good surrogate for the

level of agency costs, Hypothesis 3 posits that the interaction between

plan adoption (ADOPT) and a firm's past earnings performance (PERF)

will be negative. Alternatively, a firm's ownership control status may

provide a better measure of the degree of agency costs, with

manager-controlled firms exhibiting larger agency costs than

owner-controlled firms. Hypothesis 4 then posits that the interaction

between plan adoption (ADOPT) and a firm's ownership control status

(CONTROL) will be negative.

5.2.1 Value Line model-based results

Tables 5.5 and 5.6 present the main regression model results.

With regard to Hypothesis 1, the significance and positive sign of the

ADOPT variable coefficient for the sales based dependent variable

measures which include the adoption year provides reasonably broad

support for the hypothesis that the introduction of long-term

compensation plans is associated with a unexpected increases in capital

spending levels. While the magnitude of the adoption effect is

greatest in the year of adoption, the effect remains significant (.01

level) for the two-year period commencing with the adoption year. For

the two-year period beginning the year after adoption, the ADOPT

variable coefficient remains positive, but it is no longer

significant. Given the long-run nature of Value Line's forecast, this

result is not surprising. The longer the time which has elapsed from

the date of prediction, the greater the noise in the dependent variable

measure engendered by unexpected changes in conditions affecting

capital spending decisions. Moreover, the effect of adoption is

consistent across both operational definitions of capital spending,

capital investment alone and in combination with R&D expenditures.

While the significance of the adoption effect is lost when the

unexpected capital spending levels are deflated by PP&E (Table 5.6),

rather than deflated sales (Table 5.5), the direction of the adoption

effect remains positive.

Because Value Line's capital spending predictions implicitly

consider these variables, no hypotheses were introduced concerning the

main effects of a firm's past performance (PERF) and ownership control

status (CONTROL) on unexpected capital spending levels. However, the

PERF effect is positive and significant across many of the dependent

variable measures, suggesting that this variable is underweighed by the

Value Line model of expected investment outlays. Moreover, the same

can be said for the CONTROL variable, although the degree and level of

significance are reduced.

Hypothesis 3 involves the interaction between adoption (ADOPT) and

a firm's past performance (PERF). The coefficients here are uniformly

negative as hypothesized, and for the PP & E deflated measures, highly

significant. The magnitude of the performance interaction is

surprisingly large, particularly for the PP & E deflated measures of

the dependent variable. This suggests that long-term compensation

plans are substantially more effective in increasing capital spending

levels when adopted by poorly performing firms, as opposed to those

firms whose past earnings performance is good.

Concerning the ownership control status interaction, Hypothesis 4,

the results are stronger than those of the past performance

interaction. Once again the hypothesized negative direction of the

effect is found consistently across all of the dependent variable

measures. However, in contrast to the performance interaction

findings, it is the dependent variable measures deflated by sales for

which the CONTROL interaction is significant. It is particularly

important to note that the CONTROL interaction effect is significant in

all cases where the ADOPT effect is significant. Because this result

is consistent with the motivation and incentives hypothesis, it

validates the applicability of this theory in the settings tested.

However, the applicability of the the alternative signalling and

screening-sorting theories is not negated.

Tables 5.7 and 5.8 report the regression model results when the

experimental sample group is limited to those firms which introduced

long-term compensation plans in the adoption year, as opposed to merely

extending a previously adopted plan. While these findings are

consistent with and support those based on the full experimental sample

group (Tables 5.5 and 5.6), it is somewhat surprising that there is no

strengthening of the adoption effect. Once again the adoption effect

is found to be uniformly positive (VLCP3 is an exception), and

significant in the case of those sales deflated dependent variable

measures which include the adoption year. While the significance of

the PERF interaction coefficients dissolves when the experimental group

is limited to initial adoptors, the CONTROL interaction results closely

reflect those of the entire adopting group. It is gratifying that the

R-squares reported in Tables 5.7 and 5.8 indicate substantial

improvement in the ability of the model to explain the variation in

unexpected capital spending levels when only initial adoptors comprise

the experimental group.

In an attempt to make the effect of the adoption variable more

discriminating, a continuous measure of the magnitude of the adopted

plan (MADOPT) was extracted from the proxy statements. Tables 5.9 and

5.10 present these findings. As expected, the direction of the MADOPT

variable is uniformly positive across all dependent variables,

indicating that unexpected capital spending is positively associated

with the magnitude of the incentive award. Moreover, the MADOPT

coefficients are significant in the case of the same four sales

deflated dependent variable measures which provide the significant

ADOPT effect (Table 5.5). With regard to the PERF and CONTROL

interactions however, the directions are inconsistent across the

various dependent variable measures. Nevertheless, in the case of the

four measures with significant MADOPT effects, the CONTROL interaction

coefficients are uniformly negative. For the remaining sales deflated

measures, the CONTROL interactions are positive, and in the case of

VLCS3 significant. There is no apparent explanation for this

exceptional relationship. Finally, the increased R-squares (in

comparison to those of Tables 5.5 and 5.6) indicate that characterizing

the adoption on the basis of the magnitude of the plan (MADOPT) offers

some improvement over the dichotomous classification (ADOPT) in

explaining the unexpected capital spending variation.

In the earlier discussion of the adoption effect, Hypothesis 2 was

introduced to examine the relative effectiveness of restricted stock

versus performance plans in promoting increased levels of capital

spending. As may be seen in Tables 5.11 and 5.12, the effect of

adopting performance plans on unexpected capital spending levels is

positive, and significant across all of the sales-based dependent

variable measures except VLCRS3. As is the case with the entire set of

long-term compensation plan adoptions (Table 5.6), the direction of the

adoption effect for the PP&E based dependent variable measures is

positive as hypothesized, but in no case significant. A review of

Tables 5.13 and 5.14 indicates that for restricted stock plans, the

adoption effect is again consistently positive (VLCP3 and VLCRP3 are

exceptions), and significant in the year of adoption for both the sales

and PP & E deflated dependent variable measures. Furthermore, the

restricted stock adoption effect is significant for the two-year period

beginning with the year of adoption for the sales deflated measures.

When considering the adoption-performance and adoption-control

interaction results exhibited in Tables 5.11 through 5.14, some

further differences between performance and restricted stock plans

become evident. The accounting earnings-based performance plan

findings demonstrate significant results for the PERF interaction

across ten of the twelve dependent variable measures. On the other

hand, while consistently negative as hypothesized, the PERF interaction

restricted stock plan results are not significant in any of the cases.

While not surprising given the previously described findings, the

CONTROL interaction results are more consistent across plan types. The

direction of the effect is negative as hypothesized across all but one

of the dependent variable measures for both types of compensation plans

(restricted stock plan's VLCRP3 is the exception). Significant effects

are generally limited to the sales deflated measures, although

performance plan's VLCRPI and restricted stock plan's VLCP1 are also

significant. For performance plans (Tables 5.11 and 5.12), the CONTROL

interaction effect is significant in the case of five of the six sales

deflated measures, and weakly significant for the other (VLCS3).

Notably the CONTROL interaction is significant in every case that the

ADOPT effect is significant. Regarding restricted stock plans (Tables

5.13 and 5.14), the CONTROL interaction effect is significant for three

of the six sales deflated measures. Coincidently, the adoption effect

is significant in each of these cases. (In the case of VLCRS2, the

ADOPT effect is significant and the CONTROL interaction is weakly


The reduced number of observations indicated in Tables 5.11

through 5.14, in comparison to those of Tables 5.5 and 5.6, reflect the

use of specific sub-sets of the adopting firm group. Additionally, the

increased R-squares is undoubtedly due to the increased homogeneity of

the adopting firm groups. Not surprisingly, some differences between

restricted stock and performance plans apparently do exist in terms of

the association between their introduction and the unexpected capital

spending levels of firms.

To statistically compare the incentive effects of performance and

restricted stock plans, Chow tests were performed. Generally the

results revealed no differences between the two plan types. However,

for three dependent variable measures, VLCP1, VLCP2, and VLCRP1,

significant differences were evidenced. While the adoption effect is

significant for restricted stock plans in two of these cases (VLCPI and

VLCRP1), it is not significant for performance plans, nor the full

sample of adopting firms. With regard to the sales deflated dependent

variable measures which exhibited significant adoption effects for the

full sample, no significant differences were found between the

incentive effects of performance plans and those of restricted stock


In summary, the results based on the Value Line capital spending

predictions provide reasonably strong support for the adoption effect,

Hypothesis 1. The related arguments maintained that prior to the

adoption of long-term plans, the managers may have been rejecting

profitable capital spending projects in order to enhance short-run

earnings. The results indicate that adopting these plans can lead to

unexpected increases in capital spending levels in the year of

adoption, and the two-year period beginning with the adoption year.

Given the early period negative earnings consequences of accepting

these projects [Sunder (1980)], one implication of the significant

adoption effect is that the unexpected earnings (without knowledge of

the adoption and related capital spending increases) of adopting firms

in the post-adoption period should be lower than those of non-aaopting


Tests of this implication were performed on the full sample, with

post-adoption earnings per share expectations based on pre-adoption

levels. While the results are directionally consistent with this

contention in the year following adoption and the two-year period

beginning with the adoption year, they are not statistically

significant. Moreover, in the year of adoption the findings, while

insignificant, are in the opposite direction. At least two factors may

contribute to the weakness and inconsistency of these results. First,

while the direction of the early period earnings impact for both

capital investment and R & D expenditures is equivalent, the timing of

their impacts differ. R & D expenditures lead to reduced earnings in

the year of expenditure, due to the requirements of the FASB's

Statement No. 2. On the other hand, Sunder (1980) demonstrates that

while capital investment projects will negatively affect profits in the

year following outlay, they will be associated with higher earnings in

the outlay year. Therefore, while increased R & D expenditures in the

adoption year will lead to reduced adoption year earnings, capital

investment increases in that year will be associated with offsetting

earnings increases. These counter-balancing earnings effects must be

disentangled before the contended relationship between post-adoption

earnings and plan adoption will be evident. Secondly, a naive earnings

expectation model was utilized. Given the complexity of earnings,

using a more completely developed earnings expectation model may better

enable the relationship to be detected.

With regard to Hypothesis 2, some differences between performance

and restricted stock plans in terms of their adoption effect and its

interaction with PERF and CONTROL are evident. However, because these

differences are confined to dependent variable measures which do not

exhibit significant adoption effects for the full adopting firm sample,

no further inferences are appropriate. Furthermore, while the past

performance interaction (Hypothesis 3) findings are consistently

negative as hypothesized, and generally significant for the PP & E

deflated dependent variable measures, it is the ownership control

status interaction (Hypothesis 4) results which most strongly sustain

the motivation and incentives theory of compensation. The CONTROL

interaction coefficient is significant in the hypothesized negative

direction in every case that the adoption effect is found significant

(the only exception is the MADOPT results, Table 5.9). The next

section presents the econometric model-based results.

5.2.2 Econometric model-based results

The regression model results based on expectations generated by

the econometric models are presented in Tables 5.15 and 5.16. In

general the findings are somewhat disheartening in that the results are

at times inconsistent with the Value Line based results. On the other

hand, when the dependent variable descriptive statistics and

correlations (Tables 5.3 and 5.4) are considered, the findings are not


In general the weak and inconsistent findings point to the need

for further development of the econometric models. However, for the

model which defines capital spending as capital investment deflated by

PP & E (EMCP_ in Table 5.16), the findings provide reasonably strong

and consistent support for the hypothesized adoption effect. The ADOPT

coefficient is uniformly positive across the three dependent variable

measures, and is significant for the two measures which include the

adoption year. Futhermore, the uniformly negative direction of this

model's PERF and CONTROL interaction coefficients is consistent with

both Hypotheses 3 and 4, and the Value Line model based results. With

regard to the model which defines capital spending as capital

investment deflated by sales (EMCS_ in Table 5.15), the ADOPT

coefficients are uniformly positive as hypothesized, yet in all cases

insignificant. Moreover, the PERF and CONTROL interaction findings for

this model are in no cases significant, and the coefficients are mixed

in direction.

The two models which include R & D expenditures in their

definition of capital spending (EMCRS in Table 5.15 and EMCRP in

Table 5.16) exhibit adoption effects which are in all cases

insignificant, yet uniformly negative (in opposition to the

hypothesized direction). The PERF and CONTROL interactions of these

two models spur little interest given the insignificance of the

adoption effect. In no cases are they significant, and their

directions are inconsistent.

The PERF and CONTROL variables main effects are generally

insignificant across the four econometric models. Interestingly, the

two exceptions (EMCP1 and EMCP2, Table 5.16) are also the two dependent

variable measures which provide a significant ADOPT effect. In

combination with the mixed direction of these coefficients, this lack

of significance suggests that the econometric models implicitly

consider these factors. (No hypotheses were introduced for these main

effects, as this implicit consideration was assumed.)

The inconsistency of the results for the four econometric models

may be explained in large part by the limited availability of R & D

expenditure data. It is the results of the two models which include

R & D expenditures in their definition of capital spending that are

particularly inconsistent. While fifteen or more observations are

desirable for constructing time series models, R & D expenditure data

is available beginning in 1971, only. Therefore, the predictions of

these two models are based on a series of between seven and eleven

years of annual data. It is not surprising then that the results which

use the predictions of these models for capital spending expectations

are weak and incongruous. The other two econometric models which

define capital spending as capital investment only are constructed on

the basis of a series of annual data which are between sixteen and

twenty years in length. While a longer series may be preferred, the

costs in terms of reduced predictive ability associated with the

potentially changing nature of the process determining capital spending

levels (non-stationarity) were considered to be too large. While the

results of these two models are weak, suggesting the need for further

model development, the adoption effects are uniformly positive (as

hypothesized) and significant in the two cases described.

Given the generally poor quality of these results and the evidence

of prediction model weaknesses, further tests using the econometric

model dependent variable measures were not deemed to be useful. As

such, no findings are presented concerning the different sub-sets of

the adoptor sample groups (initial adoptions, performance plans, and

restricted stock plans), nor the different levels/measures of the

adoption variable (MADOPT).


VLCSI = [CS/Sales]0 [VL/Sales0]
EMCS1 = [CS/Sales]0 EM0
VLCS2 = [CS/Sales]01 [2VL/(Sales0 + Sales )]
EMCS2 = [CS/Sales] 0 EM,
VLCS3 = [CS/Sales]1,2 [2VL/(Sales1 + Sales2)]
EMCS3 = [CS/Sales],2 EM1,2
VLCRS1 = [(CS + RD)/Sales]0 [VL/Sales0]
EMCRS1 = [(CS + RD)/Sales]0 EMO
VLCRS2 = [(CS + RD)/Sales]0,1 [2VL/(Sales0 + Salesl)]
EMCRS2 = [(CS + RD)/Sales]0, EMO,1
VLCRS3 = [(CS + RD)/Sales],2 [2VL/(Sales + Sales-]
EMCRS3 = [(CS + RD)/Sales] EM1,2
VLCP2 = [CS/PPE]O1 [2VL/(PPE0 + PPE1)]
EMCP2 = [CS/PPE]0,1 EM0,1
VLCP3 = [CS/PPE] ,2 [2VL/(PPE + PPE2)]
EMCP3 = [CS/PPE],2 EM ,2
VLCRP2 = [(CS + RD)/PPE]O [2VL/(PPE0 + PPE )]
EMCRP2 = [(CS + RD)/PPE]0, EMO,1
VLCRP3 = [(CS + RD)/PPE]I,2 [2VL/(PPE1 + PPE2)]
EMCRP3 = [(CS + RD)/PPE],2 EM,2



where the subscripts refer to years, with "0" the adoption year and "1"
and "2" the following two years, and

CS = Capital investment expenditures

RD = Research and development expenditures

Sales = Sales (net)

PPE = Property, plant and equipment (net)

VL = Value Line prediction of plant and equipment outlays

EM = Econometric model prediction.

NOTE: The form of the econometric model expectations (EM) varies
with the form of the actual capital spending levels in the
above definitions. As an example, the EM of the dependent
variable measure EMCRP2 is the weighted average expected
capital investment plus research and development expenditure
level deflated by PP & E for the two year period commencing
with the adoption year.


Dependent N Mean Std. Dev. Minimum Maximum

Adopting firms

VLCS1 89 .015 .057 -.098 .311
VLCS2 89 .013 .046 -.099 .196
VLCS3 88 .012 .059 -.121 .430
VLCRS1 89 .030 .060 -.082 .311
VLCRS2 89 .028 .050 -.099 .196
VLCRS3 88 .029 .063 -.121 .430
VLCP1 89 .027 .178 -.536 .754
VLCP2 89 .020 .157 -.513 .449
VLCP3 88 .008 .172 -.714 .493
VLCRP1 89 .088 .203 -.536 .754
VLCRP2 89 .081 .185 -.513 .637
VLCRP3 88 .071 .203 -.714 .600

Non-adopting firms

VLCSI 40 -.006 .050 -.144 .081
VLCS2 40 .003 .051 -.113 .208
VLCS3 40 .007 .055 -.120 .248
VLCRS1 40 .002 .053 -.144 .081
VLCRS2 40 .012 .053 -.113 .208
VLCRS3 40 .017 .058 -.120 .248
VLCP1 40 .036 .145 -.275 .364
VLCP2 40 .041 .128 -.257 .362
VLCP3 40 .036 .119 -.287 .368
VLCRP1 40 .082 .167 -.206 .497
VLCRP2 40 .088 .161 -.180 .613
VLCRP3 40 .091 .185 -.195 .935

v ;3 r- ;r >o r0) 3 ^ -4 ; N vf o m r- v m v r, 3 n 3 v :'
on c H rH co 0 Q0 u) V o or- c0 N CO rN o
N '-H 00"- " "' "O-I i'-- I I i'O








0)2 rNO NO H1 ON NO r0'-m 0 Nr r-R D ;" T 3O N CO
ITl O0 f) 0 1 (N (N Or- 0 -1 1-0 00 00 HO ov N wH HO HD cO 'T HN rH
O U) 0 0 0 r H CO N ) 0 O 1-1 0 N 0 H- r1-1 0 H -1 0 4 r-4

OLn H r O N D f o 0o 0 o o r al co 0,
n N) r1 0 *n CN- (N in M 0 NH rN ON 1 ri T N M
oH rHN rO ON 001 O0 0N O OUl o0 o0c out

-4 r cN o01 o o r o o o o o m n o
N U o o ( N w o r-i N un o N n T .o N o m0 r,

NOr r00 N O 00 O NO roO m H H 0 00 o o O

-r aO "-- r-O N R c o "ON m o o 00 w-- i00 '
-10 O" '- '-- "- '10 on C! O" oco o"o o


( l U)


Dependent N Mean Std. Dev. Minimum Maximum

Adopting firms

EMCS1 78 .007 .052 -.124 .311
EMCS2 78 .005 .044 -.134 .160
EMCS3 78 .004 .078 -.279 .452
EMCRS1 61 .000 .041 -.105 .170
EMCRS2 61 .000 .045 -.121 .160
EMCRS3 61 -.001 .077 -.192 .411
EMCP1 85 .023 .119 -.402 .422
EMCP2 85 .013 .112 -.303 .231
EMCP3 85 -.004 .157 -.557 .338
EMCRP1 62 .009 .112 -.375 .264
EMCRP2 62 .017 .130 -.420 .298
EMCRP3 62 .003 .175 -.873 .366

Non-adopting firms

EMCS1 41 -.022 .103 -.476 .108
EMCS2 40 -.016 .118 -.572 .172
EMCS3 40 -.020 .182 -.663 .219
EMCRS1 26 .001 .032 -.059 .075
EMCRS2 26 -.001 .030 -.088 .043
EMCRS3 26 -.002 .051 -.208 .087
EMCP1 49 -.004 .192 -.626 .537
EMCP2 49 -.040 .217 -.818 .381
EMCP3 48 -.072 .261 -.940 .293
EMCRP1 25 -.001 .096 -.156 .202
EMCRP2 25 -.006 .110 -.293 .300
EMCRP3 25 .000 .206 -.645 .654

o Nz 0 rri H - rN 01 CNO mco n i-4 n m p 1I 'Dl 3 I a m

moe mo mm mm mo or oe aOCM mo mH oa oa0
o rto o c re- o n o r r .I o m m v -I o om nN o o0
C 1 : 1 C 1 C : *U *1 1 i C *^

'N 0 m m D R N 0 n rl- 1 o co rH, 0 N O c co 00 C ;H
o r o 0 U) m M (N V Ln C O D rn O N o r-1 m m o VN r- m IN f, O

00 mm mn 0N0 OHr -IH aa NN ON (O (N Nr
m c No r- ma oo o cy, NOr, co rm o 00
D al U m m IT D i a o al 0- CN IN -f -1 0 o C) f r-.i i "r N I c r

0 Nom o ^o om ao ar o H a ro mo- a o m
r o aH r O O ao o a o a0 o m a N o o o






I iL


-a E



044 40


NM 0G 1 m 0) m
- 0 0 H-1 0 0

c4 7

un> 00 N 4140 NM rn

v r o m G v No
Lf 4O HO ON N 4C N

o o o o 0 o o .

*N oN N N N H

r-8 Co o) 14 ma N^ Iv 0 L
v In In i m Co N m
C 0 H Cl 0 4 i .
4I I C -

4' nO fN *C 10 N*

1 -1 o 0) -T t In

'N 'N NH qN 'N -H

41I mi0 00 4' N no) v Im
'* N *H

a N c > In co Co o o -I
I0O 4 O H N 0 O N ) a
Oi 0m i 0 0 o

S N 'H 1 *c I

4 M w

H) N ul Q2 P4



Ln I
0 W

in Z 44





40 '

C 3
>0 H
1 1 0

40 C

04 -M H
0,lu *

u U 0) >Ir4 4
,0 H 1(a 1
0 1

4H 40


40 0 0 0
4.4 di -i~
-0 -
r-rl 11 C0

4 4-4 *
0 -H HN
00 O 4 0- -
00 -H N 0m

.9 Ic 4
*H H >(0,404
4-41 44 40di
HH 401 >] >,
C1C 4 4
040' 30 n 4'O '

rdo (a m > vr
1H 40 C: C C
400 40 >i0 r

41 4

1444 44 u
0 -H -
HUI J &H -C .
rd > 4 0444

d0 4) f4 '04040
-I -I 0 0>i>i

4040 40
..C S -H
4.144.4 44-4


44: '0
-HH4 40..

*e g *H

H -H
* ZW- i
-4 40

N1 N N 4N CN N
3 (N (N CN (N N CN

o 0 0 0


rH1 0 0 0 o 0 0 0
00 I '. 0

o Ho n 1 co C) .a) o 0m "
vOU ) S H m a 0 N N 3
Sa .A * N N *N 04
00 oL i m i Nr 1o i 1

| 4 lq

p o o
co Z .4 co o 4J

000 0

o 0 NO co 01 %oD0 on r o N c0 0
CI o 8 IO (O LD in lr co Mc t
.0 0 o 0 0 n 0H 0H
S a N W ,m

t+ 0

o c ; mc o m o m co > >
'- Q o o o c o I r'f

r o co v' ' a)' c) 4J4

SO O NO 00 u 0u u u W
a) .f *f N N wu*

H -1 >-
0 *H .0 0 H

"0 0 0 a 0 0 *H
41 11
Q> s > t >~ S *> >
5l /
44 00m
H '-
00 U 44.
OH U) -'I


o a o a a a 0
0) 0) 0) 0) Ot 0)

CN H r- H n


1 * r ,

O I + (
0 o o w m Lo co? o
) '' E- 0 0 0 O 0 0 0 0 .
d 1 ** o0 *o CN .0


:DU0 O 1 rIm 0 m Nn


W H 140 C 0' 0 0.
Q O 0* N *

ca m mm to a ao
z 4. 0 Q z

* O O O O O O
E- 0 4. 4. 4 >

H .*N * N .0 .m H

co M N r- :3 0-* m v In r +4J
U2 Io mD I m N N
+ 00

M 0 D rI- n ) v co rl wIo n I r.
Il L Hn H-i oi r n cH n -. L 4 c
( C, C4 '.4 4 4- H ul M M

0 HN I N (1 CI) C) C) 00

0 0 0 0 (

> c ,4 o m co N C
+ (n a% fU N co m (N 0) ci 0 W 4

0 m O^ CN CO C; 0} CN r Ct; 0 0 4

&a44a 111

"S > *> *

0 0 0 0 0 0

N 0^ 0-1 0 0

3 +ImO

0y z 0 oN n r, i oo a) rai oo O i
I ) G* 0-4 c r' rCJ 01 t 0 1 m
0 E* 0 0 m 0 0 .
E 0o * .0 .0 .0 .0

LO 0a

00 0I NI WI ON CI 0 I
E- i-a m 8^ BE :^ i ^ m
vlU H q I< C O O Or O

Z al 0 co 'i N6 l Ho
0 H 0 I- I'-' I I- I 1 '-' C

*i) m -
as + 0

Q 0
<0 -4 -. -

00' 0 (0

4 4)
4 + 00

v3) 'N 1 -0 ON 4 C 0 0 01 00 0 C
ru R 'D 'co v -N C-V (n 00 4- .6
o 0 -4 0 m m 0 N .,m *l 0 4 0 4i 'H
4 01 l fl 'l H1 4' 4H
03 + 00

Sa -o 4o -o o o

'0,-I NW 0 04 (nL >>
ao Ho o . 22

a. O 10 U 'N 0)0 4 u )4


4 H3 *H


z 0 0 0 0 0 0

H 1 0 n m, H
N O 0 0 0 N
& H H H H H H

H o
co 4

I0 oI o H O co cc N O O
O .0 O0 HN 00 ~0O N0
S 0 .0 .0 H *H v.0 .H t

u ) V *o *I 'I I
II P4 I -H

w 0 + 0

w A 'I 4 0 O 0 O *
COO 0. 4J

00 H 0 0
0 '0 H f (0 0

0 0
-T B 0Q 3)

H* *4 44
+ 00

P a) 0)
v co n co cD y ml t rr > >
CO Q A 3 OP- I

06 (0N 0u00 OH O ON H O
C; X *.;O *O *O 0
0+ 0o

0 0 0

.N 0a .H 0 N *H


a>s > > > > 0

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs