Chris James

8/2/2019 Chris James

1/19

Journal of Financial Economics 19 (1987) X7-235. North-Holland

SOME EVIDENCE ON THE UNIQUENESS OF BANK LOANS*

Christopher JAMESfJnioersi[r of Oregotl, Eugene, OR 97402. LS.1

Received April 1986. final version received June 1987

This paper presents evidence that banks provide some special service with their lending activity

that is not available from other lenders. I find evidence that bank borrowers. not CD holders. bearthe cost of reserve requirements on CDs. In addition. I ftnd a positive stock price response to theannouncement of new bank credit agreements that is larger than the stock price responseassociated with announcements of private placements or public straight debt offerings. Finally. Ifind significantly negative returns for announcements of private placements and straight debtissues used to repay bank loans.

1. Introduction

Although the economic rationale for commercial banks and other financialintermediaries is not well understood, recent theories of financial interme-diation have focused on the role of banks in information production andtransmittal [see. for example, Leland and Pyle (1977). Campbell and Kracaw(1980). and Diamond (1984)]. Banks and other intermediaries, the argumentgoes. have a cost advantage over other outsiders in producing and transferringinformation, either because of something intrinsic in the intermediation process.as Leland and Pyle and Diamond suggest. or because information productionand the provision of transaction and other intermediary services are comple-mentary activities. An implication of this view is that bank loans are differentfrom publicly placed debt because banks know more about a companysprospects than other investors do.

The emphasis on information transmission contrasts sharply with an alter-nate hypothesis about the role of banks in the economy. The alternative holdsthat their special function is to provide transaction services through theissuance of demand deposits. On the asset side, banks are assumed to besimply passive portfolio managers [see Fama (1980)].

*A portion of this study was completed while I was visiting the University of Michigan. Thanks

to James Bricklev. Larrv Dann. Mark Flannerv. Ronald Lease, Wavne Mikkelson. Greg Niehaus.Megan Partch, Peggy Wier. seminar participants at the University of Oregon, Umversity ofRochester. and New York University. an anonymous referee, and especially Rene Stulz (theeditor) for helpful comments.

0304-405X/87/%3.50 Z 1987. Elsevier Science Publishers B.V. (North-Holland)


2/19

118 C. James. The unrqueness of bank louns

This paper provides evidence on whether commercial banks provide anyspecial service with their lending activity that is not available from otherlenders (i.e.. on whether bank loans are unique). This evidence comes first

from an examination of the incidence of the reserve requirement tax, andsecond from an analysis of the stock price response to announcements of bankloans, private placements of debt, and public straight debt issues.

My first examination extends research by Fama (1985) who studies theincidence of reserve requirements on bank certificates of deposit (CDs). Heargues that because close substitutes for bank CDs, such as commercial paperor bankers acceptances, exist and because CDs provide no special transactionservices, reserve requirements on CDs must be borne by bank borrowers. Insupport of this conjecture, Fama finds no significant differences between the

average yields on CDs and on high-grade commercial paper or bankersacceptances. Fama concludes that because bank borrowers bear the cost ofreserve requirements there must be something special about bank loans thatdistinguishes them from other types of privately placed and publicly placeddebt.

A problem with Famas conclusion is that the reserve requirement tax couldbe at least partially offset by a subsidy from the Federal Deposit InsuranceCorporation (FDIC) in the form of deposit insurance supplied at less thanactuarially fair prices. A more powerful test of the incidence of the reserve tax,

provided here, examines the behavior of CD rates around changes in reserverequirements. when no offsetting changes in deposit insurance prices occur.My results support Famas conclusion that the reserve tax is borne by bankborrowers.

A second source of evidence on the uniqueness of bank loans comes from acomparison of stock price responses to the announcements of bank loanagreements and other types of debt offerings. In analyzing the function ofcommercial banks, Kane and Malkiel (1965) and more recently Fama (1985)and Bernanke (1984), argue that bank loans are a form of inside debt, becausebanks have information about the borrower that is not available to othersecurities holders. As inside debt, bank loans are a way of avoiding theunderinvestment problem associated with information asymmetries. Specifi-cally, in the context of the Myers and Majluf model (1984), loans by banks (asinside debt) are similar to financial slack (internally generated funds). Onetestable implication of the bank-debt-as-inside-debt hypothesis is that becausebank loans avoid the information asymmetries associated with public debtofferings, a non-negative stock price response will be associated with theirannouncement. For similar reasons, if private placements acquired by insurancecompanies are also inside debt, a non-negative stock price response is expect-

ed.I examine the stock price response to publicly announced bank credit

agreements, private placements, and publicly placed straight debt issues.


3/19

C. James, The umqueness of hank loans 219

Abnormal performance is positive and statistically significant for bank loanannouncements and nonpositive for publicly placed straight debt issues. Theseresults are similar to those reported by Mikkelson and Partch (1986). Sur-

prisingly, a negative and statistically significant stock price response isobserved for debt placed privately with insurance companies. Most notably.I find a negative stock price response for private placements and straightdebt issues used to repay bank loans. These results suggest that bank loansare unique, but they are not fully consistent with the inside-debt argument,as I discuss below.

The remainder of the paper is organized into five sections. In section 2. Ianalyze the incidence of the reserve requirement tax. In section 3, I describemy sample of bank credit agreements and debt offerings. The stock price

effects associated with these borrowing arrangements are examined in section4. In section 5, several explanations for the observed stock price behavior areexplored. A brief conclusion is provided in section 6.

2. Incidence of the reserve requirement

The current reserve requirement on short-term CDs with original maturityof less than 180 days is 3%. In the absence of any special service provided tobank borrowers or any special service to CD holders. a reserve requirementtax would result in the elimination of CD financing. In a competitive depositmarket other depositors (non-CD holders) will not bear the tax, because if abank attempted to shift the tax to them, other banks not issuing CDs wouldbid them away. Bank stockholders cannot be expected to pay the tax becausenon-bank lenders (who are not subject to the reserve tax) would have a higherrisk-adjusted return.

Fama (1985) argues that CD holders do not bear the reserve requirementtax and that therefore bank loans are special. This conclusion is based on hisfinding no significant difference between the yield on CDs and the yields oncommercial paper and bankers acceptances. Famas evidence is not fullyconvincing, for two reasons. First, the reserve tax could be borne not byborrowers but by the FDIC through the provision of deposit insurance at lessthan actuarially fair prices. Second, CDs are insured only to $100,000, whereasthe typical denomination is $1 million, so their rates may contain a defaultpremium. If the default risk on CDs is greater than for commercial paper,observed yields on the two securities may be identical even though CD ownerspay the reserve tax.

An alternative method of examining the incidence of the reserve tax is to

examine the behavior of CD yields in relation to the yields on other moneymarket instruments around changes in reserve requirements. Without anycontemporaneous change in insurance costs, an increase in reserve require-


4/19

22 0 C . Jumer. The uniqueness of hank loans

Table 1

Average annual yields to maturity on high-grade certificates of deposit, commercial paper. andTreasury btlls and yield spreads (January 1977 to December 1984. sample size = 471).=

Instrument

Panel A: Aceruge utwud~ields ( rceek& dutu)

3% reserverequirement period

Jan. 1977-Nov. 1978July 1980-Dec. 1984

5% reserverequirement period

Nov. 1978-July 1980

90-day CDs 10.67 11.8930-day CDs 10.52 11.5990-day commercial paper 10.59 11.8330-day commercial paper 10.40 11.5190-day Treasury bills 9.74 10.6630-day Treasury bills 9.11 10.01

Purei B: A veruge umtuul~reld spreud (rn percent ) between CDs und orher monqr marker insrrumenrs(stun&xi errors in purenrheses )

3% reserverequirement period 5% reserve

Jan. 1977-Nov. 1978Spreadh

requirement period Entire sampleJuly 1980-Dec. 1984 Nov. 1978-July 1980 period

SPCDTB,, 0.931 1.211 0.992(0.04) (0.07) (0.03)

SPCDTB,,, 1.412 1.581 1.455

(0.05) (0.09) (0.04)SPCDCP,,,, 0.073 0.038 0.065

(0.01) (0.02) (0.01)

SPCDCP,,, 0.117 0.078 0.110(0.01) (0.02) (0.01)

Yields are based on weekly (Friday close) price quotes from the traders at Bank of America forhigh-grade CDs and for dealer-placed commercial paper rated Al-Pl. All data were obtained fromData Resources Inc. DRI-FACS tile.

hSPCDTB,, = average annual yield spread between 90-day CDs and 90-day Treasury bills.SPCDTB?,, = average annual yield spread between 30-day CDs and 30-day Treasury bills,SPCDCP,, = average annual yield spread between 90-day CDs and 90-day commercial paper.SPCDCPJ,, = average annual yield spread between 30-day CDs and 30-day commercial paper.

ments should reduce the yield on CDs in relation to other yields if depositorspay the reserve tax.

Changes in the reserve requirements applied to CDs during the 1978-1980period provide an opportunity to examine the incidence of the reserve tax.Effective November 2, 1978, and continuing through July 24, 1980, the FederalReserve imposed a supplemental reserve requirement of 2% on all CDs inexcess of $100,000. In addition, during this period, marginal reserve require-

ments of from 5% to 10% were imposed on CDs in excess of a base amount.

See table A7 of the Federal Reserue Bullerin for a list of reserve requirements.


5/19

C . Jumes. The u n r q u e n e ss ojhunk louns 221

The effect of these changes was to raise the reserve requirement on all largeCDs from 3% to 5% and as high as 15% for CDs issued in excess of the baseamount.

To examine the effect of these changes and the incidence of the reserve tax. Iobtained weekly secondary market price quotes and computed yields forthirty- and ninety-day high-grade CDs, commercial paper, and Treasury bills.These data were obtained from Data Resources, Inc. (DRI), for the periodJanuary 1. 1977 to January 1, 1985.

Table 1 presents the average annual yield on CDs, commercial paper, andTreasury bills during the period in which a 3% reserve requirement waseffective and the period in which a minimum 5% reserve requirement wasimposed. Table 1 also contains the average spreads between CDs and other

money market instruments for the 1977-1984 period. I find no statisticallysignificant difference in the average spread between CDs and commercialpaper or Treasury bills during the two periods. Assuming a competitivebanking industry, the evidence presented in table 1 supports Famas conclusionthat bank borrowers and not CD holders bear the reserve tax.

3. Description of the sample and methodology

3.1. Random sample of firms

I am not aware of any source that provides information by company on newbank loan agreements. To obtain a sample of these financing events. I selected300 companies at random from the population of firms contained in the 1983Center for Research on Security Prices (CRSP) daily return file that werelisted on the first trading day in 1974. I included companies in the sample ifthey were listed in the Moodys Industrial, Transportation, or Utilities manuals.(Excluded from the sample were financial companies.) I then searched theWall Street Journal Index for information on each firm over the ten-yearperiod 1974-1983 to identify all public straight debt offerings for cash, privateplacements of debt, and bank borrowing agreements that did not coincide withother financing, dividend, or earnings announcements.

The bank loan agreements in the sample consist of new credit agreementsand the expansion of existin g agreements. They include both extensions oflines of credit (commitments to lend) and term loans. The typical agreement.however, involves a line of credit where, at the firms option, borrowing can beconverted into a term loan.

Privately placed debt agreements consist of debt sold for cash to a restrictednumber of institutional investors. Most (approximately 70%) of the agreements

involve an insurance company as the lender.The total sample consists of 207 financing announcements. There are eighty

announcements of bank loan agreements, thirty-seven announcements of


6/19

222 C. James. The unqueness of bank louns

Table 2

Distribution by year of announcements of bank credit agreements. privately placed debt, andpublicly placed straight debt for a random sample of 300 NYSE- and AMEX-traded non-financial

firms (1974-1983).

Year ofannouncement

Bank loanagreements

Privatelyplaced debt

Publicstraight debt

1974 91975 111976 71977 81978 11979 81980 111981 9

1982 101983 6

Total 80

4 57 137 87 48 61 91 101 9

1 160 10

37 90

private placements, and ninety announcements of public straight debt offerings.Table 2 presents the distribution of announcements by type of event by yearfor the period 1974-1983. Although there is no discernible time pattern for thenumber of bank loan or straight debt announcements, the number of privateplacement announcements decreases substantially after 1978.

3.2. Descriptive statistics

Table 3 contains summary statistics for the debt offerings in my sample.Row 1 contains the amount of each type of offering. As table 3 indicates.public debt offerings are larger on average than private offerings. For bankloan agreements and private placements, the loan amounts reported mayoverstate the amount actually borrowed by the firm. In many cases these arecommitments to lend, and the entries in row 1 of table 3 are based on the

amount of the commitment.3

The number and dollar volume of privately placed debt is reported in the fnc,estnrenf DedersDtgesr. The dollar value (in millions) and number of privately placed bond issues during theperiod are:

1974 1975 1976 1977 1978

Dollar value $8.214 11.856 17.811 21,797 18.511Number of issues 696 685 717 1.017 900

1979 1980 1981 1982 1983

Dollar value 15,270 10,75010,860 10.397 10.360

Number of issues 786 640 556 531 525

Private placements have many of the same features as bank loan agreements. The borrower istypically given an option to borrow up to some prespecified amount over a period of one to fiveyears. See Zinbarg (1975).


7/19

C. James, The uniqueness of bank loans 223

Table 3

Descriptive statistics for commercial baak loans. privately placed debt, aad publicly placedstraight debt for a random sample of 300 NYSE- aad AMEX-traded non-fiaaaaal firms

(1974-1983).=

Descriptive measure

Commercialbank loans

(samplesize = 80)

Mean Median(Range)

T y p e of borrowing

Privately Publicplaced debt straight debt

(sample (samplesize = 37) size = 90)

Mean Median Mean Median(Range) (Range)

Debt amount (millions ofdollars)

Firm size (millions ofdollars)b

Debt amount/market valueof common stock

Maturity of debt(years)

Number of firms

Number of firms withpublicly traded debtoutstaadingd

72.0 35.0(4-800)

675 212(28.6-10,311)

0.72 0.46(0.04-2.6)

5.6 6.0(0.6-12)

52

25

32.3 25.0(5-120)

630 147(20.2-6.365)

0.52 0.25(0.04-2.6)

15.34 15.0(3-25)

34

16

106.2 75.0(lo-1,cOO)

2,506 1,310(47-59,540)

0.26 0.15(0.02-1.5)

17.96 20.0(l-40)

43

30

Statistics given in the first row are the mean followed by the median. The range is provided inthe second row.

Firm size is for December 31 of the year immediately preceding the security offering orborrowing. Firm size equals the book value of all liabilities aad preferred stock plus the marketvalue of common stock outstanding. The market value of common stock is the product of thenumber of shares outstanding aad the closing price per share at year-end preceding the aaaouace-meat. Closing prices are from the Security Owners Stock Guide. The book value of liabilities andthe number of shares outstanding are from Moodys manuals.

Maturity of the loan or debt offering is from the Wall Street Journal article. No information onmaturity was provided for twenty-four baak loans, two private placements, aad nine straight debtofferings. For bank loans that are convertible to terar loans, the maturity of the term loan is used.

d Firms are classified as having publicly traded debt if the Moodys manual report the firra hadrated debt outstanding at year-end preceding the financing announcement.

Firms using private placements and bank loans are on average smaller thanfirms using public offerings of straight debt. The average firm size in both thebank loan sample and the private placement sample is about 25% of theaverage firm size in the straight debt sample. This finding is consistent withBrealey and Myerss (1985) view that private placements and bank loanstypically involve small and medium-sized companies.

Row 4 of table 3 presents the average maturity of each type of borrowing.

Bank loans are of considerably shorter maturity than either privately placeddebt or straight debt. Indeed, the longest-term bank loan is twelve years, lessthan the median maturity of either privately placed or publicly placed debt.


8/19

224 C. Jumes. The unrqurrms o/ hudi loans

3.3. Methodology

The market model is used to obtain estimates of abnormal stock returnsaround the announcement of the financing events. The announcement isdefined as the date of the first report of the borrowing agreement or debtoffering in the Wall Street Journal. The market model was estimated on dailyreturns for the period that begins 120 trading days before and ends 120trading days following the announcement (event) date, excluding 41 tradingdays centered around the event date. The abnormal stock return or predictionerror for firm j over day t is defined as

PE,,=R,,- (~,+&L)rwhere R,, is the rate of return of security j over period t, R,, is the rate ofreturn on the CRSP equal-weighted market index over period t, and G, and fi,,are ordinary least squares estimates of firm js market model parameters.

The daily prediction errors are averaged over all firms within a particulargroup to produce a daily portfolio average prediction error:

APE, = I/N ; PE,!./=I

where N is the number of firms in the sample. I calculate a two-dayannouncement period abnormal return by summing the prediction errors forday - 1 and day 0. This procedure incorporates the possibility that theannouncement may have been made during trading hours the previous dayand reported with a one-day lag.

Tests of statistical significance of the average prediction errors are based onstandardized prediction errors. The two-day standardized prediction error forfirm j is defined as

SPE, = i PE,,/S,,r= -1

where

and V, is the residual variance of the market model regression for firm j, Mis the number of days in the estimation period (199), and R, is the meanmarket return over the estimation period.


9/19

C. James. The unrqueness of hunk loans 223

Table 4

Average two-day percentage prediction errors (APE) on the announcement of commercial bankloans. privately placed debt. and publicly placed straight debt offerings for a random sample of

300 NYSE- and AMEX-traded non-financial firms (1974-1983).

Type of event APE Z-value

Proportionnegativeb

(sample size)

Bank loan agreement

Private placement

Public straight debt

Bank loan agreementborrowing indicated

Bank loan agreementno borrowing indicated

1.93% 3.96 0.346(80)

-0.91% - 1.87 0.56(37)

-0.11% - 0.40 0.56(90)

1.71% 3.20 0.35d(71)

3.68% 1.71 0.23r(9)

The null hypothesis is that the average standardized prediction error equals zero. Z =fi( ASPE,). where ASPE, is the average standardized prediction error and N is the number oftirms in the sample.

The null hvpothesis is that the proportion of negative prediction errors equals 0.5. The teststatistic is a Wilcoxon signed ranks statistic.

Loan agreements in which the WaN Srreer Journal article describing the agreement indicatesborrowing has occurred or is expected to occur under the loan agreement.

Sign test statistic is significant at 0.05 level.Sign test statistic is significant at 0.01 level.

The average standardized prediction error is

ASPE,= ; i SPE,,./I

Assuming the individual prediction errors are cross-sectionallythe following Z-statistic can be computed:

Z = m( ASP&),

independent,

which is asymptotically distributed unit normal under the hypothesis that theaverage standardized prediction error equals zero.

4. Stock price response to borrowing arrangements

Table 4 reports the average stock price response to the announcement ofbank loan agreements, private placements, and public straight debt offerings.The average prediction error for bank loan agreements is positive and statisti-


10/19

226 C. James. The uniqueness fh a n k loans

tally significant at the 0.01 level. In addition, 66% of the prediction errors arepositive.4 There is no statistically significant difference between announce-ments of bank loan agreements in which immediate borrowing is indicatedand announcements in which no immediate borrowing is indicated.

The positive stock price response to bank loan agreements contrasts withthe non-positive response to public offerings of securities reported by otherresearchers.5 As table 4 indicates, I also find a non-positive stock priceresponse associated with the announcement of a public offering of straightdebt. The average two-day prediction error associated with straight debtofferings is - 0.11 percent, not statistically different from zero at the 0.10 level.

If the positive response to bank loan agreements is the result of some benefitfrom the intermediation process, but a benefit not unique to commercial

banks, one would expect to observe a similar response to debt placed privatelywith insurance companies. As table 4 indicates, however, the average two-dayprediction error associated with the announcement of privately placed debt is-0.91 percent, which is significantly different from zero at the 10% level(p-value of 0.063). Moreover, the difference between the average predictionerror of bank loan agreements and of privately placed debt agreements isstatistically significant at the 0.01 level.

5. Interpretation of the average stock price response

The difference in abnormal performance among announcements of bankloans, private placements, and straight debt offerings may arise because thesedebt offerings (or the borrowers using them) differ systematically in someimportant feature, such as the maturity of the issue or the purpose of theborrowing, that is unrelated to the identity of the lender. Alternatively, bankloans may differ from other types of borrowing because banks provide somespecial service with their lending activity. A testable implication of the secondexplanation is that the share price response to the announcement of bankloans will differ from the share price response to announcements of privateplacements or public debt offerings with characteristics similar to commercialbank loans.

In this section I examine the share price response associated with an-nouncements of bank loans, private placements, and straight debt offeringsgrouped by stated purpose of the borrowing, the maturity of the offering, thedefault risk of the borrower, and the size of the borrower.

4Mikkelson and Partch (1986) also report a positive and statistically significant response to theannouncement of bank credit agreements. They, however, focus on public securities offerings anddo not explore differences in the stock price response associated with bank loan agreements andprivate placements.

See Dann and .Mikkelson (1984). Mikkelson and Partch (1986). Eckbo (1986). Asquith andMullins (1986). and ~Masulis and Korwar (1986).


11/19

C. James. The uniqueness of bank loans 227

-!/P

I-2 f9_ _


12/19

28 C. Junw. The umqurnm o)huttX lotim

3.1. Analvsis of borrowing agreements b.v stated purpose

One explanation for the positive abnormal performance

bank loan agreements is based on the asymmetric information

associated with

model of ,Myersand Majluf (1984). Bank loans may serve as a form of inside debt if bankshave inside information about the value of the firms growth prospects andbank loan rates reflect this information. Myerss and Majlufs model pertainsto new financing, however, and offers no prediction about borrowing for otherpurposes. Examining the stock price response to bank loans grouped by statedpurpose provides one test of the inside-debt hypothesis.

All borrowing announcements are placed into one of five purpose categories:(1) refinance debt, (2) capital expenditures, (3) general corporate purposes. (4)repayment of bank loans, and (5) no purpose given. The classification bypurpose is based on information contained in the Wall Street Journal articledescribing the announcement. Where several purposes are stated, borrowing isclassified by the first purpose listed or, where indicated, the primary purposeof the borrowing.

The average two-day prediction errors for bank loans, private placements.and straight debt offerings grouped by purpose are presented in table 5. Table5 also includes the average maturity of each type of borrowing. The averageprediction errors for bank loans are positive for all stated purposes. althoughgeneral corporate purposes and the repayment or refinancing of bank debt are

the only two categories in which the average prediction errors are statisticallydifferent from zero at the 0.01 level. There are, however. no significantdifferences (at the 0.10 level) between the mean returns for bank loansclassified by purpose.

In only one category, the repayment of bank loans, is the average predictionerror for private placements significantly different from zero. The averageprediction error for this category is negative and appears to be the majorcomponent of the negative average prediction error associated with privateplacements reported in table 4. Moreover, the average prediction error for

private placements used to repay bank loans is statistically different (at the0.01 significance level) from that of private placements used for other pur-poses.

In the sample of straight debt offerings, only the repayment of bank loanscategory has an average prediction error significantly different from zero at the0.10 level. The average two-day prediction error is - 1.63 percent (p-value =0.08).

Two findings in this section are of particular interest. First. there is nosignificant difference between the share price response to bank loans used torefinance debt (either existing bank loans or other debt offerings) and bankloans used for capital expenditures. The same conclusion is reached if thecapital expenditures and general corporate purpose categories are combined.Therefore, the positive average abnormal returns associated with the an-


13/19

nouncement of new bank loans cannot be attributed solely to avoidance ofinformation asymmetries associated with new investments. The second findingis the statistically significant decrease in share price for privately placed debt

and straight debt used to refinance bank loans. This result is curious: why domanagers use private placements to refinance bank loans, given the adverseshare price reaction? One possible explanation is the difference in maturitybetween bank loans, private placements. and public debt offerings. This issueis explored in the next section.

5.2. Analysis of borrowing arrangements by maturity of the offers

The difference in average abnormal performance among borrowing ar-

rangements may be attributed to differences in the average maturity of theissue. As table 3 indicates, bank loans have a shorter average maturity than doprivate placements and straight debt offerings in the sample.

Maturity of the debt issue may be important in explaining the differences inabnormal performance for several reasons. First, as suggested by Merton(1974) and Ho and Singer (1982). short-term debt may be less risky thanlong-term debt. In particular, Ho and Singer demonstrate that holding themarket value of debt constant, an increase in the time to maturity of the debtwill increase the elasticity of the value of the bond with respect to the value of

the firm.6 Myers and Majluf (1984) predict that the stock price response to theannouncement of a new security issue depends on the sensitivity of the valueof new securities to changes in firm value. This implies that the absolute valueof the stock price response to the announcement of a debt offering shouldincrease with the time to maturity of the offering.

Flannery (1986) provides a second reason for the importance of maturity.He argues that a firms choice of maturity can provide a signal aboutmanagements assessment of earnings prospects. Flannery shows that, withtransactions costs associated with new debt issues, managers who believe theirfirm is undervalued by outsiders can signal the true value of the firm by issuingshort-term debt (i.e., debt repayable before cash flows are realized). When theundervalued firms true prospects are revealed, refunding occurs at a lowerdefault risk premium. Overvalued firms, on the other hand, find a short-termdebt strategy more expensive because any initial cost savings from issuingshort-term debt are more than offset by higher transaction costs of refinancingand higher subsequent refinancin g costs (in terms of a higher default riskpremium).

With a discount bond. to maintain a constant market value of debt as its maturity increasesthe promised terminal payment to debt holders must also increase. In addition, note that theelasticity of risky debt equals the weighted average of the elasticity of equity for the unlevered tirmand the elasticity of riskless debt (which is zero). An increase in the maturity of debt makes theexpected payoff characteristics of debt more similar to those of equity (by raising the terminalpayment) and therefore increases the elasticity of debt.


14/19

2 30 C . Jones. T h e un,qurnes~ ofban!, lounr

Easterbrook (1984) and Fama (1985) provide a third reason why maturitymight matter. Both authors focus on the agency costs of monitoring managers.Easterbrook argues that the costs of monitoring are lower if the firm is

frequently in the market for new capital. The issuance of new securitiestriggers a review of the firms earnings prospects by intermediaries (investmentbankers and commercial banks). These intermediaries send reliable signals toexisting as well as new claimants on the firm about the firms ability to meetfixed-pay-off contracts. The intermediaries send reliable signals by bondingperformance. directly through their own investment or indirectly through thevalue of their reputation. Fama (1985, p. 36) argues that bank loans avoidduplication of information costs:

Bank loans usually stand last or close to last in the line of priority amongcontracts that promise fixed pay-offs. Bank loans are short-term and therenewal process triggers periodic evaluation of the organizations abilityto meet low-priority fixed pay-off contracts. Positive renewal signals frombank loans mean that other agents with higher fixed pay-off claims neednot undertake similar costly evaluations of their claims.

A firms decision to commit to periodic evaluations can therefore provide apositive signal of managements assessment of the firms earnings prospects.

The hypothesis that the positive share price response associated with bankloan announcements is due safely to the shorter maturity of bank loans I callthe maturity hypothesis. If the difference in abnormal performance is due solelyto the shorter maturity of bank loans, one would expect to observe a positiveshare price response for public straight debt offerings and private placementswith maturities similar to those of bank loans.

The maturity hypothesis is not necessarily inconsistent with the hypothesisthat banks provide some special service to borrowers. For example, Black(1975). Fama (1985), and Kane and Malkiel (1965) argue that banks have acost advantage in making loans to depositors. The inside information providedby a continuing deposit history is particularly valuable, they argue, in makingand monitoring repeating short-term loans. This argument explains why banksmay have lower costs to originate short-term repeating loans but does notexplain why firms use private placements or publicly placed long-term debt torefinance bank loans. If a continuing relationship between the bank and itsloan customers results in lower costs of refinancing, banks also should have acomparative advantage in making long-term loans to these customers. There-

Rozeff (1982) presents a similar argument in his analysis of the determinants of dividendpayout ratios. He argues that dividend payments are a device that reduces the agency cost of

equity by requiring the firm to acquire external funds more frequently. The suppliers of new fundsrequire the firm to supply new information about the firms earnings prospects. The agency costsavings from higher dividend payments are offset by higher transactions costs associated with newfinancing. These two opposing influences produce an optimum dividend payout ratio.


15/19

C. James. The umqueness of bank loans 31

fore, although a change in a firms earnings prospects may result in a shift inits maturity preference, it is not clear why this action also results in a changein the intermediary used (e.g., from banks to insurance companies). One

explanation is that banks are constrained from making long-term loans.* Thisconstraint could arise from regulatory pressure or a preference by banks formatching the maturity of their assets with the maturity of their liabilities.

To test the maturity hypothesis, I divided the straight debt and privateplacement announcements into two groups, one consisting of offerings with amaturity of less than ten years and a second consisting of offerings with amaturity of ten years or more. I then analyzed the stock price response toborrowing announcements in the two groups.

My results are reported in table 6. Although the average prediction errors

are larger for short-term offerings than for longer-term borrowing, thedifference in average returns is not statistically significant. As an additionaltest of the maturity hypothesis I estimated the relation between the two-dayprediction error and the maturity of the offering for each type of borrowingarrangement, using weighted least squares. The weights used in the regressionanalysis are the reciprocal of the standard error of each firms abnormalreturns. My results reveal no statistically significant relation between the shareprice response to the announcement of the offering and the maturity of theoffering. These results, together with those reported in table 6, are inconsistentwith the maturity hypothesis.

5.3. Other explanations

The other potential explanations for the differences in abnormal perfor-mance among borrowing arrangements are: differences in the risk of the debtissued, differences in the size of borrowing firms, and differences in the size ofdebt offering in relation to the size of borrowing firm. Smith and Warner(1977) argue that private placements contain more detailed restrictive covenantsand are more likely to be used by riskier firms than is publicly placed debt.Differences in default risk may explain the differences in abnormal returnsthat I find. Alternatively, abnormal performance may be related to firm size.The announced ability to borrow may be good news for small firms (whichborrow primarily from banks), but not much news at all for large firms (which

The presence of a supply constraint is suggested by the lack of activity in the long-termcommercial loan market. In my sample, only one bank loan has a maturity of more than ten years.The Federal Reserve Boards Sumq of the Terms of Bunk Lending indicates banks specialize inshort-term loans. The survey for August 1985 indicates only 12% of commercial loans made havematurities of more than one year. These loans have an average maturity of four years.

Although no federal regulations limit the maturity of commercial loans. a factor used in bankexaminations to determine asset quality and capital requirements is the maturity mismatch of abanks assets and liabilities. See Spong (1985).


16/19

Table 6

Average two-day percentage prediction errors ( APE) on the announcement of private placementsand straight debt offerings classified by maturity for a random sample of 300 NYSE- and

AlVEX-traded non-finanhal firms (1974-1983).

Type of event APE Z-valuesSample

sizeAverage

maturityb

Straight debt. maturityless than 10 years

Straight debt. maturitygreater than 10 years

Private placements. maturityless than 10 years

Private placements. maturitygreater than 10 years

0.766%

-O.LIlQc

- 0.3%_

- 1.011oc

1.625

- 0.537

-0.193

-2.002

25

51

5

32

5.4 years

21 years

5.6 years

17 years

The null hypothesis is that the average standardized prediction error equals zero. Z =fi( ASPE, ). where ,I.SPE, is the average standardized prediction error and $ is the number offirms i n the sample.

Maturity of the loan or debt offering is from the IV&[ Srreer Journal article describing theoffering.

Significantly different (at the 0.10 level) from the APE for straight debt issues with maturitygreater than ten years at the 0.10 level.

use publicly placed debt) that have other ways of disseminating information.Bank borrowing may therefore be simply a proxy for firm size. Finally, therelative size of the offering may be an important determinant of the stock priceresponse if it serves as a proxy for changes in leverage.

As a proxy for the default risk of the borrower, I obtained for each firm therating of its most recently issued debt prior to each announcement in mysample. Debt ratings are from the ~Moous manual. Panel A in table 7provides the proportion of firms with debt outstanding in three ratingcategories: AA or better, A, and BA\ and below. The proportion of firms ineach rating category, as well as the proportion of firms with rated debt issimilar for the private placement and bank loan samples. A higher proportionof the straight debt offerings is in the AA or better and A rated categories. Ifthe rating of outstanding debt provides a proxy for default risk, firmsannouncing new bank loans and private placements have a higher default riskthan those announcing straight debt offerings.

Myers and Majluf (1984) predict that abnormal performance is related tothe sensitivity of the value of the securities issued to changes in the firm value.Default risk can affect the sensitivity. Panel B of table 7 provides two-dayaverage prediction errors for each type of borrowing grouped by rating. Foreach type of borrowing arrangement, the abnormal returns are larger the

higher the debt rating. This result is consistent with the prediction of Myerssand Majlufs model. The results in table 7 are nor consistent, however. with thehypothesis that differences in default risk explain the difference in stock price


17/19

Table 7

Debt rating for LI random sample of 300 NYSE- and AMLY-traded non-financial tirmsannouncing bank loan agreements, private placements. and straight debt offerings (1974-1983).and average two-da? percentage prediction errors for firms grouped by rating of outstanding debt.

Type of event

Bank loan agreements

Private placements

Public straight debtofferings

Punel .A Debt rurlngl

Proportionof firms Proportion

rated AA or of firmsbetter rated A

0.12 0.10(5) (4)

0.12 0.20(2) (3)

0.31 0.41(20) (27)

Proportion Proportionof firms of firms

rated BAA with ratedor below debt

0.78(25)

0.68(11)

0.28(18)

0.48(34)

0.47(16)

0.69(65)

Punel B: Average two-duj.predicrion erron bj, debt rut7t7gc

Rated A Rated BAAor better or below Not rated

Bank loan agreements

Private placements

Public straight debtofferings

-

3.89% 1.77% 1.76%(2X) (1.92) (2.184)

1.18% 0.30% - 2.037(1.68) (0.211) (- 2.90)

0.404 - 0.328 - 1.084(1.71) (- 1.42) (- 1.45)

Rating refers to the bond rating of the most recently issued debt prior to announcement.Ratings were obtruned from .Moo&s manuals.

hSample stze is in parentheses.' Z - va l u e in parentheses: the null hypothesis is that the average standardized prediction error

equals zero. 2 = v?( ASPE, ). where ASPE, is the average standardized prediction error and .V isthe number of firms in the sample.

response to different types of borrowing agreements. The proportion of firmsin each rating category is similar for bank loans and private placements, butthe abnormal return associated with bank loans is positive (on average and ineach rating category), whereas the abnormal return for private placements isnegative.

As table 3 indicates. firms in the bank loan sample are smaller than firms inthe straight debt sample. To determine whether differences in firm size canexplain differences in abnormal performance I estimate the following cross-sectional equation:

STRET, = al + a,STMVCS, + a3fssue I + a,Issue II + E,.

where STRET, is the two-day standardized prediction error for firm i;


18/19

SZ-MVCS, is the market value of common stock divided by the standard errorof the two-day prediction errors for firm i; Issue I equals 1 if issue is a privateplacement, zero otherwise; Issue II equals 1 if issue is a straight debt offering.

zero otherwise; and E, is the error term.The results are presented below (t-statistics in parentheses):

STRET, = 0.305 + 1.17E-9STMVCS, - 0.554 Issue I - 0.306 Issue II.

(1.75) (1.61) (-1.97) (-1.83)

R= 0.05.

The results indicate no statistically significant relation between the stockprice response to the borrowing announcement and the size of the firm aftercontrolling for issue type. These results indicate that differences in abnormalreturns among borrowing agreements are not the result of differences in firmsize.

I obtain similar results when firm size is measured as the sum of the marketvalue of common stock and the book value of all other liabilities. In addition.I find no statistically significant relation between abnormal returns and firmsize within each type of borrowing arrangement. Finally, I obtain similarresults when the relative size of the offer, defined as the ratio of the amount of

the offering to the market value of the firms outstanding common stock, issubstituted for the size variable in eq. (1).

6. Summary and conclusions

Significant positive abnormal returns accrue to stockholders of firms an-nouncing new bank loan agreements, whereas negative abnormal returnsaccrue to stockholders of firms announcing private placements. In addition.negative and statistically significant abnormal returns are associated with the

announcement, of private placements and straight debt issues used to retirebank debt.One possible explanation for the difference in abnormal performance is that

bank loans differ in some important feature such as maturity. Alternatively.bank loans may differ from other types of borrowing because of some specialservice provided by banks with their lending activity. An analysis of differ-ences in the maturity, borrower default risk, borrower size, and purpose of theborrowing indicates that differences in abnormal performance are not duesolely to differences in characteristics of the loan or characteristics of theborrowers. This result, together with the evidence concerning the incidence ofreserve requirements, suggests that banks provide some special service notavailable from other lenders. Further research is needed to identify thatunique service or unique attribute of bank loans, and to explain its relation tothe market value of the firm.


19/19

References

Asquith. Paul and David ~lulhns. 1986. Equity issues and offenng dilution. Journal of FinancialEconomics 15. 61-90.

Bemanke. Ben. 1983. Non-mowtar]; etfrcts of the tinancial crisis in the propagation of the greatdepression. American Economic Review 73. 257-276.

Black. Fisher. 1975. Bank fund management in an et?icient market. Journal of Financial Economics2. 323-33Y.

Brealey. Richard and Stewart Myers, 1984. Princtples of corporate tinance (McGraw Hill, NeuYork).

Campbell. Tim and Wilham Kracaw. 1980. Information productton. market stgnaling. and thetheoF of intermediation. Journal of Finance 35. 863-882.

Dann. Larc and Wayne bfikkelson. 1984, Convertible debt issuance. capital structure change andtinancing-related information: Some new evidence. Journal of Financial Economics 13. 157-186.

Diamond. Douglas. 1984. Financial intermediation and delegated monitoring. Review of EconomicStudies 51. 393-414.

Easterbrook. Frank. 1984. Two agency-cost explanations of dividends. American EconomicReview 74. 650-660.

Eckbo. Espen. lY86. Valuation effects of corporate debt offerings. Journal of Financial Economics15. 119-152.

Fama. Eugene. 1980. Banking and the theory of finance, Journal of .Monetary Economics 10.10-19.

Fama. Eugene. 1985. Whats ditTerent about banks?, Journal of Monetary Economics 15. 29-36.Flanne?, >fark J.. 1986. Asymmetric information and risky debt maturit?; choice. Journal of

Finance 41. 19-38.Ho. Thomas. S.Y. and Ronald Singer, 1982. Bond indenture provisions and the risk of corporate

debt. Journal of Financial Economics 10. 375-406.Kane. Edward and Burton Malkiel. 1965, Bank portfolio allocation, deposit variability and the

availabilit! doctrine. Quarterly Journal of Economics 79, 113-134.Leland. Ha\nes and DaLid Pyle, 1977, Information asymmetries. financial structure and financial

intermediariss. Journal of Finance 32, 371-387.kIasulis. Ronald and Ashok Korwar. 1986. Seasoned equity offerings: An empirical investigation.

Journal of Financial Economics 15, 91-118..Llerton. Robert. 1974. On the pricing of corporate debt: The risk structure of interest rates,

Journal of Finance 29. 449-465.Mikkelson. Xavne and Megan Partch. 1986. Valuation effects of securities offerings and the

issuance process, Journal of Financial Economics 15. 31-60.Myers. Stelvart and Nicholas Majluf, 1984. Corporate financing and investment decisions when

tirms hare information that investors do not have. Journal of Financial Economics 13.157-187.RozeH. &lichael S.. 1982. Growth, beta and agency costs as determinants of dividend payout

ratios. Journal of Financial Research 5, 249-259.

Smith. Clifford and Jerold Warner. 1979. On financial contracting: An analysis of bond covenants.Journal of Financial Economics 7. 111-161.

Spong. Kenneth. 1985. Banking regulation: Its purposes, implementation and effects (FederalReserve Bank of Kansas City. KS).

Zinbarg. Edivard. 1975. The private placement loan agreement. Financial Analyst Journal 31.33-52.

Chris James

Documents

Transcript of Chris James