I

R. A/ton Gilbert is an assistant vice president at the FederalReserve Bank of St. Louis. Kevin L. Kliesen provided researchassistance.

I Do Bank Holding Companies

II

Act as “Sources of Strength”for Their Bank Subsidiaries?

HOSE WHO WISH to determine why banksfail typically focus on the characteristics ofbanks and their local markets that make themvulnerable to losses.1 A key factor often over-looked, however, is capital injections by share-holders. A bank whose losses exceed its capitalneed not fail, if its shareholders (existing ornew) inject sufficient additional funds to restoreits capital ratio to a level acceptable to thebank’s supervisory agency. Likewise, theshareholders of a bank with a positive, butrelatively low, capital ratio can always removeit from its “problem bank” status by injectingsufficient capital.

The 1980s were marked by large numbers ofbank failures and large shares of commercialbank assets written off as losses compared withthe previous four decades. Taking advantage ofthis opportunity for comparison, this paperfocuses on the incentives of bank holding com-panies (BHC5) relative to those of other investorsto inject additional capital into troubled banks.

‘Bovenzi, Marino and McFadden (1983) and Demirguc-Kunt(1989).

2Other types of BHC actions to aid troubled bank sub-sidiaries, such as providing new management, are not

We test the hypothesis that capital injectionsinto troubled bank subsidiaries of BHCs arelarger than the capital injections into troubledindependents.

The hypothesis underlying this study is thatBHCs have strong incentives to maintain “favor-able” reputations in the financial markets andwith the Federal Reserve Board (Board). In par-ticular, BHCs want to maintain favorable reputa-tions with the Board because, under generalcriteria specified in the Bank Holding CompanyAct, the Board has the authority to approve ordeny applications by BHCs to acquire additionalsubsidiaries. If BHCs inject relatively more capitalinto their troubled banks, then it is argued that,at least compared to owners of independentbanks, BHCs are “sources of strength” for theirbank subsidiaries. If BHC affiliation has no in-fluence on the amount of capital injected intotroubled banks, then BHCs provide no moresupport in this form for their subsidiary banksthan other bank owners.2

considered in this paper. Thus, BHCs may be sources ofstrength in other ways, even if they do not inject morecapital into their troubled banks than other bank owners.

1 3

B. Alton Gilbert

II

IIIIIIIIIS .~MMtaRV/FFRRIjARV 1991

THE “SOURCE OF STRENGTH”PRINCIPLE~ A GENERALDISCUSSION

The Federal Reserve Board issued a statementon April 22, 1987, about the obligations of BHCstoward their troubled bank subsidiaries. In sum-mary, it said:

The statement reaffirmed a long-standing Boardpolicy that holding companies should use theiravailable resources to provide adequate capitalfunds to their subsidiary banks during periods offinancial stress or adversity. The Board issued thepolicy statement to remind holding companies ofits expectation that they provide financial andmanagerial strength to their subsidiary banks.’

BHCs may take various types of actions to aidtheir troubled bank subsidiaries besides capitalinjections. This statement, however, indicatesthat the Board expects BHCs to employ thismethod of assistance if losses reduce the banks’capital ratios below acceptable levels.

The Board also has cited the principle ofsource of strength in denying some applicationsfor the formation of BHCs. In these cases, theBoard projected that the dividends from banksubsidiaries necessary to sevice the debt of thenew BHCs would reduce the capital ratios ofthe acquired banks to unacceptably low levels.~

The Financial Institutions Reform, Recoveryand Enforcement Act of 1989 (FIRREA) makes aweaker version of the source-of-strength princi-ple a legal requirement for multi-bank holdingcompanies that own both solvent and insolventsubsidiary banks. FIRREA requires that multi-bank holding companies use the capital of theirsubsidiary banks to cover the losses of each in-dividual bank subsidiary. Thus, it has eliminatedthe option of a multi-bank holding company to

abandon an individual subsidiary bank, therebyleaving its negative net worth to be absorbed byuninsured depositors or by the Federal DepositInsurance Corporation while retaining its sol-vent bank subsidiaries. FIRREA does not requireBHCs to use the capital of the parent organiza-tion or nonbank subsidiaries to cover negativenet worth of their bank subsidiaries. The Fed-eral Reserve Board’s standard for “source ofstrength,” although not a legal requirement,covers a broader range of circumstances, in-cluding those in which the capital of a BHC’ssubsidiary bank, after absorbing losses, is lowbut still positive. Also, the Board expects BHCsto use all of their resources to aid their troubl-ed bank subsidiaries, not just the capital of theirbank subsidiaries.

THE GENERAL HYPOTHESIS ANDITS IMPLICATIONS

This paper investigates banks whose recentlosses are so large that they must either receivecapital injections or reduce their assets to meetthe minimum capital standards. During 1985-88,the years covered by this study, the minimumcapital requirements of commercial banks werespecified in terms of the ratio of their primarycapital to their total assets. The major com-ponents of primary capital are (1) equity capital(investment by shareholders plus retained earn-ings) and (2) loan loss reserve.

The minimum requirements set by the federalsupervisory agencies for all banks in 1985 calledfor primary capital to be greater than or equalto 5.5 percent of total assets. The supervisoryauthorities indicated that even higher capitalratios would be required for banks whose as-sets were of relatively poor quality.’ Since thebanks in this study incurred large losses relative

‘Board of Governors (1987), pp. 71-72. See Duncan (1987)and Bureau of National Affairs (1987) for legal interpreta-tions of the Board’s statement. The Board also attemptedto force MCorp to act as a source of strength for itsbankrupt subsidiaries. See Bureau of National Affairs(1989). In mid-May 1990, the Fifth Circuit U.S. Court ofAppeals ruled against the Federal Reserve in its suitagainst MCorp. See Quint (1990). Despite this ruling,BHCs still have incentives to maintain their reputations infinancial markets, and the Federal Reserve still hasauthority to approve or deny any subsequent BHCapplications.

4These denials are based on projections of the paymentsnecessary to service the acquisition debt and the earningsof the banks to be acquired, not on the records of theBHCs in acting as sources of strength in the past. For a

recent case in which the Board applied this principle in de-nying an application by a BHC, see the Federal ReserveBulletin (April 1990), pp. 257-58.

‘Gilbert, Stone and Trebing (1985). The bank supervisoryauthorities also specified a minimum ratio of total capitalto total assets of 6 percent. Total capital includes primarycapital plus long-term debt of the bank that is subor-dinated to its deposits. This paper does not consider theadequacy of the total capital ratio or the incentives forbanks to issue subordinated debt to meet the total capitalratio.

4 I

II1I

a

I 5

to their capital, their minimum primary capitalratio is assumed to be C percent.°

Capital injections into independent banksgenerally involve the sale of additional bankstock to existing or new shareholders, BHCsthat own their bank subsidiaries outright, how-ever, often inject capital into them directly,leaving the number of shares outstanding un-changed.’ When a BHC injects capital directly,the transaction has the following effects on thebalance sheet of the BHC: a reduction in “cash”and an increase in another category of assetscalled “investment in subsidiaries.” On the sub-sidiary’s balance sheet, the transaction involvesan increase in both its cash and its “equity capi-tal.” In this paper, capital injections are mea-sured as the sum of funds raised by issuing ad-ditional shares and injecting capital directly.

WIGs and the Source of StrengthHypothesis

BHCs have reasons to inject capital into theirtroubled subsidiaries over and above those thatapply to individuals who own shares in troubledindependent banks. These reasons differ some-what depending on whether the subsidiarywould actually fail or simply have low but posi-tive capital ratios ~vithout a capital injection.

First, consider the incentives of BHCs. BHCswish to convince financial market participantsthat they are strong, reliable organizations, inorder to reduce their borrowing costs andpossibly boost the price of their stocks. In sev-eral cases, BHCs have covered the losses oftheir subsidiaries’ creditors just to maintain

their reputations in the financial markets.’ Aspreviously discussed, BHCs also are concernedabout their reputations with the FederalReserve Board. Their ability to acquire sub-sidiaries could be jeopardized if the Board viewstheir failure to support subsidiaries as a lack ofwillingness to act as sources of strength.

Of course, individuals who own banks alsoare concerned about their reputations with in-vestors and with bank supervisors. The failureof their banks will make it more costly for themto raise funds in the future to buy other banks.In fact, individuals with poor reputations withbank supervisors may be barred from buyingbanks,’ Thus, whether BHCs have greater incen-tives to recapitalize their troubled banks thanindividuals who own troubled independents isan empirical issue.

Up to now, the discussion has focused onbanks that would fail without capital injections.Most banks in this study, however, had positiveprimary capital after absorbing their losses,even without capital injections. These banks canraise their capital ratios by reducing their assetsor receiving capital injections. BHCs also havereasons for injecting capital into these banksthat may not apply to individuals who own stockin troubled independents. Increases in the capitalratios of their subsidiary banks achieved throughreductions in assets, rather than by injections ofcapital, may not demonstrate the BHCs’ abilityand willingness to act as sources of strength.’°Thus, if BHCs do not inject additional capital,they may be denied permission to acquire addi-tional subsidiaries in the future.”

IIIIIIIIIIIIII

°Lossesrecognized in the current year reduce primarycapital, since they reduce either the loan loss reserve orequity capital. Loans that are not collectible are declaredloan losses by bank management and charged against theloss reserve, thereby reducing that component of primarycapital. A bank increases its loan loss reserve in the cur-rent year though an expense item that is charged againstcurrent income. Negative earnings in the current year,perhaps because of a relatively large provision for loanlosses, reduce equity capital.

‘A BHC would inject capital into a bank subsidiary ratherthan buy additional shares only if it owned all of thebank’s stock. Qtherwise, a direct injection of capital is agift to other shareholders that increases the value of theirshares even though they have not increased their invest-ment in the bank.

‘For a description of cases in which BHCs covered thelosses of their subsidiaries, rather than forcing thecreditors of their subsidiaries to absorb them, see Cornyn,et al (1986), pp. 187-91. The authors interpret these ac-tions as attempts by BHCs to portray themselves asstrong, reliable organizations that honor their obligations.

‘Individuals must apply to the supervisory authorities forpermission to buy controlling interest in banks. If in-dividuals refuse to inject capital into their failing banks, thesupervisory authorities may block their acquisitions ofbanks in the future. See Spong (1985), pp. 94-95.

10The Board’s record in acting on 81-IC applications to ac-quire subsidiaries makes it difficult to provide a moredefinitive description of the Board’s standards for sourceof strength. In most cases in which the Board has statedthe principle of source of strength as the basis for denials,the basic issue is excessive debt in the formation of one-bank holding companies. Few denials have been based onthe failure of BHCs to act as sources of strength for theirbank subsidiaries.

“Note that the Board statement quoted above refers to theuse of “available resources to provide adequate capitalfunds to their subsidiary banks during periods of financialstress or adversity.” This statement could be interpretedas indicating that a BHC would not be considered a sourceof strength if it fails to inject capital into a subsidiary bankwith a low but positive capital ratio.

a JANUARY/FEBRUARY 1991

The Issue of SizeBHCs differ greatly in terms of the magnitude

of their assets and the number of their subsidi-aries. Many BHCs, for example, have only asingle subsidiary. The incentives for the ownersof such a one-bank holding company to injectcapital into their troubled subsidiary are likelyto be similar to those of individuals who own atroubled independent. The failure of a BHC’ssole subsidiary would have no adverse effectson the BHC’s cost of funds because it would beout of business as well. Moreover, a BHC withonly one subsidiary has certainly not shownthat it considers permission to acquire addi-tional subsidiaries to be a valuable privilege.

For multi-bank holding companies, the incen-tive to inject capital is presumably greater, thelarger are the total assets of all subsidiaries inthe BHC to the assets of its troubled subsidiary.The prospective penalties imposed on a BFIC byboth the financial markets and the Board for afailure to inject capital into a troubled subsidi-ary are likely to be positively related to the totalassets of the BHC.” Thus, in modeling the de-terminants of capital injections into troubledbanks, variables designed to reflect affiliationwith BHCs reflect the assets of BHCs relative tothe assets of their troubled bank subsidiaries.

Assessing the Financial Strength ofSHIlls

Capital injections by BHCs also are assumed tobe influenced by their own financial conditions.If primary capital ratios at the other banks in aBHC exceed the levels required by bank super-visors, the BHC can channel capital from them,via dividends, to the troubled subsidiary. Alter-natively, a BHC with most of its subsidiaries instrong financial condition could raise funds inthe financial markets at lower interest rates(reflecting lower risk premiums) than BHCs thathave larger numbers of troubled banks.

In contrast, if the primary capital ratios of thebanks in a BHC are generally below requiredlevels, banks have little or no excess capital topay out as dividends to the BHC to channel toits more troubled subsidiaries. Moreover, such aBHC would have less incentive to promote its

reputation with the Board as a source of strengthby attempting to come to the aid of one of itsbank subsidiaries; the relatively low capital ra-tios of its other subsidiaries clearly indicate thatthe BHC is not in a position to act as a sourceof financial strength.

A FORMAL SPECIFICATION OFTHE BANK CAPITAL INJECTIONMODEL

Equation 1 presents the model of the deter-minants of capital injections into troubled banksused in this study.

+ + +(1) INJ = frnHcslo, RFICSSO, RHCSIOO,

+ + + +RHCS100+, FSHC, NINJ, ROA)

The variables used are defined below: the signsabove the variables in equation 1 show the ex-pected signs of their estimated coefficients.

INJ the ratio of the capital injected into abank to the total assets of the bank at the endof the prior year.

RHCS1O = dummy variable with a value of uni-ty if the ratio of total banking assets of a BHCto the assets of its troubled bank subsidiary(RHCS) is greater than unity but less than 10,zero otherwise.

RHCSSO = dummy variable with a value of uni-ty if RHCS is greater than or equal to 10 butless than 50, zero otherwise.

RHCS100 = dummy variable with a value ofunity if RHCS is greater than or equal to 50 butless than 100, zero otherwise.

RHCS100+ = dummy variable with a value ofunity if RHCS is greater than or equal to 100,zero otherwise.

FSHC = the ratio of the sum of primary capitalof banks in the BHC, other than the troubledbanks included in this study, to the sum oftheir total assets.

NINJ = the ratio of the capital injection neces-sary to make primary capital equal to C percent

“While it is impossible to determine the value of future ac-quisitions to a 81-IC without knowing its plans, areasonable guess might be that BHCs that have grownlarge through acquisitions would place higher value on theprivilege to make additional acquisitions than BHCs that

have made fewer acquisitions. Under this assumption, thepenalty for not investing in a troubled subsidiary in theform of foregone opportunities for future acquisitions isproportional to the total assets of the BHC.

6 I

aFEDERAL RESERVE BANK OF ST. LOUIS

I 7

of total assets as of the end of the prior year, totheir total assets at the end of the prior year.

ROA = the ratio of net income of the banks inthe county of a troubled bank, other than thetroubled bank itself, to their total assets. Incomeis measured over the calendar year prior to theyear in which the bank becomes a troubledbank, and total assets are measured at the endof that year.

Identjfication of Troubled Banksand the Measure of capitalInjections

Equation 1 is designed to explain capital injec-tions into a specific group of banks — thosewhose primary capital ratios initially exceeded 6percent but who had losses in the current yearthat drove their primary capital below 6 per-cent. To raise their capital ratios up to 6 per-cent, these banks must receive capital injectionsor reduce their assets. The capital injected intoeach bank is measured as the sum of capital in-jections over four quarters, beginning in thequarter in which the losses reduced their pri-mary capital below 6 percent of their total as-sets in the initial period.

Each bank included in the study had, at theend of the prior year, primary capital that ex-ceeded C percent of its total assets, as illustratedin inequality 2.

(2) 0.06A0 < C,,

where

A, = total assets in the last quarter of theprior year,

C, = primary capital in the last quarter of theprior year.

Then, during some quarter of the currentyear, each bank in the study had losses suffi-cient to reduce its primary capital (net of anycapital injection in that quarter) below C per-cent of its total assets at the end of the prioryear (A0), as illustrated in inequality 3.

(3) 0.06A, > [C, — 1,1,

where

C, = primary capital in the first quarter ofthe current year (quarter t) in whichthis inequality holds,

I, = capital injections in the first quarter of

the current year (quarter t) in which thisinequality holds.

Capital injections in quarter t are subtractedfrom the right side of inequality 3 because somebanks inject capital immediately to cover at leastpart of their losses; in fact, if they inject enoughcapital, their primary capital would not actuallyfall below 6 percent of A,. Yet these banksshould be included in the analysis because theyreceived large capital injections to offset largelosses.

In deriving values of the dependent variable,capital injections are summed over four quarters,primarily because of the typical timing of thecapital injections. Most capital injections occurredin the fourth quarter of the year in which lossesreduced the primary capital of a bank enoughto satisfy inequality 3. Of the 256 banks in thisstudy that received capital injections, 238 re-ceived capital injections during this quarter,regardless of the actual quarter in which the in-equality first held. Moreover, 221 of these re-ceived their only capital injection in that fourthquarter. Given this timing, a four-quarter periodstarting with the quarter in which each banksatisfies inequality 3 is adequate for examiningcapital injections into troubled banks. Thus,capital injections are measured as the sum ofcapital injections in quarter t through n-3. Theequation for calculating the dependent variable,INJ, is specified in equation 4:

3(4) INJ = 1 I,÷~/A,

j=0

Because the capital injections are summedover four quarters, an additional constraint isimposed on the balance sheets of the banks in-cluded in this study: their primary capital inquarter t+ 3 (net of capital injections in quarterst through t+3) must be less than 6 percent ofA0, their total assets at the end of the year be-fore their large losses. This is because banksmay have increased their primary capital inquarters t+ 1 through H- 3 in ways that do notinvolve explicit capital injections; among theseare positive earnings that are retained, recover-ies on loans previously charged off as losses andchanges in accounting practices. Such changesin primary capital after quarter t will affect theamount of capital injections needed to meettheir capital requirements. This condition isstated in inequality 5:

(5) 0.OCA0 > [C,~,— I I,÷~l.j=0

IIIIIIIIIIIIIa JANUARY/FEBRUARY 1991

Finally, some banks that satisfy inequalities 3and 5 are excluded from this study becausethey were involved in mergers or were pur-chased by new owners around the time of theirlosses.” including these banks would have po-tentially biased the results of this study- Sup-pose, for instance, that troubled banks sold tonew owners are acquired by BHCs that injectedadditional capital into their new bank subsidi-aries. Including these banks in the study wouldexaggerate the significance of BHCs as sourcesof strength for their existing bank subsidiaries.

In deriving the dependent variable, INJ, thedollar value of capital injections is divided by A,to create a measure of capital injections that isunaffected by contemporaneous changes in totalassets. Using the total assets existing when thecapital injections were made could bias theestimate of the effects of affiliation with BHCson the size of capital injections.

Suppose, for example, that BHCs inject enoughcapital into their troubled subsidiaries to meetcapital requirements without reducing theirassets, while shareholders of independent bankschoose combinations of capital injections andreductions in assets. Deflating capital injectionsby the total assets when such injections aremade would bias downward the estimate of theeffect of affiliation with BHCs on the size ofcapital injections. Deflating by A, avoids thispotential bias.”

Relative Size of BHC’s and TheirTroubled Bank Subsidiaries

Equation I includes dummy variables for dif-ferent levels of RHCS — the ratio of the totalbanking assets of the BHC to the assets of itstroubled subsidiary. These dummy variablesapply only to multi-bank holding companies. Nosubsidiaries of one-bank holding companies areincluded in the study because all that met theinequality conditions were sold around the timethey became troubled banks.

If BHCs as a group inject more capital intotheir troubled bank subsidiaries than otherbank owners, the coefficients on the dummyvariables for RHCS will be positive and statisti-cally significant. Moreover, as explained previ-ously, the coefficients are expected to be largerfor higher values of RHCS.

Financial Strength of the BIB?(FSHC)

FSHC reflects the financial strength of BFICs;it is measured as the ratio of the primary capi-tal of the banking subsidiaries of a BHC — otherthan that of their troubled bank subsidiaries inthis study — to their total assets. The primarycapital and total assets values used are those inthe fourth quarter of the year prior to the yearin which the subsidiary first satisfies inequality3. For troubled banks that are not subsidiariesof BHCs, FSHC has a value of zero.

“The following rules are designed to exclude banks thatwere sold to new owners around the time they becametroubled banks. Each bank that was a subsidiary of a BHCin 1985 must have been a subsidiary of the same BI-IC inDecember 1983. Similarly, banks that were not affiliates ofBI-ICs in 1985 must also have been independent banks inDecember 1983. In a similar fashion, subsidiaries of BHCsincluded in the study for 1986 were affiliates of the sameBI-ICs as of December 1984 or were independent banks inboth periods. The rules for including banks in the study for1987 and 1988 have similar timing.

Troubled independent banks that were sold to newowners are not excluded from the study because it is moredifficult to obtain information on their ownership than onthat of BHCs. There are 113 banks that met the othercriteria for inclusion in the study that were excludedbecause of changes in their ownership around the timethey became troubled banks. These 113 include 24 inwhich multibank holding companies bought troubled in-dependent banks. The multibank holding companies in-jected capital into 19 of these banks.

14The possible distortions that would result from scalingcapital injections by total assets as of the time of thecapital injections can be illustrated by considering twobanks that are identical in every way other than theresponse of their shareholders to losses. In December1984, Bank A and Bank B had primary capital of $6 and

total assets of $100. In the first quarter of 1985, each hada loss of $2, reducing primary capital to $4 and totalassets to $98. For the rest of 1985, loan losses and net in-come were zero for each bank. During 1985, shareholdersinject $2 into Bank A, returning primary capital to $6 andtotal assets to $100. Bank B receives a capital injection of$1 during 1985 and reduces its assets by $15, raising itsprimary capital ratio to approximately 6 percent [5/(100—2+ 1—15) = 0-05951.

Since the capital injection into Bank A is twice that forBank B, the value of the dependent variable should betwice as large for Bank A. If capital injections weredeflated by total assets as of the time of the capital injec-tions, however, the ratio would be 0.02 for Bank A (2/100)and 0.0119 for Bank B (1/84).

6 I

I

II

I

I

UFEDERAL RESERVE BANK OF St LOUtS

C’apital Iqjection Necessary toMaintain the Prior Level of TotalAssets (NLVJJ

NINJ is defined and calculated as the capitalinjection necessary to raise the troubled bank’sprimary capital to 6 percent of A, divided byA,,15 The calculation of the values of NINJ is il-lustrated in equation 6:

3(C) NINJ = 0.06 — EC,÷,— 1 ~

j=0NINJ is included as an independent variable in

the equation for capital injections to avoid possi-ble biases in estimating the effect of BHC affilia-tion on capital injections. For instance, supposethat BHC subsidiaries require capital injectionsof 4 percent of total assets (A,) to meet theircapital requirements without reducing theirassets, while independent banks need only 1percent. Including the independent variableNINJ controls for such differences.

Expected Future Profits

Bank losses may cause shareholders to lowerthe future profits they expect their banks toearn for a given level of assets, making themreluctant to add capital into these banks. Alter-natively, their expected profits may be unaf-fected by current losses; in these cases, currentshareholders may add capital to offset part orall of the reduction in primary capital.

The average rates of return on assets (BOA)of other banks in the counties of the troubledbanks are used to indicate the prospects forfuture profits. BOA is included to determinewhether shareholders of troubled banks locatedin areas in which other banks have achievedrelatively high BOA are more likely to injectcapital into their banks.

THE DATA

Location and Size of the Banks

The study includes all banks located in the 20states in table 1 in the years 1985-88 that meet

“Evidence in other studies is consistent with the hypothesisthat capital requirements influence capital injections. DahIand Shrieves (1990) found that banks with capital ratiosbelow required levels were more likely to receive capitalinjections than other banks. Mingo (1975) estimatedchanges in bank equity capital as a function of several in-dependent variables, including the capital requirements ofbank supervisors. Capital requirements were significant in

9

the various criteria. These 20 states permit mul-tibank holding companies but do not permitstatewide branching. The restriction on state-wide branching is important for the constructionof BOA, since data on the profit rates of individ-ual branches are not available.

Information in table 1 highlights the deterio-rating condition of banks in Texas over the per-iod. In 1985, about 30 percent of the banks inthe study were Texas banks; by 1987 and 1988,about two-thirds of them were located in Texas.

Table 2 shows the asset size of the banks inthe study. While the vast majority had totalassets under $100 million, a few had total assetsin excess of $1 billion.

Distribution of Banks by CapitalInjections ‘Vecessani to MeetC~apitalRequirements (NJNJJ

The distribution of banks by values of NINJ,the capital injection ratio necessary to raise theprimary capital ratio to 6 percent without re-ducing total assets, is shown in table 3. Of the1358 banks, 76 had such large losses that their

explaining changes in equity capital. Wall and Peterson(1987, 1988) reported that changes in the equity capitalratios of banks were influenced predominantly by capitalrequirements of bank supervisors. These studies are con-sistent with the hypothesis that capital requirements in-fluence capital injections by bank shareholders.

IIIIIIIIIIIIIIIIIIa 1A,manv,rornnnanv loqi

primary capital would have been negative with-out capital injections. At the other extreme, 145banks required capital injections that were lessthan 1 percent of their total assets. Table 3shows that values of NINJ are not clustered atthe extreme values.

Affiliation with Bank HoldingCompanies

Table 4 indicates that about half of the Texasbanks and about 56 percent of the banks locatedin the other states are subsidiaries of multibankholding companies. The rest were independent.

THE RESULTS

‘Fable 4 provides a comparison of the in-cidence of capital injections into independentbanks and subsidiaries of BHCs, ignoring otherdeterminants. The patterns of capital injectionfrequency over time were different for banks in

Texas than for those in other states. For mdc-pendent banks and subsidiaries of BHCs in Tex-as, the percentages of banks receiving capital in-jections declined over time. In each year, how-ever, a higher percentage of BHC subsidiariesthan independent banks received capital injec-tions, and the percentages were significantlyhigher in 1986, 1987 and for the 1985-88 periodas a whole.

For banks in other states, there was no con-sistent pattern over time in the incidence ofcapital injections. As with the Texas banks, thepercentage of banks that received capital injec-tions was higher each year for BHC subsidiariesthan for independent banks. The percentagesare significantly higher in 1985, 1987 and forthe 1985-88 period as a whole. Thus, thesedirect comparisons show that, in general, alarger proportion of BHC-owned troubled banksreceived capital injections than did troubledindependents.

10

\ \\\//‘~, \/// ~7 ~ ~ k~/,, ~~ ~ ~ ~C/? ‘? ~

~ t~\/ j*~ft4v — ~ -..~-\-. __j~ ,r\/I?/t~_)~$t ~ ~ 4/

1~~~ ~t*bI~2 ~~i/ ~ / /2//\/~~\ /\/ ~ ~ ty ~/~ -::~/~ t ~ 1::~~~11.4J ~ ~

\ç ~ /~ ~ ~ ~ — C ~ ~ ‘ ~ ~ ~ y ‘ ~ ~ ~ c > ~‘ ~ /~~arSta~t ~ ‘~ /

III

~\ ~/ d ~‘‘ ~- ~7~>\~\ ~ J~O/ ~- ~ ,7~ / ~ ~ ~ .< //~ 7/~ /~z ~Aá* ~4~I ~ ~ ~ ~ <~ M j/ ,~ < ~ ~ ~/ /<

4 ~s mssw~ J$t~~ ~ ~ ~ ~ ~ ~‘ ~ ~ ~— ~ / /

~ \~ ~ / ~ / /~ ~ ~/~fr ~ /~ /\~//~ ~ ~ ~~~/\ ~~\%/~

‘/ / !~~*~ssS~7 / - /~ <S ~/~/ ~e ~s~*~tt Its ~ /<r ~ ~r\ ~ <‘ ~ : ; :;~~s~,//\

~ ~/ ~*stAIø~** .nr/O~//~ ~/ ~ ~< <~/ ~/ Th-~ ,~ ~ ~ ~ ~ ~

~\ /~ / ~ ~!~lst1~

/ç~/ ~ ~ ;i ~ 2i_+a~$_~~/~/ k ~, 1~~

~ /\~ :/~ ~ ~6 ~{~/~/~/ / ;_ t I c ~32$~~ ~ ~ // ~ ie

/ 7 ~ aøti~~ ~ ~

>:~~i~~ 1~~-y~ ~; ~ ~ / Ct

<>/?2 ~ /~r~r~~ / ~ ~ ~ t ~ ~ ~ :‘i~

IIII

I“~ ~ ~\\ ~~// ~ ~ ~ \ ~ ~ ~ A’ ~ ~ ~ ~ , “\, \<~

:: ~ Afl~/’~~ ~ ~ ~ ~/flj~ 4~

~~ : ~ j~~Lcnc*S~ n:e4*~~~ :~ii~i

I1I

II1

II

rrnrnM nrc~rnwrRM-w nr QT ~ nine

mlAuvv~ue3dIAMvnNw•f

s4uornIJJaoG~OWW!1S9~14TT9JdOJ~‘T3PJUI‘SUOI~OOIUTjfl1d~OJoS~tUPu!mJO4OpI -JiJTU!01Moqsup~jdxapuns!sATTt~u2T~qoJ.JoOJIITJaqioJOqTOI4MaP33!pU!01pOUW!SapS!i:uo1~~nb~-Puaq~sassnos~pwosu!p~p~qsoqj.suonoolux°iqnaqiu~ut&oqsXJflOfldXaiousioioi~jJaqlopfl1d~0aAqlsodpeqs~u~q~c9oqijo99Z~iuo~paTnq!~uwaqiCnuSOOUOJOJJ!Peseq~punj

‘pasnaidwi~saqiUIsuolToalu!IPT!dPOOJOZp~iqia~oeq~tuofSO!.12!piSqflS~!oqTJOJqlWuaaTss~u~qp~~q’~o~Taq2JoJoqulnua9J21rosn~o~q40S~OJflOSs~paaSJHHTPt~aTra~puiJCPWf

‘spoq~amuoissoaWoas~iojop~~nsu~‘s~sXpiu~~U!S~DUOJOJJ!P~U~OiJ!U~iSA~J~OflSfW4Saqj.uo!SSOJ2aJ~~qoJ.~u!snpai~uuisa8!11uoiwnb~

•1’°iqn~qpaisa99nssuoq

-notuipnjd~auoToOjjO311cr~XPJOJa~q~suodsa~

~~~c/J\/~~~~~~~j~~:z~~t41~

/~,//~~‘~/_~~e_t:~~t4*~i*usiMSS~sdOwnfl*~j z~~

:m~tzs4:~~:~~I ;\:~14*0~c;z~‘/~\49~~~*S~W*tSWd~ ~\~N~~~~~~/~\>/y~:\~>/c\~~/>~/P~~~:~~~~,~~t~~::~e~~/~~~/,~\/~~~\/7~4

~\~\/~\~/~iy~—~/\~-/

~2I~2/~/~<p~‘~I2~~;f~/\7$L~q~~‘~~7\p~~A~~C/!~~~~/4~~Øgj~\~/4//

~:~:~~~/;/ç//~~~~z_y‘~~***~~~naaq~/ N

~\~///~//?~>//~y//A~~~/~~~(:i~/,/~~~~~~P~~~‘~ft,/\,4cç*/~~,~~~‘~.~

/pç7NN’1~*7~~*5CTNN/7~$i/N<N’

~:‘1~~/~N’~2. /N’N/~N,7N/NNNNNN

7NNN/N~tNNN’3N~N/n~øa1N/N

7NYNNN~/NtN~~N;~Nx~~w~PCNP>N°N~NNN

jNN/,4NN’~“~‘7’N~~N

Np—“NN/~//NN/NNN~N1’~~LN~fN

7:1~I~ N”~~N’7~N N,N>/N>N>NN’>>>;N/NN,>>

N~NNNN‘>/IN/~N,~>>‘>/N~N>NN’>>~>N

NNNNNN,NNNNNNNNNN>N;,~~N>N~N

N/N>N>/>~r,~NN>NN/NN/P/~NZ>>~>,<>/NN/

NNN‘/NN>>>E~PPd~t<NN> NN<<N‘>/N,>N~>>C<N’N,>>>N

~~~::N:‘NN’~I 7N~N/7N>N/NN~//NNNNN/N>>

/NNNNN,N7N/N>>~N>NN,>>NNNNN*NNft~~N7N>>N7

/N/N/N7/NN’N>>N;/>~,7,N>>7/<~>NN>NN<>7N>,>NNN>~N,>N/N~7fl/>N7>P7N~N/N

N7t:7~N’~~<~7’NNN’N,N~N,NNN~NN/>‘N>>‘N’

7>//NN/7N7N<N</NNN’:~>N>>>’~/7

N/N/N,,,N/NNP>>/N/N>/N>N/7N;,NNP7N>/~‘/~

/~/~Ntflt>/NN//NNN/N/N/>>//N’N‘~‘

p,/7fl7N7>>~~N/N4p.>/NN’7

<>$~~>NN

-~~/>‘N~~Tc</j/Np,IN~/7/N’/7

;N>NN,NN>~/N7/N>”//N~La~:_~INç~N7N’p.>>/

/>>N<~NNN~$Ø$</NNNNNP7NNNN>NNPNNP>P>>

NNIN\N~Ø:j~~/N’NNPtN~N~~’I

NN>>>,N/<N,NN<

/~~N

N>77>>NNN/N/N>NN77N’NNN7p,,~/,~N’NN/NNN’NN,.NNNNN~//NN/

LI

SiflOl~J.Sdo)4NVEEAtdESEHIVt-2JcIBd

N~~~~~~~/~~/\\~~~~~:?~~I~1~~~ae~a~\\>~~ ~\N\\~\\N~~\/>\~\\~~\\\/////*~\N/~/~\/~~\\~\~\

~n~qtflJt’~::~qii~~wt~~N~t~1tGd~

1~t~pwm~ajflIaJt~\\~st~o*~nt~/~\

\\~1~\~\/~~\/~~~\\çfiq~Q(~>~

//~\NIt~(9~lpaMa*S

M~a~aaLw=\N\~n~a~Fus~~2wJ4Inta4emmm~,~a!!teA*qg/\\\~~~?sb44~qj,/\?/~?7/

q4~iatritEIc1t1S4~~/tttQaut,~~,,//~\//t\fltttfltt~Sf3fln~~

~Ii~s~i1Aiflt~3iiSI1a~ttnnfl/r/r—~,.

/r~Ar~—~•~t~~a±,.‘~-t~~~lc/—‘~s~n*f2~s/~~

:~~~Ift~~/~2~~N

?~c>jii?~af*s[1nr~~~~\et4~~‘~/~,

N~/\~Z~9ØI~t/~~\:‘~~:~~/B~~\art~s~fl~N~~ç~yI*p/ø~(A~~

/~~ki~2:;~~:+c~ ~7Y~hI:Sa~A~~~~~ /~/~\?/~~~~/~~~,_7/~\///N~

2~/N/~/~~øttr~*~WIMP~NN

:~~:~~~17::::~~~

~~:cT~~~~~~~N,NN:

‘~Nft~:/N~~/‘~/~\~N~~~~~N~~~\/~\N~~~:N

;N:?~~~\:N: :~:1?~N~~/~N-~-r’~‘~N‘~

~/<~~~~/~NNN~’N\rN/~/7<~NN~Nh~\\~4~N

N~/Nc:~N~~iN?~2

>N,N~)~NN?O/N/~</

<~N~~NN~IN//NN)~N

:N~?~NNNN~N~N/,N/NNNNN

~N

N~N:NN/NN~~:;:N:/NN~

N~~N//N/N/\/N~N/N~/NN~N/~N~//NN//

Nz/44~NNø3~/NNN

~N,~)N~~N~NN~NN~N/

N/7NN~N/NN/NNNN/NI:N:NN~NNNN/NI 1~~~7t//7~c/NNk7/N7~/NN\N~~7NN

NN/~NNN/NNN:P

NNN/\//NNNNN////N////N/

N~~NtN:/NNN/N/NN~t9*4t*y~N/////~N’NNNNN~NN/SON/~pt$BN~31 N~/

,JN7N/7N~~’)rN:T//IN:: //////N//1~S%3N

NN/~N///N

NNN

N~/:~~/N/N://NS%SLJ~UyUOWSJ~$q~~/

I _________________________________________

I K

II ~ ~II N N

II Results for the estimation of equation 1 for1985 through 1988 are presented in table 5.Separate equations are estimated for Texas

I banks because the substantial deterioration ofearnings and capital adequacy of the entireTexas industry in recent years may have in-fluenced Texas shareholder ‘ incentive to in)ect

I capital into troubled banks. NFhe results indicate that the effects of affilia-tion with BHCs are qualitatively imilar for

I troubled banks whether in Te as or elsenhere. NBHC owned troubled banks do not receivelarger capital injections than independent banks, N

I holding constant the other determinants of capi-tal injections, when the total banking as ets ofthe BHCs are less than 10 times the total assets N

I of their troubled bank subsidiaries (RHCS1O). Inboth equations, the coefficients on BHCSIOO and NRHCS100 + are positive and statistically signifi-cant, indicating that if the BHC’s total banking

I assets are at least 50 times a large as the assetsof its troubled subsidiary, it injects more capitalinto the subsidiary than do the owners of mdc-

I penden banks. Outside Texas, this condition Nalso holds for BHCs with total banking assets at Nlea t 10 times the assets of their troubled banks’subsidiaries. Thus, the decision to inject capital

I depends on the size of BHC relative to the izeof their troubled sub idiaries.

Fhe coefficients on the variable N NJ are posi-

I tive and significant in each equation, a thehypothesis suggested. How ver, the variable ____________________________________________

}SHC — the weighted average capital ratio of

I the banking subsidiaries of BHCs, other than

tho e included in the study — does not help cx- N

a

ROA as a Determinant of CapitalInjections

The coefficient on BOA is significant for theTexas regression but insignificant in the other.An examination of the ROA variable suggeststhat this reflects the sharp drop in profitabilityof Texas banks relative to that at non-Texasbanks over the years covered by this study. Thedecline in profitability among Texas banks (table6) induced a corresponding decline in capital in-jections by Texas bank owners (table 4)16

Alternative SpecificationTable 5 represents one way to assess the im-

pact of BHC affiliation on capital injections intotroubled banks. It is possible, however, that thecoefficients on dummy variables for the ratioRHCS in fact may reflect more than simply BHCaffiliation. Capital injections may actually be in-fluenced by the absolute size of BHCs and, in-

14

Equation 7 examines this possibility by incor-porating the asset size of troubled banks andBHCs into the equation:

(7) INJ = f(S50-D1, 575-Di, S75+D1, SSOD2,575-D2, S75 + ~D2, BHC100, BHCI000,BHCI000+, NINJ, ROA).

The additional independent variables are:

Sso — dummy variable with a value of unity ifthe total assets of the troubled bank are greaterthan 825 million but less than or equal to $50million, zero otherwise

S75 — dummy variable with a value of unity ifthe total assets of the troubled bank are greaterthan $50 million but less than or equal to $75million, zero otherwise

S75 + — dummy variable with a value of unityif the total assets of the troubled bank aregreater than $75 million, zero otherwise

Dl — dummy variable with a value of unity if abank is an independent bank, zero otherwise

‘°Oneway to separate the effects on capital injections ofvariation in RCA across counties from the effects ofchanges in average RCA over time is to add dummyvariables for individual years as independent variables. Inregressions not reported here, dummy variables for in-dividual years are added as independent variables to theequations estimated in table 5. In those regressions, the

coefficient on RCA is insignificant for the Texas equationand remains insignificant in the other equation. Thus, afterallowing for different capital injection ratios in differentyears, differences in RCA across the counties in which thetroubled banks were located did not help explain dif-ferences in capital injection ratios across troubled banks.

III

plain differences in capital injections amongtroubled banks. When FSHC was added as anindependent variable to the equations in table 5,its estimated coefficient was not significantlydifferent from zero in either equation.

dependent of BHC affiliation, by the size oftroubled banks themselves. Such separate sizeeffects might well confound the interpretationof the RHCS dummy variables.

a

I 15

D2 — dummy variable with a value of unity if abank is a BHC subsidiary, zero otherwise

BHC100 — dummy variable with a value of uni-ty if the troubled bank is a subsidiary of a BHCwith total banking assets of $100 million or less,zero otherwise

BHC1000 — dummy variable with a value ofunity if the troubled bank is a subsidiary of aBHC with total banking assets greater than $100million but less than or equal to $1 billion, zerootherwise

BHCI000 + — dummy variable with a value ofunity if the troubled bank is a subsidiary of aBHC with total banking assets greater than $1billion, zero otherwise.

Table 7 indicates that capital injection ratiosare larger for subsidiaries of the larger BHCs.Among Texas banks, capital injections increaseif they are subsidiaries of BHCs with total assetsover $100 million. For banks in the other states,capital injections are larger if banks are in sub-sidiaries with total assets in excess of $1 billion.

The BHC effect on capital injections, however,is not just a matter of BHC size. Table 7 also in-dicates that, holding constant the size of theBHC, the larger BHC subsidiaries have lowercapital injections ratios. Coefficients on dummyvariables for the size of independent banks arenot statistically significant. Thus, tables 5 and 7yield the same implications for the effect ofBHC affiliation on capital injections: it is the sizeof BHCs relative to the size of their troubledbank subsidiaries that influences capital injec-tion ratios.

Economic Sign fficance of Affilia~tion with BilL’s for CapitalInjections

The empirical results support the hypothesisthat BHC subsidiaries receive larger capital in-jections than other troubled banks. But are theeffects of affiliation with BHCs on capital injec-tions large enough to be economically signifi-cant? If the BHC effect is estimated to yield onlya few extra dollars for a typical troubled bank,the statistically significant effects could be dis-missed easily as being economically irrelevant.

Economic significance can be gauged by esti-mating the difference between the capital injec-tion ratios of a BHC subsidiary and an indepen-dent bank with identical values of the other

independent variables. These effects could beestimated alternatively using the results in tablesS or 7. Table 7 is used because its maximumlikelihood test statistics are larger, indicatinggreater explanatory power of the equations.

Estimation of the size of the effect of affilia-tion with BHCs involves an analysis of the sizeof the regression coefficient. As explained in theinsert on Tobit analysis, the regression coeffi-cients must be multiplied by the probability ofthe dependent variable being greater than zero(the fraction at the bottom of the table) to esti-mate the effects of a unit change in an indepen-dent variable on the size of the dependent vari-able. In table 7, the fraction is 0.2817 for thesecond equation (Texas banks) and 0.5147 forthe fourth equation (banks located elsewhere).These equations are used in the analysis of theeconomic significance of affiliation with BHCson capital injections.

The estimates of BHC effects on the size ofcapital injections are presented in table 8. Theseeffects are assumed to be zero unless the re-gression coefficients are significantly differentfrom zero at the 5 percent level. They are cal-culated as follows: Suppose a troubled bank inTexas, with total assets of less than $25 million,is a subsidiary of a BHC whose total assets aregreater than $100 million but less than or equalto $1 billion, Compared with an independentbank of the same size and characteristics otherthan BHC affiliation, the capital injection is esti-mated to be higher at the BHC subsidiary by1.67 percent of its total assets, which is calcu-lated as 0.0592 (the coefficient on BHCI000)times 0.2817, the adjustment to regression coef-ficients appropriate for Tobit analysis. The effectfor a subsidiary of a BHC of similar size thathas total assets between $25 million and $50million is estimated as follows:

(0.0592 — 0.0399)0.2817 0.0054.

These estimates have a consistent pattern: in agiven size range, capital injection ratios arelarger for banks that are subsidiaries of largerBHCs, and smaller for larger bank subsidiaries.

The effects of BHC affiliation can be stated indollar terms by assuming certain asset sizes forthe representative BHC subsidiaries. For in-stance, suppose the troubled bank has totalassets of $tS million and is a subsidiary of aBHC with total assets between $100 million and$1 billion. Its capital injection is estimated to be$250,500 larger than that into a similar sized in-

I

IIIIIIIIIIIIIIIS IA~~IIAOV,crcon, ~AOV 1001

16

I 17

dependent bank.17 While this paper presents nocriterion for economic significance, these effectsare too large to be dismissed as economicallyinsignificant.

Holding Company Affiliation’s In-fluence on the Incidence of BankFailure

The influence of RHCS on capital injections af-fects how one interprets some recent studies ofthe determinants of bank failure. Belongia andGilbert (1990) find that increases in RHCS reducethe probability of bank failure. Gajewski (1989)reports a similar finding: the larger the numberof bank subsidiaries in a BHC, the lower theprobability of failure by its subsidiary banks.This study indicates that the lower probabilityof failure by bank subsidiaries of large,multibank holding companies results,in part, from the larger capital injections bymultibank holding companies into their troubledbanks.

CONCLUSIONThe Federal Reserve Board expects BHCs to

serve as sources of strength for their bank sub-sidiaries. BHCs can do so by injecting capital in-to their troubled subsidiaries when losses re-duce their capital ratios below levels acceptableto bank supervisors.

This paper examines whether BHCs have in-jected relatively more capital into their troubledsubsidiaries than the owners of similarly troubledindependent banks, holding constant the valuesof other determinants of capital injections. Theempirical results indicate that the BHC effectdepends on the total banking assets of the l3HCsas well as the assets of their troubled bank sub-sidiaries. BHCs with total assets that are at least50 times larger than those of their troubled sub-sidiaries tend to inject more capital into thosesubsidiaries than other bank owners inject intotheir banks.

These results are consistent with the viewthat BHCs weigh the effects of capital injectionson their reputations in the financial markets,and with the Board, against the opportunitycost of such injections. The results are also con-sistent with recent studies which found that thesubsidiaries of relatively large BHCs have alower probability of failure than other bankswith similar characteristics—their lower failurerate reflects large capital injections into suchbanks

REFERENCES

Belongia, Michael 11, and R. Alton Gilbert. “The Effects ofManagement Decisions on Agricultural Bank Failures,”American Journal of Agriculforal Economics (November1990), pp. 901-10.

“The estimate of the size of the extra capital injection dueto affiliation with a BHC in this case also can becalculated using the results in table 5. The bank sub-sidiary has total assets of $15 million. Suppose the BHChas total assets of $750 million. Results from table 5 in-

dicate that this Texas bank subsidiary would receive anextra capital injection of $202,808 because of its affiliationwith the BHC. This amount equals 0.0468 (coefficient onRHCS 100) times 0.2889 (predicted probability of INJ>0)times $15 million.

IIII

a

18 1Board of Governors of the Federal Reserve System. Annual Gajewski, Gregory R. “Assessing Risk of Bank Failure,” Pro-

Report, 1987. ceedings of a Conference on Bank Structure and Competi-tion (Federal Reserve Bank of Chicago, 1989).

Bovenzi, John, James A. Marino, and Frank E. McFadden.“Commercial Bank Failure Prediction Models,” Federal Gilbert, R. Alton, Courtenay C. Stone and Michael E. Trebing.Reserve Bank of Atlanta Economic Review (November ‘The New Bank Capital Adequacy Standards,” this RevIew1983), pp. 14-26. (May 1985), pp. 12-20. I

Kmenta, Jan. Elements of Econometrics (Macmillan, 1986).Bureau of National Affairs. “Fed May Lack Authority to Re-

quire BHCs to Aid Bank Subsidiaries, Critics Warn,” Bank- Mingo, John J. “Regulatory Influence on Bank Capital Invest-ing Report, May 25, 1987, pp. 923-24. ment,” Journal of Finance (September 1975), pp. 1111-21. I

_______ “Court Tells Fed to go to Bankruptcy Court to Quint, Michael. “Fed’s Uncertain Power Cver Banking Com-Enforce Actions Against MCorp,” Banking Report, June 12, panies,” New York Times, August 3, 1990.1989, pp. 1289-90. Spong, Kenneth. Banking Regulation: Its Purposes, Imple-

mentation, and Effects (Federal Reserve Bank of KansasCornyn, Anthony G., et al. “An Analysis of the Concept of City, 1985).

Corporate Separateness in BHC Regulation from anEconomic Perspective,” Proceedings of a Conference on Tobin, James. “Estimation of Relationships for LimitedBank Structure and Competition (Federal Reserve Bank of Dependent Variables,” Econometrica (January 1958), pp. 1Chicago, 1986), pp. 174-212. 24-36.

DahI, Drew, and Ronald E. Shrieves. “The Impact of Regula- Wall, Larry D., and David R. Peterson. “The Effect of Capitaltion on Bank Equity Infusions,” Journal of Banking and Adequacy Guidelines on Large Bank Holding Companies:’Finance (December 1990), pp. 1209-28. Journal of Banking and Finance (December 1987), pp. J

581-600.Demirg~ic-Kunt,Ash. “Deposit-Institution Failures: A Review _______ “Capital Changes at Large Affiliated Banks:’

of Empirical Literature:’ Federal Reserve Bank of Cleveland Journal of Financial Services Research (June 1988), pp.Economic Review (Quarter 4, 1989), pp. 2-18. 253-75.

Duncan, John P. C. “Fed Raises Ante for Holding Companies Wonnacott, Thomas H., and Ronald J. Wonnacott. Intro- Iwith Troubled Subsidiaries,” American Banker (June 16, ductory Statistics for Business and Economics (John Wiley1987), pp. 4, 19. and Sons, 1990). I

IIIIIIIIIa