Title: Noncallable debt
CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00099571/00001
 Material Information
Title: Noncallable debt evidence and effect
Physical Description: x, 239 leaves : ; 28 cm.
Language: English
Creator: Kish, Richard John, 1952-
Publisher: Richard John Kish
Place of Publication: Gainesville, Fla.
Publication Date: 1988
Copyright Date: 1988
 Subjects
Subject: Corporate debt   ( lcsh )
Corporations -- Finance   ( lcsh )
Options (Finance)   ( lcsh )
Finance, Insurance, and Real Estate thesis Ph.D
Dissertations, Academic -- Finance, Insurance, and Real Estate -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
 Notes
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
Bibliographic ID: UF00099571
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 001103040
oclc - 19657784
notis - AFJ9130

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




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs