Citation
Noncallable debt

Material Information

Title:
Noncallable debt evidence and effect
Creator:
Kish, Richard John, 1952- ( Dissertant )
Livingston, Miles ( Thesis advisor )
Brown, David T. ( Reviewer )
McClave, James T. ( Reviewer )
Place of Publication:
Gainesville, Fla.
Publisher:
University of Florida
Publication Date:
Copyright Date:
1988
Language:
English
Physical Description:
x, 239 leaves : ; 28 cm.

Subjects

Subjects / Keywords:
Assets ( jstor )
Call options ( jstor )
Debt ( jstor )
Finance ( jstor )
Interest rates ( jstor )
Maturity tests ( jstor )
Sample size ( jstor )
Standard deviation ( jstor )
Taxes ( jstor )
Yield to maturity ( jstor )
Corporate debt ( lcsh )
Corporations -- Finance ( lcsh )
Dissertations, Academic -- Finance, Insurance, and Real Estate -- UF
Finance, Insurance, and Real Estate thesis Ph.D
Options (Finance) ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Abstract:
Financial theory claims that callable debt dominates noncallable debt. Yet our evidence shows that a substantial amount of noncallable debt exists, suggesting a deficiency i the theory. Over the period 1977 through 1986 over $108 billion worth of noncallable debt was offered. During this ten year period, noncallable debt offerings accoutned for 16.9 percent of the dollar value and 12.1 percent of the total number of all publicly placed corporate debt issued. The dominance of callable debt has been motivated in a variety of ways, for example uncertain interest rates. When there is a substantial decrease in rates, callable debt provides a firm's managers with the ability to refund an issue and take advantage of lower interest expenses. Other reasons offered for the dominance of callable debt include agency problems associated with information asymmetries between borrowers and lenders, different risk tolerances of equityholders and debtholders, the need for managers to signal private information, differential tax rates between borrowers and lenders of funds, maturity preferences, and the opportunity to remove an undesirable protective covenant in the bond indenture between borrowers and lenders, different risk tolerances of equityholders and debtholders the need for managers to signal private information, differential tax rates between borrowers and lenders of funds, maturity preferences, and the opportunity to remove an undesirable protective covenant in the bond indenture. This study examines four questions related to noncallable debt issues. 1. Are there identifiable market factors that dictate when and how much noncallable debt is offered? 2. Are there factors common to the firms that issue noncallable debt that differentiate them from firms that issue callable debt? 3. Does the issuance of callable debt impact the market value of the firm's equity differently than the issuance of noncallable debt? 4. What is the value of the call option and does it vary across time? This study identifies several market and firm tendencies related to the use of noncallable debt. The answer to the third question depends upon how the debt issues were segmented. For issues in my data set, the average value of a call option is approximately 60 basis points, a value that varies across time. This mean value of the call is supported through both regression analysis and direct comparison of matched callable/noncallable debt pairs.
Thesis:
Thesis (Ph. D.)--University of Florida, 1988.
Bibliography:
Includes bibliographical references.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Richard John Kish.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
024146245 ( AlephBibNum )
19657784 ( OCLC )
AFJ9130 ( NOTIS )

Downloads

This item has the following downloads:

noncallabledebte00kish ( .pdf )

noncallabledebte00kish_Page_021.txt

noncallabledebte00kish_Page_250.txt

noncallabledebte00kish_Page_041.txt

noncallabledebte00kish_Page_065.txt

noncallabledebte00kish_Page_106.txt

noncallabledebte00kish_Page_209.txt

noncallabledebte00kish_Page_201.txt

noncallabledebte00kish_Page_241.txt

noncallabledebte00kish_Page_237.txt

noncallabledebte00kish_Page_166.txt

noncallabledebte00kish_Page_203.txt

noncallabledebte00kish_Page_218.txt

noncallabledebte00kish_Page_160.txt

noncallabledebte00kish_Page_153.txt

noncallabledebte00kish_Page_170.txt

noncallabledebte00kish_Page_076.txt

noncallabledebte00kish_Page_011.txt

noncallabledebte00kish_Page_131.txt

noncallabledebte00kish_Page_031.txt

noncallabledebte00kish_Page_235.txt

noncallabledebte00kish_Page_151.txt

noncallabledebte00kish_Page_224.txt

noncallabledebte00kish_Page_244.txt

noncallabledebte00kish_Page_173.txt

noncallabledebte00kish_Page_114.txt

noncallabledebte00kish_Page_057.txt

noncallabledebte00kish_Page_089.txt

noncallabledebte00kish_Page_229.txt

noncallabledebte00kish_Page_188.txt

noncallabledebte00kish_Page_059.txt

noncallabledebte00kish_Page_183.txt

noncallabledebte00kish_Page_220.txt

noncallabledebte00kish_Page_211.txt

noncallabledebte00kish_Page_045.txt

noncallabledebte00kish_Page_193.txt

noncallabledebte00kish_Page_247.txt

noncallabledebte00kish_Page_169.txt

noncallabledebte00kish_Page_196.txt

noncallabledebte00kish_Page_063.txt

noncallabledebte00kish_Page_100.txt

noncallabledebte00kish_Page_115.txt

noncallabledebte00kish_Page_016.txt

noncallabledebte00kish_Page_246.txt

noncallabledebte00kish_Page_124.txt

noncallabledebte00kish_Page_023.txt

noncallabledebte00kish_Page_212.txt

noncallabledebte00kish_Page_210.txt

noncallabledebte00kish_Page_111.txt

noncallabledebte00kish_Page_177.txt

noncallabledebte00kish_Page_154.txt

noncallabledebte00kish_Page_040.txt

noncallabledebte00kish_Page_072.txt

noncallabledebte00kish_Page_074.txt

noncallabledebte00kish_Page_049.txt

noncallabledebte00kish_Page_155.txt

noncallabledebte00kish_Page_010.txt

noncallabledebte00kish_Page_107.txt

noncallabledebte00kish_Page_080.txt

noncallabledebte00kish_Page_136.txt

noncallabledebte00kish_Page_190.txt

noncallabledebte00kish_Page_240.txt

noncallabledebte00kish_Page_239.txt

noncallabledebte00kish_Page_073.txt

noncallabledebte00kish_Page_028.txt

noncallabledebte00kish_Page_014.txt

noncallabledebte00kish_Page_090.txt

noncallabledebte00kish_Page_104.txt

noncallabledebte00kish_Page_071.txt

noncallabledebte00kish_Page_159.txt

noncallabledebte00kish_Page_058.txt

noncallabledebte00kish_Page_192.txt

noncallabledebte00kish_Page_147.txt

noncallabledebte00kish_Page_161.txt

noncallabledebte00kish_Page_064.txt

noncallabledebte00kish_Page_095.txt

noncallabledebte00kish_Page_219.txt

noncallabledebte00kish_Page_122.txt

noncallabledebte00kish_Page_084.txt

noncallabledebte00kish_Page_129.txt

noncallabledebte00kish_Page_009.txt

noncallabledebte00kish_Page_163.txt

noncallabledebte00kish_Page_195.txt

noncallabledebte00kish_Page_200.txt

noncallabledebte00kish_Page_156.txt

noncallabledebte00kish_Page_231.txt

noncallabledebte00kish_Page_158.txt

noncallabledebte00kish_Page_036.txt

noncallabledebte00kish_Page_042.txt

noncallabledebte00kish_Page_048.txt

noncallabledebte00kish_Page_087.txt

noncallabledebte00kish_Page_130.txt

noncallabledebte00kish_Page_043.txt

noncallabledebte00kish_Page_134.txt

noncallabledebte00kish_Page_025.txt

noncallabledebte00kish_Page_035.txt

noncallabledebte00kish_Page_061.txt

noncallabledebte00kish_Page_185.txt

noncallabledebte00kish_Page_051.txt

noncallabledebte00kish_Page_144.txt

noncallabledebte00kish_Page_050.txt

noncallabledebte00kish_Page_069.txt

noncallabledebte00kish_Page_052.txt

noncallabledebte00kish_Page_127.txt

noncallabledebte00kish_Page_217.txt

noncallabledebte00kish_Page_105.txt

noncallabledebte00kish_Page_092.txt

noncallabledebte00kish_Page_204.txt

noncallabledebte00kish_Page_020.txt

noncallabledebte00kish_Page_208.txt

noncallabledebte00kish_Page_143.txt

noncallabledebte00kish_Page_146.txt

noncallabledebte00kish_Page_202.txt

noncallabledebte00kish_Page_083.txt

noncallabledebte00kish_Page_046.txt

noncallabledebte00kish_Page_024.txt

noncallabledebte00kish_Page_013.txt

noncallabledebte00kish_Page_184.txt

noncallabledebte00kish_Page_125.txt

noncallabledebte00kish_Page_216.txt

noncallabledebte00kish_Page_226.txt

noncallabledebte00kish_Page_197.txt

noncallabledebte00kish_Page_082.txt

noncallabledebte00kish_Page_150.txt

noncallabledebte00kish_Page_109.txt

noncallabledebte00kish_Page_230.txt

noncallabledebte00kish_Page_121.txt

noncallabledebte00kish_Page_141.txt

noncallabledebte00kish_Page_101.txt

noncallabledebte00kish_Page_145.txt

noncallabledebte00kish_Page_148.txt

noncallabledebte00kish_Page_067.txt

noncallabledebte00kish_Page_133.txt

noncallabledebte00kish_Page_175.txt

noncallabledebte00kish_Page_022.txt

noncallabledebte00kish_Page_157.txt

noncallabledebte00kish_Page_243.txt

noncallabledebte00kish_Page_066.txt

noncallabledebte00kish_Page_249.txt

noncallabledebte00kish_pdf.txt

noncallabledebte00kish_Page_085.txt

noncallabledebte00kish_Page_075.txt

noncallabledebte00kish_Page_027.txt

noncallabledebte00kish_Page_179.txt

noncallabledebte00kish_Page_223.txt

noncallabledebte00kish_Page_017.txt

noncallabledebte00kish_Page_199.txt

noncallabledebte00kish_Page_181.txt

noncallabledebte00kish_Page_029.txt

noncallabledebte00kish_Page_126.txt

noncallabledebte00kish_Page_214.txt

noncallabledebte00kish_Page_098.txt

noncallabledebte00kish_Page_088.txt

noncallabledebte00kish_Page_037.txt

noncallabledebte00kish_Page_068.txt

noncallabledebte00kish_Page_070.txt

noncallabledebte00kish_Page_116.txt

noncallabledebte00kish_Page_236.txt

noncallabledebte00kish_Page_142.txt

noncallabledebte00kish_Page_189.txt

noncallabledebte00kish_Page_234.txt

noncallabledebte00kish_Page_238.txt

noncallabledebte00kish_Page_242.txt

noncallabledebte00kish_Page_222.txt

noncallabledebte00kish_Page_079.txt

noncallabledebte00kish_Page_164.txt

noncallabledebte00kish_Page_118.txt

noncallabledebte00kish_Page_135.txt

noncallabledebte00kish_Page_206.txt

noncallabledebte00kish_Page_097.txt

noncallabledebte00kish_Page_030.txt

noncallabledebte00kish_Page_032.txt

noncallabledebte00kish_Page_165.txt

noncallabledebte00kish_Page_123.txt

noncallabledebte00kish_Page_078.txt

noncallabledebte00kish_Page_245.txt

noncallabledebte00kish_Page_053.txt

noncallabledebte00kish_Page_102.txt

noncallabledebte00kish_Page_225.txt

noncallabledebte00kish_Page_039.txt

noncallabledebte00kish_Page_194.txt

noncallabledebte00kish_Page_227.txt

noncallabledebte00kish_Page_015.txt

noncallabledebte00kish_Page_119.txt

noncallabledebte00kish_Page_012.txt

noncallabledebte00kish_Page_096.txt

noncallabledebte00kish_Page_152.txt

noncallabledebte00kish_Page_180.txt

noncallabledebte00kish_Page_171.txt

noncallabledebte00kish_Page_110.txt

noncallabledebte00kish_Page_138.txt

noncallabledebte00kish_Page_026.txt

noncallabledebte00kish_Page_233.txt

noncallabledebte00kish_Page_099.txt

noncallabledebte00kish_Page_149.txt

noncallabledebte00kish_Page_047.txt

noncallabledebte00kish_Page_168.txt

noncallabledebte00kish_Page_033.txt

noncallabledebte00kish_Page_186.txt

noncallabledebte00kish_Page_034.txt

noncallabledebte00kish_Page_213.txt

noncallabledebte00kish_Page_132.txt

noncallabledebte00kish_Page_081.txt

noncallabledebte00kish_Page_103.txt

noncallabledebte00kish_Page_019.txt

noncallabledebte00kish_Page_128.txt

noncallabledebte00kish_Page_162.txt

noncallabledebte00kish_Page_228.txt

noncallabledebte00kish_Page_172.txt

noncallabledebte00kish_Page_232.txt

noncallabledebte00kish_Page_094.txt

noncallabledebte00kish_Page_120.txt

noncallabledebte00kish_Page_112.txt

noncallabledebte00kish_Page_056.txt

noncallabledebte00kish_Page_108.txt

noncallabledebte00kish_Page_038.txt

noncallabledebte00kish_Page_182.txt

noncallabledebte00kish_Page_176.txt

noncallabledebte00kish_Page_044.txt

noncallabledebte00kish_Page_191.txt

noncallabledebte00kish_Page_221.txt

noncallabledebte00kish_Page_137.txt

noncallabledebte00kish_Page_139.txt

noncallabledebte00kish_Page_062.txt

noncallabledebte00kish_Page_093.txt

noncallabledebte00kish_Page_174.txt

noncallabledebte00kish_Page_054.txt

noncallabledebte00kish_Page_113.txt

noncallabledebte00kish_Page_091.txt

noncallabledebte00kish_Page_117.txt

noncallabledebte00kish_Page_198.txt

noncallabledebte00kish_Page_205.txt

noncallabledebte00kish_Page_060.txt

noncallabledebte00kish_Page_055.txt

noncallabledebte00kish_Page_140.txt

noncallabledebte00kish_Page_086.txt

noncallabledebte00kish_Page_018.txt

noncallabledebte00kish_Page_248.txt

noncallabledebte00kish_Page_077.txt

noncallabledebte00kish_Page_167.txt

noncallabledebte00kish_Page_215.txt

noncallabledebte00kish_Page_187.txt

noncallabledebte00kish_Page_207.txt

noncallabledebte00kish_Page_178.txt


Full Text











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



NONCALLABLE DEBT: EVIDENCE AND EFFECT


By

Richard John Kish

August 1988

Chairman: Dr. Miles Livingston
Major Department: Finance, Insurance, and Real Estate

Financial theory claims that callable debt dominates

noncallable debt. Yet our evidence shows that a substantial

amount of noncallable debt exists, suggesting a deficiency

in the theory. Over the period 1977 through 1986 over $108

billion worth of new noncallable debt was offered. During

this ten year period, noncallable debt offerings accounted

for 16.9 percent of the dollar value and 12.1 percent of the

total number of all publicly placed corporate debt issued.

The dominance of callable debt has been motivated in a

variety of ways, for example uncertain interest rates. When

there is a substantial decrease in rates, callable debt

provides a firm's managers with the ability to refund an

issue and take advantage of lower interest expenses. Other

reasons offered for the dominance of callable debt include

agency problems associated with information asymmetries

ix











between borrowers and lenders, different risk tolerances of

equityholders and debtholders, the need for managers to

signal private information, differential tax rates between

borrowers and lenders of funds, maturity preferences, and

the opportunity to remove an undesirable protective covenant

in the bond indenture.

This study examines four questions related to

noncallable debt issues.

1. Are there identifiable market factors that

dictate when and how much noncallable debt

is offered?

2. Are there factors common to the firms that

issue noncallable debt that differentiate

them from firms that issue callable debt?

3. Does the issuance of callable debt impact the

market value of the firm's equity differently

than the issuance of noncallable debt?

4. What is the value of the call option and does

it vary across time?

This study identifies several market and firm

tendencies related to the use of noncallable debt. The

answer to tne third question depends upon how the debt

issues were segmented. For issues in my data set, the

average value of a call option is approximately 60 basis

points, a value that varies across time. This mean value of

the call is supported through both regression analysis and

direct comparison of matched callable/noncallable debt pairs.

x















CHAPTER I
INTRODUCTION


Financial theory claims that callable debt dominates

noncallable debt. Yet our evidence shows that a substantial

amount of noncallable debt exists, suggesting a deficiency

in the theory. Over the period 1977 through 1986 over $108

billion worth of new noncallable debt was offered. During

this ten year period, noncallable debt offerings accounted

for 16.9 percent of the dollar value and 12.1 percent of the

total number of all publicly placed corporate debt

issued.1

The dominance of callable debt has been motivated in a

variety of ways, for example uncertain interest rates. When

there is a substantial decrease in rates, callable debt

provides a firm's managers with the ability to refund an

issue and take advantage of lower interest expenses.2

Other reasons offered for the dominance of callable debt

include agency problems associated with information

asymmetries between borrowers and lenders, different risk

tolerances of equityholders and debtholders, the need for

managers to signal private information, differential tax

rates between the borrower and the lender of funds, maturity

preferences, and the opportunity to remove an undesirable

protective covenant in the bond indenture.3

1











The significant amount of noncallable debt issued

during the period 1977 through 1986 warrants examination of

four questions.

1. Are there identifiable market factors that

dictate when and how much noncallable debt

is offered?

2. Are there factors common to the firms that

issue noncallable debt that differentiate

them from firms that issue callable debt?

3. Does the issuance of callable debt impact the

market value of the firm's equity differently

than the issuance of noncallable debt?

4. What is the value of the call option and does

it vary across time?

The basic concepts used throughout this study are

summarized below.

Definition of a call. Callability gives the firm's

management the right to redeem outstanding bonds before

maturity at a stated price. Without a call provision, debt

can only be retired through an open market purchase.

Elements of a call provision. A typical call

provision includes (1) a statement that the bond issue is

redeemable in whole or in part at the option of the issuer,

(2) the date(s) that the call option is exercisable, (3) the

price the issuer is required to pay the bondholder for early

redemption, and (4) any restrictions that are imposed on the










early redemption, i.e. refunding versus redeeming

restrictions.4

Call restrictions. Nearly all callable debt has

deferments and restrictions on refunding and redeeming. A

call deferment is a restriction in the indenture prohibiting

the bond from being called prior to a specified date.

Deferment constrains the lender, which reduces the value of

the call. Deferments are typically 10 years for corporate,

5 years for utilities, and calls are restricted to the 5

years prior to maturity for long-term government issues.

However this is a recent phenomenon; prior to 1958 call

options were almost always unrestricted, i.e. exercisable by

the issuer the day after the initial sale of the bonds.

A nonredeemable issue restricts the issuer from

exercising the call option under any circumstances for the

time specified by the call deferment. A nonrefundable issue

prohibits bond replacements financed by funds borrowed at a

lower cost. For example, Bristol-Meyers Company called $25

million of its $75 million nonrefundable 8 5/8% debentures

in 1973 using internally generated funds. The indenture

restricted refunding before 1980, but no restrictions where

placed upon redeeming the issue if funds became available

from a source other than floating a lower coupon bond.5

The following chapter reviews the literature on

callable bonds. Summary statistics follow in Chapter III.

A logit model based upon the tendencies found in the summary











statistics for predicting whether a firm should issue

callable or noncallable debt is provided in Chapter IV.

Chapter V tests the effects of debt issues on the firm's

equity value. A measure of the value of the call option is

provided in Chapter VI. Finally, Chapter VII ties the

analysis together in a conclusion.


Notes

1. Figures calculated from debt data provided by Moody's
Bond Survey summary pages for Corporate Bond Offerings for
the years 1977 through 1986.

2. For example Pye (1966), Elton and Gruber (1972), Bodie
and Friedman (1978), Bodie and Taggart (1978, 1980), and Van
Home (1980, 1984) all list the uncertainty of interest
rates as tha primary reason for the existence of callable
bonds.

3. For example see Barnea, Haugen, and Senbet (1980) for
agency costs associated with asymmetric information, risk
taking, and growth; Bodie and Friedman (1978) for risk
taking; Ross (1977), Leland and Pyle (1976), and Flannery
(1982) for signaling; Boyce and Kalotay (1979b), Van Home
(1984), and Marshall and Yawitz (1980) for taxes; Robbins
and Schatzberg (1986) for maturity; Kidwell (1976) on
standard indentures; and Pye (1966) on restrictive
convenants.

4. Both Hess and Winn (1962) and Bodie and Friedman (1978)
provide extended discussions on the call option.

5. Boyce and Kalotay (1979b) provide additional details on
call restriction.















CHAPTER II
LITERATURE REVIEW


Most research on corporate finance assumes that

noncallable debt is an insignificant portion of the debt

market. For example, Barnea, Haugen, and Senbet (1980)

claim that almost all corporate debt is callable. This

"almost all" statement is made without substantiation.

Other examples are plentiful. Boyce and Kalotay (1979b) and

Van Horne (1984) both claim that virtually all corporate

bonds have a call feature written into the indenture

agreement. Thus, much of the literature addresses the

question of why corporations almost never issue noncallable

long-term bonds. For example, see Bodie and Taggart

(1979).1

Several quotes reinforce the assumed insignificance of

noncallable debt. "Throughout the twentieth century,

corporate bonds issued in the United States have not been

pure bonds; in almost all cases the security purchased by an

investor has consisted of a pure bond less an option,

retained by the issuer, to call the bond at a specified

price" (Bodie and Friedman, 1978, p.20). "Most attention

[within corporate financing research] centers on the almost

invariable inclusion in corporate bonds of a call" (Boyce

and Kalotay, 1979, p. 825). Robbins and Schatzberg state

5











that "financial theory has found only moderate success in

explaining the routine inclusion of call provisions among

the covenants of corporate bonds [and later declares

that there exists a] near universality of call

provisions in corporate bond contracts" (Robbins and

Schatzberg, 1986, p. 935).

The insignificance assumption is also common in the

leading finance textbooks. The following quotes offer a

representative sample of this treatment of noncallable

debt. Brigham contends that "Most bonds contain a call

provision which gives the issuing corporation the right to

call the bonds for redemption" (Brigham, 1986, p. 382 and

Weston and Brigham, 1987, p. 710). Ross and Westerfield

state that "Almost all publicly issued corporate long-term

debt is callable" (Ross and Westerfield, 1988, p. 333).

This is not just a recent phenomenon; an early edition of

Van Horne's textbook states that "Nearly all corporate bond

issues provide for a call feature, which gives the

corporation the option to buy back bonds at a stated price

before their maturity" (Van Horne, 1977, p. 570).

Investment texts also adhere to this viewpoint. For

instance, Radcliffe writes that "Long-term corporate debt

obligations are usually term bonds with maturities of five

years or more [and that] most corporate bonds are sold

with a deferred call provision" (Radcliffe, 1987, p. 38).











Call Option Hypotheses

The existing literature suggests that the use of the

call option is motivated by (at least) four separate, but

not mutually exclusive, hypotheses: flexibility, agency,

tax, and indenture.

Flexibility Hypothesis

The flexibility hypothesis states that the call

provision provides the firm's management with increased

flexibility in the face of market uncertainties, primarily

interest rate uncertainty. The call option allows the

issuer to replace a higher cost obligation with a lower cost

bond as interest rates decline.2 Refunding is profitable

provided the interest savings outweigh the call premium,

flotation costs, and legal expenses.

In an efficient market, interest savings are priced

out. From the debtholder's viewpoint, callable debt

consists of a long position in a noncallable debt security

and a short position in a call option retained by the

equityholders. The value of the call option to the

debtholder equals the value to the issuer. Therefore at

issuance, the market price reflects the expected value of

the call privilege.3

There seems no reason, a priori, that corporate

management is better able to forecast future interest rates

than debt buyers, a majority of which are professional money

managers for financial institutions, pension funds, and life










insurance companies. Thus, interest cost reduction as a

reason for the routine inclusion of the call option rests on

a tenuous assumption.4

Refunding allows management to extend or shorten the

maturity of its debt.5 Other flexibility benefits include

the opportunity to simplify financial structure, to reduce

the debt to equity ratio of the firm, to lengthen the firm's

average debt maturity, to eliminate indenture restrictions,

or to retire unwanted bonds when a corporation accumulates

large cash balances for which it has no immediate needs.6

However, these benefits can be obtained by purchasing bonds

in the open market.

Ross (1977) justifies the call option as a tool

available to management for altering the capital structure

of the firm. A counterview, offered by Stiglitz (1979), is

that financial policies are irrelevant provided the

investment opportunity set remains unchanged by financial

policy. Thus, the call option is redundant. The basic

arguments of the flexibility hypothesis by author are

summarized in Table 2-1.

Agency Hypothesis

The above arguments for the existence of call options

are constructed assuming symmetric information across

claimholders. Not surprisingly, it is difficult to motivate

call options in a perfect market setting. If capital

structure is irrelevant in such a framework, then the











Table 2-1
Proponents of the Flexibility Hypothesis


Author (Date)
Claim (Variables tested)


Bowlin (1966), Jen and Wert (1967), and Pye (1966)
The use of callable debt is related to a potential for
interest savings sometime in the future. Periods with
high rates should see a proportionate increase in the
number of calls issued. (As TSEC, SPREAD, and UNCER
increase the use of the call feature should also
increase.)

Ross (1977)
The call option allows the manager to alter the
capital structure of the firm. (High DA and DE ratios
should be associated with callable debt.)

Stiglitz (1979)
Financial policies are irrelevant even with risky
debt, provided the investment opportunity set remains
unchanged by financial policy. The maturity structure
of debt as well as the other complexities, such as the
call option, do not affect firm value. Predicts no
abnormal change in the equity value of the firm when
debt is issued. (The CAR's of the firm should equal 0
when debt is issued.)

Van Home (1984)
The use of the call option is directly related to the
interest rate cycle. During periods with low interest
rates, there is no need for the automatic inclusion of
the call option on debt issues. (As MAT, TSEC,
SPREAD, and UNCER increase the use of the call feature
should also increase.)


Note: Symbols are defined in the preface and within each
chapter when the variable is used.









10
specifics of debt contracts cannot matter. See for example

Modigliani and Miller (1958) and Krause (1977).

However, in the presence of private action and private

information, it is well known that capital structure and the

specific nature of debt contracts effect the value of the

firm. Call options can be motivated as an endogenous

contractual response to agency problems between debt and

equity claimants.

Call options mitigate some of the self-serving

incentives of equityholders in the levered firm. Risky

debtholders broaden their collateral base anytime a firm

improves its position. Therefore, when the firm makes a

profitable investment, part of the benefit goes to

debtholders. Since equityholders are unable to reap the

full benefits of additional investments, certain positive

net present value projects are not undertaken.7 The

agency cost of debt in this case is the foregone projects.

If the bonds are callable, the debtholders' gain when

firm value rises is limited to the call price. Thus,

investment incentives are more consistent with firm value

maximization. The potential gain to bondholders is fixed

and equity captures the full marginal benefit of any new

project.

Barnea, Haugen, and Senbet (1980) point out that

callable debt also mitigates incentives to increase risk

beyond the levels consistent with firm value maximization in









11
order to shift wealth from debtholders to equityholders. As

wealth is shifted away from debtholders, the value of the

securities declines, and so does equities option to call.

As a result, any wealth shifts are partially offset by

declines in the value of the call. However, debt with the

option to convert, completely mitigates equities incentives

to shift risk, casting some doubt on the role of call

options in mitigating incentives to shift risk.

In an empirical study, Thatcher (1985) shows that

refunding protection offers the best solution to the agency

costs associated with risky debt. Support is offered in the

form of simple means tests and a discriminant analysis on

the difference between firms that issue debt with a regular

call feature versus those that use a two tiered call

provision. Other claims made within the Thatcher study are

that default risk is a major reason for the inclusion of the

call option and that firms with high debt to asset ratios

have an increased probability of using the call feature.

It should be noted that short-term debt provides

investment incentives similar to callable debt. With

short-term debt, the terms of the debt contract are

frequently renegotiated to reflect the firm's current

investment policy.8 The essential difference between

short-term debt and long-term callable debt arises only with

respect to uncertainty about the date of prepayment. The

extent to which these kinds of securities are substitutes is

examined in detail within Chapter VI.









12
Flannery (1986) develops a model where managers signal

private information about the value of the firm's assets

through the choice of debt maturity. Low private valuation,

"type" firms prefer to issue long-term debt to avoid the

transaction costs and the revaluation of credit worthiness

that occurs with each rollover. High valuation types,

desiring to reveal their true worth, are willing to pay the

transaction costs associated with rolling over short-term

debt in order to avoid being pooled with low quality firms

and being forced to have a higher cost of debt. Short-term

noncallable or long-term callable debt are credible signals

of high firm quality. While Flannery's model does not cast

light on the issue of which firms issue callable versus

short-term noncallable debt, it does predict that the

issuance of short-term noncallable or callable debt will be

viewed more favorably by the capital market then issuance of

long-term straight debt.9

In a signaling model hinging on managerial contracts,

Robbins and Schatzberg (1986) also find that short-term debt

is a signal for the "good" prospects of a firm, but that

long-term callable debt has superior risk sharing

attributes. The following claim is supported by a numerical

example, "The manager of a firm with Good News, in signaling

the firm's better prospects through callable bonds, can

achieve a reduction in the risk to his or her compensation

compared to what can be achieved through the issuance of











short-term debt" (Robbins and Schatzberg, 1986, p. 945).

They further argue that both noncallable debt and equity are

dominated securities and therefore signal bad news.

Research dealing with agency costs as an explanation

for the existence of the call option is summarized in Table

2-2 by author.

Tax Hypothesis

According to Boyce and Kalotay (1979) both the issuer

and the buyer benefit from callable debt at the expense of

the government. The difference in the average tax rates

between a profitable corporate borrower (high marginal tax

rate) and the typical lender (low marginal tax rate)

generates a preference for callable bonds. The exercise of

the call results in a reduction of the tax liability of the

issuer which is not offset by the additional taxes paid by

the lender.

The marginal corporate tax rate during the time of the

Boyce and Kalotay study was approximately 50%, whereas the

dominate lenders during the same time period had marginal

tax rates less than 50%. Corporate bondholders are

typically tax-exempt (pension funds) or in the low tax

brackets (life insurance companies).

Marshall and Yawitz (1980) support the tax argument by

pointing out that call premiums are deductible from ordinary

income as an expense to the borrower, but treated by the

lender as a capital gain. The change in the current tax











Table 2-2
Proponents of the Agency Hypothesis


Author (Date)
Claim (Variables Tested)


Barnea, Haugen, and Senbet (1980)
Callable debt is used as a means of resolving agency
problems associated with informational asymmetry,
managerial risk incentives, and foregone growth
opportunities. Shortening the maturity of debt and
issuing long-term debt with a call option perform
identical tasks in eliminating agency problems. (The
issuance of short-term debt should produce positive
CAR's.)

Barnea, Haugen, and Senbet (1985)
Noncallable debt is harmful to the equityholders of
the firm. (The issuance of noncallable debt should
produce negative CAR's and the issuance of callable
debt should produce positive CAR's.)

Bodie and Taggart (1978)
Through modeling, the authors show the dominance of
callable debt to both equity and noncallable debt.
Predicts that growth opportunities may give
equityholders a definite preference for callable
debt. (As PREGROWTH and POSTGROWTH increase, the use
of the call feature should also increase.)

Flannery (1986)
Using the signaling argument, the author models the
fact that good firms use short-term debt as a signal
that they are in fact good firms as transaction costs
increase. (As CPROFIT, PM, INTCOV and RATE increase
the use of the call feature on debt issues should also
increase. Short-term debt and long-term callable debt
should produce positive CAR's.)

Robbins and Schatzberg (1986)
Shows through modeling that callable debt dominates
short-term debt, equity, and noncallable debt of any
maturity length. (The CAR's from long term callable
debt should be greater than the CAR's from short-term
callable debt. The CAR's from noncallable debt should
signal bad news and therefore should be negative.)










15
Table 2-2--continued


Author (Date)
Claim (Variables Tested)


Thatcher (1985)
Callable debt is used as a method for reducing agency
costs due to future investment opportunities.
Predicts that high growth firms issue callable debt.
Also predicts that firms with high debt ratios, high
concentrations of long-term debt, and high probability
of default dictate the use of the call option,
specifically the two tiered call option. (High
PREGROWTH, POSTGROWTH, DA, DE, and low PM should
increase the use of callable debt.)


Note: Symbols are defined in the preface and within each
chapter when the variable is used.










laws weakens the tax motivation. Since a difference still

exists between the corporate and individual tax rates, the

tax hypothesis is not without some merit. The tax arguments

are summarized in Table 2-3.

Indenture Hypothesis

The indenture hypothesis cites standardization as a

major factor for the inclusion of the call option by most

major corporations.10 Standard indentures enhance a

bond's liquidity and therefore increase its value to the

buyer. Common features cited historically include deferment

periods of 10 years for industrials including banks and

finance companies, 5 years for utilities, and restricted use

of calls to the last 5 years of maturities for long-term

government issues.

Kidwell (1976), studying the use of the call option by

state and local governments, found that the inclusion of the

call provision was determined by precedents established in

the market place, statutory requirements, and the historical

tradition of issuers. The expected economic savings

resulting from lower interest rate levels was not a

significant factor. Interest rate levels did not appear to

affect the decision to include a call provision in a bond

issue. Whether these conclusions can be applied to the

corporate sector is discussed in later chapters. A summary

of Kidwell's position related to the indenture hypothesis

can be found in Table 2-4.











Table 2-3
Proponents of the Tax Hypothesis


Author (Date)
Claim (Variables Tested)


Boyce and Kalotay (1979a, 1979b)
The tax effect, which results from the difference in
marginal tax rates between a profitable corporate
borrower and the typical lender dictates the use of
callable debt. (High ATAX and MTAX dictates the use
of callable debt.)

Marshall and Yawitz (1980)
The U.S. tax laws create a bias in favor of the
inclusion of call provisions on corporate debt,
therefore firms issuing callable debt should have a
higher marginal tax rate than firms issuing
noncallable debt. (High ATAX and MTAX dictates the
use of callable debt.)


Note: Symbols are defined in the preface and within each
chapter when the variable is used.










18
Table 2-4
Proponent of the Indenture Hypothesis


Author (Date)
Claim (Variables Tested)


Kidwell (1976)
Callable debt is issued with greater frequency as the
length of the call deferment period increases and as
interest rates increase. (Callable debt issues
increase as the length of the call deferment
increases. High TSEC is not a factor in the issuance
of noncallable debt.)


Note: Symbols are defined in the preface and within each
chapter when the variable is used.











Call Deferments and Subtitution

Previous research has examined the determinants of

differences in the value of the call option across various

issues. For example, the interest rate environment, coupon

rate, and length of deferment period have been examined as

determinants of the value of the call option. A question

also arises concerning the "maturity" of a callable bond.

When the bond is called the actual, realized maturity is

less then the stated maturity. The effective maturity the

market places on the bond is determined by comparing the

realized yields on callable bonds with the realized yields

of straight debt. If the realized yields are the same, the

bonds are close substitutes.

Value of Deferment

The value to the bondholder of deferment protection

has been estimated in three ways. First, various investment

bankers and regulatory agencies opinions about the value of

deferment were surveyed. Second, simple yield comparisons

were made across bonds with different periods. Finally,

regression techniques are used to more accurately control

for nondifferment effects on yields.

Hess and Winn (1962), in two call deferment surveys,

reveal conflicting results. The first survey included life

insurance companies, banks, and trust companies. The

question asked was what value should be placed on a call

deferment attached to a 5 1/4% 30 year Aa rated utility











bond. The median values place on 5 and 10 year deferments

were 15 and 25 basis points respectively. For a

nonrefundable feature, the estimated cost was 30 to 36 basis

points.

The second survey involved three federal regulatory

agencies: the Securities and Exchange Commission, the

Federal Power Commission, and the Interstate Commerce

Commission. The question addressed in this survey dealt

with the cost of including a call feature that was

immediately callable. The consensus was that no substantial

evidence exists that immediately callable bonds had higher

yields then bonds with a deferment period.

Simple comparisons of the bond market from 1926

through 1959 failed to disclose any significant relationship

between the offering yield and the call features of

corporate bonds. During the period 1926 through 1943, Hess

and Winn analyzed 572 corporate bonds of which 5 were

noncallable and 11 had call deferments of 5 years or more.

Comparing immediate callables with bonds having call

deferments of one year or more failed to reveal any value

placed on the call privilege by the debt market. A similar

finding was found during their study of 332 corporate bonds

issued during the period 1944 through 1955. Of the 332

issues only 13 contained call deferments, 12 of which were

private placements and not rated.









21
The third phase of the Hess and Winn study covered the

period 1956 through 1959. During this 4 year phase, debt

issues were restricted to publicly issued industrial and

utility bonds rated A or better in issue amounts of $5

million or larger. A total of 361 debt issues met this set

of criteria, of which 105 contained call deferments. Again

they failed to find a significant value for the call

option. Only in the latter half of 1959 was a detectable

value for the call option uncovered. During the last half

of 1959, bonds with 5 year call deferments carried yields 13

to 20 basis points lower than freely callable issues with

the same ratings issued during approximately the same time

period.

The third method for measuring the magnitude of the

yield differential required for call deferments of various

lengths was the focus of Pye's regression analysis in 1976.

Yield was regressed against a set of dummy variables for

time, 5 year call deferment, and ratings (Aa and A with Aaa

used as the base case). The only other independent variable

was maturity. The results revealed an approximate cost of 4

to 13 basis points of the 5 year call deferment during low

(2 3/4% 4%) and high (4% 5 1/4%) interest rate levels

respectively for the period 1959 through 1966.

Pye's regression results were reinforced with a

simulation based on a transitive probability matrix of one

year interest rates from 1900 to 1966. The model relied









22
upon two basic assumptions,(1) that investors rank lotteries

by their expected values and (2) that the one period

interest rate follows a Markov chain. Estimated costs on 5

year call deferments were similar to his direct

estimations--5 to 10 basis points. Projections for longer

deferments ranged from 40 to 70 basis points for low (2

1/8%) and high (5 7/8%) interest rate periods respectively.

Substitution

Substitution is the opportunity to exchange debt

issues without a material loss in realizable yields. When

analyzing the preference in maturity structure, Jen and Wert

(1967) found that the difference in the promised yields

between long-term callable and short-term noncallable debt

is explained by the risk difference in the possibility of

the call. Although a difference in the promised yields was

detected, the realized yield differences were negligible.

This lack of difference in the realized yields on long-term

callable and short-term noncallable debt support the

substitution of the two debt types.

Bodie and Friedman (1978) note that intermediate term

notes, which are typically not callable until maturity,

broaden the range of an investor's selection. These

noncallable issues often offer maturities which equal the

periods of call protection for the callable bonds, thereby

offering alternatives to long-term callable bonds. The key

to the substitution effect is that even though the promised










yields at the time of issuance are higher for the callable

bonds, the realized yields are essentially identical. Bodie

and Friedman's position is summarized in Table 2-5.

Two differences in the use of short- and long-term

debt that might mitigate their substitutability for issuers

are refinancing and rollover risks. With short-term debt,

several refinancing may be necessary to fund a project.

Each refunding includes flotation costs; therefore the total

flotation costs of rolling over several short-term debt

offerings may be higher than that associated with a single

long-term debt issue. The use of short-term debt in

anticipation of refinancing at lower rates runs the risk of

an incorrect forecast. If rates increase, refinancing with

short-term debt must take place at higher rates. Long-term

callable debt provides protection against an increase in

rates by locking in the borrowing rate for the maturity of

the debt instrument.

Summary

Key attributes associated with the call option include

increased flexibility, a reduction of agency costs, tax

benefits, and the use of standard indentures. The central

flexibility argument deals with the uncertainty of interest

rates, but simplifying financial structure, changing debt

maturities, eliminating indenture restrictions, and retiring

unwanted bonds also have merit. Agency arguments revolve

around the opportunity to mitigate the self-serving










24
Table 2-5
Proponent of the Substitution Effect


Author (Date)
Claim (Variables Tested)


Bodie and Friedman (1978)
Short and intermediate term debt issues are
substitutes for long term callable debt. (Average
call deferments should be approximately equal to the
average maturities of noncallable debt issues.


Note: Symbols are defined in the preface and within each
chapter when the variable is used.









25
incentives of equityholders in a levered firm and to signal

private information to potential investors. If tax

differentials exist, then the issuance of callable debt

becomes a positive sum gain between the borrower and lender

of funds. The gains are at the expense of the government in

the form of reduced taxes. Finally, standard indentures

established by precedents in the market place, statutory

requirements, or historical tradition of issuers are also

hypothesized as a reason for the inclusion of the call

option. A summary of the various viewpoints, as well as the

aspect of the bond issue most effected by their position, is

provided in Table 2-6. Testable implications are outlined

within the next three chapters.

The literature review outlined three methods used in

prior research for valuing the call deferment. They

included surveys, matched pair comparisons, and regression

analysis. Matched pairs of callable and noncallable debt

will be used in Chapter IV to estimate the value of the call

option, thus eliminating the need to extrapolate the cost of

the call option from call deferments. This value is

supported by regression analysis. Finally, the substitution

of long-term callable and short-term noncallable debt issues

was suggested based upon the realizable yields available on

both sets of debt instruments.











Table 2-6
Variables to Test Hypothesized Explanations
For the Call Option

Authors (Date) Tablea: Variables Tested

Barnea, Haugen, and Senbet (1980) A: 1, 2
Barnea, Haugen, and Senbet (1985) A: 1
Bodie and Friedman (1978) S: 6
Bodie and Taggart (1978) A: 10
Boyce and Kalotay (1979a, 1979b) T: 15
Flannery (1986) A: 1, 7, 11, 12
Jen and Wert (1967) F: 14, 17, 17
Kidwell (1976) I: 6, 14, 16
Marshall and Yawitz (1980) T: 15
Pye (1966) F: 14, 16, 17
Robbins and Schatzberg (1986) A: 1
Ross (1977) A: 4, 5
Stiglitz (1979) F: 1
Thatcher (1985) A: 4, 5, 8, 10, 11
Van Home (1984) F: 9, 14, 16, 17

aTable under which additional information can be found:
F=Flexibility (Table 2-1), A=Agency (Table 2-2),
T=Tax (Table 2-3), I=Indenture (Table 2-4), and
S=Substitution (Table 2-5)
Variable tested under the authors) hypothesized claim:
1 = CAR (Cumulative average excess returns)
2 = CLASS (Firms classification)
3 = CPROFIT (Percentage change in profits)
4 = DA (Debt to asset ratio)
5 = DE (Debt to equity ratio)
6 = DEF (Call deferment)
7 = INTCOV (Interest coverage)
8 = MARKET (Measure of the relationship between the
existing debt issue and current debt outstanding)
9 = MAT (Maturity)
10 = GROWTH (PREGROWTH-Average growth in the 3 years prior
to the debt issue and POSTGROWTH-Average growth in the
year of the debt issue)
11 = PM (Profit margin)
12 = RATE (Debt rating class)
13 = SIZE (Natural logarithm of the dollar amount of the
debt issue)
14 = SPREAD (The difference between the yield on the debt
issue and a comparable treasury issue)
15 = TAX (Corporate tax rate)
16 = TSEC (Treasury security yield)
17 = UNCER (Change in interest rates over the 3 weeks prior
to the debt issue date)











Notes
1. Marshall and Yawitz (1980) find that the majority of
research on bond refunding is concerned with explaining why
the call provisions are normally included on corporate bond
issues.

2. For example, see Bowlin (1966), Pye (1966, 1976), Jen
and Wert (1967), and Van Horne (1980, 1984).

3. See Barnea, Haugen, and Senbet (1985), Kraus (1973), and
Myers (1971) for the effects of callable debt in efficient
markets.

4. A weakness pointed out by Kraus (1973) is that a policy
of issuing only noncallable bonds is as good as that of
issuing callable bonds and subsequently following an optimal
refunding policy within informationally efficient markets.
Bodie and Taggart (1978) state that in an efficient capital
market, callable and noncallable corporate bonds should
coexist with the price differential between them reflecting
the value of the call option. Note that even if management
were systematically smarter, the market would fail.

5. Elton and Gruber (1972) find that if the new issue has a
later maturity than the issue it is refunding, then the new
issue assures the borrower funds for a longer period of time
at a lower interest rate.

6. Cash surpluses could result from a number of
circumstances such as exceptionally profitable operations,
compulsory sale of property to a public body or to satisfy
requirements for mergers, loss of insured property through
catastrophe, and liquidation of inventories and
receivables. In the case of a finance company, cash may
accumulate because of restrictions on installment credit or
similar restrictions.

7. An example of the externalities caused under the agency
heading is the manner in which growth opportunities are
handled. Bodie and Taggart (1978) suggest that when a firm
has future discretionary investment opportunities, the call
feature may not be a zero sum game between the debtholders
and the equityholders. Risky debtholders participate in
changes in the firm's fortunes.

8. Leland and Pyle (1976) use the entrepreneur's equity
stake as a signal. The greater the equity stake of the
entrepreneur, the more reliable the signal that the firm is
"good". See Campbell and Kracaw (1980), Ross (1977), and
Leland and Pyle (1977) for resolution of the moral hazard
problem. See Allen, Lamy, and Thompson (1987) for a










28
discussion of relationship between the information revealed
by management and the perceived riskiness of the firm's
debt.

9. See Myers and Majluf (1984) for a summary of related
signaling arguments.

10. See Smith and Warner (1979) for a discussion on bond
covenants.















CHAPTER III
SUMMARY STATISTICS


Evidence suggests that noncallable debt has been

erroneously overlooked under the mistaken notion that it

accounts for an insignificant portion of the debt market.

The percentage of issues that are noncallable during the ten

year period, 1977 through 1986, ranges from 2.9% to over 20%

(average 12.1%). As a percentage of the dollar amount of

issues during the same time period, the range is even higher

from 3.5% to over 25% (average 16.9%). The bar graphs in

Figures 3-1 and 3-2 illustrate the proportion of noncallable

debt measured by the number and the dollar amount of issues

during the ten year sample period.

The dollar amount of noncallable debt is substantial,

ranging from a low of $900 million in 1979 to over $36

billion in 1986 (average $10.8 billion). The yearly dollar

amounts for both the callable and the noncallable debt

issues are graphed in Figure 3-3. The majority, but not

all, of this noncallable debt has short to intermediate term

maturities. Figure 3-4 graphs the frequency distributions

for the maturity levels of both callable and noncallable

issues over for the total sample period.

At this point we examine the following two questions:

(1) Are there identifiable market factors that dictate when

29















21%
20% -
19%-
18%
17%
16%
15%
S 14%
S 13%
12%
1 1%7
10% /


0 8%
7%
6%
5%
4%
3% /// / //

2%
1% /// / //

0%
1977 1978 1979 1980 1981 1982 1983 1984 1985 1986

Year Debt Issued





Figure 3-1
Noncallable Debt
As a Percentage of the Total Debt Issues
During the Ten Year Period 1977 1986


Source: Percent values based upon debt issues reported in
Moody's Bond Survey (1977-1986).














26% -

24% -

22% -

20%/

18%

16%
n 6/ / / / /// ///
S 14%/

o 12%

S10%-



6%

4%

2% //


1977 1978 1979 1980 1981 1982 1983 1984 1985 1986

Year Debt Issued





Figure 3-2
Noncallable Debt
As a Percentage of the Total Dollar Debt Issued
During the Ten Year Period 1976 1986


Source: Percent values based upon debt issues reported in
Moody's Bond Survey (1977-1986)





































1977 1978 1979 1980 1981 1982 1983 1984 1985 1986
Year Debt Issued




Figure 3-3
Dollar Amount of Debt Issues by Year
Callable versus Noncallable Debt


Note: //// Callable Debt
\\\\ Noncallable Debt

Soruce: Dollar amounts of debt based upon debt issues
reported in Moody's Bond Survey (1977-1986).



































5- 9.9


10-14 9


15-19.9


20-24.9


I2-
25-29.9


Maturity in Years




Figure 3-4
Maturity Distribution of Debt Issues 1977-1986
Callable versus Noncallable Debt


Note: //// callable Debt
\\\\ Noncallable Debt


Source: Maturity frequencies compiled from Moody's Bond
Survey (1977-1986)


>= 30


\\ \ //\\1 v/
\ //\x //

v \l //\\! ; //

\\ Ill // \ /

\\/\\ //1\'. V/ // / //
,-7 //\/\\/ /// //
// ',-" \ -"\ / \1 // //









34
and how much noncallable debt is offered? and (2) Are there

factors common to the firms that issue noncallable debt that

differentiate them from firms that issue callable debt? The

analysis of the tendencies identified is continued within

the logit regression found in Chapter IV.

Testable Implications

The testable implications for callable/noncallable

debt issues follow from Table 2-6, which is reproduced as

Table 3-1. This summary of the authors discussed in the

literature review contains a list of the variables needed

for testing the call option hypotheses.

Flexibility Tests

Tests of the flexibility hypothesis focus on the level

of interest rates as well as their volatility. As the level

or volatility of rates increase, the value and hence the

popularity of the call option should increase. A simple

test of this hypothesis is to see whether there is an

increase in the proportion of callable versus noncallable

debt issued during periods of high uncertainty.

Volatility is proxied by the variability of interest

rates during the three weeks prior to the debt issue. The

proxy for the level of interest rates is the yield on a

fixed maturity Treasury bill. If this yield increases, the

use of the call option should also increase. The spread

between the 3 and 30 year Treasury bills indicates the shape

of the yield curve. If the long-term Treasury is greater











Table 3-1
Variables to Test Hypothesized Explanations
For the Call Option

Authors (Date) Tablea: Variables Tested

Barnea, Haugen, and Senbet (1980) A: 1, 2
Barnea, Haugen, and Senbet (1985) A: 1
Bodie and Friedman (1978) S: 6
Bodie and Taggart (1978) A: 10
Boyce and Kalotay (1979a, 1979b) T: 15
Flannery (1986) A: 1, 7, 11, 12
Jen and Wert (1967) F: 14, 17, 17
Kidwell (1976) I: 6, 14, 16
Marshall and Yawitz (1980) T: 15
Pye (1966) F: 14, 16, 17
Robbins and Schatzberg (1986) A: 1
Ross (1977) A: 4, 5
Stiglitz (1979) F: 1
Thatcher (1985) A: 4, 5, 8, 10, 11
Van Horne (1984) F: 9, 14, 16, 17

aTable under which additional information can be found:
F=Flexibility (Table 2-1), A=Agency (Table 2-2),
T=Tax (Table 2-3), I=Indenture (Table 2-4), and
S=Substitution (Table 2-5)
Variable tested under the authors) hypothesized claim:
1 = CAR (Cumulative average excess returns)
2 = CLASS (Firms classification)
3 = CPROFIT (Percentage change in profits)
4 = DA (Debt to asset ratio)
5 = DE (Debt to equity ratio)
6 = DEF (Call deferment)
7 = INTCOV (Interest coverage)
8 = MARKET (Measure of the relationship between the
existing debt issue and current debt outstanding)
9 = MAT (Maturity)
10 = GROWTH (PREGROWTH-Average growth in the 3 years prior
to the debt issue and POSTGROWTH-Average growth in the
year of the debt issue)
11 = PM (Profit margin)
12 = RATE (Debt rating class)
13 = SIZE (Natural logarithm of the dollar amount of the
debt issue)
14 = SPREAD (The difference between the yield on the debt
issue and a comparable treasury issue)
15 = TAX (Corporate tax rate)
16 = TSEC (Treasury security yield)
17 = UNCER (Change in interest rates over the 3 weeks prior
to the debt issue date)









36
(smaller) than the short-term Treasury, then the yield curve

is rising (falling). If the yield curve is falling

(rising), the market anticipates interest rates will fall

(rise) in the future.

Agency Tests

Callable bonds, as opposed to straight debt, provide

equityholders incentives to maximize the total value of the

firm with respect to investment decisions. In particular,

callability internalizes the external benefits to risky debt

holders when a firm undertakes a value enhancing project.

This implies an advantage to using callable debt.

The likelihood that the call option will be used

depends on the following variables. First, the more risky

or lower rated the debt, the more bondholders can gain in

the absence of a call option. Second, the lower the firm's

profit level, the greater the possibility for improving the

firm's position. Third, when a great deal of the value of

firms with potentially high growth lies in future projects,

the greater the call option's value. Finally, firms that

are highly levered, in addition to the increased likelihood

that they have low debt ratings, have a larger co-insurance

problem because they have a larger amount of debt

outstanding.

Flannery (1986) predicts that there is information

content in the choice of debt financing. Firms with good

private information issue short-term or callable debt while









37
firms with bad private information issue long-term straight

debt. An ex-post verification of the signaling hypothesis

occurs if a firm's profitability increases after the

issuance of callable and short-term debt versus long-term

noncallable debt. Later, the profitability results are

compared with the stock price announcement effects.

Tax Test

Tax motivations for callable debt suggest that the

marginal and average tax rates of firms issuing callable

debt should be higher then those issuing noncallable debt.

Issuers are concerned about expected tax rates over the life

of the issue. Current tax rates proxy the firm's expected

tax rates.

Substitution Test

Certain arguments suggest that noncallable debt is a

substitute for callable debt. The maturity on the

noncallable debt is hypothesized as being equal to the

maturity of the call deferment period. Thus, the maturity

of the average noncallable bond is compared with the average

call deferment period as a means of testing whether the two

investment options are similar.

Data

Information on each debt issue was gathered for

testing whether market factors are in fact unique to either

callable or noncallable debt issues. Additional data were

gathered about each of the issuing firms to test the











hypothesis about why certain firms issue callable debt and

others noncallable debt.

Debt Offerings

Moody's Bond Survey is the primary source of data on

the individual debt offerings. Information on each debt

offering includes company name, date of issue, size of

offering (in millions of dollars), years to maturity,

coupon, yield to maturity, firm classification (utility,

industrial, financial, transportation, or international),

restrictions (length of call deferments, nonrefunding, and

nonredeeming), initial call price, sinking fund, converting

provisions, and debt rating.

Data include both noncallable and callable debt issued

over the ten year period 1977 through 1986. This period was

chosen to test for the popularity of callable debt during

different interest rate cycles, i.e. periods with stable

rates, increasing rates, and declining rates. Figure 3-5

shows the fluctuation in rates during the sample period

through graphs of the average yields on Aaa and Baa

corporate bonds and 3 year Treasury bills. Stable interest

rates were common during two periods within our sample,

January 1977 through June 1979 and during the year 1986.

Rates were increasing over two separate time periods. The

first occurred during the three year period July 1979

through June 1982. The period July 1983 through June 1984

was the second time span of increasing rates. Declining







































77-1 78-1 79-1 80-1 81-1 82-1 83-1 84-1 84-1 86-1

Date (Year-Month)





Figure 3-5
Average Aaa and Baa Corporate Bonds
and
3-year U.S. Treasury Bill Yield Distributions
1977 1986


Note: Top line -
Middle line -
Bottom line -


Average Baa Corporate Bond Yields
Average Aaa Corporate Bond Yields
Average 3-yr U.S. Treasury Bill Yield


Source: Annual Statistical Digest 1977-1986
Board of Governors, Federal Reserve System











interest rate periods were represented from July 1982

through June 1983 and from July 1984 through December 1985.

Financial Data

Financial data on the individual corporations that

issued debt during the years 1977 through 1986 came from the

Industrial Compustat Tapes.1 Data were gathered for the

year the debt was issued and the three years prior to this

date. Corporate data include average tax rate, total debt

outstanding, total assets, dividends and earnings per share,

gross and net property, plant, and equipment, sales, profit

margin, and the firm's debt to equity and debt to assets

ratios.

Treasury Yields

Treasury yields are compiled from the annual

Statistical Digest.2 These yields were matched with the

debt issues by date of issuance and maturity for analysis in

the logit chapter.

Methodology

Tests utilized to determine significant differences

between the issuance of callable and noncallable debt

include the t-test for means and proportions, the F-test for

variances, and the nonparametric Wilcoxon rank sum test for

probability distributions. The comparisons of proportions,

means, variances, and population distributions are used as a

quick and simple method for detecting differences in the

attributes of callable and noncallable debt. The analysis









41
of the attributes, taking into consideration the presence of

the other variables, is continued within Chapter IV.

Means Tests

The standard t-test is utilized to test for the

differences between the population proportions and means of

the callable and noncallable debt issued. The null and

alternative hypotheses tested when comparing proportions of

the various attributes are defined in equations 3-1 and 3-2.

H0: P,i Pn,i = 0 (3-1)
No difference in the proportions from the
callable and the noncallable samples for
attribute i exist.

HI: Pc,i Pn,i # 0 (3-2)

There is a difference in the proportions from
the callable and the noncallable samples for
attribute i

where P = proportion; c = callable;
n = noncallable; i = attribute.

The attributes checked within the proportion analysis

include industry type (industrial, transportation, utility,

finance, international), bond rating (Aaa, Aa, Caa,

Not Rated), maturity, yield, coupon, dollar amount of the

issue, and the inclusion of a sinking fund, conversion

feature or floating rates. A similar set of hypotheses is

defined to test the difference in the means and standard

deviations of the financial data plus the yield, coupon,

dollar amount of the debt issue, and maturity. See note 3

for the set of hypotheses for the means and standard

deviation tests.










Wilcoxon Rank Sum Test

Since many of the attributes tested under the standard

t-test may violate the assumption of normality, the Wilcoxon

rank sum test is also used.4 The Wilcoxon is used to test

the hypothesis that the probability distributions associated

with two populations are equivalent. The key attribute of

this test is that no assumptions have to be made about the

shape of the population probability distributions. The only

assumptions needed are (1) the two samples are random and

independent and (2) the observations can be ranked in order

of magnitude. The hypotheses tested by the Wilcoxon Rank

Sum Test are defined by statements 3-3 and 3-4.

H0: The two sampled populations, callable
and noncallable debt issues, have identical
probability distributions. (3-3)

HI: The probability distribution for the
noncallable debt issues is shifted to the
right or to the left of the probability
distribution for the callable debt issues. (3-4)

Summary Statistics and Test Results

Summary statistics are provided for two sets of

variables: general variables and variables identified for

specific tests. General variables are those variables not

specifically identified for testing a specific hypothesis.

They are used for detecting subtle differences between the

callable and noncallable debt sectors. Included within the

general category are coupons, yields, maturity, size of

issue, number of issues per firm, firm classifications, and











other miscellaneous factors. Specific test variables

include debt ratings, debt ratios, profitability, various

growth components, tax rates, and the level and volatility

of interest rates.

General Variables

Coupons and yields. The average difference between

the coupons attached to the debt issues of the two groups is

insignificant, 10.56% versus 10.53% for the callable and

noncallable groups respectively. This is analyzed in more

detail in Chapter VI, where I account for differences in

maturity, rating, and time of issuance. A comparison of the

coupon between callable and noncallable debt in 1981

actually shows the mean of the average coupons on

noncallable debt is greater than the mean of the average

coupons on callable debt. This abnormality in mean rates is

partially explained by the time of issuance. The

noncallable debt issues during 1981 were clustered in the

last quarter of the year when interest rates were high.

This is in contrast to the issuance of callable debt which

is fairly regular throughout the year. The mean, standard

deviation, and the various difference tests for the overall

period are reported in Table 3-2. For the frequency

distributions of coupon rates on an annual basis and the

distribution of issues by month see Appendix A, Table A-1


and A-2 respectively.











Table 3-2
Mean Coupon Summary for the Period 1977-1986

Callable Noncallable

Mean 10.56% 10.53%
Standard Deviation 2.84% 3.06%
Sample Size 5763 issues 796 issues
t-scorea 0.2492
F-scoreb 1.16 **
z-scorec 0.6293


aTest for a difference in the means.
bTest for a difference in the standard deviations.
CTest for a difference in distributions.
(nonparametric Wilcoxon test statistic)

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.











Promised yields at issuance are also examined to

detect if systematic deviations of the issue price for the

two types of securities exist. The mean (standard

deviation) for the callable and noncallable groups are

11.37% (5.49%) and 10.96% (2.51%) respectively. The

difference between the mean yields for the callable and

noncallable debt issues is significant when the overall ten

year period is tested and within most of the individual

sample years. See Table 3-3 for the summation of the mean,

standard deviation, and test results for the overall sample

period. The same attributes on a yearly basis are shown in

Appendix A, Table A-3.

Another way of illustrating the differences in the

yields between the two groups is by comparing the frequency

distributions of their respective yields. For example, in

1977 none of the noncallable debt issues carried promised

yields of over 10%, but six percent of the callable sector

is found within that segment. Another example of the yield

differences between the two groups is in the 1986 results.

Debt yields less than 9% averaged 49.1% versus 68.9%

respectively for the callable and the noncallable groups.

The frequency distributions for yields summarized for

the overall ten year period is found in Table 3-4. Annual

frequency tables are located in Appendix A, Table A-4. By

comparing the overall summary table with those found in the

Appendix, it is evident that the yield differences hold up











Table 3-3
Mean Yield Summary for the Period 1977-1986

Callable Noncallable

Mean 11.37% 10.96%
Standard Deviation 5.49% 2.51%
Sample Size 5585 issues 766 issues
t-scorea 3.5437***
F-scoreb 4.77 **
z-scorec 1.4900*


aTest for a difference in the means.
bTest for a difference in the standard deviations.
CTest for a difference in distributions.
(nonparametric Wilcoxon test statistic)

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.












Table 3-4
Yield Frequency Summary for the Period 1977-1986


Callable Debt


# issues
390
312
916
975
657
653
693
405
333
272
155
55
34
8
23


% issues
6.6%
5.3
15.6
16.6
11.2
11.1
11.8
6.9
5.7
4.6
2.6
0.9
0.6
0.1
1.1


Noncallable Debt


# issues
81
67
127
106
92
105
94
54
49
29
19
4
4
0
1


% issues
9.7%
8.1
15.3
12.7
11.1
12.6
11.3
6.5
5.9
3.5
2.3
0.5
0.5
0.0
0.1


t
-2.88***
-2.78***
0.23
3.06***
0.10
-1.24
0.41
0.43
-0.26
1.65**
0.63
1.68**
0.38
2.83***
1.87**


5881 100.0% 832 100.0%


Percentage
Yields
< 7
7 7.9
8 8.9
9 9.9
10 10.9
11 11.9
12 12.9
13 13.9
14 14.9
15 15.9
16 16.9
17 17.9
18 18.9
19 19.9
> = 20


Total


*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.










across all 10 years of the sample. Appendix A, Table A-5,

reinforces these differences across firm classification.

Estimates of the value of the call option based upon the

differences between callable and noncallable debt issues

matched by rating and maturity are discussed in Chapter VI.

Maturity. Since the option to call long-term debt is

a means of reducing the true maturity of a debt issue, one

might expect callable debt to have longer maturities than

noncallable debt on average. For the entire 10 year period,

the average maturity for the callable issues is 19.0 years

with a standard deviation of 9.1 years. For the noncallable

group, the average is significantly less. The noncallable

mean maturity is 8.0 years with a standard deviation of 6.1

years. These differences are statistically significant at

over the 1% significance level for the entire ten year

period and for each individual year. Null hypothesis 3-3,

that the two sampled populations have identical probability

distributions, is also rejected at a 1% significance level.

This result reinforces the differences between the

maturities of the callable and noncallable samples detected

by the t-test. See Table 3-5 for the summary statistics for

the overall period and Appendix A, Table A-6 for the annual

results.

Maturity tests are further broken down by industry

classifications: financial (F), industrial (C), and utility

(U). Within all three categories, the null hypotheses were











Table 3-5
Mean Maturity Summary for the Sample Period 1977-1986


Callable Noncallable

Mean 19.0 years 8.0 years
Standard Deviation 9.1 years 6.0 years
Sample Size 5841 issues 837 issues
t-scorea 33.8707***
F-scoreb 2.30 ***
z-scorec 33.6720***


aTest for a difference in the means.
bTest for a difference in the standard deviations.
CTest for a difference in distributions (Wilcoxon).

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.











rejected. Since the individual test on the firm

subcategories yield similar results to those outlined for

the overall sample, the specific results are not reported.

Another way of highlighting the differences in the

maturities is through the use of frequency tables. Table

3-6 shows the frequency distribution of maturities for the

overall ten year period. The yearly distributions for the

maturity frequencies can be found in Appendix A, Table A-7.

The maturities in both the table and the appendix are

grouped into five year clusters.

Two of the key factors shown in the frequency tables

are the significant differences in the low and high ends of

the distributions. The callable issues with maturities of 5

years or less range from 0 to 4.9% with an average of 1.9%

for the entire ten year period. Contrasting this is the

noncallable group which ranges from 0 to 35% with an average

of 20.6%. At the other extreme is the percentage of long

term issues within each group. Callable issues with

maturities of at least 20 years average 55.0% with a range

of 33.7 to 75.8% over the years 1977 through 1986. Over the

same maturity group, the noncallable issues averaged 5.3%

with a range of 0 to 13.6%.

The maturity structure is expanded across the three

main firm classifications (industrial, finance, and utility)

in Table 3-7. The results reinforce the differences in the

maturity structure of the callable and the noncallable











51
Table 3-6
Summary of Debt Issues Maturities for the Period 1977-1986

Maturity Callable Debt Noncallable Debt
in Years # issues % issues # issues % issues t
< 5 136 1.9% 216 20.6% -22.79***
5 9.9 681 12.3 332 51.8 -29.89***
10 14.9 1463 22.2 205 20.3 1.13
15 19.9 655 8.8 17 2.1 6.06***
20 24.9 773 14.3 25 1.9 10.38***
25 29.9 719 12.4 13 2.0 9.05***
> = 30 1475 28.3 26 1.4 17.49***

Total 5902 100.0% 834 100.0%

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.











Table 3-7
Maturity By Firm Classification Summary
Number of Issues per Maturity Class

Industrial Financial Utility
Call Ncall Call Ncall Call Ncall

< 5 23 46*** 107 153*** 2 8***
5 9.9 183 74*** 369 160*** 73 33***
10 14.9 495 57 653 80** 255 11
15 19.9 224 2*** 372 9*** 32 0
20 24.9 342 11*** 327 6*** 52 2
25 29.9 333 5*** 278 4*** 66 0*
> = 30 222 7*** 548 2*** 666 4***

Total 1822 202 2655 414 1146 58

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.











groups regardless of firm classification. For example, in

the industrial category the callable group has only 1.3% of

the reported issues with maturities 5 years or less while

the noncallable group has 22.8%. Another extreme example is

found in the utility category. Percentages of 58.1 and 6.9

are found within the 30 years or greater maturity set for

the callable and noncallable groups respectively.

Size of issue. A summary of the dollar amounts of the

individual issues shows a difference between the callable

and noncallable groups. The callable debt issues averaged

$90.0 million with a standard deviation of $106.4 million.

This is contrasted by the mean and standard deviation of the

noncallable group, $132.9 million and $110.1 million. The

nonparametric Wilcoxon rank sum test was also performed.

Test results for the overall period and the yearly periods

are shown in Table 3-8 and Appendix A, Table A-8

respectively. The results show that on average, noncallable

debt issues are larger than callable debt issues.

Frequency distributions of the individual issues by

amounts were examined. For example, 36.50% of the callable

debt issues were $50 million or less while only 12.46% of

the noncallable issues were found in this category. Table

3-9 summarizes the issue amount frequency distribution for

the overall sample period. This is expanded in Appendix A,

Table A-9 which shows the frequency table for the dollar

amount of issues on a yearly basis and Table A-10 which











Table 3-8
Mean Issue Amount Summary for the Period 1977-1986

Callable Noncallable
Mean $ 90.0d $132.9d
Standard Deviation $106.4 $110.1
Sample Size 5890 issues 818 issues
t-scorea -10.7706***
F-scoreb 1.07
z-scorec -16.3127***


aTest for a difference in the means.
bTest for a difference in the standard deviations.
CTest for a difference in distributions
Snonparametric Wilcoxon test statistic).
Dollar amounts in millions.

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.











55
Table 3-9
Issue Amount Frequency Summary for the Period 1977-1986


Callable Debt


# issues
2243
1454
1513
446
134
43
37
10
6
5
2
9


% issues
36.5%
25.9
27.1
7.3
1.5
0.8
0.4
0.2
0.1
0.1
0.0
0.1


Dollar
Amount
< 50
50 99
100 199
200 299
300 399
400 499
500 599
600 699
700 799
800 899
900 999
>= 1000

Total


Noncallable Debt


# issues
117
164
348
153
27
13
8
0
1
0
0
1


% issues
12.5%
25.2
42.3
15.6
1.7
1.3
0.9
0.0
0.2
0.0
0.0
0.4


834 100.0%


5902 100.0%


t
13.62***
0.42
-9.14***
-7.91***
-0.29
-1.41*
-1.67**
1.10
-0.71
0.98
0.53
-2.07**


*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.











gives a further breakdown of the dollar amount frequencies

by firm classification. Results are consistent across the

various classifications.

Number of issues per company. The number of debt

issues per company summarized on a yearly basis shows an

insignificant difference between the callable and

noncallable groups. The proportions of firms with only 1

debt issue were between 70 and 80% (76.3% versus 71.6% for

callable and noncallable respectively). This similarity can

be found throughout the frequencies for the other categories

of issues per firm. See Table 3-10 for the overall

frequency distribution of issues per firm and Appendix A,

Table A-11, for the frequency table on issues per firm by

year. Thus, the number of times a firm enters the debt

market does not seem to effect the decision of whether a

firm issues callable or noncallable debt.

Total assets. The overall sample, limited to debt

issues from firms on the Compustat data tapes show a wide

dispersion in the average asset values between the callable

and noncallable sectors. The means for the callable and

noncallable sectors are $9.7 billion and $21.7 billion,

respectively. Large firms are more likely to issue

noncallable debt then small firms. This difference holds up

across the various firm classifications with the difference

narrowing in both the financial and utility classes. See

Table 3-11 for a broader summation of the means, standard











Table 3-10
Number of Debt Issue per Firm for the Period 1977-1986

# Issues Callable Debt Noncallable Debt
per Companya #issues % issues # issues % issues t
1 2691 76.3 338 71.6 2.40***
2 457 13.0 96 17.7 -3.00***
3 138 3.9 24 4.4 -0.57
4 110 3.1 16 3.0 -0.21
5 36 1.0 7 1.3 -0.57
6+ 93 2.6 11 2.0 0.84

Total 3525 100.0 542 100.0

aNumber of issues based on a yearly basis.

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.










58
Table 3-11
Total Assets of Issuing Firms during the Period 1977-1986


Callable Noncallable

Mean $9.7 $21.7
Standard Deviation 24.5 32.7
Sample Size 2106 issues 464 issues
t-scorea -7.4744***
F-scoreb 1.76 ***
z-scorec -5.4396***


aTest for a difference in the means.
bTest for a difference in the standard deviations.
CTest for a difference in distributions.
(nonparametric Wilcoxon test statistic)

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.

Dollar values in millions of dollars.











deviations, and test statistics in regards to the total

assets carried by the debt issuing firms and Appendix A,

Table A-12 for the summary results by firm classification.

Firm classification. Moody's classifies firms

offering debt by five broad categories: industrials (C),

financial (F), transportation (T), utility (U), and

international (I). Summary statistics show that the

proportion of issues within the callable and noncallable

sectors varies across firm classifications. For instance,

only 45.0% of the callable issues were financial, compared

with 49.6% for the noncallable issues. A major difference

in firm type proportions occurs within the international

sector. Only 2.5% of the callable debt issues were offered

by internationals compared with 16.3% for the noncallable

group. A summary of firm classifications is shown in Table

3-12 for the overall ten year period.

The differences between the proportions of the two

groups, callable and noncallable, are highly significant

under each firm classification except for the proportions of

debt offered within the transportation sector. The

insignificance in the difference of proportions within the

transportation sector is not surprising given the small

number of firms from the sample found within the category.

Significance levels are also found to be fairly consistent

for each of the individual years within the sample.

Appendix A, Table A-13, offers the test results and summary

proportions of the firm classification on a yearly basis.










60
Table 3-12
Firm Classification Summary for the Sample Period 1977-1986

Callable Debt Noncallable Debt
# issues % issues # issues % issues t
Industrial 1824 30.9% 202 24.2% 3.94***
Utility 1146 19.4 58 7.0 8.79***
Financial 2654 45.0 414 49.6 -2.54***
Transportation 130 2.2 24 2.9 1.22
International 148 2.5 136 16.3 -18.56***

Total 5902 100.0% 834 100.0%

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.











Another test of debt type distributions within the

firm classifications utilizes SEC classifications. A

breakdown of the firms within the sample which are listed on

the Compustat tapes are summarized by industry code. The

number of firms within this subsample differs from the

original sample used. Only 3313 debt offerings are included

of which 2770 are callable and 543 noncallable. The

decrease in the number of firms is due to the fact that the

industry code classifications were obtained from the

Compustat data tape and that not all of the firms in the

original sample are listed with the Compustat file. The SEC

classifications reinforce the findings from using the broad

classifications C, F, I, T, and U. For example, the number

of firms found in the 6000-6999 sector verifies that a

significant portion of the Compustat sample firms issuing

noncallable debt are financial. The proportions for the

callable and noncallable debt offerings for firms listed on

the Compustat tapes within this 6000 level of industrial

codes are 28.2% and 60.8% respectively. Appendix A, Table

A-14 summarizes the frequency of issues by industry code for

the ten year sample period.5

The significance of the difference of the proportions

within the financial sector offers support for the signaling

argument. The finance industry is hypothesized as having

the most to gain by withholding the exact nature of their

asset makeup due to the fact that it could be easily











duplicated. Therefore, a means of signaling quality to the

investment public is needed. A number of signals have been

suggested in the financial literature. The elimination of

the call feature on the debt issue by substituting

short-term debt offers a means for reinforcing these

signals.6

Other factors. The number of debt issues carrying

floating rates is relatively small over the time period

under study. It averaged only 2.69% and 6.11% for callable

and noncallable debt categories respectively. This

difference is significant at the one percent level. The

prevalence of floaters among noncallable debt is not

surprising. If a bond contains a floating rate, its coupon

rate moves with the general level of interest rates. Thus,

its price does not deviate much from par, i.e. its duration

is close to zero. Hence, callability is basically

worthless. Floating rate debt is a recent phenomenon and

the frequencies should increase within both categories over

time, especially if the market experiences another period of

widely fluctuating rates.

Similarly, the use of convertible debt differs

depending on the type of debt offered. The percentiles are

11.01% versus 0.84% for the callable and noncallable groups

respectively. This difference is also highly significant.

Again the results are not surprising. Convertible debt and

callable debt are both argued to mitigate agency problems.










Granted one deals with under investing and the other risk

shifting. However, when managers have a great deal of

latitude, it is not surprising that both are used. See

Table 3-13 for a summary of the float and convertible

features of the debt issues within the sample period.

Specific Test Variables

In this section, tests of the variables hypothesized

to effect the decision to issue callable versus noncallable

debt are analyzed. First, variables proxying the advantage

of callable debt in mitigating the underinvestment agency

problem are examined. The proxies include debt ratios,

profitability, and growth. Next, tax rates of issuers are

examined to determine if, as predicted, firms that issue

callable debt have higher corporate tax rates. Finally, the

interest rate environment is examined to see it the value of

the call influences the probability that it will be

included.

Ratings. Debt ratings are reported by the broad

classifications Aaa, Aa, A, through Caa. Although the major

ratings Aa, A, through B, are further classified into three

subcategories by Moody's starting in 1982, only the broad

classifications are reported.7 A difference in the

quality ratings of the debt offerings for the noncallable

and callable groups is detected for the overall ten year

period as well as for the individual years.


















Floating Rate
Debt
Total Firms
Industrials
Financial
Utilities

Covertible
Debt
Total Firms
Industrials
Financial
Utilities


Table 3-13
Floating Rate and Converting Debt Summary
For the Period 1977-1986

Callable Debt Noncallable Debt
#issues % issues # issues % issues t


2.7%
0.7
8.6
0.3


6.1%
2.0
9.3
0.0



0.8%
2.0
0.9
0.0


650 11.0%
339 26.9
64 8.9
10 1.6


-5.28***






9.27***


*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.











For the ten year period, the percentage of the debt

issues rated A or better are 80.3% and 44.7% for the

noncallable and callable issues respectively. This

difference is significant at the 1% level. The percentages

within the callable sample with an A or better rating range

from a high of 79.4% in 1977 to a low of 33.4% in 1986. The

percentages within the noncallable group are more consistent

throughout the period, ranging in the high seventies and

above for the entire ten year period. The low of 77.6% for

the noncallable group occurred in 1985, while the high of

100% is shown in 1978 and 1979.

Another major difference in ratings between the

callable and noncallable groups is shown by the proportion

of nonrated issues within each group. The proportion of

nonrated debt is 28.3% to 1.6% for the callable and

noncallable groups respectively. If the nonrated group is

not included in the percentage calculations, the differences

between the remaining classes are reduced. For instance,

the proportion of the noncallable group in the A or better

category is practically unchanged at 81.6%. The callable

group's percentage for the same class, A or better,

increases to 62.3%. The difference is still significant at

the 1% confidence level.

The summary of ratings with and without the nonrated

group generates similar results for the individual years

within the study. The ranges for the A or better category









66
excluding nonrated debt issues were 57.8% through 79.8% and

77.6% through 100% for the callable and noncallable debt

groups respectively. This reinforces the fact that the

differences detected within the debt ratings remain

regardless of whether or not the nonrated sector is included

in the analysis.

An interesting aside is that the number of nonrated

debt instruments reported by Moody's is almost nonexistent

before 1981. In the callable debt group, the nonrated bond

percentages ran from approximately 0% through 1980 to a high

of 47.3% in 1986. The noncallable group had approximately

0% in all years except 1981 (12.2%), 1982 (3.6%), and 1983

(3.3%). Reasons for a debt issue not being rated include:

(1) An application for rating was not received or accepted

by Moody's. (2) The issue or issuer belong to a group of

securities or companies that are not rated as a matter of

policy by Moody's. (3) There is a lack of essential data

pertaining to the issue or issuer. (4) The issue was

privately placed, in which case the rating is not published

in Moody's publications. Note that reason number 4 does not

pertain to the sample used in this dissertation.

Table 3-14 shows the breakdown of the ratings for the

callable and noncallable debt groups as an absolute number

and as a percentage of the issues within each rating

category. A similar rating summary can be found in Appendix

A, Table A-15, for the callable and noncallable issues on a

yearly basis.











Table 3-14
Rating Summary on the Sample Period 1977-1986

Callable Debt Noncallable Debt
# issues % issues # issues % issues t
Aaa 755 12.8% 151 18.0% -4.17***
Aa 724 12.3 214 25.6 -10.41***
A 1161 19.7 307 36.7 -11.16***
Baa 545 9.2 104 12.4 -2.93***
Ba 322 5.5 30 3.6 2.28**
B 706 12.0 18 2.2 8.58***
Caa 22 0.4 0 0.0 1.77**
Nonrated 1668 28.3 13 1.6 16.17***

Total 5903 100.0% 837 100.0%

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.









68
The differences between the proportion of debt issues

within all of the individual rating categories for callable

versus noncallable issues are statistically significant at

greater than a 5% significance level when null hypothesis

3-1 is tested on the entire 10 year period. The t-tests,

segmented on a yearly basis, also support the fact that a

greater proportion of higher grade debt instruments were

issued in the noncallable category versus the callable

category. Thus, the results show that the null hypothesis

is rejected for the overall ten year period and for most of

the individual years within the ten year period.

The respective t-values resulting from testing the

proportion of callable and noncallable debt within category

i (where i equals the debt ratings Aaa, Aa, through Caa and

nonrated debt) are reported on a yearly basis in Appendix A,

Table A-15. A comparison of the yearly results with the

overall summary found in Table 3-13 show a consistency in

the debt rating across the sample period.

The results obtained within the rating analysis

indicate that high quality firms are more likely to issue

noncallable debt then lower quality firms. One reason for

this type of behavior deals with the externality problem.

If debt is highly rated, the firm's managers do not have to

worry about co-insuring the debt. The call option fails to

buy anything. A second reason is linked to the ability of

high quality firms to obtain favorable rates when the debt









69

contract is offered. Therefore, a need for later recall is

lessened.

Debt ratios. The debt to asset and debt to equity

ratios gage the relationship of the amount of debt to the

supporting base. The results show that even though the

callable debt firms on average have lower asset bases, they

have a higher average debt to asset ratio (0.27 versus

0.23). Thus evidence exists that firms issuing callable

debt have less debt coverage for the amount of debt issued

when measured against an asset base.

The tendency for having a higher debt ratio is

supported by the debt to equity ratios. Callable debt

issuing firms have an average debt to equity ratio of 1.63

compared to 1.54 for firms issuing noncallable debt. The

debt ratio summary, shown in Table 3-15, support the

signaling arguments of debt choice. The debt ratios by firm

classifications are summarized in Appendix A, Table A-16.

Profit measures. Various measures of profitability

are gathered to detect differences between the two debt

issuing groups. First, an ex-post test of the signaling

hypothesis involves comparing precentage changes in

profitability. The percentage changes measure the profits

before and after the debt issue. Four specific comparisons

were made: callable versus noncallable, short-term

noncallable/long-term callable versus long-term noncallable,

short-term noncallable/long-term callable versus long-term











Table 3-15
Debt Ratios of Issuing Firms during the Period 1977-1986

Debt to Asset Ratio


Callable


Noncallable


Mean
Standard Deviation
Sample Size
t-scorea
F-scoreb
z-scorec

Debt to Equity Ratio


Mean
Standard Deviation
Sample Size
t-scorea
F-scoreb
z-scorec


0.276 0.229
0.160 0.169
2105 issues 463 issues
5.6384***
1.12
5.7254***


Callable


Noncallable


1.634 1.544
12.019 1.391
2105 issues 463 issues
0.3352
74.62 ***
3.9340***


aTest for a difference in the means.
bTest for a difference in the standard deviations.
CTest for a difference in distributions.
(nonparametric Wilcoxon test statistic)

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.









71
noncallable/short-term callable, and short-term noncallable

versus long-term callable. The percentage change in profit

tests fails to uncover any statistical difference within any

of the comparison groups. Results are shown in Table 3-16.

Other profit variables include dividends and earnings

per share and profit margin. A statistically significant

difference is detected for the variable dividends per share,

but as shown in the logit analysis in Chapter IV this

difference is not useful for predictive purposes. Earnings

per share and profit margins before and after taxes failed

to show a statistically significant difference. Test

results for the second set of variables are shown in

Appendix A, Table A-17. This lack of difference in

profitability is inconsistent with the theory that predicts

an information content in the choice of debt financing.

Growth. Several growth measures are compiled

including growth in assets, debt, earnings and dividends per

share, property, plant, and equipment (net and gross), and

sales. Two types of growth variables are calculated, pre

and post. Pregrowth is the average growth rate of the

various variables over the three year period prior to the

debt issue examined.8 Postgrowth is calculated ex-post as

the growth rate over the year the debt was issued.

Postgrowth is used as a proxy for unanticipated growth.

Pregrowth rates for asset, debt, and sales are found

to be significantly different for the noncallable and












Table 3-16
Percentage Change in Profitability
Before and After the Debt Issue
During the Period 1977-1986


Noncallable versus Callable
Mean
Standard Deviation
Sample Size
t-scorea
F-scoreb
z-scorec


S-T Noncallable/L-T
Mean
Standard Deviation
Sample Size
t-scorea
F-scoreb
z-scorec

S-T Noncallable/L-T
Mean
Standard Deviation
Sample Size
t-scorea
F-scoreb
z-scorec


S-T Noncallable versus
Mean
Standard Deviation
Sample Size
t-scorea
F-scoreb
z-scorec


-0.44% -0.65%
5.67 2.53
663 issues 2250 issues
-0.3467
2.63 ***
-0.1342


Callable versus L-T Noncallable
-0.45% -0.77%
4.89 2.76
1961 issues 79 issues
-0.9699
3.13 ***
-0.0995

Callable vs S-T Noncallable/S-T Callable
-0.45% -0.65%
4.89 3.24
1961 issues 952 issues
-1.3665
2.28 ***
-0.5684


L-T Callable
-0.51
2.22
584 issues 1
0.4875
6.50 ***
0.7902


0.42
5.66
377 issues


aTest for a difference in the means.
bTest for a difference in the standard deviations.
cTest for a difference in distributions.
(nonparametric Wilcoxon test statistic)

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.











callable debt firms. The postgrowth variable show

significant differences for assets, property, plant, and

equipment (both gross and net figures), and sales. Table

3-17 summaries the means and representative test results.

The results from the individual firm classification mirror

the results found in the overall sample and therefore are

not reported.

Tax rate. The average tax rates between firms issuing

callable and noncallable debt are approximately the same.

The mean values are 31.8% and 33.8% for the callable and

noncallable debt issuing firms. Using the standard t-test,

no significant difference between the mean values of the two

groups is detected. This result is not supported by the

Wilcox rank sum test which finds a significant difference in

the distributions of the two debt classes when average tax

rates are analyzed. This difference in test results is

primarily due to the wide variation within the standard

deviation of the tax rates for the callable debt firms.

The analysis of the marginal tax rates shows

statistically significant differences between the firms

issuing callable and noncallable debt. The t-test indicate

significant difference favoring the marginal tax rates of

the noncallable debt issuing firms which is contradictory to

the tax hypothesis. This results is questionable due to the

skewness of the distributions caused by a limited upper

bound.











Table 3-17
Growth Rates During the Period 1977-1986


Callable Noncallable
Average Growth Rates
During the 3 years Period prior to the Debt Issue

Total Assets 18.9% 11.0%***
Dividends per Share 3.4 2.4
Earnings per Share 5.3 0.5
Total Debt 47.6 18.8 *
Property, Plant, and Equipment
Gross 27.7 30.6
Net 31.1 31.9
Sales 18.7 10.0 ***

Average Growth Rates
During the year of the Debt Issue

Total Assets 27.5% 19.1%***
Dividends per Share 2.0 13.0
Earnings per Share -13.2 -13.3
Total Debt 153.4 103.1
Property, Plant, and Equipment
Gross 8.8 51.8***
Net 9.8 49.3***
Sales 6.3 3.7***


*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.

Dollar values in millions of dollars.









75
The Wilcox rank sum test is a more accurate measure of

the relationship between the marginal tax rates of the two

types of firms because of this violation of the normalcy

assumption. The Wilcox test supports the tax hypothesis,

but as will be shown within the logit analysis in Chapter

IV, this difference does not add to the predictive power of

the logit model. Table 3-18 contains the summary results

from the tests on both the average and marginal tax rates.

The results are consistent across the firm classifications

as shown in Appendix A, Table A-18. Thus, the tests of the

tax hypothesis show inconclusive results.

Interest rate level. The level of interest rates are

measured using 3 and 30 year Treasury yields. A frequency

distribution of callable and noncallable debt issues by

interest rate levels is found in Table 3-19 for the 3 year

Treasury yields and in Appendix A, Table A-19 for the 30

year Treasury yields. Neither distribution shows any

tendencies for an increase in the issuance of callable debt

as interest rates increase. The test of null hypothesis 3-1

at all levels of interest rates show insignificant results.

A test of the issuance of callable debt due to the

shape of the yield curve, whether increasing or decreasing

shows significant results. Callable debt dominates the

issuance of noncallable debt when the yield curve is

decreasing, i.e. when the difference between 30 and 3 year

Treasuries is between -1 and 0%. But when the yield curve











Table 3-18
Average and Marginal Tax Rates for Issuing Firms
During the Period 1977-1986


Average Tax Rates

Mean
Standard Deviation
Sample Size
t-scorea
F-scoreb
z-scorec

Marginal Tax Rates

Mean
Standard Deviation
Sample Size
t-scorea
F-scoreb
z-scorec


Callable


Noncallable


31.8% 33.8%
172.2 21.4
2106 issues 464 issues
-0.5103
65.00 ***
-3.0489**


Callable


Noncallable


32.3% 37.6%
14.2 9.99
2096 issues 463 issues
-9.5576***
2.04 ***
6.3402***


aTest for a difference in the means.
bTest for a difference in the standard deviations.
CTest for a difference in distributions.
(nonparametric Wilcoxon test statistic)

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.

Dollar values in millions of dollars.












Table 3-19
Issue Frequency Summary by Interest Rate Level
For the Period 1977-1986


Callable Debt Noncallable Debt
# issues % issues # issues % issues
1115 22.9% 172 23.0%
615 12.6 71 9.5
613 12.6 86 11.5
888 18.3 114 15.2
528 10.9 90 12.0
371 7.6 80 10.7
233 4.8 59 7.9
262 5.4 43 5.7
155 3.2 19 2.5
69 1.4 13 1.7
13 0.3 2 0.3


Interest
Rate Level
<= 7
7 7.9
8 8.9
9 9.9
10 10.9
11 11.9
12 12.9
13 13.9
14 14.9
15 15.9
>= 16

Total


t
-0.01
0.85
0.31
0.84
-0.31
-0.82
-0.82
0.09
0.17
-0.08
0.00


749 100.0%


4862 100.0%


Note: Interest rate level based on 3 year Treasury yields.
*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.










is rising, noncallable debt dominates. A rising yield curve

is defined as a positive spread. See Table 3-20 for the

frequency distribution of debt issues based upon the spread

between 30 and 3 year Treasuries.

The uncertainty of interest rates, measured as the

average absolute change in 3 year Treasury rates over the

three week period prior to the issuance of the debt, is a

marginally significant factor in the decision of whether or

not to include the call feature. Only during periods of low

uncertainty is one type of debt dominated by the other. The

percentage of noncallable debt issues dominates callable

percentages when the average absolute change in 3 year

Treasuries is less than 25 basis points. Although this

difference is statistically significant, Chapter IV shows

that the difference is not useful in the prediction of debt

types. Uncertainty of rates, tested using 30 year

Treasuries, produces similar results. Both sets of

uncertainty frequencies are reported in Table 3-21.

Substitution. The substitution hypothesis is analyzed

from the perspective of restrictions on the call deferment.

The hypothesis is that noncallable debt has the same

maturity as the time of the call restriction. The average

call deferment period is compared to the average maturity

for the noncallable debt sector using the null hypothesis

defined below.


HO: Average
Call Deferment


= Average Maturity
Noncallable Debt Issue











Table 3-20
Issue Frequency Summary by Interest Spread
For the Period 1977-1986

Interest Callable Debt Noncallable Debt
Spread # issues % issues # issues % issues t
< = -1.0 299 6.2% 19 2.6% 0.82
-1.0 0.0 804 16.7 50 6.7 2.29**
0.0 1.0 2586 53.7 455 60.7 -2.40**
> = 1.0 1126 23.4 225 30.1 -1.74**

Total 4815 100.0% 749 100.0%

Note: Interest spread equal 30 year Treasury yield minus
3 year Treasury yield.

TSPREAD is a proxy for the shape of the yield curve.
If TSPREAD is negative (positive), then the yield
curve is falling (rising).

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.











Table 3-21
Issue Frequency Summary by Uncertainty Level
For the Period 1977-1986

3 year Treasuries:
Uncertainty Callable Debt Noncallable Debt
Level # issues % issues # issues % issues t
< = 0.00 126 2.6% 4 0.5% 0.46
0.00 0.25 3749 77.1 640 85.4 -2.22**
0.25 0.50 783 16.1 82 11.0 1.18
0.50 0.75 112 2.3 18 2.4 -0.02
>= 0.75 92 1.9 5 0.7 0.26

Total 4862 100.0% 749 100.0%

30 year Treasuries:
Uncertainty Callable Debt Noncallable Debt
Level # issues % issues # issues % issues t
< = 0.00 142 2.9% 5 0.7% 0.49
0.00 0.25 4062 83.5 667 89.0 -2.46**
0.25 0.50 608 12.5 73 9.8 0.63
0.50 0.75 50 1.0 4 0.5 0.11
> = 0.75 0 0.0 0 0.0 0.00

Total 4862 100.0% 749 100.0%


Note: Uncertainty level calculated as the average
absolute change in 3 and 30 year Treasury yields
over the 3 week period prior to the debt issue.

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.











Where the average call deferment is calculated as
the average of the maximum length of time between
the refunding or redeeming restrictions of the
callable debt issues.

For the overall period of the sample, the mean of the

refunding restrictions are 5.14 years versus 7.37 years for

the average maturity of the noncallable debt issues

excluding zero coupon debt. Using the t-test for comparing

the means of these two values shows a difference at greater

than a 1% significance level. Although there exists a

statistically significant difference between the call

deferment and noncallable maturities, the difference is

negligible from a practical standpoint. Since not all

callable debt is actually called at the first opportunity, a

substitution effect is realistic. Thus, the difference in

maturity structures fails to refute Bodie and Friedman's

(1978) claim that noncallable debt is a substitute, although

an imperfect one, for long term callable debt. A summary of

the test results can be found in Table 3-22.

Summary

Many of the attributes of issues and issuers are

different across the two debt sectors, callable and

noncallable. The signaling argument is supported by the

results reported within the ratings and firm classification

sections. Tendencies detected within the rating section

indicate that on average the ratings of noncallable debt

issues are higher than those found within the callable debt











Table 3-22
Call Deferment vs Noncallable Maturity
Summary for the Period 1977-1986

Callable Noncallable

Mean 5.14 yrs 7.37 yrs
Standard Deviation 3.11 5.52
Sample Size 2070 issues 538 issues
t-scorea 9.0066 ***
F-scoreb 3.16 **
z-scorec 6.5324 ***


aTest for a difference in the means.
bTest for a difference in the standard deviations.
CTest for a difference in distributions.
(nonparametric Wilcoxon test statistic)

*Significant at 10% level using a one-tail test.
**Significant at 5% level using a one-tail test.
***Significant at 1% level using a one-tail test.









83
sector. Within the firm classification section, support was

found for financial firms to issue noncallable debt as one

of the signals available for conveying private information

without revealing its underlying asset components.

The maturity and yield sections supported the general

belief that on average the maturities and yields of callable

debt is greater than these found within the noncallable

sector. Size of the firm is also detected as a significant

factor in that the larger the debt issue and the larger the

firm size, the greater the probability that an issue is

noncallable. Nonsignificant differences are detected for

both the average tax rates and the profitability between

firms issuing the two types of debt. The level of rates is

insignificant as a factor influencing the issuance of

callable and noncallable debt. But the shape of the yield

curve is significant with callable debt dominating during

downward sloping yield curves and noncallable dominating

during rising yield curves. Uncertainty of rates was shown

to be marginally significant. General tendencies and test

results are summarized in Tables 3-23 and 3-24

respectively. These major differences are the initial

starting point in the variable selection for the logit

analysis in Chapter IV.











Table 3-23
Callable and Noncallable Debt Issue Trends
Period 1977-1986


Callable Noncallable
Ratings (A or Better Rating) 44.7% 80.3%
Trend: Noncallable debt has a higher rating on average
versus callable debt.

Firm Classification (Financial) 45.0% 49.6%
Trend: Financial firms, the industry type with the
most to gain from signaling are more prevalent
within the noncallable debt sector.

Maturity 19.0 yrs. 8.0 yrs.
Trend: Callable debt has an average maturity over
twice that found on noncallable debt issues.
The average call restriction, 5.14 years, was
considerably shorter then the average
noncallable debt maturity.

Yields 10.56% 10.53%
Trend: As can be expected the granting of the call
option is not a free good.

Size of Debt Issue $90.0 million $132.9 million
Trend: There is a tendency for only large firms to be
able to obtain favorable rates on noncallable
debt.

Total Assets $9.7 billion $21.7 billion
Trend: Reinforces the fact that small firms are
reluctant issue noncallable debt.

Average Tax Rate 31.8% 33.8%
Trend: No significant difference in the tax rates of
the two firm type was detected.

Profit Margin (after tax) 7.5% 7.7%
Trend: No significant difference in the two type of
firms was detected.











Table 3-24
Level and Sign of Significance
Between Callable and Noncallable Debt Variables
Period 1977-1986

Dominant Support
Variable Trend Prop Means Std Dev Dist Theory


General
Coupons
Yield
Maturity
Size
# issues
Assets


Variables:
(c>n)
(c>n)
(c>n)
(c (c=n)
(c

Test Variables:
Ratings (c Classificationa **
Debt Ratios
D/A (c>n)
D/E (c>n)
Tax Rates
Average (c>n)
Marginal
(cn distri
Profitability
Change in Profits (c DPS (c EPS (c PM (c

ns ** ns na
*** ** na
*** *** *** na
*** ns *** na
na
*** *** *** na


yes
yes

*** ns *** yes
ns *** *** yes


bution


ns *** **
*** *** ***
is)

ns *** ns
ns *** ns
*** ns *
ns *** ns


of
utilities.


aFinancial firms issued a greater proportion
noncallable debt than either industrials or












Table 3-24--continued

Dominant Support
Variable Trend Prop Means Std Dev Dist Theory


Test Variables (cont.)
Growth (Pre)
Assets (c>n)
DPS (c=n)
EPS (c=n)
Debt (c>n)
Property, Plant, & Equipment
Net (c Gross (c Sales (c>n)
Growth (Post)
Assets (c>n)
DPS (c=n)
EPS (c=n)
Debt (c>n)
Property, Plant, & Equipment
Net (c Gross (c Sales (c>n)
Interest Rates
Level (c=n) ns
Yield Curve
Falling (c>n) *
Rising (c Uncertanty (c=n) ns


*** *** *** yes
ns ns ns no
ns ns ns no
* *** ns inc

ns *** *** no
ns *** *** no
*** *** *** yes

*** *** *** yes
ns *** ns no
ns *** ns no
* *** ns yes

ns *** *** no
ns *** *** no
*** *** ** yes

no

yes
yes
no


ns Not significant at any practical level.
Significant at 10% level using a one-tail test.
** Significant at 5% level using a one-tail test.
*** Significant at 1% level using a one-tail test.
inc Inconclusive results.











Notes

1. Standard & Poor's Compustat Services, Inc. (Annual Data
Tapes.)

2. Annual Statistical Digest is published by the Board of
Governors of the Federal Reserve System annually.

3. The null hypothesis for comparing the means from the
callable and noncallable groups is:

H0: meanc,i = meani

Similarly, the null hypothesis used for testing equal
variances between the callable and noncallable groups is:

H0: varc,i = varn,i

4. See McClave and Benson (1980) for a full description of
the Wilcoxson Rank Sum Test, pages 675-679.

5. The Compustat tapes do not include the international
firms from the original sample. A listing of the meaning of
the SIC classifications can be summarized as follows:
0000-0999 Agricultural
1000-1999 Mining (Metal, Coal, Oil, and
Gas) and Construction.
2000-2999 Food, Drugs, and Lumber Related
Products.
3000-3999 Production: Rubber, Leather,
Glass, Concrete, Metal Goods,
Machinery, Appliances, etc.
4000-4999 Services: Transportation
(Railroads, Trucking, Water,
and Air), Communication
(Telephone, Cable, Radio, and,
TV), Electric, and Gas.
5000-5999 Merchandising: Auto, Sporting,
Durable Goods, Drugs, Apparel,
Drugs, Grocery, and Jewelry.
6000-6999 Finance: Banks, S & L's, CU's,
Insurance, and Real Estate.
7000-7999 Personnel: Hotel, Advertising,
Data Processing, Consulting,
Auto Repair, and Theaters.
8000-8999 Medical: Hospitals, Nursing
Homes, Labs, Education.

6. Another explanation for the large percentage of
noncallables within the financial sector is due to the
matching of asset and liability maturities. If matched
completely, there is no need for the call feature. But the










88
fact that many noncallables also exist for nonfinancial
firms weakens this argument as the sole reason for
noncallable debt.

7. Beginning on April 26, 1982, five of Moody's nine
corporate bond ratings were expanded to included numerical
modifiers 1, 2, and 3. The five bond ratings with the
expanded format include: Aa, A, Baa, Ba, and B. For
example, an Aal rating indicates that the security meets all
of Moody's criteria for a double-A rating and that it ranks
at the high end of that rating category. The modifier 2
indicates that the security is in the mid-range of its
category, and a 3 indicates that the bond is nearer the low
end of its generic category. Moody's emphasize that the
numerical modifiers are only refinements of the defined
categories and that the relative positions of all of Moody's
corporate bond rating symbols, and their definitions remain
unchanged. See pages 2181-2182 of Moody's Bond Survey
(April 26, 1982) for a broader description of the rating
categories.

8. Five year average growth rates were also calculated with
similar results obtained.















CHAPTER IV
LOGIT REGRESSION ANALYSIS


The examination of the first 2 questions raised in the

introduction, (1) Are there identifiable market factors that

dictate when and how much noncallable debt is offered? and

(2) Are there factors common to the firms that issue

noncallable debt that differentiate them from firms that

issue callable debt?, was initiated in Chapter III. The

logit regression is an extension of the summary statistics

analysis. By testing the significance of these tendencies

in a logit regression, the effects of interactions are

considered. Highly correlated variables are redundant for

the prediction process. Thus, the logit analysis is used to

determine if any of the statistically significant

differences uncovered in Chapter III are useful in the

prediction of the debt type issued.

Testable Implications

The testable implications for callable/noncallable

debt issues outlined in Chapter III which are applicable

within this chapter are repeated for clarity. Also Table

3-1 is reproduced as Table 4-1 for easy reference of the

variables utilized in the logit model as they relate to the

various authors in the literature review.

89











Table 4-1
Variables to Test Hypothesized Explanations
For the Call Option

Authors (Date) Tablea: Variables Tested

Barnea, Haugen, and Senbet (1980) A: 1, 2
Barnea, Haugen, and Senbet (1985) A: 1
Bodie and Friedman (1978) S: 6
Bodie and Taggart (1978) A: 10
Boyce and Kalotay (1979a, 1979b) T: 15
Flannery (1986) A: 1, 7, 11, 12
Jen and Wert (1967) F: 14, 17, 17
Kidwell (1976) I: 6, 14, 16
Marshall and Yawitz (1980) T: 15
Pye (1966) F: 14, 16, 17
Robbins and Schatzberg (1986) A: 1
Ross (1977) A: 4, 5
Stiglitz (1979) F: 1
Thatcher (1985) A: 4, 5, 8, 10, 11
Van Home (1984) F: 9, 14, 16, 17

aTable under which additional information can be found:
F=Flexibility (Table 2-1), A=Agency (Table 2-2),
T=Tax (Table 2-3), I=Indenture (Table 2-4), and
S=Substitution (Table 2-5)
Variable tested under the authors) hypothesized claim:
1 = CAR (Cumulative average excess returns)
2 = CLASS (Firms classification)
3 = CPROFIT (Percentage change in profits)
4 = DA (Debt to asset ratio)
5 = DE (Debt to equity ratio)
6 = DEF (Call deferment)
7 = INTCOV (Interest coverage)
8 = MARKET (Measure of the relationship between the
existing debt issue and current debt outstanding)
9 = MAT (Maturity)
10 = GROWTH (PREGROWTH-Average growth in the 3 years prior
to the debt issue and POSTGROWTH-Average growth in the
year of the debt issue)
11 = PM (Profit margin)
12 = RATE (Debt rating class)
13 = SIZE (Natural logarithm of the dollar amount of the
debt issue)
14 = SPREAD (The difference between the yield on the debt
issue and a comparable treasury issue)
15 = TAX (Corporate tax rate)
16 = TSEC (Treasury security yield)
17 = UNCER (Change in interest rates over the 3 weeks prior
to the debt issue date)











Flexibility Tests

The level and uncertainty of interest rates enhance

the value of the call option. Therefore, the variables

related to the level and uncertainty of rates within the

market should be positively related to the probability of

witnessing a call option.

Agency Tests

Asymmetric information is a broad category and as such

different sets of variables are utilized to test subsets of

the theory. First, high risk/low rated debt implies a

greater gain to bondholders in the absence of a call

option. Second, low profitability increases the

possibilities for future improvements in the firm's

position. Third, high growth associated with future

projects implies increased potential opportunities.

Finally, high leverage/low debt ratings increases

co-insurance problems. This set of tendencies implies an

advantage to using callable debt and hence the increased

likelihood of attaching the call option to the debt issue.

Tax Test

Tax motivations for callable debt suggest that the tax

rates of firms issuing callable debt should be higher then

those issuing noncallable debt. Both average and marginal

tax rates are utilized in the analysis.











Data

The debt offering data discussed in Chapter III is

restricted to include only those issues where the issuing

firm's financial data is also available on the Compustat

Tapes. This reduces the data set to 2570 debt issues of

which 2106 are callable and 464 noncallable. Mean summary

data and frequency probabilities are provided in Tables 4-2

and 4-3 respectively. The tendencies identified in Chapter

III hold with this subset of data. The additional data on

Treasury security yields are compiled from the Annual

Statistical Digest as explained in the previous chapter.

Based on the existing literature, the logit model uses

the explanatory variables shown in Table 4-4. All the firm

level variables are obtained from the Compustat tapes except

where indicated. Three binary variables for the broad

classifications of firm type (CLASS) are industrial, utility

and finance. Due to the small number of firms in the sample

from the international and transportation sectors, firms

falling into these two classifications were dropped from

this part of the analysis.

The two book value ratios, debt to equity (DE) and

debt to assets (DA) proxy the firm's leverage. MARKET is

the size of the debt issue relative to the total amount of

consolidated funded debt the firm has outstanding, including

the relevant issue. MARKET is calculated as

MARKET = current debt
(current debt + prior debt)

The measurement of the prior debt is based on book value.












Table 4-2
Mean Value Summary Statistics
For Potential Logit Regression Variables


Debt Type
Variable Callable Noncallable t value Prob


*DA
DE
INTCOV
(Before Tax)
(After Tax)
*MARKET
*MAT
PREGROWTH
TA
DPS
EPS
Debt
GPPE
NPPE
* Sales
POSTGROWTH
* TA
DPS
EPS
Debt
GPPE
NPPE
* Sales
PM
SPREAD
*SIZE
TAX
ATAX
* MTAX
*TSEC
SPREADD
UNCER


Statistics


0.28
1.63

4.59
3.14
0.25
18.46 yrs

18.90%
3.36%
5.29%
47.98%
27.72%
31.14%
18.72%

27.53%
2.04%
-13.23%
153.43%
8.84%
9.78%
6.29%
0.12
1.00%
4.39

31.85%
32.31%
10.35%
0.45%
0.06%


0.23
1.54

5.34
3.59
0.13
7.53 yrs

11.00%
2.41%
0.46%
18.80%
30.62%
31.98%
10.04%

19.10%
13.00%
-13.26%
103.07%
51.86%
49.28%
3.71%
0.12
1.14%
4.62

33.83%
37.65%
9.54%
0.68%
0.08%


5.6384 0.0001
0.3352 0.7375


-0.2855
-0.2954
12.3906
32.7362

4.1457
0.9102
1.5488
2.7026
-0.2435
-0.2435
3.5273

4.1619
-0.7746
0.0048
1.1801
-3.0633
-2.9699
5.8058
-0.3468
-1.3912
-5.9008

-0.5103
-9.5576
7.2416
-8.7070
-0.9322


based on logit data set.


* There exists a highly significant


difference between the


callable and the noncallable samples for this variable.


0.7754
0.7679
0.0001
0.0001

0.0001
0.3631
0.1216
0.0069
0.8077
0.9447
0.0004

0.0001
0.4391
0.9962
0.2382
0.0024
0.0031
0.0001
0.7288
0.1642
0.0001

0.6099
0.0001
0.0001
0.0001
0.3513











Table 4-3
Frequency Statistics
For Potential Logit Binary Dummy Variables


Total Debt Callable Debt
# of (%total) # of (%total
issues issues /%call)


Rating Groups (2570 issues):
LOW (Debt rating Ba or lower)
670 (26.07%) 640 (24.90%/30.39%) 30
MODERATE (Debt rating A or Baa)
1217 (47.35%) 935 (36.38%/44.40%) 282
HIGH (Debt rating Aaa or Aa)
683 (26.58%) 531 (20.66%/25.21%) 152

Maturity Groups (2557 issues):
Short Maturities (Maturities < 10 years)
615 (24.05%) 302 (11.81%/14.41%) 313
Long Maturities (Maturities > 10 years)
1942 (75.95%) 1794 (70.16%/85.59%) 148

Firm Classifications (2570 issues):
Industrials
1219 (47.43%) 1075 (41.83%/51.04%) 144
Financial
724 (28.17%) 460 (17.90%/21.84%) 264
Utilities
545 (21.21%) 509 (19.81%/24.17%) 36


Noncallable Debt
# of (%total
issues /%noncall)


(1.17%/6.47%)

(10.97%/60.79%)

(5.91%/32.76%)



(12.24%/67.90%)

(5.79%/32.10%)



(5.60%/31.03%)

(10.27%/56.90%)

(1.40%/7.76%)


Calculations based upon logit regression data set.

Percent figures reported as percentages of the total data
set used and as percentages of the type of debt, e.g.
callable or noncallable debt subsets.











Table 4-4
Functional Form of the Logic Regression Equation


NCALL = f(CLASS, DA, DE, INTCOV, MARKET, MAT,
POSTGROWTH, PREGROWTH, PM, RATE, SIZE,
SPREAD, TAX, TSEC, SPREAD, and UNCER)

Dependent Variable:
NCALL = A binary variable for the presence (0) or
absence (1) of a call feature being placed on
the debt debt issue.


Independent Variables:


CLASS

DA
DE
INTCOV
MARKET

MAT
POSTGROWTH
PREGROWTH

PM
RATE
SIZE

SPREAD

TAX
TSEC
SPREAD

UNCER


= A set of binary variables for firm
classifications.
= Debt to asset ratio.
= Debt to equity ratio.
= Interest coverage ratio.
= A measure of the current debt issue to the
existing debt outstanding.
= Maturity of the debt issue in years.
= Growth during the year the debt was issued.
= Average growth during the 3 years prior to the
debt issue.
= Profit margin.
= Debt ratings.
= The natural logarithm of the dollar amount of
the debt issue.
= The difference in the yield of the debt issue
and a comparable treasury issue.
= The corporate tax rate of the issuing firm.
= The yield on a treasury security issue.
= The difference in the yield of a 30 year and a
3 year Treasury security.
= Average change in interest rates over the 3
weeks prior to the debt issue.)









96
The interest coverage ratio, earnings before interest

and taxes divided by interest expense, is a measure of the

firm's ability to make its interest payments when due.

Interest coverage is measured on both a before and an after

tax basis (INTCOVBT and INTCOVAT). MAT is the maturity of

the debt issue in years. Profit margin (PM), net income

divided by sales, is a proxy for the firm's expected

profitability at the time of the debt issue. The profit

margin for the firm is measured ex-post in the year after

the debt was issued. SIZE is the natural logarithm of the

issue size in millions of dollars.

The growth rates for each firm measured for various

attributes are defined for two time periods. PREGROWTH is

the average growth rate of a selected variable over the

three year period prior to the actual debt issue date.

POSTGROWTH is the growth of the same variable during the

current year measured ex-post. The growth variables tested

include assets, dividends and earnings per share, gross and

net property, plant and equipment, sales, and long-term

debt. Both growth rates are proxies for future growth, but

PREGROWTH is know at the time of the debt issue and

POSTGROWTH is not known until after the debt is actually

issued.

Dummy variables are created for the issuer's debt

ratings, Aaa, Aa, A, Baa, Ba, B, Caa, and nonrated. An

alternative set of rating variables HIGH, MOD and LOW are











also utilized. HIGH is a binary variable for Moody's

highest ratings, Aaa or Aa. Similarly, MOD and LOW are

binary variables for Moody's moderate debt ratings (A or

Baa) and low debt ratings (Ba or lower including nonrated

debt), respectively. When either complete set of ratings

are utilized Aaa or HIGH are deleted.

Proxies for the level and variability of interest

rates are TSEC, SPREAD, SPREAD, and UNCER. TSEC is defined

as the yield to maturity on a fixed maturity Treasury

security measured in the secondary markets on the same date

the sample debt instrument was issued. The two maturities

utilized are 3 and 30 years. A measure for the variability

of interest rates, SPREAD, is calculated as the difference

between the yield of the debt issue and the yield to

maturity on a U.S. Treasury security having the same

maturity as the debt issue on the same date the debt

instrument was issued. SPREAD is defined as

SPREAD = Yieldit YieldT,t,

where i equals the debt issue used, t equals the
offering sate, and T equals the matched treasury
security.

The yield curve is proxied by TSPREAD, defined as the

difference between 30 and 3 year Treasuries,

TSPREAD = Yield30 yr,t Yield3 yr,t,

a negative (positive) value indicates a falling (rising)

yield curve.









98
The final interest variable, UNCER, is defined as the

mean absolute change in rates on 3 year U.S. Treasury

securities during the 3 weeks prior to the debt issue.

Formally UNCER is calculated as follows,

UNCER = [IX_3-X_21|+X_2-X_-1+XI-XO ]/3,

where X0 equals the interest rate on the 3 year
U.S. Treasury bond index reported in the Annual
Statistical Digest, Xt is the rate on the same
index reported t weeks before the date the bond
was issued, and | | signifies absolute value.

The average tax rate (ATAX) is defined as the firm's

total tax expense, including all income taxes paid to

federal, state, and foreign governments, divided by the

firm's pretax income. MTAX is the firm's marginal tax rate

and is based on the firm's pretax profits reported by the

Compustat tapes and derived from the tax schedules in effect

at the time of the debt issue.

Logit Model1

A logit regression model is utilized to test what

attributes determine the choice between issuing callable and

noncallable debt. The model is formally classified as a

univariate dichotomous model since it is only concerned with

the occurrence or nonoccurrence of the inclusion of the call

option when debt is issued.

The logit model uses the logistic distribution as a

probability function. One of the basic benefits of this

distribution is that it constrains the dependent variable to

lie between 0 and 1. The model coefficients are estimated




Full Text

PAGE 1

NONCALLABLE DEBT: EVIDENCE AND EFFECT BY RICHARD JOHN KISH A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1988 pipsrTY GF FLORIDA UBRAIiiw

PAGE 2

To my wife, Janine, in gratitude for her unfailing love, support, and encouragement. The many sacrifices she willingly made over the past 3 years helped make this study possible.

PAGE 3

ACKNOWLEDGMENTS Special thanks are due Professor Miles Livingston who chaired my supervisory committee. His guidance and support ensured the successful completion of this dissertation. I also thank Professor David T. Brown for his assistance and suggestions throughout the dissertation process and Professor James T. McClave for his help with the statistical tests.

PAGE 4

TABLE OF CONTENTS PAGE ACKNOWLEDGMENTS. . ill SYMBOLS VI 1 ABSTRACT .... IX CHAPTER I INTRODUCTION Notes 4 II LITERATURE REVIEW ... c D Call Option Hypotheses 7 Flexibility Hypothesis. ....... 7 Agency Hypothesis ." 8 Tax Hypothesis '.'.'.'. 13 Indenture Hypothesis '.'.'. 16 Call Deferments and Substitution ." ." .' .' 19 Value of Deferment ." 19 Substitution 22 Summary Notes .' 27 III SUMMARY STATISTICS 29 Testable Implications 34 Flexibility Tests . . 34 Agency Tests '.".'.' 36 Tax Test '.'.["'" 37 Substitution ""**_„ Data ::::::: l] Debt Offerings 38 Financial Data . 40 Treasury Yields ...... 40 Methodology ... Means Tests '.",",'. 41 Wilcoxon Rank Sum Test. ....... 42 Summary Statistics and Test Results.' . 42 General Variables 43 Specific Test Variables ....... 63 Summary ftl Notes '''•'.'.'.'.'.'.'.'. 87

PAGE 5

CHAPTER PAGE IV LOGIT REGRESSION ANALYSIS 89 Testable Implications 89 Flexibility Tests . . 91 Agency Tests . 91 Tax Test * ata ...::::: II Logit Model g8 Hypothesized Effects .... '..'.. 99 Results 102 Summary *"'*'.. Notes utes 118 V EVENT STUDY Data 122 Event Study Model 124 Hypotheses and Test Results. .' ." .' '. [ "128 Total Debt Issues !.'.'." 128 Callable versus Noncallable Debt. .' '. 131 Short versus Long Maturity Debt ... 132 Correctly versus Incorrectly Identified Firms 136 Short versus Long Maturity Callable Debt 1 39 Short versus Long Maturity Noncallable Debt 142 Additional Tests 145 S u r ar y .'.'.':: 148 Notes .,,-.. bl VI VALUATION OF THE CALL OPTION. 153 Regression Analysis 155 Identification 1 55 Valuation '.'.'' 159 Direct Estimation '.'.'.'.' 165 Summary 169 VII SUMMARY AND CONCLUSION 170 Introduction 170 Summary of Results ..........' 170 Future Research ,,, 1/6 Conclusion n 1/6 APPENDICES A SUMMARY TABLES 179

PAGE 6

PAGF B AVERAGE AND CUMULATIVE AVERAGE EXCESS RETURN TABLES 203 C AVERAGE YIELDS BY DATE, RATING, AND MATURITY. 215 REFERENCES _._ 235 BIOGRAPHICAL SKETCH 23g VI

PAGE 7

AR ATAX CALL CAR CLASS CPROFIT DA DE INTCOV MARKET MAT MTAX NCALL POSTGROWTH ( ) PREGROWTH ( ) PM SYMBOLS Average excess stock returns. Average tax rate for the issuing firm. A binary for the presence (1) or absence (0) of the call option. Cumulative average excess returns. Broad classifications for firm types which include financial (F) industrial (C) international (I), transportation (T) and utility (U) Percentage change in profits. The debt to asset ratio. The debt to equity ratio. The interest coverage ratio where INTCOVBT and INTCOVAT indicates the before and after tax ratios respectively. A measure of the current debt issue to the existing debt outstanding. The maturity of the debt issue in years. The marginal tax rate for the issuing firm. A binary for the presence (0) or absence (1) of the call option. The growth rate of the variable ( ) during the year the debt issue was floated. The average growth rate of the variable ( ) during the 3 years prior to the year the debt issue was floated. The profit margin of the issuing firm. VII

PAGE 8

RATE SIZE SPREAD T3 T30 TSEC TSPREAD UNCER The rating assigned to the debt issue reported as separate ratings (Aaa, Aa, through nonrated debt) or as rating groups (HIGH-Aaa or Aa, MOD-A or Baa, and LOW-Ba or lower including nonrated debt) The natural logarithm of the dollar amount of the debt issue. The difference between the yield on the debt issue and a comparable treasury issue. The yield on a 3 year Treasury security. The yield on a 30 year Treasury security. The yield of a Treasury security that matches the debt issue by maturity. The difference between the yields on 3 year and 3 year Treasury securities. The change in interest rates over the 3 week period prior to the date the debt was issued. vm

PAGE 9

J^J . f Dlssert ation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy NONCALLABLE DEBT: EVIDENCE AND EFFECT By Richard John Kish August 198 8 Chairman: Dr. Miles Livingston Major Department: Finance, Insurance, and Real Estate Financial theory claims that callable debt dominates noncallable debt. Yet our evidence shows that a substantial amount of noncallable debt exists, suggesting a deficiency in the theory. Over the period 1977 through 1986 over $108 billion worth of new noncallable debt was offered. During this ten year period, noncallable debt offerings accounted for 16.9 percent of the dollar value and 12.1 percent of the total number of all publicly placed corporate debt issued. The dominance of callable debt has been motivated in a variety of ways, for example uncertain interest rates. when there is a substantial decrease in rates, callable debt provides a firm's managers with the ability to refund an issue and take advantage of lower interest expenses. other reasons offered for the dominance of callable debt include agency problems associated with information asymmetries ix

PAGE 10

between borrowers and lenders, different risk tolerances of equityholders and debtholders, the need for managers to signal private information, differential tax rates between borrowers and lenders of funds, maturity preferences, and the opportunity to remove an undesirable protective covenant in the bond indenture. This study examines four questions related to noncallable debt issues. 1. Are there identifiable market factors that dictate when and how much noncallable debt is offered? 2. Are there factors common to the firms that issue noncallable debt that differentiate them from firms that issue callable debt? 3. Does the issuance of callable debt impact the market value of the firm's equity differently than the issuance of noncallable debt? 4. What is the value of the call option and does it vary across time? This study identifies several market and firm tendencies related to the use of noncallable debt. The answer to the third question depends upon how the debt issues were segmented. For issues in my data set, the average value of a call option is approximately 60 basis points, a value that varies across time. This mean value of the call is supported through both regression analysis and direct comparison of matched callable/noncallable debt pairs.

PAGE 11

CHAPTER I INTRODUCTION Financial theory claims that callable debt dominates noncallable debt. Yet our evidence shows that a substantial amount of noncallable debt exists, suggesting a deficiency in the theory. Over the period 1977 through 1986 over $108 billion worth of new noncallable debt was offered. During this ten year period, noncallable debt offerings accounted for 16.9 percent of the dollar value and 12.1 percent of the total number of all publicly placed corporate debt issued. The dominance of callable debt has been motivated in a variety of ways, for example uncertain interest rates. When there is a substantial decrease in rates, callable debt provides a firm's managers with the ability to refund an issue and take advantage of lower interest expenses. 2 Other reasons offered for the dominance of callable debt include agency problems associated with information asymmetries between borrowers and lenders, different risk tolerances of eguityholders and debtholders, the need for managers to signal private information, differential tax rates between the borrower and the lender of funds, maturity preferences, and the opportunity to remove an undesirable protective covenant in the bond indenture. 3 1

PAGE 12

2 The significant amount of noncallable debt issued during the period 1977 through 1986 warrants examination of four questions. 1. Are there identifiable market factors that dictate when and how much noncallable debt is offered? 2 Are there factors common to the firms that issue noncallable debt that differentiate them from firms that issue callable debt? 3. Does the issuance of callable debt impact the market value of the firm's equity differently than the issuance of noncallable debt? 4. What is the value of the call option and does it vary across time? The basic concepts used throughout this study are summarized below. Definition of a call Callability gives the firm's management the right to redeem outstanding bonds before maturity at a stated price. Without a call provision, debt can only be retired through an open market purchase. Elements of a call provision A typical call provision includes (1) a statement that the bond issue is redeemable in whole or in part at the option of the issuer, (2) the date(s) that the call option is exercisable, (3) the price the issuer is required to pay the bondholder for early redemption, and (4) any restrictions that are imposed on the

PAGE 13

3 early redemption, i.e. refunding versus redeeming restrictions. 4 Call restrictions Nearly all callable debt has deferments and restrictions on refunding and redeeming, a call deferment is a restriction in the indenture prohibiting the bond from being called prior to a specified date. Deferment constrains the lender, which reduces the value of the call. Deferments are typically 10 years for corporates, 5 years for utilities, and calls are restricted to the 5 years prior to maturity for long-term government issues. However this is a recent phenomenon; prior to 1958 call options were almost always unrestricted, i.e. exercisable by the issuer the day after the initial sale of the bonds. A nonredeemable issue restricts the issuer from exercising the call option under any circumstances for the time specified by the call deferment. A nonrefundable issue prohibits bond replacements financed by funds borrowed at a lower cost. For example, Bristol -Meyers Company called $25 million of its $75 million nonrefundable 8 5/8% debentures in 1973 using internally generated funds. The indenture restricted refunding before 1980, but no restrictions where placed upon redeeming the issue if funds became available from a source other than floating a lower coupon bond. 5 The following chapter reviews the literature on callable bonds. Summary statistics follow in Chapter III. A logit model based upon the tendencies found in the summary

PAGE 14

4 statistics for predicting whether a firm should issue callable or noncallable debt is provided in Chapter IV. Chapter V tests the effects of debt issues on the firm's equity value. A measure of the value of the call option is provided in Chapter VI. Finally, Chapter VII ties the analysis together in a conclusion. Notes 1. Figures calculated from debt data provided by Moody's Bond Survey summary pages for Corporate Bond Offerinqs for the years 1977 through 1986. 2. For example Pye (1966), Elton and Gruber (1972), Bodie and Friedman (1978), Bodie and Taggart (1978, 1980), and Van Home (1980, 1984) all list the uncertainty of interest rates as the primary reason for the existence of callable bonds 3. For example see Barnea, Haugen, and Senbet (1980) for agency costs associated with asymmetric information, risk taking, and growth; Bodie and Friedman (1978) for risk taking; Ross (1977), Leland and Pyle (1976), and Flannery (1982) for signaling; Boyce and Kalotay (1979b) Van Home (1984), and Marshall and Yawitz (1980) for taxes; Robbins and Schatzberg (1986) for maturity; Kidwell (1976) on standard indentures; and Pye (1966) on restrictive convenants. 4. Both Hess and Winn (1962) and Bodie and Friedman (1978) provide extended discussions on the call option. 5. Boyce and Kalotay (1979b) provide additional details on call restriction.

PAGE 15

CHAPTER II LITERATURE REVIEW Most research on corporate finance assumes that noncallable debt is an insignificant portion of the debt market. For example, Barnea, Haugen, and Senbet (1980) claim that almost all corporate debt is callable. This "almost all" statement is made without substantiation. Other examples are plentiful. Boyce and Kalotay (1979b) and Van Home (1984) both claim that virtually all corporate bonds have a call feature written into the indenture agreement. Thus, much of the literature addresses the guestion of why corporations almost never issue noncallable long-term bonds. For example, see Bodie and Taggart (1979) 1 Several guotes reinforce the assumed insignificance of noncallable debt. "Throughout the twentieth century, corporate bonds issued in the United States have not been pure bonds; in almost all cases the security purchased by an investor has consisted of a pure bond less an option, retained by the issuer, to call the bond at a specified price" (Bodie and Friedman, 1978, p. 20). "Most attention [within corporate financing research] centers on the almost invariable inclusion in corporate bonds of a call" (Boyce and Kalotay, 1979, p. 825). Robbins and Schatzberg state 5

PAGE 16

6 that "financial theory has found only moderate success in explaining the routine inclusion of call provisions among the covenants of corporate bonds . [and later declares that there exists a] . near universality of call provisions in corporate bond contracts" (Robbins and Schatzberg, 1986, p. 935) The insignificance assumption is also common in the leading finance textbooks. The following guotes offer a representative sample of this treatment of noncallable debt. Brigham contends that "Most bonds contain a call provision which gives the issuing corporation the right to call the bonds for redemption" (Brigham, 1986, p. 382 and Weston and Brigham, 1987, p. 710). Ross and Westerfield state that "Almost all publicly issued corporate long-term debt is callable" (Ross and Westerfield, 1988, p. 333). This is not just a recent phenomenon; an early edition of Van Home's textbook states that "Nearly all corporate bond issues provide for a call feature, which gives the corporation the option to buy back bonds at a stated price before their maturity" (Van Home, 1977, p. 570). Investment texts also adhere to this viewpoint. For instance, Radcliffe writes that "Long-term corporate debt obligations are usually term bonds with maturities of five years or more [and that] . most corporate bonds are sold with a deferred call provision" (Radcliffe, 1987, p. 38).

PAGE 17

7 Call Option Hypotheses The existing literature suggests that the use of the call option is motivated by (at least) four separate, but not mutually exclusive, hypotheses: flexibility, agency, tax, and indenture. Flexibility Hypothesis The flexibility hypothesis states that the call provision provides the firm's management with increased flexibility in the face of market uncertainties, primarily interest rate uncertainty. The call option allows the issuer to replace a higher cost obligation with a lower cost bond as interest rates decline. 2 Refunding is profitable provided the interest savings outweigh the call premium, flotation costs, and legal expenses. In an efficient market, interest savings are priced out. From the debtholder's viewpoint, callable debt consists of a long position in a noncallable debt security and a short position in a call option retained by the equityholders. The value of the call option to the debtholder equals the value to the issuer. Therefore at issuance, the market price reflects the expected value of the call privilege. 3 There seems no reason, a priori, that corporate management is better able to forecast future interest rates than debt buyers, a majority of which are professional money managers for financial institutions, pension funds, and life

PAGE 18

insurance companies. Thus, interest cost reduction as a reason for the routine inclusion of the call option rests on a tenuous assumption. 4 Refunding allows management to extend or shorten the maturity of its debt. 5 Other flexibility benefits include the opportunity to simplify financial structure, to reduce the debt to equity ratio of the firm, to lengthen the firm's average debt maturity, to eliminate indenture restrictions, or to retire unwanted bonds when a corporation accumulates large cash balances for which it has no immediate needs. 6 However, these benefits can be obtained by purchasing bonds in the open market. Ross (1977) justifies the call option as a tool available to management for altering the capital structure of the firm. A counterview, offered by Stiglitz (1979), is that financial policies are irrelevant provided the investment opportunity set remains unchanged by financial policy. Thus, the call option is redundant. The basic arguments of the flexibility hypothesis by author are summarized in Table 2-1. Agency Hypothesis The above arguments for the existance of call options are constructed assuming symmetric information across claimholders. Not surprisingly, it is difficult to motivate call options in a perfect market setting. If capital structure is irrelevent in such a framework, then the

PAGE 19

Table 2-1 Proponents of the Flexibility Hypothesis Author (Date) Claim (Variables tested) Bowlin (1966), Jen and Wert (1967), and Pye (1966) The use of callable debt is related to a potential for interest savings sometime in the future. Periods with high rates should see a proportionate increase in the number of calls issued. (As TSEC, SPREAD, and UNCER increase the use of the call feature should also increase. ) Ross (1977) The call option allows the manager to alter the capital structure of the firm. (High DA and DE ratios should be associated with callable debt.) Stiglitz (1979) Financial policies are irrelevant even with risky debt, provided the investment opportunity set remains unchanged by financial policy. The maturity structure of debt as well as the other complexities, such as the call option, do not affect firm value. Predicts no abnormal change in the equity value of the firm when debt is issued. (The CAR's of the firm should equal when debt is issued.) Van Home (1984) The use of the call option is directly related to the interest rate cycle. During periods with low interest rates, there is no need for the automatic inclusion of the call option on debt issues. (As MAT, TSEC, SPREAD, and UNCER increase the use of the call 'feature should also increase.) Note: Symbols are defined in the preface and within each chapter when the variable is used.

PAGE 20

10 specifics of debt contracts cannot matter. See for example Modigliani and Miller (1958) and Krause (1977). However, in the presence of private action and private information, it is well known that capital structure and the specific nature of debt contracts effect the value of the firm. Call options can be motivated as an endogenous contractual response to agency problems between debt and eguity claimants. Call options mitigate some of the self-serving incentives of eguityholders in the levered firm. Risky debtholders broaden their collateral base anytime a firm improves its position. Therefore, when the firm makes a profitable investment, part of the benefit goes to debtholders. Since equityholders are unable to reap the full benefits of additional investments, certain positive net present value projects are not undertaken. 7 The agency cost of debt in this case is the foregone projects. If the bonds are callable, the debtholders' gain when firm value rises is limited to the call price. Thus, investment incentives are more consistent with firm value maximization. The potential gain to bondholders is fixed and equity captures the full marginal benefit of any new project. Barnea, Haugen, and Senbet (1980) point out that callable debt also mitigates incentives to increase risk beyond the levels consistent with firm value maximization in

PAGE 21

11 order to shift wealth from debtholders to equityholders. As wealth is shifted away from debtholders, the value of the securities declines, and so does equities option to call. As a result, any wealth shifts are partially offset by declines in the value of the call. However, debt with the option to convert, completely mitigates equities incentives to shift risk, casting some doubt on the role of call options in mitigating incentives to shift risk. In an empirical study, Thatcher (1985) shows that refunding protection offers the best solution to the agency costs associated with risky debt. Support is offered in the form of simple means tests and a discriminant analysis on the difference between firms that issue debt with a regular call feature versus those that use a two tiered call provision. Other claims made within the Thatcher study are that default risk is a major reason for the inclusion of the call option and that firms with high debt to asset ratios have an increased probability of using the call feature. It should be noted that short-term debt provides investment incentives similar to callable debt. With short-term debt, the terms of the debt contract are frequently renegotiated to reflect the firm's current investment policy. 8 The essential difference between short-term debt and long-term callable debt arises only with respect to uncertainty about the date of prepayment. The extent to which these kinds of securities are substitutes is examined in detail within Chapter VI.

PAGE 22

12 Flannery (1986) develops a model where managers signal private information about the value of the firm's assets through the choice of debt maturity. Low private valuation, "type" firms prefer to issue long-term debt to avoid the transaction costs and the revaluation of credit worthiness that occurs with each rollover. High valuation types, desiring to reveal their true worth, are willing to pay the transaction costs associated with rolling over short-term debt in order to avoid being pooled with low quality firms and being forced to have a higher cost of debt. Short-term noncallable or long-term callable debt are credible signals of high firm quality. While Flannery 's model does not cast light on the issue of which firms issue callable versus short-term noncallable debt, it does predict that the issuance of short-term noncallable or callable debt will be viewed more favorably by the capital market then issuance of long-term straight debt. 9 In a signaling model hinging on managerial contracts, Robbins and Schatzberg (1986) also find that short-term debt is a signal for the "good" prospects of a firm, but that long-term callable debt has superior risk sharing attributes. The following claim is supported by a numerical example, "The manager of a firm with Good News, in signaling the firm's better prospects through callable bonds, can achieve a reduction in the risk to his or her compensation compared to what can be achieved through the issuance of

PAGE 23

13 short-term debt" (Robbins and Schatzberg, 1986, p. 945). They further argue that both noncallable debt and eguity are dominated securities and therefore signal bad news. Research dealing with agency costs as an explanation for the existence of the call option is summarized in Table 2-2 by author. Tax Hypothesis According to Boyce and Kalotay (1979) both the issuer and the buyer benefit from callable debt at the expense of the government. The difference in the average tax rates between a profitable corporate borrower (high marginal tax rate) and the typical lender (low marginal tax rate) generates a preference for callable bonds. The exercise of the call results in a reduction of the tax liability of the issuer which is not offset by the additional taxes paid by the lender. The marginal corporate tax rate during the time of the Boyce and Kalotay study was approximately 50%, whereas the dominate lenders during the same time period had marginal tax rates less than 50%. Corporate bondholders are typically tax-exempt (pension funds) or in the low tax brackets (life insurance companies) Marshall and Yawitz (1980) support the tax argument by pointing out that call premiums are deductible from ordinary income as an expense to the borrower, but treated by the lender as a capital gain. The change in the current tax

PAGE 24

14 Table 2-2 Proponents of the Agency Hypothesis Author (Date) Claim (Variables Tested) Barnea, Haugen, and Senbet (198 0) Callable debt is used as a means of resolving agency problems associated with informational asymmetry, managerial risk incentives, and foregone growth opportunities. Shortening the maturity of debt and issuing long-term debt with a call option perform identical tasks in eliminating agency problems. (The issuance of short-term debt should produce positive CAR s ) Barnea, Haugen, and Senbet (1985) Noncallable debt is harmful to the equityholders of the firm. (The issuance of noncallable debt should produce negative CAR's and the issuance of callable debt should produce positive CAR's.) Bodie and Taggart (1978) Through modeling, the authors show the dominance of callable debt to both equity and noncallable debt. Predicts that growth opportunities may give equityholders a definite preference for callable debt. (As PREGROWTH and POSTGROWTH increase, the use of the call feature should also increase.) Flannery (1986) Using the signaling argument, the author models the fact that good firms use short-term debt as a signal that they are in fact good firms as transaction costs increase. (As CPROFIT, PM, INTCOV and RATE increase the use of the call feature on debt issues should also increase. Short-term debt and long-term callable debt should produce positive CAR's.) Robbins and Schatzberg (1986) Shows through modeling that callable debt dominates short-term debt, equity, and noncallable debt of any maturity length. (The CAR's from long term callable debt should be greater than the CAR's from short-term callable debt. The CAR's from noncallable debt should signal bad news and therefore should be negative.)

PAGE 25

15 Table 2-2 — continued Author (Date) Claim (Variables Tested) Thatcher (1985) Callable debt is used as a method for reducing agency costs due to future investment opportunities. Predicts that high growth firms issue callable debt Also predicts that firms with high debt ratios, high concentrations of long-term debt, and high probability of default dictate the use of the call option, specifically the two tiered call option. (High PREGROWTH, POSTGROWTH, DA, DE, and low PM should increase the use of callable debt.) Note: Symbols are defined in the preface and within each chapter when the variable is used.

PAGE 26

16 laws weakens the tax motivation. Since a difference still exists between the corporate and individual tax rates, the tax hypothesis is not without some merit. The tax arguments are summarized in Table 2-3. Indenture Hypothesis The indenture hypothesis cites standardization as a major factor for the inclusion of the call option by most major corporations. 10 Standard indentures enhance a bond's liquidity and therefore increase its value to the buyer. Common features cited historically include deferment periods of 10 years for industrials including banks and finance companies, 5 years for utilities, and restricted use of calls to the last 5 years of maturities for long-term government issues. Kidwell (1976), studying the use of the call option by state and local governments, found that the inclusion of the call provision was determined by precedents established in the market place, statutory requirements, and the historical tradition of issuers. The expected economic savings resulting from lower interest rate levels was not a significant factor. Interest rate levels did not appear to affect the decision to include a call provision in a bond issue. Whether these conclusions can be applied to the corporate sector is discussed in later chapters. A summary of Kidwell 's position related to the indenture hypothesis can be found in Table 2-4.

PAGE 27

17 Table 2-3 Proponents of the Tax Hypothesis Author (Date) Claim (Variables Tested) Boyce and Kalotay (1979a, 1979b) The tax effect, which results from the difference in marginal tax rates between a profitable corporate borrower and the typical lender dictates the use of callable debt. (High ATAX and MTAX dictates the use of callable debt.) Marshall and Yawitz (1980) The U.S. tax laws create a bias in favor of the inclusion of call provisions on corporate debt, therefore firms issuing callable debt should have a higher marginal tax rate than firms issuing noncallable debt. (High ATAX and MTAX dictates the use of callable debt.) Note: Symbols are defined in the preface and within each chapter when the variable is used.

PAGE 28

Table 2-4 Proponent of the Indenture Hypothesis Author (Date) Claim (Variables Tested) Kidwell (1976) Callable debt is issued with greater frequency as the length of the call deferment period increases and as interest rates increase. (Callable debt issues increase as the length of the call deferment increases. High TSEC is not a factor in the issuance of noncallable debt.) Note: Symbols are defined in the preface and within each chapter when the variable is used.

PAGE 29

19 Call Deferments and Subtitution Previous research has examined the determinants of differences in the value of the call option across various issues. For example, the interest rate environment, coupon rate, and length of deferment period have been examined as determinants of the value of the call option. A question also arises concerning the "maturity" of a callable bond. When the bond is called the actual, realized maturity is less then the stated maturity. The effective maturity the market places on the bond is determined by comparing the realized yields on callable bonds with the realized yields of straight debt. If the realized yields are the same, the bonds are close substitutes. Value of Deferment The value to the bondholder of deferment protection has been estimated in three ways. First, various investment bankers and regulatory agencies opinions about the value of deferment were surveyed. Second, simple yield comparisons were made across bonds with different periods. Finally, regression techniques are used to more accurately control for nondif ferment effects on yields. Hess and Winn (1962), in two call deferment surveys, reveal conflicting results. The first survey included life insurance companies, banks, and trust companies. The question asked was what value should be placed on a call deferment attached to a 5 1/4% 30 year Aa rated utility

PAGE 30

20 bond. The median values place on 5 and 10 year deferments were 15 and 25 basis points respectively. For a nonrefundable feature, the estimated cost was 30 to 36 basis points. The second survey involved three federal regulatory agencies: the Securities and Exchange Commission, the Federal Power Commission, and the Interstate Commerce Commission. The question addressed in this survey dealt with the cost of including a call feature that was immediately callable. The consensus was that no substantial evidence exists that immediately callable bonds had higher yields then bonds with a deferment period. Simple comparisons of the bond market from 1926 through 1959 failed to disclose any significant relationship between the offering yield and the call features of corporate bonds. During the period 192 6 through 194 3, Hess and Winn analyzed 572 corporate bonds of which 5 were noncallable and 11 had call deferments of 5 years or more. Comparing immediate callables with bonds having call deferments of one year or more failed to reveal any value placed on the call privilege by the debt market. A similar finding was found during their study of 332 corporate bonds issued during the period 1944 through 1955. Of the 332 issues only 13 contained call deferments, 12 of which were private placements and not rated.

PAGE 31

21 The third phase of the Hess and Winn study covered the period 1956 through 1959. During this 4 year phase, debt issues were restricted to publicly issued industrial and utility bonds rated A or better in issue amounts of $5 million or larger. A total of 361 debt issues met this set of criteria, of which 105 contained call deferments. Again they failed to find a significant value for the call option. Only in the latter half of 1959 was a detectable value for the call option uncovered. During the last half of 1959, bonds with 5 year call deferments carried yields 13 to 20 basis points lower than freely callable issues with the same ratings issued during approximately the same time period. The third method for measuring the magnitude of the yield differential required for call deferments of various lengths was the focus of Pye's regression analysis in 1976. Yield was regressed against a set of dummy variables for time, 5 year call deferment, and ratings (Aa and A with Aaa used as the base case) The only other independent variable was maturity. The results revealed an approximate cost of 4 to 13 basis points of the 5 year call deferment during low (2 3/4% 4%) and high (4% 5 1/4%) interest rate levels respectively for the period 1959 through 1966. Pye's regression results were reinforced with a simulation based on a transitive probability matrix of one year interest rates from 1900 to 1966. The model relied

PAGE 32

22 upon two basic assumptions, (1) that investors rank lotteries by their expected values and (2) that the one period interest rate follows a Markov chain. Estimated costs on 5 year call deferments were similar to his direct estimations— -5 to 10 basis points. Projections for longer deferments ranged from 40 to 70 basis points for low (2 1/8%) and high (5 7/8%) interest rate periods respectively. Substitution Substitution is the opportunity to exchange debt issues without a material loss in realizable yields. When analyzing the preference in maturity structure, Jen and Wert (1967) found that the difference in the promised yields between long-term callable and short-term noncallable debt is explained by the risk difference in the possibility of the call. Although a difference in the promised yields was detected, the realized yield differences were negligible. This lack of difference in the realized yields on long-term callable and short-term noncallable debt support the substitution of the two debt types. Bodie and Friedman (1978) note that intermediate term notes, which are typically not callable until maturity, broaden the range of an investor's selection. These noncallable issues often offer maturities which equal the periods of call protection for the callable bonds, thereby offering alternatives to long-term callable bonds. The key to the substitution effect is that even though the promised

PAGE 33

23 yields at the time of issuance are higher for the callable bonds, the realized yields are essentially identical. Bodie and Friedman's position is summarized in Table 2-5. Two differences in the use of shortand long-term debt that might mitigate their substitutability for issuers are refinancings and rollover risks. with short-term debt, several refinancings may be necessary to fund a project. Each refunding includes flotation costs; therefore the total flotation costs of rolling over several short-term debt offerings may be higher than that associated with a single long-term debt issue. The use of short-term debt in anticipation of refinancing at lower rates runs the risk of an incorrect forecast. If rates increase, refinancing with short-term debt must take place at higher rates. Long-term callable debt provides protection against an increase in rates by locking in the borrowing rate for the maturity of the debt instrument. Summary Key attributes associated with the call option include increased flexibility, a reduction of agency costs, tax benefits, and the use of standard indentures. The central flexibility argument deals with the uncertainty of interest rates, but simplifying financial structure, changing debt maturities, eliminating indenture restrictions, and retiring unwanted bonds also have merit. Agency arguments revolve around the opportunity to mitigate the self-serving

PAGE 34

24 Table 2-5 Proponent of the Substitution Effect Author (Date) Claim (Variables Tested) Bodie and Friedman (1978) Short and intermediate term debt issues are substitutes for long term callable debt. (Average call deferments should be approximately egual to the average maturities of noncallable debt issues. Note: Symbols are defined in the preface and within each chapter when the variable is used.

PAGE 35

25 incentives of equityholders in a levered firm and to signal private information to potential investors. if tax differentials exist, then the issuance of callable debt becomes a positive sum gain between the borrower and lender of funds. The gains are at the expense of the government in the form of reduced taxes. Finally, standard indentures established by precedents in the market place, statutory requirements, or historical tradition of issuers are also hypothesized as a reason for the inclusion of the call option. A summary of the various viewpoints, as well as the aspect of the bond issue most effected by their position, is provided in Table 2-6. Testable implications are outlined within the next three chapters. The literature review outlined three methods used in prior research for valuing the call deferment. They included surveys, matched pair comparisons, and regression analysis. Matched pairs of callable and noncallable debt will be used in Chapter IV to estimate the value of the call option, thus eliminating the need to extrapolate the cost of the call option from call deferments. This value is supported by regression analysis. Finally, the substitution of long-term callable and short-term noncallable debt issues was suggested based upon the realizable yields available on both sets of debt instruments.

PAGE 36

Table 2-6 Variables to Test Hypothesized Explanations For the Call Option 26 Authors (Date) Table a : Variables Tested Barnea, Haugen, and Senbet (1980) Barnea, Haugen, and Senbet (1985) Bodie and Friedman (1978) Bodie and Taggart (1978) Boyce and Kalotay (1979a, 1979b) Flannery (1986) Jen and Wert (1967) Kidwell (1976) Marshall and Yawitz (1980) Pye (1966) Robbins and Schatzberg (198 6) Ross (1977) Stiglitz (1979) Thatcher (1985) Van Home (1984) A

PAGE 37

27 Notes 1. Marshall and Yawitz (1980) find that the majority of research on bond refunding is concerned with explaining why the call provisions are normally included on corporate bond issues. 2. For example, see Bowlin (1966), Pye (1966, 1976) Jen and Wert (1967), and Van Home (1980, 1984). 3. See Barnea, Haugen, and Senbet (1985), Kraus (1973) and Myers (1971) for the effects of callable debt in efficient markets. 4. A weakness pointed out by Kraus (1973) is that a policy of issuing only noncallable bonds is as good as that of issuing callable bonds and subsequently following an optimal refunding policy within informationally efficient markets Bodie and Taggart (1978) state that in an efficient capital market, callable and noncallable corporate bonds should coexist with the price differential between them reflecting the value of the call option. Note that even if management were systematically smarter, the market would fail. 5. Elton and Gruber (1972) find that if the new issue has a later maturity than the issue it is refunding, then the new issue assures the borrower funds for a longer period of time at a lower interest rate. 6. Cash surpluses could result from a number of circumstances such as exceptionally profitable operations, compulsory sale of property to a public body or to satisfy requirements for mergers, loss of insured property through catastrophe, and liquidation of inventories and receivables. In the case of a finance company, cash may accumulate because of restrictions on installment credit or similar restrictions. 7. An example of the externalities caused under the agency heading is the manner in which growth opportunities are handled. Bodie and Taggart (1978) suggest that when a firm has future discretionary investment opportunities, the call feature may not be a zero sum game between the debtholders and the equityholders. Risky debtholders participate in changes in the firm's fortunes. 8. Leland and Pyle (1976) use the entrepreneur's equity stake as a signal. The greater the equity stake of the entrepreneur, the more reliable the signal that the firm is "good". See Campbell and Kracaw (1980), Ross (1977), and Leland and Pyle (1977) for resolution of the moral hazard problem. See Allen, Lamy, and Thompson (1987) for a

PAGE 38

2 8 discussion of relationship between the information revealed by management and the perceived riskiness of the firm's debt. 9. See Myers and Majluf (1984) for a summary of related signaling arguments. 10. See Smith and Warner (1979) for a discussion on bond covenants.

PAGE 39

CHAPTER III SUMMARY STATISTICS Evidence suggests that noncallable debt has been erroneously overlooked under the mistaken notion that it accounts for an insignificant portion of the debt market. The percentage of issues that are noncallable during the ten year period, 1977 through 1986, ranges from 2.9% to over 20% (average 12.1%). As a percentage of the dollar amount of issues during the same time period, the range is even higher from 3.5% to over 25% (average 16.9%). The bar graphs in Figures 3-1 and 3-2 illustrate the proportion of noncallable debt measured by the number and the dollar amount of issues during the ten year sample period. The dollar amount of noncallable debt is substantial, ranging from a low of $900 million in 1979 to over $3 6 billion in 1986 (average $10.8 billion). The yearly dollar amounts for both the callable and the noncallable debt issues are graphed in Figure 3-3. The majority, but not all, of this noncallable debt has short to intermediate term maturities. Figure 3-4 graphs the frequency distributions for the maturity levels of both callable and noncallable issues over for the total sample period. At this point we examine the following two questions: (1) Are there identifiable market factors that dictate when 29

PAGE 40

30 21%

PAGE 41

31 1981 1982 Year Debt Issued Figure 3-2 Noncallable Debt As a Percentage of the Total Dollar Debt Issued During the Ten Year Period 1976 1986 Source: Percent values based upon debt issues reported in Moody's Bond Survey (1977-1986)

PAGE 42

32 180 170 160 150 140 130 120 1 10 100 90 30 70 60 50 40 30 20 10 1981 1982 Year Debt Issued Figure 3-3 Dollar Amount of Debt Issues by Year Callable versus Noncallable Debt Note: //// Callable Debt \\\\ Noncallable Debt Soruce: Dollar amounts of debt based upon debt issues reported in Moody's Bond Survey (1977-1986)

PAGE 43

33 35% 30% 10-14.9 15-19.9 20-24.9 25-29.9 Maturity in Years Figure 3-4 Maturity Distribution of Debt Issues 1977-1986 Callable versus Noncallable Debt Note: //// callable Debt \\\\ Noncallable Debt Source: Maturity frequencies compiled from Moody's Bond Survey (1977-1986)

PAGE 44

34 and how much noncallable debt is offered? and (2) Are there factors common to the firms that issue noncallable debt that differentiate them from firms that issue callable debt? The analysis of the tendencies identified is continued within the logit regression found in Chapter IV. Testable Implications The testable implications for callable/noncallable debt issues follow from Table 2-6, which is reproduced as Table 3-1. This summary of the authors discussed in the literature review contains a list of the variables needed for testing the call option hypotheses. Flexibility Tests Tests of the flexibility hypothesis focus on the level of interest rates as well as their volatility. As the level or volatility of rates increase, the value and hence the popularity of the call option should increase. A simple test of this hypothesis is to see whether there is an increase in the proportion of callable versus noncallable debt issued during periods of high uncertainty. Volatility is proxied by the variability of interest rates during the three weeks prior to the debt issue. The proxy for the level of interest rates is the yield on a fixed maturity Treasury bill. if this yield increases, the use of the call option should also increase. The spread between the 3 and 30 year Treasury bills indicates the shape of the yield curve. If the long-term Treasury is greater

PAGE 45

Table 3-1 Variables to Test Hypothesized Explanations For the Call Option 35 Authors (Date) Table 3 : Variables Tested Barnea, Haugen, and Senbet (1980) Barnea, Haugen, and Senbet (198 5) Bodie and Friedman (1978) Bodie and Taggart (1978) Boyce and Kalotay (1979a, 1979b) Flannery (1986) Jen and Wert (1967) Kidwell (1976) Marshall and Yawitz (1980) Pye (1966) Robbins and Schatzberg (1986) Ross (1977) Stiglitz (1979) Thatcher (1985) Van Home (1984) A

PAGE 46

36 (smaller) than the short-term Treasury, then the yield curve is rising (falling). if the yield curve is falling (rising), the market anticipates interest rates will fall (rise) in the future. Agency Tests Callable bonds, as opposed to straight debt, provide eguityholders incentives to maximize the total value of the firm with respect to investment decisions. In particular, callability internalizes the external benefits to risky debt holders when a firm undertakes a value enhancing project. This implies an advantage to using callable debt. The likelihood that the call option will be used depends on the following variables. First, the more risky or lower rated the debt, the more bondholders can gain in the absence of a call option. Second, the lower the firm's profit level, the greater the possibility for improving the firm's position. Third, when a great deal of the value of firms with potentially high growth lies in future projects, the greater the call option's value. Finally, firms that are highly levered, in addition to the increased likelihood that they have low debt ratings, have a larger co-insurance problem because they have a larger amount of debt outstanding. Flannery (1986) predicts that there is information content in the choice of debt financing. Firms with good private information issue short-term or callable debt while

PAGE 47

37 firms with bad private information issue long-term straight debt. An ex-post varification of the signaling hypothesis occurs if a firm's profitability increases after the issuance of callable and short-term debt versus long-term noncallable debt. Later, the profitability results are compared with the stock price announcement effects. Tax Test Tax motivations for callable debt suggest that the marginal and average tax rates of firms issuing callable debt should be higher then those issuing noncallable debt. Issuers are concerned about expected tax rates over the life of the issue. Current tax rates proxy the firm's expected tax rates. Substitution Test Certain arguments suggest that noncallable debt is a substitute for callable debt. The maturity on the noncallable debt is hypothesized as being equal to the maturity of the call deferment period. Thus, the maturity of the average noncallable bond is compared with the average call deferment period as a means of testing whether the two investment options are similar. Data Information on each debt issue was gathered for testing whether market factors are in fact unique to either callable or noncallable debt issues. Additional data were gathered about each of the issuing firms to test the

PAGE 48

38 hypothesis about why certain firms issue callable debt and others noncallable debt. Debt Offerings Moody 's Bond Survey is the primary source of data on the individual debt offerings. Information on each debt offering includes company name, date of issue, size of offering (in millions of dollars), years to maturity, coupon, yield to maturity, firm classification (utility, industrial, financial, transportation, or international), restrictions (length of call deferments, nonrefunding, and nonredeeming) initial call price, sinking fund, converting provisions, and debt rating. Data include both noncallable and callable debt issued over the ten year period 1977 through 1986. This period was chosen to test for the popularity of callable debt during different interest rate cycles, i.e. periods with stable rates, increasing rates, and declining rates. Figure 3-5 shows the fluctuation in rates during the sample period through graphs of the average yields on Aaa and Baa corporate bonds and 3 year Treasury bills. Stable interest rates were common during two periods within our sample, January 1977 through June 1979 and during the year 1986. Rates were increasing over two separate time periods. The first occurred during the three year period July 1979 through June 1982. The period July 1983 through June 1984 was the second time span of increasing rates. Declining

PAGE 49

39 Figure 3-5 Average Aaa and Baa Corporate Bonds and 3-year U.S. Treasury Bill Yield Distributions 1977 1986 Note: Top line Average Baa Corporate Bond Yields Middle line Average Aaa Corporate Bond Yields Bottom line Average 3-yr U.S. Treasury Bill Yield Source: Annual Statistical Digest 1977-1986 Board of Governors, Federal Reserve System

PAGE 50

40 interest rate periods were represented from July 1982 through June 1983 and from July 1984 through December 1985. Financial Data Financial data on the individual corporations that issued debt during the years 1977 through 1986 came from the Industrial Compustat Tapes. 1 Data were gathered for the year the debt was issued and the three years prior to this date. Corporate data include average tax rate, total debt outstanding, total assets, dividends and earnings per share, gross and net property, plant, and equipment, sales, profit margin, and the firm's debt to equity and debt to assets ratios. Treasury Yields Treasury yields are compiled from the annual Statistical Digest. 2 These yields were matched with the debt issues by date of issuance and maturity for analysis in the logit chapter. Methodology Tests utilized to determine significant differences between the issuance of callable and noncallable debt include the t-test for means and proportions, the F-test for variances, and the nonparametric Wilcoxon rank sum test for probability distributions. The comparisons of proportions, means, variances, and population distributions are used as a quick and simple method for detecting differences in the attributes of callable and noncallable debt. The analysis

PAGE 51

41 of the attributes, taking into consideration the presence of the other variables, is continued within Chapter IV. Means Tests The standard t-test is utilized to test for the differences between the population proportions and means of the callable and noncallable debt issued. The null and alternative hypotheses tested when comparing proportions of the various attributes are defined in eguations 3-1 and 3-2. H : p c,i ~ p n,i = (3-1) No difference in the proportions from the callable and the noncallable samples for attribute i exist. H l s P c,i p n,i (3-2) There is a difference in the proportions from the callable and the noncallable samples for attribute i where P = proportion; c = callable; n = noncallable; i = attribute. The attributes checked within the proportion analysis include industry type (industrial, transportation, utility, finance, international), bond rating (Aaa, Aa, . caa, Not Rated), maturity, yield, coupon, dollar amount of the issue, and the inclusion of a sinking fund, conversion feature or floating rates. A similar set of hypotheses is defined to test the difference in the means and standard deviations of the financial data plus the yield, coupon, dollar amount of the debt issue, and maturity. See note 3 for the set of hypotheses for the means and standard deviation tests.

PAGE 52

42 Wilcoxon Rank Sum Test Since many of the attributes tested under the standard t-test may violate the assumption of normality, the Wilcoxon rank sum test is also used. 4 The Wilcoxon is used to test the hypothesis that the probability distributions associated with two populations are equivalent. The key attribute of this test is that no assumptions have to be made about the shape of the population probability distributions. The only assumptions needed are (1) the two samples are random and independent and (2) the observations can be ranked in order of magnitude. The hypotheses tested by the Wilcoxon Rank Sum Test are defined by statements 3-3 and 3-4. H Q : The two sampled populations, callable and noncallable debt issues, have identical probability distributions. (3-3) H-l: The probability distribution for the noncallable debt issues is shifted to the right or to the left of the probability distribution for the callable debt issues. (3-4) Summary Statistics and Test Results Summary statistics are provided for two sets of variables: general variables and variables identified for specific tests. General variables are those variables not specifically identified for testing a specific hypothesis. They are used for detecting subtle differences between the callable and noncallable debt sectors. Included within the general category are coupons, yields, maturity, size of issue, number of issues per firm, firm classifications, and

PAGE 53

43 other miscellaneous factors. Specific test variables include debt ratings, debt ratios, profitability, various growth components, tax rates, and the level and volatility of interest rates. General Variables Coupons and yields The average difference between the coupons attached to the debt issues of the two groups is insignificant, 10.56% versus 10.53% for the callable and noncallable groups respectively. This is analyzed in more detail in Chapter VI, where I account for differences in maturity, rating, and time of issuance. A comparison of the coupon between callable and noncallable debt in 1981 actually shows the mean of the average coupons on noncallable debt is greater than the mean of the average coupons on callable debt. This abnormality in mean rates is partially explained by the time of issuance. The noncallable debt issues during 1981 were clustered in the last quarter of the year when interest rates were high. This is in contrast to the issuance of callable debt which is fairly regular throughout the year. The mean, standard deviation, and the various difference tests for the overall period are reported in Table 3-2. For the frequency distributions of coupon rates on an annual basis and the distribution of issues by month see Appendix A, Table A-l and A-2 respectively.

PAGE 54

Table 3-2 Mean Coupon Summary for the Period 1977-1986 44 Mean Standard Deviation Sample Size t-score a F-score z-score c Callable Noncallable 10.56% 2.84% 5763 issues 0.2492 1.16 ** 0.6293 10.53% 3.06% 796 issues fo Test for a difference in the means. c Test for a difference in the standard deviations, Test for a difference in distributions. (nonparametric Wilcoxon test statistic) *Significant at 10% level using a one-tail test Significant at 5% level using a one-tail test. Significant at 1% level using a one-tail test.

PAGE 55

45 Promised yields at issuance are also examined to detect if systematic deviations of the issue price for the two types of securities exist. The mean (standard deviation) for the callable and noncallable groups are 11.37% (5.49%) and 10.96% (2.51%) respectively. The difference between the mean yields for the callable and noncallable debt issues is significant when the overall ten year period is tested and within most of the individual sample years. See Table 3-3 for the summation of the mean, standard deviation, and test results for the overall sample period. The same attributes on a yearly basis are shown in Appendix A, Table A-3 Another way of illustrating the differences in the yields between the two groups is by comparing the frequency distributions of their respective yields. For example, in 1977 none of the noncallable debt issues carried promised yields of over 10%, but six percent of the callable sector is found within that segment. Another example of the yield differences between the two groups is in the 1986 results. Debt yields less than 9% averaged 49.1% versus 68.9% respectively for the callable and the noncallable groups. The frequency distributions for yields summarized for the overall ten year period is found in Table 3-4. Annual frequency tables are located in Appendix A, Table A-4 By comparing the overall summary table with those found in the Appendix, it is evident that the yield differences hold up

PAGE 56

Table 3-3 Mean Yield Summary for the Period 1977-1986 46 Mean Standard Deviation Sample Size t-score a Callable 11.37% 5.49% 5585 issues Noncallable 10.96% 2.51% 766 issues F-score z-score c 3 5437*** 4.77 ** 1.4900* fa Test for a difference in the means. c Test for a difference in the standard deviations, Test for a difference in distributions. (nonparametric Wilcoxon test statistic) Significant at 10% level using a one-tail test. Significant at 5% level using a one-tail test. ***Significant at 1% level using a one-tail test.

PAGE 57

Table 3-4 Yield Frequency Summary for the Period 1977-1986 47 Percentage

PAGE 58

48 across all 10 years of the sample. Appendix A, Table A-5, reinforces these differences across firm classification. Estimates of the value of the call option based upon the differences between callable and noncallable debt issues matched by rating and maturity are discussed in Chapter VI. Maturity Since the option to call long-term debt is a means of reducing the true maturity of a debt issue, one might expect callable debt to have longer maturities than noncallable debt on average. For the entire 10 year period, the average maturity for the callable issues is 19.0 years with a standard deviation of 9.1 years. For the noncallable group, the average is significantly less. The noncallable mean maturity is 8.0 years with a standard deviation of 6.1 years. These differences are statistically significant at over the 1% significance level for the entire ten year period and for each individual year. Null hypothesis 3-3, that the two sampled populations have identical probability distributions, is also rejected at a 1% significance level. This result reinforces the differences between the maturities of the callable and noncallable samples detected by the t-test. See Table 3-5 for the summary statistics for the overall period and Appendix A, Table A-6 for the annual results. Maturity tests are further broken down by industry classifications: financial (F) industrial (C) and utility (U) Within all three categories, the null hypotheses were

PAGE 59

Table 3-5 Mean Maturity Summary for the Sample Period 1977-1986 49 Mean Standard Deviation Sample Size t-score a Callable 19.0 years 9.1 years 5841 issues Noncallable 8.0 years 6.0 years 837 issues F-score z-score c 33.8707*** 2 30 *** 33.6720*** fc Test for a difference in the means. c Test for a difference in the standard deviations. Test for a difference in distributions (Wilcoxon) Significant at 10% level using a one-tail test. Significant at 5% level using a one-tail test. ***Significant at 1% level using a one-tail test.

PAGE 60

50 rejected. Since the individual test on the firm subcategories yield similar results to those outlined for the overall sample, the specific results are not reported. Another way of highlighting the differences in the maturities is through the use of freguency tables. Table 3-6 shows the freguency distribution of maturities for the overall ten year period. The yearly distributions for the maturity frequencies can be found in Appendix A, Table A-7 The maturities in both the table and the appendix are grouped into five year clusters. Two of the key factors shown in the freguency tables are the significant differences in the low and high ends of the distributions. The callable issues with maturities of 5 years or less range from to 4.9% with an average of 1.9% for the entire ten year period. Contrasting this is the noncallable group which ranges from to 35% with an average of 20.6%. At the other extreme is the percentage of long term issues within each group. Callable issues with maturities of at least 20 years average 55.0% with a range of 33.7 to 75.8% over the years 1977 through 1986. Over the same maturity group, the noncallable issues averaged 5.3% with a range of to 13.6%. The maturity structure is expanded across the three main firm classifications (industrial, finance, and utility) in Table 3-7. The results reinforce the differences in the maturity structure of the callable and the noncallable

PAGE 61

51 Table 3-6 Summary of Debt Issues Maturities for the Period 1977-1986 Maturity

PAGE 62

Table 3-7 Maturity By Firm Classification Summary Number of Issues per Maturity Class 52

PAGE 63

53 groups regardless of firm classification. For example, in the industrial category the callable group has only 1.3% of the reported issues with maturities 5 years or less while the noncallable group has 22.8%. Another extreme example is found in the utility category. Percentages of 58.1 and 6.9 are found within the 3 years or greater maturity set for the callable and noncallable groups respectively. Size of issue. A summary of the dollar amounts of the individual issues shows a difference between the callable and noncallable groups. The callable debt issues averaged $90.0 million with a standard deviation of $106.4 million. This is contrasted by the mean and standard deviation of the noncallable group, $132.9 million and $110.1 million. The nonparametric Wilcoxon rank sum test was also performed. Test results for the overall period and the yearly periods are shown in Table 3-8 and Appendix A, Table A-8 respectively. The results show that on average, noncallable debt issues are larger than callable debt issues. Freguency distributions of the individual issues by amounts were examined. For example, 36.50% of the callable debt issues were $50 million or less while only 12.46% of the noncallable issues were found in this category. Table 3-9 summarizes the issue amount freguency distribution for the overall sample period. This is expanded in Appendix A, Table A-9 which shows the freguency table for the dollar amount of issues on a yearly basis and Table A-10 which

PAGE 64

Table 3-8 Mean Issue Amount Summary for the Period 1977-1986 54 Mean Standard Deviation Sample Size t-score a F-score z-score c Callable $ 90. d $106.4 5890 issues Noncallablp. $132. 9 l $110.1 818 -10.7706*** 1.07 -16.3127*** issues fa Test for a difference in the means. c Test for a difference in the standard deviations. Test for a difference in distributions (nonparametric Wilcoxon test statistic) Dollar amounts in millions. Significant at 10% level using a one-tail test **Significant at 5% level using a one-tail test ***Significant at 1% level using a one-tail test'

PAGE 65

Table 3-9 Issue Amount Frequency Summary for the Period 1977-1986 55 Dollar

PAGE 66

56 gives a further breakdown of the dollar amount frequencies by firm classification. Results are consistent across the various classifications. Number of issues per company The number of debt issues per company summarized on a yearly basis shows an insignificant difference between the callable and noncallable groups. The proportions of firms with only 1 debt issue were between 70 and 80% (76.3% versus 71.6% for callable and noncallable respectively) This similarity can be found throughout the frequencies for the other categories of issues per firm. See Table 3-10 for the overall frequency distribution of issues per firm and Appendix A, Table A-ll, for the frequency table on issues per firm by year. Thus, the number of times a firm enters the debt market does not seem to effect the decision of whether a firm issues callable or noncallable debt. Total assets. The overall sample, limited to debt issues from firms on the Compustat data tapes show a wide dispersion in the average asset values between the callable and noncallable sectors. The means for the callable and noncallable sectors are $9.7 billion and $21.7 billion, respectively. Large firms are more likely to issue noncallable debt then small firms. This difference holds up across the various firm classifications with the difference narrowing in both the financial and utility classes. See Table 3-11 for a broader summation of the means, standard

PAGE 67

57 Table 3-10 Number of Debt Issue per Firm for the Period 1977-1986 # Issues per Company 3 1

PAGE 68

58 Table 3-11 Total Assets of Issuing Firms during the Period 1977-1986 Callable Noncallable Mean Standard Deviation Sample Size t-score a F-score z-score c $9.7 24.5 2106 issues -7.4744*** 1.76 *** -5.4396*** $21.7 32.7 464 issues b Test for a difference in the means. c Test for a difference in the standard deviations Test for a difference in distributions. (nonparametric Wilcoxon test statistic) Significant at 10% level using a one-tail test. Significant at 5% level using a one-tail test. ***Significant at 1% level using a one-tail test. Dollar values in millions of dollars,

PAGE 69

59 deviations, and test statistics in regards to the total assets carried by the debt issuing firms and Appendix A, Table A-12 for the summary results by firm classification. Firm classification Moody's classifies firms offering debt by five broad categories: industrials (C) financial (F) transportation (T) utility (U) and international (I) Summary statistics show that the proportion of issues within the callable and noncallable sectors varies across firm classifications. For instance, only 45.0% of the callable issues were financial, compared with 49.6% for the noncallable issues. A major difference in firm type proportions occurs within the international sector. Only 2.5% of the callable debt issues were offered by internationals compared with 16.3% for the noncallable group. A summary of firm classifications is shown in Table 3-12 for the overall ten year period. The differences between the proportions of the two groups, callable and noncallable, are highly significant under each firm classification except for the proportions of debt offered within the transportation sector. The insignificance in the difference of proportions within the transportation sector is not surprising given the small number of firms from the sample found within the category. Significance levels are also found to be fairly consistent for each of the individual years within the sample. Appendix A, Table A-13, offers the test results and summary proportions of the firm classification on a yearly basis.

PAGE 70

60 Table 3-12 Firm Classification Summary for the Sample Period 1977-1986 Callable Debt Noncallable Debt # issues % issues # issues % issues t Industrial 1824 30.9% 202 24.2% 3.94*** Utility H46 19.4 58 7.0 8.79*** Financial 2654 45.0 414 49.6 -2.54*** Transportation 130 2.2 24 2.9 1 22 International 148 2.5 136 iel3 -18.' 56*** Total 5902 100.0% 834 100.0% Significant at 10% level using a one-tail test, Significant at 5% level using a one-tail test. Significant at 1% level using a one-tail test.

PAGE 71

61 Another test of debt type distributions within the firm classifications utilizes SEC classifications. A breakdown of the firms within the sample which are listed on the Compustat tapes are summarized by industry code. The number of firms within this subsample differs from the original sample used. Only 3313 debt offerings are included of which 2770 are callable and 543 noncallable. The decrease in the number of firms is due to the fact that the industry code classifications were obtained from the Compustat data tape and that not all of the firms in the original sample are listed with the Compustat file. The SEC classifications reinforce the findings from using the broad classifications C, F, I, T, and U. For example, the number of firms found in the 6000-6999 sector verifies that a significant portion of the Compustat sample firms issuing noncallable debt are financial. The proportions for the callable and noncallable debt offerings for firms listed on the Compustat tapes within this 6000 level of industrial codes are 28.2% and 60.8% respectively. Appendix A, Table A-14 summarizes the freguency of issues by industry code for the ten year sample period. 5 The significance of the difference of the proportions within the financial sector offers support for the signaling argument. The finance industry is hypothesized as having the most to gain by withholding the exact nature of their asset makeup due to the fact that it could be easily

PAGE 72

62 duplicated. Therefore, a means of signaling quality to the investment public is needed. A number of signals have been suggested in the financial literature. The elimination of the call feature on the debt issue by substituting short-term debt offers a means for reinforcing these signals. 6 Other factors The number of debt issues carrying floating rates is relatively small over the time period under study. It averaged only 2.69% and 6.11% for callable and noncallable debt categories respectively. This difference is significant at the one percent level. The prevalence of floaters among noncallable debt is not surprising. if a bond contains a floating rate, its coupon rate moves with the general level of interest rates. Thus, its price does not deviate much from par, i.e. its duration is close to zero. Hence, callability is basically worthless. Floating rate debt is a recent phenomenon and the frequencies should increase within both categories over time, especially if the market experiences another period of widely fluctuating rates. Similarly, the use of convertible debt differs depending on the type of debt offered. The percentiles are 11.01% versus 0.84% for the callable and noncallable groups respectively. This difference is also highly significant. Again the results are not surprising. Convertible debt and callable debt are both argued to mitigate agency problems.

PAGE 73

63 Granted one deals with under investing and the other risk shifting. However, when managers have a great deal of latitude, it is not surprising that both are used. See Table 3-13 for a summary of the float and convertible features of the debt issues within the sample period. Specific Test Variables In this section, tests of the variables hypothesized to effect the decision to issue callable versus noncallable debt are analyzed. First, variables proxying the advantage of callable debt in mitigating the underinvestment agency problem are examined. The proxies include debt ratios, profitability, and growth. Next, tax rates of issuers are examined to determine if, as predicted, firms that issue callable debt have higher corporate tax rates. Finally, the interest rate enviroment is examined to see it the value of the call influences the probability that it will be included. Ratings Debt ratings are reported by the broad classifications Aaa, Aa, A, through Caa. Although the major ratings Aa, A, through B, are further classified into three subcategories by Moody's starting in 1982, only the broad classifications are reported. 7 A difference in the quality ratings of the debt offerings for the noncallable and callable groups is detected for the overall ten year period as well as for the individual years.

PAGE 74

Table 3-13 Floating Rate and Converting Debt Summary For the Period 1977-1986 64

PAGE 75

65 For the ten year period, the percentage of the debt issues rated A or better are 80.3% and 44.7% for the noncallable and callable issues respectively. This difference is significant at the 1% level. The percentages within the callable sample with an A or better rating range from a high of 79.4% in 1977 to a low of 33.4% in 1986. The percentages within the noncallable group are more consistent throughout the period, ranging in the high seventies and above for the entire ten year period. The low of 77.6% for the noncallable group occurred in 1985, while the high of 100% is shown in 1978 and 1979. Another major difference in ratings between the callable and noncallable groups is shown by the proportion of nonrated issues within each group. The proportion of nonrated debt is 28.3% to 1.6% for the callable and noncallable groups respectively. If the nonrated group is not included in the percentage calculations, the differences between the remaining classes are reduced. For instance, the proportion of the noncallable group in the A or better category is practically unchanged at 81.6%. The callable group's percentage for the same class, A or better, increases to 62.3%. The difference is still significant at the 1% confidence level. The summary of ratings with and without the nonrated group generates similar results for the individual years within the study. The ranges for the A or better category

PAGE 76

66 excluding nonrated debt issues were 57.8% through 79.8% and 77.6% through 100% for the callable and noncallable debt groups respectively. This reinforces the fact that the differences detected within the debt ratings remain regardless of whether or not the nonrated sector is included in the analysis. An interesting aside is that the number of nonrated debt instruments reported by Moody's is almost nonexistent before 1981. In the callable debt group, the nonrated bond percentages ran from approximately 0% through 198 to a high of 47.3% in 1986. The noncallable group had approximately 0% in all years except 1981 (12.2%), 1982 (3.6%), and 1983 (3.3%). Reasons for a debt issue not being rated include: (1) An application for rating was not received or accepted by Moody's. (2) The issue or issuer belong to a group of securities or companies that are not rated as a matter of policy by Moody's. (3) There is a lack of essential data pertaining to the issue or issuer. (4) The issue was privately placed, in which case the rating is not published in Moody's publications. Note that reason number 4 does not pertain to the sample used in this dissertation. Table 3-14 shows the breakdown of the ratings for the callable and noncallable debt groups as an absolute number and as a percentage of the issues within each rating category. A similar rating summary can be found in Appendix A, Table A-15, for the callable and noncallable issues on a yearly basis.

PAGE 77

Table 3-14 Rating Summary on the Sample Period 1977-1986 67

PAGE 78

68 The differences between the proportion of debt issues within all of the individual rating categories for callable versus noncallable issues are statistically significant at greater than a 5% significance level when null hypothesis 3-1 is tested on the entire 10 year period. The t-tests, segmented on a yearly basis, also support the fact that a greater proportion of higher grade debt instruments were issued in the noncallable category versus the callable category. Thus, the results show that the null hypothesis is rejected for the overall ten year period and for most of the individual years within the ten year period. The respective t-values resulting from testing the proportion of callable and noncallable debt within category i (where i equals the debt ratings Aaa, Aa, through Caa and nonrated debt) are reported on a yearly basis in Appendix A, Table A-15. A comparison of the yearly results with the overall summary found in Table 3-13 show a consistency in the debt rating across the sample period. The results obtained within the rating analysis indicate that high quality firms are more likely to issue noncallable debt then lower quality firms. One reason for this type of behavior deals with the externality problem. If debt is highly rated, the firm's managers do not have to worry about co-insuring the debt. The call option fails to buy anything. A second reason is linked to the ability of high quality firms to obtain favorable rates when the debt

PAGE 79

69 contract is offered. Therefore, a need for later recall is lessened. Debt ratios The debt to asset and debt to equity ratios gage the relationship of the amount of debt to the supporting base. The results show that even though the callable debt firms on average have lower asset bases, they have a higher average debt to asset ratio (0.27 versus 0.23). Thus evidence exists that firms issuing callable debt have less debt coverage for the amount of debt issued when measured against an asset base. The tendency for having a higher debt ratio is supported by the debt to equity ratios. Callable debt issuing firms have an average debt to equity ratio of 1.63 compared to 1.54 for firms issuing noncallable debt. The debt ratio summary, shown in Table 3-15, support the signaling arguments of debt choice. The debt ratios by firm classifications are summarized in Appendix A, Table A-16. Profit measures Various measures of profitability are gathered to detect differences between the two debt issuing groups. First, an ex-post test of the signaling hypothesis involves comparing precentage changes in profitability. The percentage changes measure the profits before and after the debt issue. Four specific comparisons were made: callable versus noncallable, short-term noncallable/long-term callable versus long-term noncallable, short-term noncallable/long-term callable versus long-term

PAGE 80

Table 3-15 Debt Ratios of Issuing Firms during the Period 1977-1986 70 Debt to Asset Ratio Mean Standard Deviation Sample Size t-scoref* F-score z-score c Debt to Equity Ratio Mean Standard Deviation Sample Size t-score a F-score z-score c Callable 0.276 0.160 2105 issues Noncallable 0.229 0.169 463 issues 5.6384*** 1.12 5.7254*** Callable Noncallable 1.634 12.019 2105 issues 0.3352 74.62 *** 3.9340*** 1.544 1.391 463 issues b Test for a difference in the means. c Test for a difference in the standard deviations. Test for a difference in distributions. (nonparametric Wilcoxon test statistic) Significant at 10% level using a one-tail test. Significant at 5% level using a one-tail test. ***Significant at 1% level using a one-tail test.

PAGE 81

71 noncallable/short-term callable, and short-term noncallable versus long-term callable. The percentage change in profit tests fails to uncover any statistical difference within any of the comparison groups. Results are shown in Table 3-16. Other profit variables include dividends and earnings per share and profit margin. A statistically significant difference is detected for the variable dividends per share, but as shown in the logit analysis in Chapter IV this difference is not useful for predictive purposes. Earnings per share and profit margins before and after taxes failed to show a statistically significant difference. Test results for the second set of variables are shown in Appendix A, Table A-17. This lack of difference in profitability is inconsistent with the theory that predicts an information content in the choice of debt financing. Growth. Several growth measures are compiled including growth in assets, debt, earnings and dividends per share, property, plant, and equipment (net and gross), and sales. Two types of growth variables are calculated, pre and post. Pregrowth is the average growth rate of the various variables over the three year period prior to the debt issue examined. 8 Postgrowth is calculated ex-post as the growth rate over the year the debt was issued. Postgrowth is used as a proxy for unanticipated growth. Pregrowth rates for asset, debt, and sales are found to be significantly different for the noncallable and

PAGE 82

Table 3-16 Percentage Change in Profitability Before and After the Debt Issue During the Period 1977-1986 72 Noncallable versus Callable Mean -0.44% -0.65% Standard Deviation 5.67 2.53 Sample Size 663 issues 2250 issues t-score* -0.3467 F-score^ 2.63 *** z-score c -0.1342 S-T Noncallable /L-T Callable versus L-T Noncallablp. Mean -0.45% -0.77% Standard Deviation 4.89 2.76 Sample Size 1961 issues 79 issues t-score -0.9699 F-score D 3.13 *** z-score c -o!o995 S-T Noncallable/L-T Calla b le vs S-T Noncal 1 able/S-T Callable Mean -0.45% -0.65% Standard Deviation 4.89 3.24 Sample Size 1961 issues 952 issues t-score a F-score b .c -1.3665 2.28 *** z-score^ -0.5684 S-T Noncallable versus L-T Callable Mean -0.51 -0.42 Standard Deviation 2.22 5.66 Sample Size 584 i ssue s 1377 issues t-score a 0.4875 F-score" 6<50 *** z-score c 0.7902 c a Test for a difference in the means. Test for a difference in the standard deviations. Test for a difference in distributions, (nonparametric Wilcoxon test statistic) Significant at 10% level using a one-tail test. Significant at 5% level using a one-tail test. Significant at 1% level using a one-tail test.

PAGE 83

73 callable debt firms. The postgrowth variable show significant differences for assets, property, plant, and equipment (both gross and net figures) and sales. Table 3-17 summaries the means and representative test results. The results from the individual firm classification mirror the results found in the overall sample and therefore are not reported. Tax rate. The average tax rates between firms issuing callable and noncallable debt are approximately the same. The mean values are 31.8% and 33.8% for the callable and noncallable debt issuing firms. Using the standard t-test, no significant difference between the mean values of the two groups is detected. This result is not supported by the Wilcox rank sum test which finds a significant difference in the distributions of the two debt classes when average tax rates are analyzed. This difference in test results is primarily due to the wide variation within the standard deviation of the tax rates for the callable debt firms. The analysis of the marginal tax rates shows statistically significant differences between the firms issuing callable and noncallable debt. The t-test indicate significant difference favoring the marginal tax rates of the noncallable debt issuing firms which is contradictory to the tax hypothesis. This results is questionable due to the skewness of the distributions caused by a limited upper bound.

PAGE 84

74 Table 3-17 Growth Rates During the Period 1977-1986 Callable Noncallable Average Growth Rates — During the 3 years Period prior to the Debt Issue Total Assets 18.9% Dividends per Share 3 4 Earnings per Share 5^3 Total Debt 47# 6 Property, Plant, and Equipment Gross Net Sales i 8 [7 11.0%*** 2.4 0.5 18.8 27.7 30.6 31.1 31.9 10. *** Average Growth Rates During the year of the Debt Issue Total Assets Dividends per Share Earnings per Share Total Debt Property, Plant, and Equipment Gross Net Sales 27.5%

PAGE 85

75 The Wilcox rank sum test is a more accurate measure of the relationship between the marginal tax rates of the two types of firms because of this violation of the normalcy assumption. The Wilcox test supports the tax hypothesis, but as will be shown within the logit analysis in Chapter IV, this difference does not add to the predictive power of the logit model. Table 3-18 contains the summary results from the tests on both the average and marginal tax rates. The results are consistent across the firm classifications as shown in Appendix A, Table A-18. Thus, the tests of the tax hypothesis show inconclusive results. Interest rate level The level of interest rates are measured using 3 and 30 year Treasury yields. A freguency distribution of callable and noncallable debt issues by interest rate levels is found in Table 3-19 for the 3 year Treasury yields and in Appendix A, Table A-19 for the 30 year Treasury yields. Neither distribution shows any tendencies for an increase in the issuance of callable debt as interest rates increase. The test of null hypothesis 3-1 at all levels of interest rates show insignificant results. A test of the issuance of callable debt due to the shape of the yield curve, whether increasing or decreasing shows significant results. Callable debt dominates the issuance of noncallable debt when the yield curve is decreasing, i.e. when the difference between 3 and 3 year Treasuries is between -1 and 0%. But when the yield curve

PAGE 86

Table 3-18 Average and Marginal Tax Rates for Issuing Firms During the Period 1977-1986 76 Average Tax Rates Mean Standard Deviation Sample Size t-score a F-score b z-score c Marginal Tax Rates Mean Standard Deviation Sample Size t-scoref* F-score b z-score c Callable 31.8% 172.2 2106 issues Noncallable 33.8% 21.4 464 issues -0.5103 65.00 *** -3.0489** Callable 32.3% 14.2 2096 issues Noncallable 37.6% 9.99 463 issues -9.5576*** 2 04 *** 6.3402*** b Test for a difference in the means. c Test for a difference in the standard deviations Test for a difference in distributions. (nonparametric Wilcoxon test statistic) Significant at 10% level using a one-tail test. Significant at 5% level using a one-tail test. ***Significant at 1% level using a one-tail test. Dollar values in millions of dollars,

PAGE 87

Table 3-19 Issue Frequency Summary by Interest Rate Level For the Period 1977-1986 77 Interest

PAGE 88

78 is rising, noncallable debt dominates. A rising yield curve is defined as a positive spread. See Table 3-20 for the frequency distribution of debt issues based upon the spread between 3 and 3 year Treasuries. The uncertainty of interest rates, measured as the average absolute change in 3 year Treasury rates over the three week period prior to the issuance of the debt, is a marginally significant factor in the decision of whether or not to include the call feature. Only during periods of low uncertainty is one type of debt dominated by the other. The percentage of noncallable debt issues dominates callable percentages when the average absolute change in 3 year Treasuries is less than 25 basis points. Although this difference is statistically significant, Chapter IV shows that the difference is not useful in the prediction of debt types. Uncertainty of rates, tested using 30 year Treasuries, produces similar results. Both sets of uncertainty frequencies are reported in Table 3-21. Substitution. The substitution hypothesis is analyzed from the perspective of restrictions on the call deferment. The hypothesis is that noncallable debt has the same maturity as the time of the call restriction. The average call deferment period is compared to the average maturity for the noncallable debt sector using the null hypothesis defined below. H Q : Average = Average Maturity Call Deferment Noncallable Debt Issue

PAGE 89

Table 3-20 Issue Frequency Summary by Interest Spread For the Period 1977-1986 79 Interest Spread

PAGE 90

Table 3-21 Issue Frequency Summary by Uncertainty Level For the Period 1977-1986 80 3 year Treasur Uncertainty Level

PAGE 91

Where the average call deferment is calculated as the average of the maximum length of time between the refunding or redeeming restrictions of the callable debt issues. For the overall period of the sample, the mean of the refunding restrictions are 5.14 years versus 7.37 years for the average maturity of the noncallable debt issues excluding zero coupon debt. Using the t-test for comparing the means of these two values shows a difference at greater than a 1% significance level. Although there exists a statistically significant difference between the call deferment and noncallable maturities, the difference is negligible from a practical standpoint. Since not all callable debt is actually called at the first opportunity, a substitution effect is realistic. Thus, the difference in maturity structures fails to refute Bodie and Friedman's (1978) claim that noncallable debt is a substitute, although an imperfect one, for long term callable debt. A summary of the test results can be found in Table 3-22. Summary Many of the attributes of issues and issuers are different across the two debt sectors, callable and noncallable. The signaling argument is supported by the results reported within the ratings and firm classification sections. Tendencies detected within the rating section indicate that on average the ratings of noncallable debt issues are higher than those found within the callable debt

PAGE 92

Table 3-22 Call Deferment vs Noncallable Maturity Summary for the Period 1977-1986 82 Mean Standard Deviation Sample Size t-scoref* F-score b z-score c Callable 5.14 yrs 3.11 2070 issues Noncallable 7.37 yrs 5.52 538 issues 9.0066 *** 3.16 ** 6.5324 *** fc Test for a difference in the means. Test for a difference in the standard deviations, Test for a difference in distributions, (nonparametric Wilcoxon test statistic) Significant at 10% level using a one-tail test. Significant at 5% level using a one-tail test. Significant at 1% level using a one-tail test.

PAGE 93

83 sector. Within the firm classification section, support was found for financial firms to issue noncallable debt as one of the signals available for conveying private information without revealing its underlying asset components. The maturity and yield sections supported the general belief that on average the maturities and yields of callable debt is greater than these found within the noncallable sector. Size of the firm is also detected as a significant factor in that the larger the debt issue and the larger the firm size, the greater the probability that an issue is noncallable. Nonsignificant differences are detected for both the average tax rates and the profitability between firms issuing the two types of debt. The level of rates is insignificant as a factor influencing the issuance of callable and noncallable debt. But the shape of the yield curve is significant with callable debt dominating during downward sloping yield curves and noncallable dominating during rising yield curves. Uncertainty of rates was shown to be marginally significant. General tendencies and test results are summarized in Tables 3-23 and 3-24 respectively. These major differences are the initial starting point in the variable selection for the logit analysis in Chapter IV.

PAGE 94

84 Table 3-23 Callable and Noncallable Debt Issue Trends Period 1977-1986 Callable Noncallable Ratings (A or Better Rating) 44.7% 80.3% Trend: Noncallable debt has a higher rating on average versus callable debt. Firm Classification (Financial) 45.0% 49.6% Trend: Financial firms, the industry type with the most to gain from signaling are more prevalent within the noncallable debt sector. Maturity 19.0 yrs 8.0 yrs. Trend: Callable debt has an average maturity over twice that found on noncallable debt issues. The average call restriction, 5.14 years, was considerably shorter then the average noncallable debt maturity. Yie lds 10.56% 10.53% Trend: As can be expected the granting of the call option is not a free good. Size of Debt Issue $90.0 million $132.9 million Trend: There is a tendency for only large firms to be able to obtain favorable rates on noncallable debt. Total Assets $9.7 billion $21.7 billion Trend: Reinforces the fact that small firms are reluctant issue noncallable debt. Average Tax Rate 31.8% 33.8% Trend: No significant difference in the tax rates of the two firm type was detected. Profit Margin (after tax) 7.5% 7.7% Trend: No significant difference in the two type of firms was detected.

PAGE 95

Table 3-24 Level and Sign of Significance Between Callable and Noncallable Debt Variables Period 1977-1986 85 Variable

PAGE 96

Table 3-24 — continued 86

PAGE 97

Notes 87 1. Standard & Poor's Compustat Services, Inc. (Annual Data Tapes.) 2. Annual Statistical Digest is published by the Board of Governors of the Federal Reserve System annually. 3. The null hypothesis for comparing the means from the callable and noncallable groups is: H, mean. = mean n,i Similarly, the null hypothesis used for testing egual variances between the callable and noncallable groups is: H, var c,i var n,i 4. See McClave and Benson (1980) for a full description of the Wilcoxson Rank Sum Test, pages 675-679. 5. The Compustat tapes firms from the original the SIC classifications 0000-0999 1000-1999 2000-2999 3000-3999 4000-4999 5000-5999 6000-6999 7000-7999 8000-8999 do not include the international sample. A listing of the meaning of can be summarized as follows: Agricultural Mining (Metal, Coal, oil, and Gas) and Construction. Food, Drugs, and Lumber Related Products Production: Rubber, Leather, Glass, Concrete, Metal Goods, Machinery, Appliances, etc. Services: Transportation (Railroads, Trucking, Water, and Air) Communication (Telephone, Cable, Radio, and, TV), Electric, and Gas. Merchandising: Auto, Sporting, Durable Goods, Drugs, Apparel, Drugs, Grocery, and Jewelry. Finance: Banks, S & L's, CU's, Insurance, and Real Estate. Personnel: Hotel, Advertising, Data Processing, Consulting, Auto Repair, and Theaters. Medical: Hospitals, Nursing Homes, Labs, Education. 6. Another explanation for the large percentage of noncallables within the financial sector is due to the matching of asset and liability maturities. If matched completely, there is no need for the call feature. But the

PAGE 98

fact that many noncallables also exist for nonfinancial firms weakens this argument as the sole reason for noncallable debt. 7. Beginning on April 26, 1982, five of Moody's nine corporate bond ratings were expanded to included numerical modifiers 1, 2, and 3. The five bond ratings with the expanded format include: Aa, A, Baa, Ba, and B. For example, an Aal rating indicates that the security meets all of Moody ^s criteria for a double-A rating and that it ranks at the high end of that rating category. The modifier 2 indicates that the security is in the mid-range of its category, and a 3 indicates that the bond is nearer the low end of its generic category. Moody's emphasize that the numerical modifiers are only refinements of the defined categories and that the relative positions of all of Moody's corporate bond rating symbols, and their definitions remain unchanged. See pages 2181-2182 of Moody's Bond Survey (April 26, 1982) for a broader description of the ratinq categories. 8. Five year average growth rates were also calculated with similar results obtained.

PAGE 99

CHAPTER IV LOGIT REGRESSION ANALYSIS The examination of the first 2 questions raised in the introduction, (1) Are there identifiable market factors that dictate when and how much noncallable debt is offered? and (2) Are there factors common to the firms that issue noncallable debt that differentiate them from firms that issue callable debt?, was initiated in Chapter III. The logit regression is an extension of the summary statistics analysis. By testing the significance of these tendencies in a logit regression, the effects of interactions are considered. Highly correlated variables are redundant for the prediction process. Thus, the logit analysis is used to determine if any of the statistically significant differences uncovered in Chapter III are useful in the prediction of the debt type issued. Testable Implications The testable implications for callable/noncallable debt issues outlined in Chapter III which are applicable within this chapter are repeated for clarity. Also Table 3-1 is reproduced as Table 4-1 for easy reference of the variables utilized in the logit model as they relate to the various authors in the literature review. 89

PAGE 100

Table 4-1 Variables to Test Hypothesized Explanations For the Call Option 90 Authors (Date) Table 3 : Variables Tested Barnea, Haugen, and Senbet (1980) Barnea, Haugen, and Senbet (1985) Bodie and Friedman (1978) Bodie and Taggart (1978) Boyce and Kalotay (1979a, 1979b) Flannery (1986) Jen and Wert (1967) Kidwell (1976) Marshall and Yawitz (1980) Pye (1966) Robbins and Schatzberg (1986) Ross (1977) Stiglitz (1979) Thatcher (1985) Van Home (1984) A

PAGE 101

91 Flexibility Tests The level and uncertainty of interest rates enhance the value of the call option. Therefore, the variables related to the level and uncertainty of rates within the market should be positively related to the probability of witnessing a call option. Agency Tests Asymmetric information is a broad category and as such different sets of variables are utilized to test subsets of the theory. First, high risk/low rated debt implies a greater gain to bondholders in the absence of a call option. Second, low profitability increases the possibilities for future improvements in the firm's position. Third, high growth associated with future projects implies increased potential opportunities. Finally, high leverage/low debt ratings increases co-insurance problems. This set of tendencies implies an advantage to using callable debt and hence the increased likelihood of attaching the call option to the debt issue. Tax Test Tax motivations for callable debt suggest that the tax rates of firms issuing callable debt should be higher then those issuing noncallable debt. Both average and marginal tax rates are utilized in the analysis.

PAGE 102

92 Data The debt offering data discussed in Chapter III is restricted to include only those issues where the issuing firm's financial data is also available on the Compustat Tapes. This reduces the data set to 2570 debt issues of which 2106 are callable and 464 noncallable. Mean summary data and frequency probabilities are provided in Tables 4-2 and 4-3 respectively. The tendencies identified in Chapter III hold with this subset of data. The additional data on Treasury security yields are compiled from the Annual Statistical Digest as explained in the previous chapter. Based on the existing literature, the logit model uses the explanatory variables shown in Table 4-4. All the firm level variables are obtained from the Compustat tapes except where indicated. Three binary variables for the broad classifications of firm type (CLASS) are industrial, utility and finance. Due to the small number of firms in the sample from the international and transportation sectors, firms falling into these two classifications were dropped from this part of the analysis. The two book value ratios, debt to equity (DE) and debt to assets (DA) proxy the firm's leverage. MARKET is the size of the debt issue relative to the total amount of consolidated funded debt the firm has outstanding, including the relevant issue. MARKET is calculated as MARKET = current debt (current debt + prior debt) The measurement of the prior debt is based on book value.

PAGE 103

93 Table 4-2 Mean Value Summary Statistics For Potential Logit Regression Variables

PAGE 104

Table 4-3 Frequency Statistics For Potential Logit Binary Dummy Variables 94 Total Debt Callable Debt Noncallable Debt # of (%total) # of (%total # of (%total issues issues /%call) issues /%noncall) Rating Groups (2570 issues) : LOW (Debt rating Ba or lower) 670 (26.07%) 640 (24 90%/30 39%) 30 (1. 17%/6.47%) MODERATE (Debt rating A or Baa) 1217 (47.35%) 935 (36 38%/44 40%) 282 (10. 97%/60. 79%) HIGH (Debt rating Aaa or Aa) 683 (26.58%) 531 (20 66%/25 21%) 152 (5. 91%/32 76%) Maturity Groups (2557 issues) : Short Maturities (Maturities < 10 years) 615 (24.05%) 302 (11. 81%/14 41%) 313 (12 24%/67 90%) Long Maturities (Maturities > 10 years) 1942 (75.95%) 1794 (70 16%/85 59%) 148 (5 79%/32 10%) Firm Classifications (2570 issues) : Industrials 1219 (47.43%) 1075 (41 83%/51 04%) 144 (5 60%/31 03%) Financial 724 (28.17%) 460 (17 90%/21 84%) 264 (10. 27%/56. 90%) Utilities 545 (21.21%) 509 ( 19 81%/24 17%) 36 ( 1 40%/7 76% ) Calculations based upon logit regression data set. Percent figures reported as percentages of the total data set used and as percentages of the type of debt, e.g. callable or noncallable debt subsets.

PAGE 105

Table 4-4 Functional Form of the Logic Regression Equation 95 NCALL = f (CLASS, DA, DE, INTCOV, MARKET, MAT, POSTGROWTH, PREGROWTH, PM, RATE, SIZE, SPREAD, TAX, TSEC, TSPREAD, and UNCER) Dependent Variable: NCALL = A binary variable for the presence (0) or absence (1) of a call feature being placed on the debt debt issue. Independent Variables: CLASS = A set of binary variables for firm classifications. DA = Debt to asset ratio. DE = Debt to equity ratio. INTCOV = Interest coverage ratio. MARKET = A measure of the current debt issue to the existing debt outstanding. MAT = Maturity of the debt issue in years. POSTGROWTH = Growth during the year the debt was issued. PREGROWTH = Average growth during the 3 years prior to the debt issue. PM = Profit margin. RATE = Debt ratings. SIZE = The natural logarithm of the dollar amount of the debt issue. SPREAD = The difference in the yield of the debt issue and a comparable treasury issue. TAX = The corporate tax rate of the issuing firm. TSEC = The yield on a treasury security issue. TSPREAD = The difference in the yield of a 30 year and a 3 year Treasury security. UNCER = Average change in interest rates over the 3 weeks prior to the debt issue.)

PAGE 106

96 The interest coverage ratio, earnings before interest and taxes divided by interest expense, is a measure of the firm's ability to make its interest payments when due. Interest coverage is measured on both a before and an after tax basis (INTCOVBT and INTCOVAT) MAT is the maturity of the debt issue in years. Profit margin (PM) net income divided by sales, is a proxy for the firm's expected profitability at the time of the debt issue. The profit margin for the firm is measured ex-post in the year after the debt was issued. SIZE is the natural logarithm of the issue size in millions of dollars. The growth rates for each firm measured for various attributes are defined for two time periods. PREGROWTH is the average growth rate of a selected variable over the three year period prior to the actual debt issue date. POSTGROWTH is the growth of the same variable during the current year measured ex-post. The growth variables tested include assets, dividends and earnings per share, gross and net property, plant and equipment, sales, and long-term debt. Both growth rates are proxies for future growth, but PREGROWTH is know at the time of the debt issue and POSTGROWTH is not known until after the debt is actually issued. Dummy variables are created for the issuer's debt ratings, Aaa, Aa, A, Baa, Ba, B, Caa, and nonrated. An alternative set of rating variables HIGH, MOD and LOW are

PAGE 107

97 also utilized. HIGH is a binary variable for Moody's highest ratings, Aaa or Aa. Similarly, MOD and LOW are binary variables for Moody's moderate debt ratings (A or Baa) and low debt ratings (Ba or lower including nonrated debt), respectively. When either complete set of ratings are utilized Aaa or HIGH are deleted. Proxies for the level and variability of interest rates are TSEC, SPREAD, TSPREAD, and UNCER. TSEC is defined as the yield to maturity on a fixed maturity Treasury security measured in the secondary markets on the same date the sample debt instrument was issued. The two maturities utilized are 3 and 30 years. A measure for the variability of interest rates, SPREAD, is calculated as the difference between the yield of the debt issue and the yield to maturity on a U.S. Treasury security having the same maturity as the debt issue on the same date the debt instrument was issued. SPREAD is defined as SPREAD = Yieldi t Yield T t where i equals the debt issue used, t equals the offering date, and T equals the matched treasury security. The yield curve is proxied by TSPREAD, defined as the difference between 3 and 3 year Treasuries, TSPREAD = Yield 30 yr/t Yield 3 yr t a negative (positive) value indicates a falling (rising) yield curve.

PAGE 108

98 The final interest variable, UNCER, is defined as the mean absolute change in rates on 3 year U.S. Treasury securities during the 3 weeks prior to the debt issue. Formally UNCER is calculated as f follows, UNCER = [|X_3-X_ 2 | + |X_ 2 -X_ 1 |+|X_ 1 -X |]/3, where X Q equals the interest rate on the 3 year U.S. Treasury bond index reported in the Annual Statistical Digest, X t is the rate on the same index reported t weeks before the date the bond was issued, and | | signifies absolute value. The average tax rate (ATAX) is defined as the firm's total tax expense, including all income taxes paid to federal, state, and foreign governments, divided by the firm's pretax income. MTAX is the firm's marginal tax rate and is based on the firm's pretax profits reported by the Compustat tapes and derived from the tax schedules in effect at the time of the debt issue. Log it Model 1 A log.it regression model is utilized to test what attributes determine the choice between issuing callable and noncallable debt. The model is formally classified as a univariate dichotomous model since it is only concerned with the occurrence or nonoccurrence of the inclusion of the call option when debt is issued. The logit model uses the logistic distribution as a probability function. One of the basic benefits of this distribution is that it constrains the dependent variable to lie between and 1. The model coefficients are estimated

PAGE 109

99 using the maximum likelihood function. Logistic regression is utilized rather than discriminant analysis since it does not require the assumption of multivariate normality. 2 The only assumption necessary is that the probability that NCALL^ = 1 (e.g. noncallable debt issue) equals: l/fl+expC-a-Xj^B) ) where a is the intercept parameter and B denotes the vector of regression parameters. Hypothesized Effects Callable bonds mitigate agency problems. When these agency problems are likely to be large, callable bonds will be used. The following variable types proxy the size of the agency problem: debt ratings, debt ratios, profitability, and growth. The debt rating variables are expected to effect the probability of issuing noncallable debt. The less risky and higher rated the debt is, the less potential the bondholders' gain in the absence of a call option. Thus, the higher the debt rating, the greater the probability that the debt issue will be noncallable. The effect of the current issue on the firm's capital structure is measured by the debt ratios, DA and DE, and the variable MARKET. Thus, DA, DE, and MARKET are expected to each have a negative impact on the probability of issuance of noncallable debt. The more debt the firm takes on, the greater the probability that the firm will desire to recall

PAGE 110

100 some portion of the debt to either modify the capital structure or eliminate restrictive convenants. The profit margin, PM, of the firm should be positively related to the issuance of noncallable debt. As the firm increases its profitability, the probability of default decreases and with it the need for recalling debt to eliminate restrictive convenants. Another measure of the firm's strength, INTCOV, should also be positively related to the issuance of noncallable debt. The higher the interest coverage ratios, the smaller the probability of default and the less the need for the call option. The growth variables, PREGROWTH and POSTGROWTH, are both expected to exert a negative effect on the dependent variable, NCALL. Growth is a proxy for the need to avoid wealth shifts. The larger the growth, the stronger the need for callable debt as a means of limiting the gain of the debtholders from investment opportunities. Taxes, both average and marginal, are predicted to have a negative effect on the dependent variable NCALL. The indication is that the higher the (average or marginal) tax rate of the firm, the greater the benefits obtainable from callable debt. Therefore, as the tax rate of the firm increases, the probability that the firm will issue callable debt should also increase. Thus, a negative relationship is predicted between the tax rate and the issuance of noncallable debt.

PAGE 111

101 An increase in interest rates or in the variability of rates are given as one of the principle reasons for issuing callable debt. As either the variability or the absolute size of interest rates increase, it is hypothesized that the proportion of callable debt issued will also increase. Thus the variables SPREAD, TSEC, and UNCER are predicted to have a negative effect on the occurrence of noncallable debt. The proxy for the shape of the yield curve, TSPREAD, should be positively related to the issuance of noncallable debt. During period of a rising yield curve, the incentive to attach a call feature on the debt issue is less than during periods with a falling yield curve. The summary statistics from Chapter III suggest that firm classification, maturity, and size of the issue are also related to the issuance of callable debt. The CLASS variables are expected to show mixed effects. Finance firms are expected to issue greater proportions of noncallable debt than the other classifications, therefore, it should be positively related to the NCALL. Conversely, both utilities and industrials are expected on average to issue long maturity debt and have a preference for callable debt. Thus, both utilities and industrials should be negatively related to NCALL. The maturity of the debt issue, MAT, is expected to have a negative effect on noncallable debt issues since the longer the maturity of the debt issue the greater the probability that the debt issue is callable.

PAGE 112

102 The natural logarithm of the size of the debt issue, SIZE, is expected to have a positive effect on the issuance of noncallable debt. The larger the issue, the greater the probability that the firm can deviate from the norm of the debt market, i.e. issue noncallable debt, and not be penalized. The natural logarithm of the debt amount is used since the effect of size is likely to decrease as the issue size increases. See table 4-5 for a summary of all the hypothesized effects. Results As defined in Table 4-4, the choice of whether to issue callable or noncallable debt is influenced by many market as well as firm specific factors. The following logit eguation is estimated, NCALL = f (CLASS, DA, DE, INTCOV, MARKET, MAT, POSTGROWTH, PREGROWTH, PM, RATE, SIZE SPREAD, TAX, TSEC, TSPREAD, UNCER) The logit regression is estimated using the maximum likelihood function for each of these proxies on an individual basis to gauge their predictive powers. The results of these regressions are shown in Table 4-6. The data set utilized during the estimation phase of the logit regression analysis was restricted to include only firms that had either issued callable or noncallable debt during any one year time period. It was further restricted during the estimation phase to include only a subsample of the

PAGE 113

103 Table 4-5 Hypothesized Effects of the Independent Variables on NCALL NCALL = f (CLASS, DA, DE INTCOV, MARKET, MAT, + + + POSTGROWTH, PREGROWTH, PM, RATE, SIZE, SPREAD, TAX, TSEC, TSPREAD, UNCER) Note: A negative (-) relationship indicates that as the independent variable increases, the probability of the debt issue being noncallable decreases. A positive (+) relationship indicates that as the independent variable increases, the probability of the debt issue being noncallable increases.

PAGE 114

Table 4-6 Logit Regressions Using Individual Variables 104

PAGE 115

105 Table 4-6 — continued Variable

PAGE 116

106 original data set to allow the model to be tested on the total data set for predictive purposes later in the analysis. Debt rating (RATE) proxies the firm's potential for co-insuring the debt issue. Unfortunately, most of the individual rating classes are eliminated due to limited dispersion or lack of predictive capabilities. Thus, the set of binary variables for the individual rating categories has limited use in the overall logit analysis. The results from estimating a second set of binary proxies for ratings (HIGH, MOD, and LOW) are more promising than the individual ratings. Both MOD and LOW are statistically significant. This supports the hypothesis that noncallable debt issues tend to have higher ratings when compared with callable issues. The capital structure and default proxies in general have no predictive powers within the model and therefore, the signs on the generated coefficients are meaningless. The debt to equity ratio (DE) the profit margin (PM) and the interest coverage ratios (INTCOV) all fail to contain any predictive power in the determination of the use of debt type. The debt to assets ratio (DA) is the only variable within this section to generate significant results. The individual regression on the variable DA does produce a negative beta which matches the predicted results, i.e. a negative relationship with the issuance of noncallable debt.

PAGE 117

107 MARKET shows a negative relationship with noncallable issues. As a debt issue becomes a larger proportion of the amount of debt the firm has outstanding, the greater the need for later recall to eliminate undesirable convenants or to change capital structure in the future. The growth variables generate results similar to the default proxies, i.e. most of the variables tested generate insignificant results. Both preand post-growth rates for dividends and earnings per share, long-term debt, and gross and net property, plant, and equipment results are statistically insignificant. Two sources for this insignificance are a lack of variation of the variable within the callable and noncallable debt samples and a lack of predictive power generated within the logistic regression. The remaining growth variables, assets and sales, generate negative coefficients suggesting that noncallable debt can be used as a means of limiting the participation of debtholders in the profits of the firm. The average tax rate, ATAX, generates the correct sign but is not significant. The means of the average tax rates from the firms issuing callable and noncallable debt also failed to show a significant difference, therefore the results were expected. Marginal tax rates, MTAX, do indicate a significant difference when the Wilcox sum rank test is used. But as the logit regression indicates, this difference has little practical value.

PAGE 118

108 The measures of the level and uncertainty of interest rates generate mixed results. TSEC, a measure of the rate of interest, shows a negative relationship for the issuance of noncallable debt when either the 3 or 30 year Treasury security is used. The next two interest rate proxies, UNCER and SPREAD, generate the wrong signs. This shortcoming is explained by lack of variability of the variables within the callable and noncallable debt categories. The lack of predictive power for both variables supports the insignificance of UNCER and SPREAD for use within the logistic regression. The final proxy, TSPREAD, generates the correct sign and is significant. The three general variables, CLASS, MAT, and SIZE, yield the predicted results. A positive relationship with the issuance of noncallable debt is detected for financial firms. This is contrasted with the negative relationship found for both the utility and industrial classifications. This supports the hypothesis that financial firms have a greater probability to use noncallable debt than the other two firm types, industrials and utilities. SIZE is positively related to the issuance of noncallable debt supporting the belief that larger issuers are better able to deviate from the norm of issuing debt with the call option attached. Also the maturity of the debt issue (MAT) is negatively related to the dependent variable NCALL. As the length of maturity increases, the probability of issuing noncallable debt decreases.

PAGE 119

109 Table 4-2 includes a summary table for the means and standard deviation for all the variables used in the first pass of the logit regression analysis and supports the lack of variation in the variables dropped. Table 4-3 shows the difference in the probabilities for ratings, maturity, and firm classifications between the two groups. It supports the keep/ reject decision of the various binary variables. Notice that the mean values and standard deviations are different than those reported in Chapter III. This is because these results are calculated on only the debt issues used in the logit analysis and not on the total data set used in Chapter III. The second criteria for narrowing down the variables used in the final regression model is to run the surviving logistic regression variables in a principal component regression. This is used only to give some guidance as to the number of independent sources of variation. Note that the "principal components regression is a method of inspecting the sample data or design matrix for directions of variability and using this information to reduce the dimensionality of the estimation problem" (Judge et. el., p. 910) The criteria used in the analysis was to retain all components whose eigenvalues exceed 0.9. The regression resulted in 8 eigenvalues that satisfied the above criteria. The next stage was to utilize the 8 logit variables that contained the highest factor pattern within this set of

PAGE 120

110 eight factors. The surviving variables included MAT, TSEC, MARKET, DA, MOD, LOW, FINANCE, and POSTGROWTH (sales) A comparison of the correlation of the variables that survived the first elimination criteria is also used to detect extreme correlation of the surviving variables. Table 4-7 shows the correlation matrix for the surviving variables. Variables that are highly correlated with each other decrease the predictive powers of the individual betas but does not detract from the overall usefulness of the logistic regression for classifying prospective debt issues as being either callable or noncallable. The final logit model contains the following variables: MAT, TSEC, MARKET, DA, MOD, LOW, FINANCE, and POSTSALES. The model was estimated using only a subsample of the original data set. The criteria was setup so that egual portions of callable and noncallable debt were utilized. The data criteria included: (1) debt issues from financial, industrial, or utility firms, (2) no zero coupon debt, and (3) every ninth callable debt issue so that approximately egual numbers of callable and noncallable debt issues were used in estimating the logit parameters. The final model, shown in Table 4-8, includes the estimated beta parameters, the standard errors of the individual betas, the chi-sguare significance test results, the significance level, and the predictive power of the individual model components. The individual variables

PAGE 121

Ill Table 4-7 Correlation Matrix For Logistic Regression Variables Variable

PAGE 122

112 Table 4-8 Logistic Regression Estimation Eguation Final Model Variable

PAGE 123

113 within the model contain the predicted sign as analyzed within the individual logit regressions. 4 Estimated coefficients do not indicate the increase in the probability of the event occurring (noncallable debt issue versus callable debt issue) given a one unit increase in the corresponding independent variable. Rather, the coefficients reflect the effect of a change in an independent variable upon the natural logarithm of [Probability^ (1-Probabilityj^) ] The amount of the increase in the probability depends upon the original probability and thus upon the initial values of all the independent variables and their coefficients. The sign of the coefficient does indicate the direction of the change. The logistic regression shows an 81.2% prediction ratio for the debt issues correctly identified during the estimation phase of the analysis. Within the correct predictions were 82.0% and 80.5% correctly predicted callable and noncallable debt issues respectively. These estimated parameters are run on the entire data set to test the predictive power of the model. Callable and noncallable predictions, using the final regression coefficients on the total data set, are summarized in Table 4-9. On the total data set, the accuracy of the model for predictive purposes was 78.9%. Various subsets of the total set highlight the strengths of the model. Broken down by short and long maturities the results show higher predictive

PAGE 124

114 Table 4-9 Prediction Results Using Final Logistic Regression Estimation Eguation On Total Data Set Total Data Set (All Maturities) Debt Type Correct Prediction Wrong Prediction Callable Debt 1214 (80.7%) 291 (19.3%) Noncallable Debt 240 (71.2%) 97 (28.8%) Total 1454 (78.9%) 388 (21.1%) Short Maturity Debt Issues (Maturity < 10 yrs.) Debt Type Correct Prediction Wrong Prediction Callable Debt 93 (41.0%) 134 (59.0%) Noncallable Debt 187 (83.5%) 37 (16.5%) Total 280 (62.1%) 171 (37.9%) Long Maturity Debt Issues (Maturity > 10 yrs.) Debt Type Correct Prediction Wrong Prediction Callable Debt 1121 (87.7%) 157 (12.3%) Noncallable Debt 53 (46.9%) 60 (53.1%) Total 1174 (84.4%) 217 (15.6%) Predictions based on final model shown in Table 4-8

PAGE 125

115 power for the long maturities, 84.4% versus 62.1%. Segmented by callable and noncallable debt, the results favor the callable sector, 80.7% versus 71.2%. Further details are provided in Table 4-9. The predictive powers of the model increase if the data set is restrictive to only firms that only issue one type of debt during any one year period. The overall percentage increases to 87.6%. Similar increases are shown in Table 4-10 for the other subsamples. An interesting aside is that when the final logit variables are run in a discriminant analysis model, the results are only slightly lower. The discriminant analysis results are shown in Table 4-11. Summary The logit analysis reinforces the tendencies uncovered within the summary statistics chapter. These include: -a tendency of financial firms to dominate the issuance of noncallable debt. (Financial firms are not the only type of firm to utilize noncallable debt. Industrials and utilities also issue noncallable debt but to a lessor extent. ) -a tendency of the current debt outstanding to influence the type of debt issued. (The greater the proportion of debt already in the capital structure of the firm, the greater the probability that additional issues will contain a call feature.)

PAGE 126

116 Table 4-10 Prediction Results Using Final Logistic Regression Estimation Eguation On Total Data Set Excluding Firms Issuing Both Callable and Noncallable Debt in a 1 Year Period Total Data Set (All Maturities) Debt Type Correct Prediction Wrong Prediction Callable Debt 1131 (84.9%) 201 (15.1%) Noncallable Debt 110 (70.1%) 47 (29.9%) Total 1241 (83.3%) 248 (16.7%) Short Maturity Debt Issues (Maturity < 10 yrs.) Debt Type Correct Prediction Wrong Prediction Callable Debt 83 (51.2%) 79 (48.8%) Noncallable Debt 75 (82.4%) 16 (17.6%) Total 158 (62.5%) 95 (37.6%) Long Maturity Debt Issues (Maturity > 10 yrs.) Debt Type Correct Prediction Wrong Prediction Callable Debt 1048 (89.6%) 122 (10.4%) Noncallable Debt 35 (53.0%) 31 (47.0%) Total 1083 (87.6%) 153 (12.4%) Predictions based on final model shown in Table 4-8

PAGE 127

117 Table 4-11 Prediction Results of Final Logit Regression Variables Used in Discriminant Analysis Model Estimated Model: Debt Type Correct Prediction Wrong Prediction Callable Debt 143 (79.0%) 38 (21.0%) Noncallable Debt 148 (86.0%) 24 (14.0%) Total 291 (82.4%) 62 (17.6%) Prediction Results: Debt Type Correct Prediction Wrong Prediction Callable Debt 995 (66.1%) 510 (33.9%) Noncallable Debt 309 (91.6%) 28 (8.4%) Total 1304 (70.8%) 538 (29.2%) The independent variables used in the discriminat analysis include: FINANCE, DA, MARKET, MAT, MOD, LOW, POSTGROWTH (Sales) and TSEC. The definition for each of these variables can be found within Chapter IV and within the Symbols page found in the preliminary section of the dissertation.

PAGE 128

118 -a tendency for the inclusion of the call feature to increase with an increase in the maturity of the debt issue. -a tendency for high growth firms to utilize the call option. -a tendency for noncallable debt to be utilized only by good firms, where good is measured by the debt ratings assigned to the firm's debt. (Good firms have less need for co-insuring the debt issue.) -a tendency for the use of the call option to increase with the volatility of interest rates. Notes 1. See Judge, Griffiths, Hill, Lutkepohl and Lee (1985) for the theoretical justification behind the logistic regression. 2. Press and Wilson (1978) compare logistic regression and discriminant analysis and conclude that logistic regression is preferable to discriminant methods based on normality assumptions when the variables do not have multivariate normal distributions within classes. 3. For example, New York State Electric & Gas Corporation 3 year bond issued on July 1987 would be matched with the 3 year Treasury Bond issued at the same point in time. The yield spread eguals 150 basis points (12% 10.5%). The spread between 3 year and 3 year treasury securities was also tried in the logit regression. This set of treasury differences also failed to provide any predictive powers to the model 4. Various maturities of treasury securities were tried within the logit regression with negligible differences. The model was also estimated without the two variables with the highest chi-sguare values to determine the predictive powers of the remaining variables. The model without MAT and TSEC had a correct prediction ratio of 70.0%. This reinforces the importance of the level of rates and the maturity of the debt issue in the determination of whether or not to attach a call option to the debt issue.

PAGE 129

119 The final model, without the binary variable for finance firms, was estimated on data sets containing only financial firms, and only industrials and utilities. Both regressions yielded similar results. The predictive power of the financial data set was 80.4% versus 82.0% for the industrial/utility data set.

PAGE 130

CHAPTER V EVENT STUDY The capital market responds differently to announcements that a firm is raising external capital, depending on whether the financing is to be met by issuing debt versus equity. In particular, raising external debt elicits no change or an insignificant decrease in the issuer's stock price, while equity issuances substantially reduce share prices. This is consistant with the notion that managers with good private information issue debt as opposed to equity because they know the firm will be able to use the interest deductions without a great fear of default. Managers with negative private information reduce leverage because they preceive the firm will be unable to use the interest deductions associated with debt. It stands to reason that the type of debt issued should also reveal manager's private information. In particular, manager's with good private information will issue short-term debt as opposed to long-term debt according to Flannery (1986). Similarly, managers with good private information prefer to issue callable debt because they know that the value of the debt is likely to be increased in the near future. Low private valuation firms prefer to issue long-term debt to avoid the transaction costs and the 120

PAGE 131

121 revaluation of credit worthiness that occurs with each rollover. High types, desiring to reveal their true worth, are willing to pay the transactions cost associated with rolling over short-term debt in order to avoid being pooled with low quality firms and being forced to have a higher cost of debt. Thus, short-term noncallable and long-term callable debt are credible signals of high firm quality. This chapter examines the capital market responses to debt issuances. As with previous studies, I find negative but insignificant wealth effects on the announcement dates. When the issues are grouped by long-term versus short-term and callable versus noncallable, I discover that contrary to the signaling associated with short-term callable debt. Consistant with signaling models, significant negative abnormal returns are associated with long-term noncallable debt. A summary of empirical research on the impact of debt issues on the firm's equity value is provided by Eckbo (1986). Masulis (1980) finds that with the exchange of debt for common stock there was a 9.79% excess return for the 2 trading days proceeding and including the day of the first announcement. McConnell and Schlarbaum (1981) found a 2.18% 2-day abnormal residual for exchange offers involving income bonds for preferred stock. Dann and Mikkelson (1984) find a -2.31% 2-day abnormal residual with the announcement of the issuance of

PAGE 132

122 convertible debt and -1.76% 2-day abnormal residual with the announcement of straight debt offerings. Eckbo's own study finds that straight debt offerings are on average associated with a zero or a negative announcement period average residual. His results showed a -0.06% 2-day abnormal residual with a t-value of -0.44 which is insignificant but suggests a non-positive impact from the announcement of debt. For this study, various stratifications of debt issues are tested to determine if any debt subsets have a significant impact on the eguity value of the firm either at the announcement date or the actual date of issuance. Data Daily stock returns were compiled from the CRSP Daily Returns tape. Issuance and announcement dates were obtained from Moody 's Bond Survey and the Wall Street Journal Index respectively. The announcement date (AD) is the earliest mention of the proposed debt issue, restricted to one year prior to the actual issuing date. The execution date (ED) is the actual date the debt was placed. A corporate debt issue is included in the sample if the following reguirements are met. 1. The issuer is on the 1987 CRSP Daily Return Tape. 2. The debt issue occurred during the period 1981 through 1986. 3 A distinct announcement date of the debt issue is found in the Wall Street Journal Index that differed from the actual issue date.

PAGE 133

123 Restricting the event study sample to firms listed on the CRSP Daily Return Tape means that only firms listed on the New York or American Stock Exchanges are utilized. The reduction in the time period for this part of the analysis is due to the small number of firms issuing noncallable debt that met the event study criteria in the years 1977 through 1980. Using the methodology of Brown and Warner (1985) a base period is established for calculating a standard against which the test period returns are measured. The base period utilized covers 180 trading days. It starts 200 trading days before and ends 2 trading days prior to the announcement date, AD, for each firm. The announcement date is the first mention of the debt issue as found in the Wall Street Journal Index or the registration date with the SEC, whichever comes first. By utilizing a 20-day buffer, the effects of any leakage from the debt announcement (either by itself or associated with an underlying economic event) is captured. The testing for excess returns is calculated at two points in time. The announcement date is used for centering all the firms on a common date for the first testing period. This period starts 20 trading days prior and continues for 20 trading days past the AD and is used to test the effects of excess returns resulting from the announcement of the debt issue. This 41 day period surrounding the AD falls

PAGE 134

124 within the range commonly used in event studies. This process is repeated for testing excess returns around the execution date. Figure 5-1 depicts the time frame associated with the event study portion of the analysis. Event Study Model The market model (5-1) developed by Sharpe and later expanded by others for estimating returns, is utilized forestimation during the base period as outlined below. *i,t = H + B^t + e i>t (5-1) Where R ijt is the daily return including dividends for the firm under study at time t and the tildes (~) indicate random variables. The daily returns used within this study are obtained from the Center for Research in Security Prices Tapes (CRSP) and are defined as: R i,t (P lft + D i,t Pi, t -1> / P i,t-1p i,t = price of common stock at time t for firm i D i t = dividend at time t for firm i Data on the realized returns including dividends for each firm are obtained on a daily basis for a period which begins 2 00 trading days before the AD and ends 2 trading days after the EX (execution date). Thus, company returns are obtained from a fixed period to the AD through a fixed period after the EX for each issue. Notice that the time period between AD and EX vary from one issue to the next, therefore the number of return observations in each array is different.

PAGE 135

125 ESTIMATING PERIOD Base t=-200 TESTING PERIOD 1 AD Test t=-21 t=-20 AD t=0 t=+2 where t (t = -210 to 20) is the number of trading days from the announcement day (AD) and t = is the announcement date as reported in the Wall Street Journal Index TESTING PERIOD 2 EX Test EX t=-20 t=+2 where t (t = -20 to 20) during the second testing period is the number of trading days from the execution date (EX) Figure 5-1 Event Time Frame

PAGE 136

126 Rm^t -*s the da i!y return at time t of a broad portfolio of stocks on the NYSE and AMEX. The specific index utilized in this study was the CRSP Value Weighted Index with Dividends (VWRETD) It is used as a proxy for the market index. The variables a.^ and B-^are defined as follows a i = E ( R i B i E (Rm>) (5-2) B i = cov ( R j,t' R m,t)/ var ( R m,t)' ( 5 -3) The final term in the model, e^ t represents the error associated with the model in predicting the true return. These base period estimates, a^ and B^ are applied to the return series within each of the test periods. Residual returns, u i t during the test periods are obtained by subtracting the predicted returns for the actual returns G i,t = [ R i,t (i + B iVt>l( 5_4 > The daily residual return illustrates the change in the realized residual returns after the base period calculations. The change reflects any new information from the announcement and actual debt issue. Since the debt issue is the only specific information incorporated in the study for each firm, any movement in residual returns measures the change in required rates of return and share prices caused by the debt issue. When the debt issuing firms are analyzed, they are compared to the general base group of firms, i.e. the market index.

PAGE 137

127 Next, the individual company residuals are crosssectionally averaged to yield average abnormal returns for each day (t = -2 to 20) for both test periods. The average residual return, AR t for all the firms on each day relative to the AD and the EX are calculated using a simple mean, N AR t = 1/N Sum (u i t ) (5-5) i=l where N is the sample size. Next, the average residuals (AR) are accumulated across time, T CAR = Sum AR t (5-6) t=l where T, the test periods revolve around the period 20 days before and 20 days after the AD and EX. A 2-day excess return is reported for each test as a means of capturing the announcement and the actual debt issue effects due to the timing of the market's trading hours. Day t=0 is the day the news of the event (either the announcement or the actual issuance) is published in the financial press. In most cases, the news is obtained on the previous day, t=-l, and reported the next day. If the debt issue is announced before the market closes, then the market's response to the news actually predates the publication by one day. If the news is obtained after the market closes, the market will respond the next day and the

PAGE 138

128 reaction is indeed on day 0. Excess returns are also reported over the periods -/+ 20 trading days, -/+ 10 trading days, -/+ 5 trading days, -/+ 1 trading day, and -1 through +5 trading days. No abnormal changes in the returns to the shareholders due to the debt issue is shown by insignificant t-values for the respective cumulative test periods. Abnormal changes should fluctuate randomly around zero and on average egual zero. But if there was a positive trend overtime, then the abnormal changes would show up as a shift above the 0% value and indicate that shareholders were better off from the debt issue than the average shareholders found in the market. Similarly, if this trend is negative, shareholders were worse off than the average shareholder. The average and cumulative average residuals within each test are reported with their corresponding t-values in Appendix B. (See notes 3 and 4 for a description of the calculations of the t-values for the average and cumulative average residuals respectively. ) Hypotheses and Test Results Total Debt Issues The full event study sample hypothesis is that shareholders' wealth is not effected by the issuance of debt when all debt issues are treated as a homogeneous group. The null and alternative hypotheses are defined in eguations 5-7 and 5-8 respectively.

PAGE 139

129 CARm = (5-7) The cumulative abnormal returns for firms issuing debt is not significantly different from 0, i.e. the issuance of debt does not impact the shareholders' wealth. H, CAR T (5-8) The cumulative abnormal returns for firms issuing debt is significantly different from 0. To test these hypotheses, CARs are calculated for the full sample of 631 debt issues during the period 1981 through 1986. These accumulated abnormal returns are tested to determine significant differences from zero caused by the impact of debt issues when evaluated in totality. Table 5-1 presents results of the cumulative abnormal returns (CAR) for the full sample of 631 firms that issued debt over the six test periods around both the announcement date and the execution date. The average and cumulative average residuals are shown in Appendix B, Table B-l along with the respective t-values. Although both negative and positive ARs and CARs are shown, none are statistically significant. Average abnormal returns, accumulated over the entire 41-day test period (day -20 to day +20) yield -1.169% with an insignificant t-value (t-value is -0.415) during the announcement period. The tests around the other 5 AD test periods also show insignificant results. Tests were also performed around the actual day the debt was issued. These results are also found in Table 5-1. The ARs and CARs, shown in Appendix B, Table B-l, show

PAGE 140

Table 5-1 Cumulative Average Residuals 130 All Debt Issues (n=631) Interval CAR t-value Announcement Period: -20 through +20 -1.1693 -0.4157 -10 through +10 -0.5296 -0.2631 5 through +5 -0.4396 -0.3017 1 through +1 0.1147 0.1508 1 through +5 -0.4186 -0.3602 1 through -0.0346 -0.0557 Execution Period: -20 through +20 -0.4798 -0.1705 -10 through +10 -0.6302 -0.3130 5 through +5 -0.7662 -0.5258 1 through +1 0.4294 0.5643 1 through +5 -0.5727 -0.4927 1 through 0.3305 0.5319 Significant at 10% level using a two-tail test, Significant at 5% level using a two-tail test. Significant at 1% level using a two-tail test.

PAGE 141

131 insignificant abnormal returns. Average abnormal returns, accumulated over the entire 41-day test period (day -2 to day +20), yield -0.479% with an insignificant t-value (t-value is -0.171). All other ED test periods also showed insignificant results. Thus, when the debt offerings are analyzed as a homogeneous group no significant impact on the firm's stock value is detected. Callable versus Noncallable Debt Wealth effects are examined across the various debt classifications. The first set of classifications are callable and noncallable debt issues. The null and alternative hypotheses are defined in equation 5-9 and 5-10. CAR C = CAR n (5-9) The cumulative abnormal returns for firms issuing callable debt (CAR C ) is not significantly different from the cumulative abnormal returns for firms issuing noncallable debt (CAR n ) H x : CAR C f CAR n (5-10) The cumulative abnormal returns for firms issuing callable debt (CAR C ) is significantly different from the cumulative abnormal returns for firms issuing noncallable debt (CAR n ) The average and cumulative abnormal returns are calculated separately for the callable and noncallable debt issues and reported in Appendix B, Table B-2 (announcement period results) and Table B-3 (execution period results) The results from the debt type partitioning (callable and noncallable debt issues) yielded results similar to the total debt sample. A test of the null hypothesis that the

PAGE 142

132 cumulative average excess returns equals (H Q : CAR= 0) cannot be rejected. The announcement period test results for hypothesis 5-9 are shown in Table 5-2. The CAR for the callable sample over the entire 41 day period yielded a -1.515% excess return with an insignificant t-value of -0.434. The noncallable sample although positive, 0.012%, was also insignificant (t-value of 0.003). The test around the actual issue day also resulted in insignificant results. The CAR's for the 41 day test period were -1.128 and 1.735 percent respectively for the callable and noncallable samples (t-values of -0.327 and 0.408). Thus, a firm's decision to issue callable or noncallable debt does not show a statistically significant impact on the shareholder's wealth. Both types of debt issues have essentially no impact on shareholder wealth. The test results from hypothesis 5-9 of equal CARs from each subsample are shown in the bottom of Tables 5-2 and 5-3 for the announcement and execution periods respectively. The tests under all 6 of the test periods for both the announcement and execution periods fail to reject the null hypothesis. Both debt classes are insignificantly different from zero and insignificantly different from each other. The calculation for the t-value for the comparison of CARs from two samples is shown in note 4 Short versus Long Maturity Debt The question has been raised that it is not the decision to issue callable or noncallable debt that impacts

PAGE 143

133 Table 5-2 Cumulative Average Residuals for Announcement Period Callable vs Noncallable Debt Issue Effects Callable Debt Issues (n=488) Interval CAR t-value -20 through +20 -1.5156 -0.4349 10 through +10 -0.3956 -0.1602 • 5 through +5 -0.4562 -0.2554 • 1 through +1 0.2845 0.3049 1 through +5 -0.5143 -0.3609 1 through 0.0779 0.102 3 Noncallable Debt Issues (n=143) Interval CAR t-value •20 through +20 0.0124 0.0029 •10 through +10 -0.9871 -0.3239 • 5 through +5 -0.3829 -0.1736 • 1 through +1 -0.4646 -0.4034 1 through +5 -0.0921 -0.0523 • 1 through -0.4190 -0.4455 Callable Versus Noncallable Debt Issues Interval Callable Noncallable C-N t-value -20 through +20 -1.5156 0.0124 -1.5280 -0.2788 -10 through +10 -0.3956 -0.9871 0.5914 0.1508 5 through +5 -0.4562 -0.3829 -0.0733 -0.0258 1 through +1 0.2845 -0.4646 0.7491 0.5054 1 through +5 -0.5143 -0.0921 -0.4222 -0.1865 1 through 0.0779 -0.4190 0.4969 0.4106 Significant at 10% level using a two-tail test, Significant at 5% level using a two-tail test. Significant at 1% level using a two-tail test.

PAGE 144

Table 5-3 Cumulative Average Residuals for Execution Period Callable vs Noncallable Debt Issue Effects 134 Callable Debt Issues

PAGE 145

135 the stockholders return, but the decision to issue short or long maturity debt instruments. To test the maturity effect on the total sample, the effects of short maturities and long maturities are compared by hypotheses 5-11 and 5-12. H CAR S = CAR-l (5-11) The cumulative abnormal returns for firms issuing short-term debt does not differ from firms issuing long-term debt. % CAR S f CAR-l ( 5 ~ 12 ) The cumulative abnormal returns for firms issuing short-term debt differs from firms issuing long-term debt. The ARs and CARs are shown in Appendix B, Table B-4 and B-5 for the announcement and execution periods respectively. The CARs for the short maturity debt issues are significantly negative from day -15 through day 5. The cause of these significant CARs can be traced back to four significant ARs on day -16, -5, -2, and +15. So although a period exists when the CARs are significantly different from 0, none of the actual test periods show significant results. For example, the CAR's for the short and long maturities around the 41 day announcement test period are -0.381 and -1.537 percent respectively. Both CAR's were insignificantly different from with t-values of -0.386 and -0.411 for the short and long maturity debt issues. The tests around the issue date also failed to reject the null hypothesis of a difference in the CAR's from 0. The 41 day CAR's are 1.301 and -1.312 (with t-values of 1.318 and

PAGE 146

136 -0.348) respectively for the short and long maturity debt issues. The test of hypothesis 5-11 concerning a difference in the CARs for the 6 test periods between the short and long maturity debt issues fail to reject the null hypothesis under both the announcement and execution period tests. For the 41 day test the t-values are 0.2984 and 0.6717 respectively for the announcement and execution period tests. Similar results are shown for the other 5 test periods in Tables 5-4 (announcement period tests) and 5-5 (execution period tests) Correctly versus Incorrectly Identified Firms The analysis from the logit regression classified firms into two groups, correctly and incorrectly identified issues. Correctly identified issues are those issues which are callable (noncallable) and the logit regression predicted they would be callable (noncallable) See Chapter IV for a complete analysis of the logit regression and the predicted results. The hypotheses tested are listed in equations 5-13 and 5-14. H CAR, CAR. (5-13) The cumulative abnormal returns for debt issues correctly identified by the logit regression analysis do not differ from those incorrectly identified. CAR-, ^ CAR. (5-14) The cumulative abnormal returns for debt issues correctly identified by the logit regression analysis differ significantly from those incorrectly identified.

PAGE 147

137 Table 5-4 Cumulative Average Residuals for Announcement Period Short vs Long Maturity Debt Issue Effects Short Maturity Debt Issues (

PAGE 148

13! Table 5-5 Cumulative Average Residuals for Execution Period Short vs Long Maturity Debt Issue Effects Short Maturity Debt Issues (n=201) Interval CAR t-value -20 through +20 •10 through +10 • 5 through +5 • 1 through +1 1 through +5 • 1 through 1

PAGE 149

139 The overall results from the logit identified samples yield results similar to the previous tests. Neither sample had ARs or CARs which were significantly different from zero for either the announcement or execution periods. Complete results are shown in Appendix B, Tables B-6 and B-7 for the announcement and execution period tests respectively. The tests of hypothesis 5-13, that the CARs from the correctly and incorrectly identified firms differ, are shown in Table 5-6 and 5-7 for the announcement and execution periods. Again the results fail to reject the null hypothesis. For example, the t-values for the 41-day test period are 0.2959 and 0.5854 for the announcement and execution periods respectively. Both t-values are insignificant. Short versus Long Maturities: Callable Debt The maturity question is also analyzed under both the callable and noncallable subsamples. The maturity test within the callable sample utilizes the hypotheses 5-15 and 5-16 to test the difference in the market's reaction to shortand long-term callable debt. H CAR c,s = CAR c,l ( 5 15 ) The cumulative abnormal returns for firms issuing short-term callable debt does not differ from firms issuing long-term callable debt. H l CAR c,s CAR c,l ( 5 ~ 16 ) The cumulative abnormal returns for firms issuing short-term callable debt differs from firms issuing long-term callable debt.

PAGE 150

Table 5-6 Cumulative Average Residuals for Announcement Period For Correctly vs Incorrectly Identified Firms Based on Logit Model 140 Correctly

PAGE 151

Table 5-7 Cumulative Average Residuals for Execution Period For Correctly vs Incorrectly Identified Firms Based on Logit Model 141 Correctly Identified Firms (n=436) Interval CAR -20

PAGE 152

142 The results shown in Appendix B, Table B-8 suggests a difference may exist between the two maturity types within the callable debt sample. Short maturity callable debt issues show significantly negative CAR's over 31 days of the announcement period test. During the same test period long maturity callable debt failed to differ significantly from zero at any point in time. The execution period test failed to detect any difference from for either the long or short maturity callable debt. (See Appendix B, Table B-9 ) This suggests that the market penalizes the shareholders when the firm pays a premium to attach a call option to short-term debt. Even with the significant CARs within the short maturity callable debt sample, the test of hypothesis 5-15 failed to detect a difference large enough to reject the null hypothesis at any reasonable significance level. The t-values for the formal tests are shown in Tables 5-8 and 5-9 for the announcement and execution periods respectively. Short versus Long Maturity: Noncallable Debt A similar set of hypotheses are defined for noncallable debt issues by equations 5-17 and 5-18. H CAR n,s = CAR n,l ( 5 17 ) The cumulative abnormal returns for firms issuing short-term noncallable debt does not differ from firms issuing long-term noncallable debt. H l CAR n,s CAR n,l (518 ) The cumulative abnormal returns for firms issuing short-term noncallable debt differs from firms issuing long-term noncallable debt.

PAGE 153

Table 5-8 Cumulative Average Residuals for Announcement Period Firms Issuing Short versus Long Maturity For Callable Debt Issues 143 Short Maturity Debt (

PAGE 154

Table 5-9 Cumulative Average Residuals for Execution Period Firms Issuing Short versus Long Maturity For Callable Debt Issues 144 Short Maturity Debt (

PAGE 155

145 The announcement period results shown in Appendix B, Table B-10 indicates a significantly negative CAR from day -1 through day 2 for the long-term noncallable debt No significant difference from was detected in the short maturities noncallable debt. During the execution period test, significantly positive CARs were detected from day through day 2 for the short-term noncallable debt issues. No further abnormalities were detected for the long-term noncallable debt issues during this test period. Results are detailed in Appendix B, Table B-ll. Null hypothesis 5-17 is rejected within the announcement period for the 41-day, 21-day, and 11-day tests. Within the execution period tests, the null hypothesis is rejected for the 41-day test only. This suggests that the market perceives a difference in the issuance of shortand long-term noncallable debt. The t-values for the formal tests are shown in Tables 5-10 and 5-11 for the announcement and execution periods respectively Additional Tests Four additional tests, revolving around the implications suggested by Robbins and Schatzberg's model, were conducted for both the announcement and execution periods. Their conclusions are that long-term callable debt dominates all other methods for raising funds. This implies

PAGE 156

Table 5-10 Cumulative Average Residuals for Announcement Period Firms Issuing Short versus Long Maturity For Noncallable Debt Issues 146 Short Maturity Debt (

PAGE 157

Table 5-11 Cumulative Average Residuals for Execution Period Firms Issuing Short versus Long Maturity For Noncallable Debt Issues 147 Short Maturity Debt (

PAGE 158

148 that the CARs associated with the issuance of long-term callable debt should exceed those from the issuance of short-term debt and any form of noncallable debt. The results shown in Tables 5-12 (announcement period tests) and 5-13 (execution period tests) fail to detect any statistical difference when the CARs from long-term callable debt are compared to short maturity debt, short-term noncallable debt, and noncallable debt under all six test periods. Summary Six sets of hypotheses were tested within this chapter to determine the effect debt issues have on the equity valueof the firm. Five of the six hypotheses could not be rejected around either the announcement or the execution day. The impact on the firm's equity value from the total debt issue sample was slightly negative but insignificantly different from zero. Callable debt issues generated an insignificant negative impact versus a insignificant positive impact for noncallable debt. The difference of the callable and noncallable debt issues was also insignificant within both the announcement and execution period tests. Similar results were also obtained when long and short maturity debt issues and correctly and incorrectly logit model identified issues were compared. The two subsamples within the callable debt sample, short and long maturity issues, also failed to detect a difference at any practical level of significance. The

PAGE 159

149 Table 5-12 Extra Cumulative Average Residuals Tests Announcement Period Long-term Callable versus Short Maturity Debt Interval L-T Call S-T Debt L-S t-value -20 through +20 -1.3838 -0.3818 -1.0020 -0.2368 -10 through +10 -0.2209 -0.9637 0.7427 0.2452 5 through +5 -0.4731 -0.6154 0.1422 0.0649 1 through +1 0.3068 -0.0960 0.4028 0.3519 1 through +5 -0.7087 0.0080 -0.7167 -0.4099 1 through 0.1003 -0.1397 0.2400 0.2568 Long-term Callable versus Short-term Noncallable Debt Interval L-T Call S-T Ncall L-S t-value -20 through +20 -1.3838 2.1644 -3.5483 -0.8278 -10 through +10 -0.2209 0.4102 -0.6311 -0.2057 5 through +5 -0.4731 0.2041 -0.6772 -0.3050 1 through +1 0.3068 -0.3069 0.6137 0.5293 1 through +5 -0.7087 0.1565 -0.8652 -0.4885 1 through 0.1003 -0.2130 0.3133 0.3309 Long-term Callable versus Noncallable Debt Interval L-T Call Ncall L-N t-value -20 through +20 -1.3838 0.0124 -1.3962 -0.2358 -10 through +10 -0.2209 -0.9871 0.7661 0.1808 5 through +5 -0.4731 -0.3829 -0.0902 -0.0294 1 through +1 0.3068 -0.4646 0.7714 0.4816 1 through +5 -0.7087 -0.0921 -0.6166 -0.2520 1 through 0.1003 -0.4190 0.5193 0.3971 Short-term Debt versus Noncallable Debt Interval s-T Ncall S-N t-value -20 through +20 -0.3818 0.0124 -0.3942 -0.0901 -10 through +10 -0.9637 -0.9871 0.0234 0.0074 5 through +5 -0.6154 -0.3829 -0.2325 -0.1027 1 through +1 -0.0960 -0.4646 0.3686 0.3117 1 through +5 0.0080 -0.0921 0.1001 0.0554 1 through -0.1397 -0.4190 0.2793 0.2893 Significant at 10% level using a two-tail test. Significant at 5% level using a two-tail test. Significant at 1% level using a two-tail test.

PAGE 160

Table 5-13 Extra Cumulative Average Residuals Tests Execution Period 150 Long-term Callable versus Short Maturity Debt Interval L-T Call S-T Debt L-S t-value 2 through +2

PAGE 161

151 noncallable debt sample did show a significant difference in the way the market perceives short and long maturity debt. Additional tests were conducted on Robbins and Schatzberg's contention that long-term callable debt dominates all other forms of raising funds. Tests within both the announcement and execution periods failed to support the dominance claim. Notes 1. CRSP stands for Center for Research in Security Prices. 2. See Brown and Warner (1985, p. 7). The t-test is: t = AR t /S A where S is the estimated standard deviation of average abnormal returns over the estimation period. 3. Brown and Warner (1985, p 29). The t-test for accumulated abnormal returns is: Sum (AR^) CAR+. t = t (Sum (S 2 )) 1 / 2 (t*S 2 ) 1 /2 t where the summation is from -20 through +2 0. 4. See Miles and Rosenfeld (1983, p. 1603). The t-test for a difference in CARs is: AR i,l AR i,2 V (1/T i,l> + <1/T if2 )]^ where AR^ = the stratum-specific average abnormal return over the interval i. T^ = the stratum-specific number of days in interval i. s p = the estimated pooled standard deviation which is defined on the next page.

PAGE 162

S p ^BaseU,].)" 1 ) 3 + ( T Base(i,2) _1 ) s 2 2 T Base ( i 1 ) +T Base ( i 2 ) ~ 2 A A = 119{S 1 2 ) + 179(S 2 2 ) V2 358 since T Base(i/1) = T Base(i;2) = 180 where A 2 S n = the estimated variance of group n's excess returns over the comparison period. 152

PAGE 163

CHAPTER VI VALUATION OF THE CALL OPTION In an attempt to determine what value, if any, the market has placed on the call option, all publicly traded debt issues reported in Moody 's Bond Survey from 1977 through 1986, excluding municipals, pollution control, and government issues are analyzed. During the ten year sample period, Moody's reported a total of 6740 debt issues of which 5903 were callable and the remaining 837 were noncallable. The unigueness of this data set is that a substantial number of noncallable debt issues exist. The need for extrapolation the value of the call option from call deferments is not needed. The opportunity to discharge a debt obligation before maturity, especially during volatile interest rate periods, is granted only in return for a higher coupon payment. A higher interest charge compensates the investor for the option 'sold' to the lender. In an informationally perfect market, the marginal participants in the debt market should be indifferent between callable and noncallable debt. An early study by Hess and Winn (1962) failed to uncover a significant value for the call option extrapolating from call deferments using data from 192 6 through 1959. Jen and Wert (1967) derived yields for noncallable 153

PAGE 164

154 debt from freely callable debt issues. Jen and Wert's analysis found no value for rates up to 5%. Above the 5% level, the value was approximately the same as a change from one rating class to the next. Using the option pricing framework and the assumptions that interest rates follow a Gauss-Wiener process and that the pure expectations hypothesis holds, Brennan and Schwartz (1977) found a wide range of compensation paid for the call option. At the low end of interest rates (2%) the cost for attaching a call feature was 19 basis points. At the other extreme of their study, 16% interest level, the cost was 327 basis points. Boyce and Kalotay (1979a) find that a tax-exempt investor, in order to compensate himself for the risk of calling, demands a coupon approximately 30 basis points higher than he would reguire on noncallable bonds. No distinction was made for noncallable debt during high and low interest rate cycles. Using market data, Pye (1976) found long-term utility bonds (maturities of 25 to 50 years) with a five year call deferment and ratings of A, Aa, or Aaa required a 4 basis point discount when applied to interest rates in the lower range of the interest rate cycle during 1959 through 1965. In the upper interest rate range of this same period, the discount was 13 basis points. Pye also used a theoretical model which produced results of 40 and 70 basis points during the low and high interest rate ranges for 3 year

PAGE 165

155 bonds. Pye showed the significance of these results by pointing out that the spread between the callable and noncallable debt issues were of the same magnitude as the rating quality improvements, A to Aaa, during the same time period. Two methods are utilized to answer the final question raised in the introduction: What is the value placed on the call option and does it vary across time? The methods include regression analysis and direct measurement within the debt market. Regression Analysis The regression analysis is divided into two distinct parts: identification and valuation. The first aspect of the regression analysis is utilized to detect the various factors that influence the yield offered on a bond when it is first issued, primarily the effect of the call option. The second phase deals with the effects directly related to the spread between callable and noncallable debt issues. The data set utilized for the regression phase of the analysis is restricted to debt issues without a zero coupon, a conversion feature, or a floating rate provision. Also the debt issues are restricted to financials, utilities, and industrials with debt ratings of Baa or higher. Identification The functional form of the regression is shown in equation 6-1.

PAGE 166

156 YIELD = f(CALL, CLASS, MAT, RATE, SIZE, TSEC, TSPREAD, UNCER) (6-1) The dependent variable, YIELD, is the offering yield of the debt instrument at the time of issue as reported in Moody's Bond Survey. Chatfield and Moyer (1986) point out that this is a better measure of the investor's yield at the time of issue versus the yield calculated from the issuer's cost which reflects the behavior of both the investor and the underwriter. The current yield is justified since all the debt issues used within the regression sample were sold at or within two percentage points of par value eliminating the need for defining elaborate methodologies for incorporating capital gains and losses. Definitions of the independent variables are shown in Table 6-1. Since the variables have already been defined in Chapter IV, no further definitions are given within the text. The hypothesized effects of the independent variables on YIELD are: CALL (+) CLASS (+/")/ MAT (+) RATE (-) SIZE (+/") TSEC (+) TSPREAD (+) and UNCER (+) The existence of the call option is expected to have a positive impact on debt yields, since the investor bears the risk of reinvestment at a lower yield before maturity if the issue is called. CALL is the major variable of interest in the analysis since it calculates the impact of the call feature.

PAGE 167

Table 6-1 Functional Form of the Yield Regression Equation 157 YIELD = f(CALL, CLASS, MAT, RATE, SIZE, TSEC, TSPREAD, UNCER) Dependent Variable: YIELD = The yield to maturity at the time of purchase. Independent CALL CLASS MAT RATE SIZE TSEC TSPREAD UNCER Varialbles: = A binary variable for the presence (1) or absence (0) of the call option. = Broad classification for firm types: FINANCE (0/1 binary for financial firms) UTILITY (0/1 binary for utilities) INDUSTRIALS (base case) = The maturity of the debt issue in years. = Broad classification for debt ratings: Aaa, Aa, and A (0/1 for absence/presence of respective debt rating with Baa as the base case. ) = The natural logarithm of the dollar amount of the debt issue. = The yield of a 3 year treasury security measured when the debt was issued. = The difference between the yields on 3 and 3 year Treasuries, a proxy for the shape of the yield curve. = The volatility of interest rates over the three week period prior to the debt issue date.

PAGE 168

158 The type of firm classification (CLASS) is defined through the two dummy variables, UTILITY and FINANCE. The utility dummy coefficient is expected to be positive reflecting the generally higher cost of financing for utility debt relative to egually rated industrial issues. The coefficient of the financial dummy is expected to be negative signifying a systematic difference in the capital costs relative to the base class of industrials. MAT is expected to be positively related to a bond's yield. The effect of issue size on the debt yields is difficult to predict. Arguments for both a negative and a positive effect on yield have been made. The positive argument revolves around the increased supply effect that a larger issue size causes yields to increase. This larger size could also be an indication of greater marketability, causing lower yields. It is possible that these two contrasting effects may have a cancelling effect on yield when the overall sample is tested. The debt ratings, incorporated as a set of dummy variables, should show up as a negative effect on yield, since ratings are measured relative to the base case of Baa. The size of the effect should increase as the debt ratings are increased. The variables associated with the level and uncertainty of rates should have a positive impact on a bond's yield. Rates in general are highly correlated.

PAGE 169

159 Therefore, as treasury yields (TSEC) increase, corporate yields should also increase. The proxy for the shape of the yield curve, TSPREAD, should also be positively related to yields. A rising yield curve indicates an increase in rates, while a falling yield curve indicates anticipation of falling rates. UNCER attempts to account for the variance in interest rates in the period leading up to the debt issue. Thus, it is expected to be positively correlated with the bond yield. The results of the yield regression are shown in Table 6-2. The relative signs of the estimated coefficients coincide with the predicted directional effects. The independent variable, CALL, has an estimated beta of 0.5956. This result is close to the estimates given in previous research and comes close to the estimates obtained in the next section by direct estimation. To offer further support for the valuation of the call option, CALL was regressed against the independent variable, YIELD. The estimated coefficient of 0.5734 is highly significant (t-value is 5.077). Thus, it is reasonable to assume that during the ten year period 1977 through 1986, the call option costs the issuer approximately 60 basis points. Valuation The second aspect of the regression analysis attempts to measure the factors that directly affect the premium paid

PAGE 170

160 Table 6-2 Yield Regression Model Variable

PAGE 171

161 for the call option. The dependent variable in this second regression is DIFF, a proxy for the call premium's cost. DIFF is defined as the difference in yields between matched pairs of callable and noncallable debt issues. The criteria for matching callable and noncallable debt issues include same maturity, same debt rating, and same month of issuance. The independent variables are proxies for the level and uncertainty of rates (TSEC and UNCER) the slope of the yield curve (TSPREAD) maturity (MAT) and the ratings attached to the debt issue (restricted to Aaa, Aa, A, and Baa only) The functional form of the regression equation for DIFF is: DIFF = f(MAT, RATE, TSEC, TSPREAD, UNCER) A summary of the meaning for each variable is provided in Table 6-3. The hypothesized effects of the independent variables on DIFF are MAT (+) RATE (-) TSEC (+) TSPREAD (-) and UNCER (+) An increase in either the level or uncertainty of interest rates should drive up the premium required on the call option. Thus the expected relationship between TSEC and UNCER with DIFF is positive. The cost of the call option is also expected to increase as the length of maturity on the debt pair increases. RATE, a set of binary variables for bond ratings, should show a negative correlation with DIFF. As the ratings improve, the difference between the yields offered

PAGE 172

162 Table 6-3 Functional Form of the Call Option Regression Equation DIFF = f(MAT, RATE, TSEC, TSPREAD, UNCER) Dependent Variable: DIFF = The difference between the yields on matched pairs of callable and noncallable debt. Independent Varialbles: MAT RATE TSEC TSPREAD UNCER = The maturity of the matched debt issues in years. = A set of binary variables for the match pair's bond rating (Aaa, Aa, and A with Baa used as the base case) = The yield of a 3 year treasury security measured when the debt was issued. = The difference between the yields on 3 and 3 year Treasuries, a proxy for the shape of the yield curve. = The volatility of interest rates over the three week period prior to the debt issue date.

PAGE 173

163 on callable and noncallable debt should decrease since the riskiness of the debt issue decreases. TSPREAD is also expected to be negatively correlated with the cost of the call option. If TSPREAD is positive, this indicates a rising yield curve and thus limited use for the call option. When TSPREAD is negative, this indicated a falling yield curve and increased opportunities for utilizing the call option in the future. The regression results are shown in Table 6-4 The regression with the full set of independent variables is included to show that all the estimated coefficients contain the hypothesized sign. The second regression shown includes only the significant variables. The regression results are misleading in that both the volatility of interest rates and the shape of the yield curve show up as insignificant factors. The lack of significance for the variables TSPREAD and UNCER is due to a lack of variability within the matched pair sample set and not with the lack of importance of these factors. The effects of maturity is small. The estimated coefficient for maturity indicates that for each year that the maturity of a debt issue is increased, the cost of the call premium increases 1 basis point. The rating proxies indicate that the higher the rating attached to the debt issue the lower the cost of the call option. The difference between a Baa and Aaa rated issue is 38 basis points.

PAGE 174

Table 6-4 Call Option Regression Model 164 Variable

PAGE 175

165 Similar differences can be estimated for the other rating classes using the estimated coefficients found in Table 6-4. The effect of the level of rates is shown by the coefficient for TSEC. As interest rates increase, the cost of the call option also increases. The results indicate that the valuation of the call option is effected by the debt rating attached to the bond, the level of interest rates, and the maturity of the issue. Although unsupported by the regression on DIFF, prior work within this study show that the uncertainty of interest rates within the market at the time of issuance as well as the slope of the yield curve impact the cost attached to the call option. The mean yields of the debt issues used within this study by rating, date of issue (month and year) and debt type (callable and noncallable) are provided in Appendix C, Tables C-l through C-2 0. These tables give an indication of the range of rates within each debt sector over the period 1977 through 1986. Direct Estimation Related to the yield regression is the direct estimation of the call options from the summary data. Some indication of the value that the market places on the call option is provided by comparing the yields of callable and noncallable issues. Two different calculations are made within this section of the analysis. The first set of calculations is the difference between the average yield for

PAGE 176

166 all noncallable and callable debt issues segmented by the rating classification on a yearly basis. The results obtained from this set of difference calculations place a 61 basis point valuation on the call option. The range of yearly values show a low of 20 basis points in 1979 to a high of 94 basis points in 1983. See Table 6-5 for the complete results on a yearly basis. Also reported are the valuation of the call option by rating classifications. The Aaa rating shows a value of the call option of 54 basis points which increases to 68 basis points on Baa rated debt issues. Valuations are not reported on ratings below the Baa class due to the limited number of matched issues within each class and thus a lack of confidence in the estimates. See Table 6-6 for the valuation estimates by rating classifications. The second method utilized uses only matches between callable and noncallable debt issues from the same firm. The advantage of the second method over the first, is that the riskiness of the debt instruments are better matched by using only debt from the same firm. A summary of the restrictions applied to the debt pairs used within method 2 include the following: 1. Debt pairs must be from the same firm. 2. Debt pairs must be within the same risk class. 3. Debt pairs must be within three years for maturity. 4. Debt pairs must be within 2 months of issue date. Method 2 offers a better estimate of the premium that must be paid for the call option since reguirement 1 forces a

PAGE 177

Table 6-5 Matched Yield Difference Summary by Year 167

PAGE 178

Table 6-6 Matched Yield Difference Summary by Ratings 168

PAGE 179

169 better control of risk. The call option's average value over the ten years of the study is 65 basis points. Again there exists a range of values over the individual years. Some of the yearly results may be misleading due to the limited number of issues used in the calculations. For instance, 1980 shows a low of 35 basis points, but this is based upon only one matched pair. See Table 6-5 for a summary of the yield spreads using the second methodology on a yearly basis. Also reported are the differences for the first three rating classes. The values are slightly higher than those found within the first method. The highest rating, Aaa, shows a value for the call of 57 basis points compared to 54 under method 1. Similar increases are shown for the Aa and A classifications in Table 6-6. Summary Using a unigue data set that contains a significant number of noncallable debt issues, an approximation of the value of the call option is made. For the overall ten-year period, the call option is valued at approximately 60 basis points. The estimates within the individual years show a wide variation. This variation is supported by the regression analysis which shows that the value of the call option is affected by the level of interest rates and the maturity and rating of the debt issue. Prior chapters have shown other key factors that directly affect the value of the call option such as issue size and strength of the issuing firm.

PAGE 180

CHAPTER VII SUMMARY AND CONCLUSION Introduction This study was designed to examine the relationship between noncallable and callable debt. No previous study, to our knowledge, has attempted to determine the differences that exist between the two debt types using actual noncallable debt issues or to document the exact amount of noncallable debt that is actually issued. Thus, this study fills a gap in the debt literature that had previously existed. The issues addressed in this research are of importance for two very practical reasons. First, the empirical validation of the existence of noncallable debt establishes an area of financing that has previously been ignored in research. Second, theoretical models which predict the effect of various debt issues are tested and rejected, suggesting the need for further examination of the underlying models. Summary of Results The four specific guestions outlined within the introduction were examine within this study and are utilized as the focus of the summary. 1. Are there identifiable market factors that dictate when and how much noncallable debt is offered? 170

PAGE 181

171 An answer to this first question can be found in Chapter III (Summary Statistics) and IV (Logit Regression Analysis) From the summary statistics, a link between the issuance of noncallable debt and the level and uncertainty of interest rates was not established. The logit analysis shows the importance of the level of interest rates, but fails to support the effect of uncertainty of rates. This lack of support is due primarily to the limited dispersion of the range within the callable and noncallable debt sectors. 2 Are there factors common to the firms that issue noncallable debt that differentiate them from firms that issue callable debt? The second question is also answered within the third and fourth chapters. Key attributes that show up in firms issuing noncallable debt include high ratings, a tendency towards finance related industries, large debt issues as well as large firm size, shorter maturities, higher profitability, lower growth, lower debt ratios (debt to equity and debt to assets) and a larger base to support the debt issue. The sample data was categorized into callable and noncallable groups for supporting the above conclusions. Within this study the means, standard deviations, and population distributions were compared by the t, F, and z tests for coupons, yields, amount of issues in millions of dollars, and maturities. Also analyzed by comparing the proportion within each of the two debt classifications were risk as measured by Moody's bond ratings, firm

PAGE 182

172 classifications (industrial, financial, utility, transportation, or international) conversion features, and floating debt. Tables 7-1 and 7-2 summarize the means and proportions respectively of the key attributes of the debt issues as an overview of the differences between the callable and noncallable debt categories. Summary statistics were also provided for the firm specific variables. These included total assets, long-term debt, dividends and earning per share, interest coverage, profit margin, property, plant, and equipment, sales, and various growth rates. A summary of the key firm variables are provided in Table 7-3. 3. Does the issuance of callable debt impact the market value of the firm's equity differently from the issuance of noncallable debt? The answer to question three is found primarily within chapter V. The event study analysis failed to show any difference in the impact from the issuance of noncallable and callable debt when tested in either the announcement or execution period. When the noncallable sample was further segmented by maturity, a significantly positive impact was detected in both the announcement and execution periods for short maturity noncallable debt. This suggests that short maturity noncallable debt is perceived by the market as good news. 4. What is the value of the call option and does it vary across time?

PAGE 183

173 Table 7-1 Mean Attributes of Callable and Noncallable Debt Issues Means:

PAGE 184

Table 7-2 Proportion of Debt Sample Segmented by Attributes 174 Proportions of Total Sample Within Each Debt Class Callable Noncallable Significance Variable Debt Debt Level Firm Classification:

PAGE 185

175 Table 7-3 Mean Values of Firm Specific Attributes Means:

PAGE 186

176 The value of the call option was measured by both regression analysis and direct valuation with the average value being approximately 60 basis points. This valuation did differ by the level of rates across time. Future Research Most studies conclude by suggesting possible avenues for continued research. This study is no different. At least three areas which require examination can be outlined. First, much more study is needed to explore the effect of debt issues on the firm's equity value from firms not limited to the New York and American Stock Exchanges. With the use of the NASTEX tapes, a whole unexplored area has been opened. Secondly, the large amount of nonrated debt issues within this study dictate further research into the reasons and effects of nonrated debt. Topics include marketability of nonrated debt, whether the nonratings were just at the time of Moody's publication or whether they carry over the life of the debt instrument. Finally, the various models outlining the dominance of callable debt, such as the one outlined by Robbins and Schatzberg (1986), need to be reexamined and reformulated. Conclusion This paper has presented a series of tests which have documented the existence of a significant amount of noncallable debt and suggested reasons why it is sometimes advantageous for a firm to issue noncallable debt. The

PAGE 187

177 analysis was performed over a broad period from 1977 through 1986 suggesting similar results could be obtained by incorporating future years into the study. The results generally support the common notions for the issuance of callable debt but fails to support the market reaction on a firm's equity value from such an issue.

PAGE 188

APPENDIX A SUMMARY TABLES 178

PAGE 189

Table A-l Summary Table: Coupon Frequencies by Year 179 Coupon CALLABLE DEBT in % 1977 1978 1979 1980 1981 1982 1983 1984 1985 1985 1 6.9 2 7 7.9 34 10 16 12 32 11 12 34 31 43 11 94 8

PAGE 190

Year Year Table A-2 Number of Debt Issues by Month for Each Year 180 Callable Debt Issues Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1977

PAGE 191

181 Table A-3 Mean Yields at Time of Issuance by Year Year

PAGE 192

182 Table A-4 Summary Table: Yield Frequency by Year Yield

PAGE 193

183 Table A-5 Yield Frequency Distribution by Firm Classification Y

PAGE 194

184 Table A-6 Summary Table: Mean Maturities by Year Year

PAGE 195

185 Table A-7 Summary Table: Maturity Frequencies by Year Maturity CALLABLE DEBT in Years 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 <

PAGE 196

186 Table A-8 Mean Amounts of Debt Issues in Million of Dollars Year Call Noncall Sample Size Call Noncall 1977 87.9 119.8 -1.73 (70.2) (71.4) -1.76* 1.04 -2.14** 248 16 1978 88.5 165.6 -2.19* (70.7) (104.5) -3.14*** 2.19* -2.80*** 211 9 1979 109.0 130.0 -0.98 (91.8) (54.5) -0.60 2.84 -1.39 232 7 1980 96.4 129.5 -2.36** (76.3) (65.8) -2.07** 1.34 -2.94*** 362 24 1981 103.0 123.4 -1.68* (95.9) (70.3) -1.31 1.86** -2.52** 381 40 1982 81.6 108.2 -2.97*** (73.8) (86.7) -3.27*** 1.38** 3.47*** 433 111 1983 78.3 95.6 -2.26** (82.6) (64.0) -1.89* 1.67*** 3.33*** 515 90 1984 94.0 124.5 -2.11** (126.4) (136.9) -2.23** 1.17 3.97*** 590 103 1985 74.8 150.8 -6.31*** (85.3) (153.3) -9.47*** 3.23*** 10.39*** 1087 170 1986 97.0 150.4 -8.17*** (134.4) (90.3) -6.07*** 2.22*** 11.57*** 1831 248 Table Code: (1) (2) (4) (6) (3) (5) (7) (8) (9) (10) (11) (1) Year (2) Mean (Callable Debt) (3) Standard Deviation (Callable Debt) (4) Mean (Noncallable Debt) (5) Standard Deviation (Noncallable Debt) (6) t-value (assuming unequal variance) (7) t-value (assuming equal variance) (8) F-value (testing equality of variances) (9) z-value (nonparametric Wilcoxon test statistic) (10) Sample size (Callable Debt) (11) Sample size (Noncallable Debt) Significant at 10% level using a one-tail test. Significant at 5% level using a one-tail test. ***Significant at 1% level using a one-tail test.

PAGE 197

187 Table A-9 Frequency Distribution of Dollar Amount of Debt Issue Amount Callable Debt Issues of Issue 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 ($million) <50

PAGE 198

188 Table A-10 Frequency Distribution of Dollar Issue Amount by Firm Type Amount

PAGE 199

189 Table A-ll Frequency Distribution of Number of Issues per Firm

PAGE 200

190 Table A-12 Total Assets of Issuing Firms during the Period 1977-1986 Industrials Mean Standard Deviation Sample Size t-scoref* F-score z-score c Callable $3.0 6.2 1070 issues Noncallable $13.7 29.4 146 issues -4.3536*** 22.20 *** -10.9462*** Financial Mean Standard Deviation Sample Size t-scoref* F-score z-score c Callable $31.2 44.5 461 issues Noncallable $29.7 35.2 262 issues 0.5307 1.60 *** 2.7857* Utilities Mean Standard Deviation Sample Size t-score^ F-score z-score c Callable

PAGE 201

191 Table A-13 Summary Table: Firm Type by Year

PAGE 202

Table A-13 — continued 192

PAGE 203

193 Table A-14 Summary Table: SIC Industry Code For the Period 1977-1986

PAGE 204

194 Table A-15 Summary Table: Ratings by Year

PAGE 205

Table A-15 — continued 195

PAGE 206

196 Table A-16 Debt Ratios of Issuing Firms during the Period 1977-1986 Debt to Asset Ratio Industrial Mean Standard Deviation Sample Size t-score^ F-score z-score c Callable Noncallable 0.262 0.218 0.154 0.108 1074 issues 144 issues 4.2926** 2.01 *** 2.8582** Debt to Equity Ratio Industrial Mean Standard Deviation Sample Size t-scoref* F-score z-score c Callable Noncallable 0.921 0.701 1.848 0.925 1074 issues 144 issues 2.3052** 3.99 *** 1.5032 Debt to Asset Ratio Financial Mean Standard Deviation Sample Size t-scoref* F-score z-score c Callable 0.200 0.194 460 issues -0.8305 1.02 -0.9431 Noncallable 0.213 0.195 264 issues Debt to Equity Ratio Financial Mean Standard Deviation Sample Size t-score^ F-score z-score c Callable 2.516 5.321 460 issues 1.6259 13.23 *** 1.7166* Noncallable 2.086 1.463 264 issues

PAGE 207

Table A-16 — continued 197 Debt to Asset Ratio Utilities Mean Standard Deviation Sample Size t-scoref* F-score z-score c Callable 0.367 0.072 509 issues -0.0540 1.44 0.4770 Noncallable 0.367 0.086 36 issues Debt to Equity Ratio Utilities Mean Standard Deviation Sample Size t-scoref* F-score z-score c Callable Noncallable 2.394 23.716 509 issues 1.0768 1783.44 *** 0.5755 1.258 0.561 36 issues ^Test for a difference in the means. b Test for a difference in the standard deviations, c Test for a difference in distributions, (nonparametric Wilcoxon test statistic) Significant at 10% level using a one-tail test, Significant at 5% level using a one-tail test. Significant at 1% level using a one-tail test.

PAGE 208

198 Table A-17 Profitability Measures of Issuing Firms During the Period 1977-1986

PAGE 209

199 Table A-18 Average and Marginal Tax Rate of Issuing Firms During the Period 1977-1986 AVERAGE TAX RATES Industrials Callable Noncallable Mean Standard Deviation Sample Size t-scoref* F-score z-score c 30.9% 225.2 1067 issues 38.1% 30.8 145 issues -0.9778 53.58 *** -2.1017* Financial Callable Noncallable Mean Standard Deviation Sample Size t-scoref* F-score z-score c 34.6% 133.2 459 issues 32.5% 14.9 262 issues 0.3332 79.93 *** -0.8764 Utilities Callable Noncallable Mean Standard Deviation Sample Size t-score^ F-score z-score c 31.4% 14.0 509 issues 1.9551* 1.18 2.3545* 26.6% 12.9 34 issues

PAGE 210

Table A-18 — continued 200 MARGINAL TAX RATES Industrials Callable Noncallable Mean Standard Deviation Sample Size t-score^ F-score b z-score c 28.8% 15.2 1075 issues -5.9210*** 1.68 *** 4.3079*** 35.3% 11.7 143 issues Financial Callable Noncallable Mean Standard Deviation Sample Size t-score^ F-score z-score c 35.6% 12.2 460 issues -4 .8571*** 2.06 *** 3.4529*** 39.3% 8.5 264 issues Utilities Callable Noncallable Mean Standard Deviation Sample Size t-scoref* F-score z-score c 37.5% 11.0 509 issues -0.9205 2.29 *** -0.1319 38.8% 7.3 36 issues ^Test for a difference in the means. Test for a difference in the standard deviations. c Test for a difference in distributions. (nonparametric Wilcoxon test statistic) Significant at 10% level using a one-tail test. **Significant at 5% level using a one-tail test. ***Signif icant at 1% level using a one-tail test.

PAGE 211

Table A-19 Issue Frequency Summary by Interest Rate Level For the Period 1977-1986 201 Interest

PAGE 212

APPENDIX B AVERAGE AND CUMULATIVE AVERAGE EXCESS RETURN TABLES 202

PAGE 213

Table B-l Abnormal and Cumulative Abnormal Returns For Firms Issuing Debt n = 631 203 Day

PAGE 214

204 Table B-2 Announcement Period Effects: AR and CAR For Firms Issuing Callable versus Noncallable Debt Day

PAGE 215

205 Table B-3 Execution Period Effects: AR and CAR For Firms Issuing Callable versus Noncallable Debt Day

PAGE 216

206 Table B-4 Announcement Period Effects: AR and CAR For Firms Issuing Short versus Long Maturity Debt Day

PAGE 217

Table B-5 Execution Period Effects: AR and CAR For Firms Issuing Short versus Long Maturity Debt 207 Day

PAGE 218

Table B-6 Announcement Period Effects: AR and CAR For Correctly and Incorrectly Identified Firms Based on the Logit Model 208 Day

PAGE 219

Table B-7 Execution Period Effects: AR and CAR For Correctly and Incorrectly Identified Firms Based on the Logit Model 209 Day

PAGE 220

Table B-8 Announcement Period Effects: AR and CAR For Firms Issuing Short versus Long Maturity For Callable Debt Issues 210 Day

PAGE 221

Table B-9 Execution Period Effects: AR and CAR For Firms Issuing Short versus Long Maturity For Callable Debt Issues 211 Day

PAGE 222

Table B-10 Announcement Period Effects: AR and CAR For Firms Issuing Short versus Long Maturity For Noncallable Debt Issues 212 Day

PAGE 223

Table B-ll Execution Period Effects: AR and CAR For Firms Issuing Short versus Long Maturity For Noncallable Debt Issues 213 Day

PAGE 224

APPENDIX C AVERAGE YIELDS BY DATE, RATING, AND MATURITY 214

PAGE 225

215 Table C-l Average Yields Noncallable Debt 1977 Maturity

PAGE 226

Table C-2 Average Yields Callable Debt 1977 216 Matu

PAGE 227

217 Table C-3 Average Yields Noncallable Debt 1978 Maturity

PAGE 228

Table C-4 Average Yields Callable Debt 1978 21! Matt

PAGE 229

219 Table C-5 Average Yields Noncallable Debt 1979 Maturity

PAGE 230

Table C-6 Average Yields Callable Debt 1979 220 Matt

PAGE 231

221 Table C-7 Average Yields Noncallable Debt 1980 Mati

PAGE 232

Table C-8 Average Yields Callable Debt 1980 222 Matu

PAGE 233

223 Table C-9 Average Yields Nonmailable Debt 1981 Maturity

PAGE 234

Table C-10 Average Yields Callable Debt 1981 224 Matu

PAGE 235

225 Table C-ll Average Yields Noncallable Debt 1982 Maturity Date/Rating Jan Aaa

PAGE 236

226 Table C-12 Average Yields Callable Debt 1982 MatL

PAGE 237

227 Table C-13 Average Yields Noncallable Debt 1983 Matt

PAGE 238

Table C-14 Average Yields Callable Debt 1983 228 Matu

PAGE 239

229 Table C-15 Average Yields Noncallable Debt 1984 Matt

PAGE 240

Table C-16 Average Yields Callable Debt 1984 230 Mati.

PAGE 241

Table C-17 Average Yields Noncallable Debt 1985 231 Maturity Date/Rating Jan Aaa

PAGE 242

Table C-18 Average Yields Callable Debt 1985 232 Matu

PAGE 243

Table C-19 Average Yields Noncallable Debt 1986 233 Matu

PAGE 244

Table C-20 Average Yields Callable Debt 1986 234 Matu

PAGE 245

REFERENCES Aivazian, V. A., and J. L. Callen (1980). "Future Investment Opportunities and the Value of the Call Provision on a Bond: Comment," Journal of Finance 35:1051-1054. Allen, D. S., R. E. Lamy, and G. R. Thompson (1987). "Agengy Costs and Alternative Call Provisions: An Emperical Investigation," Financial Management 16:37-44. Barnea, A., R. A. Haugen, and L. W. Senbet (1980). "A Rationale for Debt Maturity Structure and Call Provisions in the Agency Theoretic Framework," Journal of Finance 35:1223-1234. Barnea, A., R. A. Haugen, and L. W. Senbet (1981). "Market Imperfections, Agency Problems, and Capital Structure: A Review," Financial Management 10:7-22 Barnea, A., R. A. Haugen, and L. W. Senbet (1985). Agency Problems and Financial Contracting Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 61-111. Bodie, Z., and B. M. Friedman (1978). "Interest Rate Uncertainty and the Value of Bond Call Protection," Journal of Political Economy 86:19-43. Bodie, Z., and R. A. Taggart, Jr. (1978). "Future Investment Opportunities and the Value of the Call Provision on a Bond," Journal of Finance 33:1187-1200. Bodie, Z., and R. A. Taggart, Jr. (1980). "Future Investment Opportunities and the Value of the Call Provision on a Bond," Journal of Finance 35:1055-1056. Bowlin, 0. J. (1966) "The Refunding Decision: Another Special Case in Capital Budgeting," Journal of Finance 21:55-68. Boyce, W. M. and A. J. Kalotay (1979a). "Optimum Bond Calling and Refunding, Interfaces," The Institute of Management Science 9:36-49. Boyce, W. M. and A. J. Kalotay (1979b). "Tax Differentials and Callable Bonds," Journal of Finance 34:825-838. 235

PAGE 246

236 Brennan, M. J., and E. S. Schwartz (1977). "Savings Bonds, Retractable Bonds, and Callable Bonds," Journal of Financial Economics 5:67-88. Brigham, E. F. (1986). Fundamentals of Financial Management Dryden Press, New York, 382. Brown, S. J., and J. B. Warner (1985). "Using Daily Stock Returns," Journal of Financial Economics 8:3-31. Campbell, T. and W. Kracaw (1980). "Information Production, Market Signalling, and the Theory of Financial Intermediation," Journal of Finance 35:863-882. Dunn, L.Y., and W.H. Mikkelson (1984). "Convertible Debt Issuance, Capital Structure Change, and Financing Related Information: Some New Evidence," Journal of Financial Economics 13:157-186. Eckbo, B. E. (1986) "Valuation Effects of Corporate Debt Offerings," Journal of Financial Economics 15:119-151 Elton, E. J., and M. J. Gruber (1971). "Dynamic Programming Applications in Finance," Journal of Finance 26:473-506. Elton, E. J., and M. J. Gruber (1972). "The Economic Value of the Call Option," Journal of Finance 27:891-901. Fama, E. F. (1980). "Agency Problems and the Theory of the Firm," Journal of Political Economy 88:3-29. Flannery, M. J. (1986) "Asymmetric Information and Risky Debt Maturity Choice," Journal of Finance 41:19-37. Haugen, R. A., and L. W. Senbet (1988). "Bankruptcy and Agency Costs: Their Significance to the Theory of Optimal Capital Structure, Journal of Financial and Quantitative Analysis f 23:27-38. Hess, A. P., Jr., and W. J. Winn (1962). The Value of the Call Privilege McGregor and Werner, Inc. Washington, D.C., 1-100. Jen, F. C, and J. E. Wert (1967). "The Effect of Call Risk on Corporate Bond Yields," Journal of Finance 22:637-651. Jensen, M. C. and W. H. Meckling (1976). "Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure," Journal of Financial Economics 3:305-360. Judge, G. G., W. E. Griffiths, R. C. Hill, H. Lutkepohl, and T. Lee (1985) The Theory and Practice of Econometrics John Wiley and Sons, New York, 752-778.

PAGE 247

237 Kidwell, D. S. (1976). "The Inclusion and Exercise of Call Provisions By State and Local Governments," Journal of Money. Credit and Banking 8:391-398. Kraus, A. (1973) "The Bond Refunding Decision in an Efficient Market," Journal of Financial and Quantitative Analysis 8:793-806. Leland, H. E. and D. H. Pyle (1976). "Informational Asymmetries, Financial Structure, and Financial Intermediation," Journal of Finance 32:371-387. Marshall, W. J., and J. B. Yawitz (1980). "Optimal Terms of the Call Provision on a Corporate Bond," Journal of Financial Research 3:203-211. Masulis, R. W. (1983). "The Impact of Capital Structure Change on Firm Value: Some Estimates," Journal of Finance 38:107-126. McClave, James T. and P. George Benson (1982). Statistics for Business and Economics Dellen Publishing Company, San Francisco, 675-679. McConnell, J. J., and C. J. Muscarella (1985). "Corporate Capital Expenditure Decisions and the Market Value of the Firm," Journal of Financial Economics 14:399-422. Morris, J. R. (1976). "On Corporate Debt Maturity Strategies," Journal of Finance 31:29-37. Modigliani, F. and M. H. Miller (1958). "The Cost of Capital, Corporation Finance and the Theory of Investment," American Economic Review 48:261-297. Myers, S. C. (1971). "Discussion," Journal of Finance 26:538-539. Myers, S. C. (1977). "Determinants of Corporate Borrowing," Journal of Financial Economics 5:147-175. Myers, S. C. and N. S. Majluf (1984). "Corporate Financing and Investment Decisions When Firms Have Information That Investors Do Not Have," Journal of Financial Economics 13:187-221. Neter, J., W. Wasserman, and M. H. Kutner (1985). Applied Linear Statistical Models Richard D. Irwin, Inc., Homewood, Illinois, 361-367. Press, S. and S. Wilson (1978). "Choosing Between Logistic Regression and Discriminant Analysis," Journal of the American Statistical Association 23:699-705.

PAGE 248

2 3 8 Pye, G. (1966). "The Value of the Call Option on a Bond," Journal of Political Economy 74:200-205. Pye, G. (1976). "The Value of Call Deferment of a Bond: Some Emperical Results," Journal of Finance 22:623-636. Radcliffe, R. C. (1987). Investment — Concepts. Analysis, and Strategy Scott, Foresman and Company, Glenview, Illinois, 38. Robbins, H. R. and J. D. Schatzberg (1986). "Callable Bonds: A Risk-Reducing Signalling Mechanism," Journal of Finance 41:935-949. Ross, S. A. (1977). "The Determination of Financial Structure: The Incentive Signalling Approach," Bell Journal of Economics 8:23-40. Ross, S. A., and R. W. Westerfield (1988). Corporate Finance Times Mirror/Mosby College Publishing, St. Louis, Missouri, 333. Stiglitz, J. E. (1979), "On the Irrelevance of Corpaorate Financial Policy," American Economic Review 69:147-153. Smith, C. W. Jr., and J. B. Warner (1979). "On Financial Contracting," Journal of Financial Economics 7:117-161. Thatcher, J. S. (1985), "The Choice of Call Provision Terms: Evidence of the Existance of Agency Costs of Debt," Journal of Finance 40:549-561. Van Home, J. C. (1980). "Called Bonds: How Did the Investor Fare?," Journal of Portfolio Management 6:58-61. Van Home, J. C. (1977) Financial Management and Policy Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 570. Van Home, J. C. (1984). Financial Market Rates and Flows Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 203-220. Vu, J. D. (1986) "An Empirical Investigation of Calls of Non-Convertible Bonds," Journal of Financial Economics 18:235-263. Weston, J. F., and E. F. Brigham (1987). Essentials of Managerial Finance Dryden Press, New York, 710.

PAGE 249

BIOGRAPHICAL SKETCH Richard John Kish, the son of Paul J. and Sophia Kish, was born in Natrona Heights, Pennsylvania, on May 24, 1952. He earned a Bachelor of Science degree in secondary education with a concentration in mathematics and athletic coaching from Clarion State University, Clarion, Pennsylvania, in 1977. He earned a Master of Business Administration with a concentration in finance at the University of Florida in 1985. The requirements for the degree of Doctor of Philosophy (finance) were completed during the summer of 1988 at the University of Florida. Richard and his wife Janine currently reside in Bethlehem, Pennsylvania, where Mr. Kish is an assistant professor in the finance department of Lehigh University. 239

PAGE 250

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Miles Livingston, Chairman Associate Professor of Finance, Insurance, and Real Estate I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy.
PAGE 251

UNIVERSITY OF FLORIDA 3 1262 08285 250 9