FINANCIAL INDUSTRYFEDERAL RESERVE BANK OF DALLASAUGUST 1997

StudiesAre Capital Requirements Effective?

A Cautionary Tale from Pre-Depression Texas

Jeffery W. GuntherSenior Economist and Policy Advisor

Linda M. HooksAssistant Professor

Washington and Lee University

Kenneth J. RobinsonSenior Economist and Policy Advisor

Mexican Payments System Reforms

Sujit “Bob” ChakravortiSenior Economist

Financial Industry StudiesFederal Reserve Bank of Dallas

Robert D. McTeer, Jr.President and Chief Executive Officer

Helen E. HolcombFirst Vice President and Chief Operating Officer

Robert D. HankinsSenior Vice President

Genie D. ShortVice President

EconomistsJeffery W. GuntherRobert R. MooreKenneth J. RobinsonThomas F. SiemsSujit “Bob” Chakravorti

Financial AnalystsRobert V. BubelKaren M. CouchKelly KlemmeSusan P. TetleyEdward C. Skelton

Research Programmer AnalystOlga N. Zograf

Graphic DesignerLydia L. Smith

EditorsAnne L. CourseyMonica ReevesLee Shenkman

Financial Industry StudiesDesign & ProductionLaura J. Bell

Financial Industry Studies is published by theFederal Reserve Bank of Dallas. The views expressedare those of the authors and should not be attributedto the Federal Reserve Bank of Dallas or the FederalReserve System.

Articles may be reprinted on the condition thatthe source is credited and a copy of the publication containing the reprinted article is provided to theFinancial Industry Studies Department of the FederalReserve Bank of Dallas.

Financial Industry Studies is available free ofcharge by writing the Public Affairs Department,Federal Reserve Bank of Dallas, P.O. Box 655906,Dallas, Texas 75265–5906, or by telephoning(214) 922-5254 or (800) 333-4460, ext. 5254.

Contents

Mexican PaymentsSystem Reforms

Sujit “Bob” Chakravorti

Page 12

This article investigates the ongoing payments systemreforms begun by the Bank of Mexico in 1994. The goals of thesereforms are to reduce the amount of uncollateralized intradaycredit extended by the Bank of Mexico (previously unlimited), topromote a market-based allocation of intraday credit for interbankpayments, and to move large-value paper-based payments to elec-tronic form. The Bank of Mexico has been successful in achievingall of these goals to some extent. But despite this progress, likeother central banks around the world, the Bank of Mexico stillfaces the possibility that government guarantees may weaken market discipline in the payments system.

Are CapitalRequirements Effective?

A Cautionary Tale from Pre-Depression Texas

Jeffery W. Gunther, Linda M. Hooks,and Kenneth J. Robinson

Page 1

Capital requirements are now a primary ingredient in effortsto supervise and regulate the banking industry. Their main pur-pose is to protect the deposit insurance fund and to minimize taxpayer exposure should financial difficulties occur. Capitalrequirements are not new, however. Texas was one of the firststates to institute formal capital requirements when it introduceda deposit insurance program early in the century. But this earlyattempt at capital regulation proved ineffective in preventing acomplete breakdown of the deposit insurance system it was meantto protect. Using recently discovered examination data for Texasbanks operating in the troubled 1920s, we show that the capitalrequirements were unsuccessful largely due to a reliance on book-value capital measures that overstated the true financial conditionof banks. As some researchers have shown recently, the sametypes of problems confront current efforts to rely on measures of capital as the focus of banking supervision. This has led to recentproposals to restructure bank capital regulation, such as the pre-commitment approach.

Capital regulation is now the cornerstoneof efforts to regulate bank risk taking. In 1988,the Basle Accord was adopted by bank regula-tors in the United States, as well as in Japan,Europe, and Canada. Under this agreement,banks’ minimum capital requirements dependon the perceived credit risk exposure of theirassets and off-balance-sheet items. In the UnitedStates, capital regulation was taken a step fur-ther when Congress enacted the FederalDeposit Insurance Corporation ImprovementAct of 1991 (FDICIA). Under this legislation,with its provisions for prompt corrective actionand early closure, banks face increasingly strin-gent supervision and regulation as their capitallevels decline, including the usual requirementthat regulators close the institution if the capital-to-asset ratio reaches 2 percent.

According to many accounts, it was thecontinued and relatively unfettered operation of capital-impaired financial institutions that ex-acerbated both the bank and thrift failures of the past decade and the associated losses to thefederal deposit insurance funds.1 Conceptuallyspeaking, the owners of thinly capitalized banksmay have the incentive to take on especiallyrisky ventures, as they have relatively little tolose if the ventures should fail, but much to gain if the ventures prosper. And with the federal government guaranteeing the safety ofmost deposits, depositors would have littleincentive to discipline bank owners against such risk taking.2 Similarly, the owners of thinly capitalized banks may find it more profitable to loot their institutions than to manage them as going concerns. Again, with the safety ofdeposits guaranteed, only bank regulators couldstand in the way of such deceitful strategies.3

These factors may have played a signifi-cant role in the banking and thrift crises of the1980s, as risky financial strategies turned outpoorly or fraudulent dealings created financiallosses. The goal of the new capital-based super-vision is to protect the deposit insurance fund(and ultimately the taxpayer) from the types of risk taking, looting schemes, and financiallosses experienced during the banking and thriftdifficulties of the 1980s. This is accomplished bypreventing regulators from allowing capital-impaired banks to continue to operate withoutcompensating restrictions on their financialbehavior and by not permitting insolvent banksto remain open.

While the adoption of risk-based capitalrequirements, prompt corrective action, andearly closure represent significant steps in theimplementation of capital regulation, they are

Are CapitalRequirements

Effective? A Cautionary

Tale from Pre-Depression Texas

Jeffery W. GuntherSenior Economist and Policy Advisor

Financial Industry Studies DepartmentFederal Reserve Bank of Dallas

Linda M. HooksAssistant Professor

Washington and Lee University

Kenneth J. RobinsonSenior Economist and Policy Advisor

Financial Industry Studies DepartmentFederal Reserve Bank of Dallas

In this article, we explore how

this early attempt at capital

regulation operated and why it

ultimately proved unsatisfactory.

We are particularly grateful to Alan Barrat the Texas Department of Banking forproviding data on failures of state char-tered banks.

1 See Kane (1989) and Short and Gunther(1988).

2 See Merton (1977) and Marcus (1984).

3 Akerlof and Romer (1993) describe indetail how such looting can take place.

FEDERAL RESERVE BANK OF DALLAS 1 FINANCIAL INDUSTRY STUDIES AUGUST 1997

2

not the first attempts by federal regulators tofocus on capital as a key component of effortsto regulate insured financial institutions. InDecember 1981, the federal bank regulatoryagencies first jointly announced specific capitalrequirements applicable to insured commercialbanks. Initially, these requirements were basedon the size of the institution, with larger organi-zations required to hold a smaller percentage oftheir assets as capital. In 1985, bank regulatorsdecided to impose uniform capital-to-assetrequirements on all banks, regardless of size.4

Long before these requirements were im-plemented at the federal level, several states hadexperimented with capital requirements for state-chartered banks. Texas was one of the firststates to establish formal capital requirements.The Texas capital requirements were part of anoverall package of increased regulation forstate-chartered banks that accompanied thestate deposit insurance system introduced in the early part of the century. Similar to later capital regulations at the federal level, Texasinstituted capital requirements in an effort toprotect its deposit insurance fund from financiallosses. However, in the 1920s, little more thanone decade after the capital requirements wereintroduced, bank failures mounted and thestate-run deposit insurance program sufferedincreasing losses—losses that proved severeenough to cause the insurance program to closedown.

In this article, we explore how this earlyattempt at capital regulation operated and whyit ultimately proved unsatisfactory. Using re-cently discovered examination data available for a sample of state banks operating in Texasin the 1920s, we are able to determine that the book-value measures of capital applied toindividual banks were often grossly inflated.Judging from entries on the examinationrecords, regulators recognized the shortcom-ings in the reported financial data and initiatedattempts to adjust reported capital levels down-ward to reflect estimates of the expected lossesembedded in bank loan portfolios. When esti-mated losses began to reduce the amount of abank’s paid-in capital, regulators would thenclassify its capital position as impaired. How-ever, these adjustments were not included aspart of the banks’ balance sheets that werepublished by the state banking authorities. Andthe formal capital requirements instituted inTexas were based on reported capital, ratherthan capital adjusted for loan losses. As aresult, it was possible for banks to satisfy thecapital requirements, even as losses in their

loan portfolios were propelling them towardinsolvency.

We can find no indication, though, thatbanks were allowed to operate for any length oftime after they were classified by regulators asinsolvent. Still, while forbearance is not indi-cated in our examination data, we do find evi-dence that regulators allowed banks to operatefor several years after their adjusted capital wasdesignated as impaired.

We conclude that this early attempt atcapital regulation proved unsuccessful largelybecause the capital requirements were based onbook-value measures that did not adequatelyreflect the true financial strength of individualbanks, thereby allowing capital-impaired banksto continue operating without fully compen-sating restrictions on their financial behavior.

The failure of capital regulation in pre-Depression Texas offers some parallels to thefinancial-sector difficulties of the 1980s. Thecapital regulations in place during the 1980s didnot prevent either the severe deterioration thatoccurred in the financial strength of the FederalDeposit Insurance Corporation or the meltdownof the Federal Savings and Loan Insurance Cor-poration. In these relatively recent episodes offinancial failure, as in the collapse of the Texasdeposit insurance system in the 1920s, book-value accounting practices had the effect of artificially boosting reported capital levels, andcapital-impaired institutions were allowed tocontinue operating without fully compensatingrestrictions on their financial behavior.

The Texas experience with capital regula-tion during the 1920s also has possible implica-tions for the efficacy of capital regulation today,even under the relatively rigorous capital-basedsupervision implemented under Basle andFDICIA. In the pre-Depression Texas experi-ence, faulty book-value capital measures madecapital regulation ineffective. Similarly, theeffectiveness of today’s capital-based super-vision depends critically on the accuracy of regulatory accounting procedures. Prompt cor-rective action and early closure are at the heartof the new supervisory approach, but both arebased on book-value capital. To the extent thatbook values overstate capital levels relative tothe risk of bank activities, the new capital-basedsupervision is subject to the same criticisms thathave plagued capital regulation in the past. Andthe task of assessing capital adequacy is be-coming more rather than less difficult as thecomplexity of bank activities, including deriva-tives trading, continues to grow. Hence, thesame types of problems that beset this early4 See Keeley (1988).

FEDERAL RESERVE BANK OF DALLAS 3 FINANCIAL INDUSTRY STUDIES AUGUST 1997

attempt at capital regulation in pre-DepressionTexas persist even today as a major challenge toeffective capital regulation.

Capital requirements in pre-Depression TexasTexas was one of the first states to experi-

ment with formal capital regulations. These capital regulations were part of a broader pack-age of increased supervision and regulation forstate-chartered banks that was implemented toaccompany the introduction of state depositinsurance. Legislation that ushered in the Texasdeposit insurance system passed the state legis-lature in its second session in May 1909. Thelaw was referred to as Senate Bill 4 and becameeffective January 1, 1910. All state-charteredbanks were required to join the system underone of two plans—the depositors guaranty fundor the depositors bond security system.

The most popular of the two plans wasthe depositors guaranty fund.5 The guarantyfund covered only noninterest bearing depositspayable on demand, although there was nolimit on the amount of these deposits that wascovered by insurance. When banks joined thesystem, they were required to pay a premium of1 percent of their average daily deposits for theprevious year. Thereafter, each bank wasassessed one-fourth of 1 percent of its dailyaverage deposits annually until the fundreached $2 million, after which additional regular assessments were not required. How-ever, if the guaranty fund fell below $2 million,or in an emergency, banks would be subject toa special assessment that was not to exceedmore than 2 percent of average daily depositsfor any one year.6

Recognizing that the existence of depositinsurance might lead insured banks to increasetheir risk profiles, more stringent regulations onstate-chartered banks accompanied the imple-mentation of the guaranty fund. Senate Bill 4specified that no state bank could own morethan 10 percent of the capital stock of anotherbank; it instituted additional penalties for fraud;and limits were placed on the indebtedness to abank of its directors and officers.7

Probably the most important regulatorychange accompanying the introduction ofdeposit insurance was the implementation ofcapital requirements.8 In his classic on the history of deposit guarantees, Robb (1921, 151)points out that

Texas is credited with making the firstattempt in American history to establish an arithmetical relationship between the

deposits and the capital of a bank. With acapital of $10,000, the deposits of a bankare restricted to five times its capital andsurplus; from $10,000 to $20,000 to sixtimes its capital and surplus; from $20,000to $40,000 to seven times; from $40,000 to $75,000 to eight times; from $75,000 to $100,000, nine times; and if capital isover $100,000, to ten times its capital andsurplus.9

Unlike modern capital requirements, thisearly attempt at capital regulation based itsrequirements not on assets, but deposits.Moreover, the bill was quite specific regardinghow banks were to rectify any shortcoming intheir capital. Banks were required to file asworn statement of their average daily depositsfor the year ending on the first day ofNovember. If average daily deposits for the yearamounted to more than the relevant multiple ofcapital and surplus,

…then in any such case it shall be the dutyof the State Banking Board to require thatsuch state bank shall within sixty daysthereafter increase its capital by 25 percent thereof, and it shall be the duty of thecommissioner to immediately furnish suchstate bank with a certified copy of theorder making such requirement, and uponreceipt of such requisition the directors ofsuch State Bank shall, within the timerequired, cause such increase to be madein its capital stock, and if the same is notdone within such time, it shall be unlaw-ful for such bank to thereafter receive anydeposits at any time when its total demandand time deposits shall in the aggregateamount to more than the limitation hereinplaced upon deposits.10

Concerns about the propensity for insur-ance to increase bank risk, with potentiallyadverse consequences for the guaranty fund,seemed to be a motivating factor behind theserequirements. As Robb (1921, 151) points out,“By these additional regulations the lawattempts to counteract any tendencies towardreckless banking that the guaranty system mayengender.” These capital requirements are sum-marized in Table 1.

The 1920s: A bad time for Texas banksThe early years of the Texas depositors

guaranty fund were relatively uneventful. By1920, the fund had reached its required maxi-

5 Under the second plan, known as thedepositors bond security system, abank filed annually with the Commis-sioner of Insurance and Banking a pri-vate bond, or policy of insurance orindemnity, equal to the amount of itscapital. This latter plan was not popularwith the banks, as evidenced by the factthat less than 10 percent of bankschose to be insured under this type ofplan when the deposit insurance systembegan operations. Moreover, until verylate in the operating life of the depositinsurance system, banks were notallowed to switch between the twodeposit insurance plans.

6 In response to mounting banking diffi-culties, the minimum amount requiredin the guaranty fund was increased to$5 million in 1921.

7 For some evidence that risk taking atinsured banks in Texas during the1920s was greater than at uninsuredbanks, see Hooks and Robinson (1996).

8 Section 27 of Senate Bill 4 contains thecapital requirements on Texas state-chartered banks.

9 Robb points out that the Kansas depositinsurance system, which passed thestate legislature on March 6, 1909,established a relationship betweendeposits and capital. The Kansas lawstated, “It shall be unlawful for anybank guaranteed under the provisionsof this act to receive deposits continu-ously for six months in excess of tentimes its paid-up capital and surplus...”(State of Kansas, 1909, 103). UnlikeTexas, participation in the Kansas system was voluntary.

10 See General Laws of Texas (1909, 423).

4

mum of $2 million. From 1910 through 1920,only twenty-four state banks failed in Texas,compared with the more than 1,000 left operat-ing at the period’s end. According to Grant andCrum (1978, 103), fifteen banks were closeddue to fraud, one bank was closed due to losses on loans, one bank closed due to itsinability to meet a large withdrawal, and noinformation is available about why the otherseven banks were closed. Only small disburse-ments from the guaranty fund were needed toresolve these closed banks, and the withdrawalswere covered easily by special assessments thatreturned the fund to its legislated minimum.After allowing for recovery on the assets of thefailed banks, the special assessments averagedabout three-hundredths of 1 percent of thedeposits of participating banks.11

Beginning with a severe recession in 1920,however, the Texas economy suffered a seriesof setbacks that led to banking difficultiesthroughout the decade. Cotton was king inTexas, accounting for 70 percent of total cropacreage and about the same percentage of farm

cash income. Texas produced about one-thirdof the U.S. cotton crop and about one-fifth ofworld output. It is no surprise then that thesharp drop in cotton prices that occurred in theearly 1920s produced a severe regional reces-sion. Cotton prices reached a high in March1920 before falling almost 75 percent by April1921. The index of prices received by Texasfarmers for all crops fell 71.4 percent over thisperiod. The economy recovered in 1922 andinto 1923, only to experience the effects ofanother sharp decline in cotton prices from1923 through 1926.12

These difficult economic times produced asevere financial impact on Texas state banks,most of which were located in rural areas. Asshown in Chart 1, the liquidation rate of state-chartered banks in Texas rose sharply in the1920s, peaking at 17 percent in 1925.13

Failure of the guaranty fundAs bank failures mounted, special assess-

ments on state banks were necessary to main-tain the guaranty fund’s designated balance.Chart 2 shows how these special assessmentsincreased sharply in the early to mid-1920s,both in dollar amounts and as a percent of totaldeposits. In the early 1920s, these assessmentsamounted to about 2 percent of banks’ totaldeposits. By 1927, assessments reached almost2.5 percent of deposits.14 A total of $17 millionwas raised through special assessments from1910 through early 1927. Of course, recoverieswere made from the liquidations of the failedbanks. But, as shown in Chart 3, payments todepositors far exceeded initial liquidation valuesand subsequent recoveries from the failedbanks, resulting in large losses to the insurance

11 See Warburton (1959).

12 See Grant and Crum (1978, 125–26).

13 The liquidation rate includes both invol-untary and voluntary liquidations. Aportion of the voluntary liquidationsinvolved banks switching to a nationalcharter to avoid the increasing costsassociated with membership in thedepositors guaranty fund. Also, Grantand Crum (1978, 155) point out that thevoluntary category included liquidationsundertaken in response to considerablepressure from bank regulators for finan-cially impaired banks to surrender theircharters: “...(M)ost of the banks thatentered ‘voluntary’ liquidation, exceptfor those that were nationalized, did sobecause of financial difficulty.”

14 As noted below, membership in theguaranty fund dropped considerablyafter 1925. Thus, even though the dollaramount of assessments had begun todecline, the burden was spread over asmaller number of banks. These calcu-lations are based on total deposits asrecorded in Warburton (1959) ratherthan on average daily deposits as stipu-lated in the law because data for thislatter deposit category were not avail-able. Therefore, they should be consid-ered an approximation of the actualassessment burden. For comparison,the maximum effective assessment ratecharged by the FDIC on insured banksduring the banking difficulties of the1980s and early 1990s was one-fourthof 1 percent of total deposits.

Table 1Capital Requirements on Texas State-Chartered Banks

Capital Limit on deposits

$10,000 or less ≤ 5 × (capital + surplus)

over $10,000 but less than $20,000 ≤ 6 × (capital + surplus)

$20,000 but less than $40,000 ≤ 7 × (capital + surplus)

$40,000 but less than $75,000 ≤ 8 × (capital + surplus)

$75,000 but less than $100,000 ≤ 9 × (capital + surplus)

$100,000 or more ≤ 10 × (capital + surplus)

NOTE: Deposits are based on average daily deposits for the year ended November 1.

SOURCE: General Laws of Texas, Thirty-first Legislature, Regular, First, and Second CalledSessions, 1909, 423.

Chart 1Liquidation Rate of Texas State Banks, 1910–27Percent

SOURCE: Grant and Crum (1978).

0

5

10

15

20

192619241922192019181916191419121910

Chart 2Special Assessments on Texas State Banks, 1910–27Percent of deposits Millions of dollars

SOURCES: Warburton (1959); FDIC (1956).

0

1

2

3

4

5

1926192419221920191819161914191219100

1

2

3

4

5

Percent

Dollars

FEDERAL RESERVE BANK OF DALLAS 5 FINANCIAL INDUSTRY STUDIES AUGUST 1997

fund. Over the entire life of the guaranty fund,accumulated losses totaled $11.6 million.15

As the assessment burden on banks beganto increase, state bankers began to question thewisdom of membership in the guaranty fund.But short of switching to a national charter, theywere not allowed to leave the plan.16 In responseto increased demands from bankers, the statelegislature in 1925 revised the original legislationto allow banks to leave the guaranty fund andjoin the depositors bond security system.17 Withinthe first three months of this change in the law,more than 300 of the approximately 900 statebanks in existence switched to the bond system.By the end of 1926, approximately 800 statebanks were in the bond system, while only 75banks remained in the guaranty fund. The per-centage of eligible banks participating in theguaranty fund fell from 96 percent in 1924 to 9percent in 1926.18 Those banks that remained in the guaranty fund faced increasingly heavyassessment burdens. By August 1, 1926, the maximum 2-percent assessment had already been levied, collected, and paid out, amountingto 8.5 percent of the capital of the remaining127 banks in the guaranty fund. State bankers,the Texas Bankers Association, and the stateDepartment of Banking all favored repeal of theguaranty fund. On February 11, 1927, legislationwas signed that repealed both the depositorsguaranty fund and the depositors bond securitysystem. Liquidation of the guaranty fundrequired another four years.19

Why was capital regulation ineffective?Given that the primary goal of the capital

regulations accompanying the introduction ofdeposit insurance in Texas was to protect the

solvency of the guaranty fund, the failure of thefund in the mid-1920s suggests that the capitalregulations may have been seriously flawed. Ofcourse, other factors also contributed to thefund’s failure, such as the severity of the eco-nomic shocks that precipitated the crisis andincompetent and dishonest bank management.Nevertheless, an apparent paradox of the Texasbanking crisis of the 1920s is that reported capi-tal levels remained fairly high, even thoughbanks were failing in increasing numbers.20

As shown in Chart 4, although the aggregatecapital-to-asset ratio for Texas state-charteredbanks fell during the 1910 –27 period, even itslowest point of 17 percent is remarkably highby today’s standards.21

Similar findings hold as well for individualbanks. The Annual Reports of the Commissionerof Insurance and Banking are an importantsource of information on the condition of statebanks in Texas. We were able to obtain some of the reports that were produced from 1910 to 1922. Unfortunately, after 1922, with the creation of a separate Department of Banking,no annual reports were published. However,during the 1910–22 period, the reports that we obtained contained financial information onfifty-eight failed banks at the time of their closure.22 The reported capital-to-asset ratios ofthese failed banks range from 7.1 percent to 37 percent, with an average of 20 percent,which, by today’s standards, is a remarkablyhigh capital position for failed banks.23

Given these findings, the experience withcapital regulation in pre-Depression Texasposes several interrelated questions. Why didbanks fail when their reported capital levelswere so high? And how could the insurance

15 See Warburton (1959) and FDIC (1956).

16 According to Grant and Crum (1978,188), during the last five years of theexistence of the guaranty fund, 126state banks obtained national charters,an average of twenty-five per year. Prior to 1922, conversions averaged threeper year.

17 For a description of the depositors bondsecurity system, see footnote 5.

18 See Grant and Crum (1978, 185). Notonly did the legislation allow banks toswitch insurance plans, but it also sig-nificantly watered down the require-ments of the bond security system.Henceforth, banks would be allowed tocount certain bonds already on theirbooks as sufficient to meet the require-ments of the bond plan.

19 See Grant and Crum (1978, 186, 189).

20 As we discuss in a later section, itappears likely that liquidity pressuresoften were the factor that led banksupervisors to close these institutions.

21 Capital is defined as the sum of paid-incapital, surplus, and other capitalaccounts.

22These closures represent true failures inthe sense that they do not include anybanks that liquidated solely to convertto a national charter.

23 In an effort to ensure consistency withthe capital measures reported in Chart 4,capital is defined from the items in theAnnual Reports as the sum of paid-incapital, surplus, and undivided profits.Surplus represents the part of bankcapital that was accumulated from netearnings to serve as an additional safe-guard against losses. Texas state bankswere required to set aside 10 percent ofnet earnings to surplus until the surplusaccount equaled 50 percent of paid-incapital. Undivided profits are, in today’sterminology, retained earnings.

Chart 3Disposition of Insured Deposits at Failed Texas State Banks, 1910–27Millions of dollars

SOURCES: Warburton (1959); FDIC (1956).

0

2

4

6

8

10

192619241922192019181916191419121910

Recoveries

Initial liquidations

Losses to the fund

Chart 4Capital Ratios at Texas State Banks, 1910–27Percent of assets

SOURCE: All-Bank Statistics, United States 1896–1955, Board ofGovernors of the Federal Reserve System (1959).

0

5

10

15

20

25

30

35

192619241922192019181916191419121910

6

fund have suffered such large losses, given thatbanks failed with such high levels of capitalremaining on their books? An alternative sourceof data regarding the financial condition of thebanks at their time of failure might help answerthese questions.

Book-value capital overstated. In theFederal Reserve Bank of Dallas’ archives are ex-amination records on almost two hundred state-chartered banks in Texas from 1919–27, whichcoincides with the turbulence of the state bank-ing system and its deposit insurance fund.These records contain bank balance-sheet infor-mation, including the capital positions of thebanks. Further, these examination records alsoinclude estimates of the losses likely to beincurred by individual banks. These losses aremostly from the banks’ loan portfolios but couldalso include estimated losses on securities, realestate, and cash items. Judging from hand-written entries in these examination reports,bank examiners would deduct the amount ofthese estimated losses from bank capital. Thisadjustment would result in the examiner indi-cating on the reports for a number of banks thattheir capital was “impaired.” 24

By investigating those examination recordsthat are available for banks close to the time oftheir failure, and by deducting from capital thelosses estimated by examiners, we are able toobtain somewhat more meaningful measures of the financial condition of failed banks. Toaccomplish this, we pool together the data onall the banks for which we have an examinationreport dated one year or less before the time of

failure.25 This procedure results in a sample ofthirty-two banks.26

To gauge the extent of the losses esti-mated by examiners, we compare the reportedor unadjusted capital-to-asset ratios for our sample of failed banks, as recorded in theirexamination records, with what we call adjustedcapital-to-asset ratios that incorporate the lossesestimated by examiners. We divide the banksinto five groups of nearly equal size based ontheir unadjusted capital-to-asset ratios. Chart 5shows the average unadjusted and adjusted capital-to-asset ratio for each of the five groups.Consistent with the data found in the variousAnnual Reports of the Commissioner of Insur-ance and Banking, these banks reported highlevels of capital relative to total assets, evennear the time of their failure. The group ofbanks with the highest capital reported an aver-age capital-to-asset ratio of 35 percent, while thegroup with the lowest capital had an averagecapital ratio of 11 percent. However, once theestimated losses are deducted, the average capi-tal ratio for each of the groups falls consider-ably, often to about half its unadjusted level.These findings indicate that the book-valuecapital ratios reported by the Texas banks weregrossly overstated and thereby help to explainthe apparent paradox of how the banks couldhave failed even as their reported capital levelsremained fairly high.

The relationship between the unadjustedand adjusted capital-to-asset ratios for our banksample is shown in Chart 6. For most levels of

24 There is no indication in the examina-tion records of what is entailed in classifying a bank’s capital as impaired.However, in most cases this occurredwhen the potential losses estimatedduring the examination process ex-ceeded the level of surplus plus undi-vided profits, so that the potentiallosses began to erode the amount ofpaid-in capital contributed to a bank.

25 We use as our definition of failure thosebanks whose charter surrender wascharacterized as either “involuntary” or“voluntary” because, as noted in foot-note 13, this distinction was largelymeaningless. We do not include thosevoluntary surrenders for purposes ofswitching to a national charter, though,since this would not represent closuredue to financial difficulties. We alsoexclude two banks that were sold volun-tarily and were not in financial difficulty.

26 These failures occurred between 1924and 1928.

Chart 5Distribution of Capital Ratios ofTexas State Banks at FailurePercent of assets

NOTE: Bars represent the mean values for each group.

SOURCE: Examiner’s Report of Condition, Federal Reserve Bankof Dallas.

0

5

10

15

20

25

30

35

40

54321

AdjustedUnadjusted

Rank based on ratio of unadjusted capital to assets

Chart 6Capital Versus Adjusted Capital at Failed Texas Banks(Percent of assets)Adjusted capital

Capital

0 10 20 30 40 50–5

5

15

25

35

45

55

SOURCE: Examiner’s Report of Condition, Federal Reserve Bankof Dallas.

FEDERAL RESERVE BANK OF DALLAS 7 FINANCIAL INDUSTRY STUDIES AUGUST 1997

adjusted capital, reported or unadjusted capitalratios tend to fall in a fairly wide range of about20 percentage points, as indicated by thedashed lines.27 Of particular note is how widethe range of reported capital becomes for banksapproaching insolvency on an adjusted-capitalbasis, with reported capital ratios reaching ashigh as 30 percent.

Because the formal capital requirementsinstituted in Texas were based on reportedbook-value capital, and given the abundance ofevidence presented above indicating thatreported capital was grossly overstated, it is difficult to see how the capital regulations inplace could have been effective in their intend-ed purpose of preventing “reckless” bankingand protecting the guaranty fund. In Chart 7, thestatutory required minimum amount of capitalrelative to deposits, which, as shown in Table 1,is based on banks’ capital zones, is depicted bythe horizontal lines, while each of our thirty-twoindividual banks’ reported or unadjusted capital-to-deposit ratio, as stated in the last examinationreport available prior to failure, is shown by thevertical bars.28 Reported capital fell short of itsstatutory level relative to deposits for only fourof the thirty-two banks, even near the time offailure, suggesting that the capital requirementsprobably had little constraining effect on bankbehavior. In contrast, in Chart 8, when each

bank’s adjusted capital is compared with itslevel of deposits, now seventeen banks are inviolation of the capital requirement, and threeof these are insolvent. These results indicatethat the adjustments by examiners to bank capi-tal to reflect likely losses give a more accuratepicture of the financial health of individualbanks. Chart 8 also shows which of these bankswere classified as capital-impaired by examin-ers.29 Interestingly, of the thirty-two failed banks,twenty-four were deemed either to be capitalimpaired, as assessed by examiners, or to havehad adjusted capital ratios below the statutoryminimum. While the capital regulations them-selves were largely ineffective, regulators didattempt to take into account the prevalent overstatement of book-value capital levels byforming their own estimate of potential losses.However, given that (1) almost half of the banksin our sample were not in violation of capitalrequirements near the time of their failure, evenafter adjusting for estimated losses, (2) only threeout of the thirty-two banks’ capital fell belowzero after accounting for estimated losses, and(3) the depositors guaranty fund suffered largelosses that ultimately resulted in its liquidation,even the regulatory adjustments to bank capitalcontained in the examination reports did notprovide a very accurate picture of the economicvalue of Texas state banks.30

Was forbearance a problem? One reasonwhy these adjustments to capital, while animprovement, are still suspect is that regulatorsmay have delayed recognition of problems in

Chart 7Capital Levels of Texas Banks at Time of FailureCapital-to-deposit ratio

NOTE: The horizontal lines represent the inverse of the limit on deposits found in Table 1. The requirement for bankswith capital of $100,000 or more is 10 percent; with capital of $75,000 –100,000, 11.1 percent; with capital of $40,000 –75,000, 12.5 percent; with capital of$20,000 – 40,000, 14.3 percent.

SOURCE: Examiner’s Report of Condition, Federal Reserve Bankof Dallas.

Banks’ capital zones

0

80

70

60

50

40

30

20

10

$20,000–40,000

$40,000–75,000

$75,000–100,000

$100,000or more

109.92

Chart 8Adjusted Capital Levels of Texas Banks at Time of FailureAdjusted capital-to-deposit ratio

NOTE: See note to Chart 7.

SOURCE: Examiner’s Report of Condition, Federal Reserve Bankof Dallas.

Banks’ capital zones

–20

80

70

60

50

40

30

20

10

0

–10

$20,000–40,000

$40,000–75,000

$75,000–100,000

$100,000or more

99.35

Capital impaired

Capital unimpaired

27 The correlation between these two measures of capital is 63 percent.

28 We convert the representation of capitalrequirements from a deposits-to-capital to a capital-to-deposits ratio for ease of exposition. The capital-to-deposit ratios shown here are calculatedas paid-in capital and capital surplusrelative to total deposits. These calcula-tions are based on total deposits frombank examination records rather thanon average daily deposits as stipulatedin the law because data for this latterdeposit category were not available.

29 We should note that it is possible for a bank to be designated as capital im-paired, as judged by examiners, evenwhen its capital is fairly high relative todeposits, if the bank had suffered a largedecline in deposits. See footnote 24.

30 It is possible that losses incurred by theinsurance fund in resolving failed banksmight overstate the extent of the banks’insolvency if the banks possessed avalue as a going concern that was lostin the resolution process. But it isunlikely that such factors can accountfor the large losses suffered by theinsurance fund.

8

the hopes that banks’ financial health wouldrecover. Grant and Crum (1978, 157) report several cases in which insolvent banks wereallowed to continue operating.31 They quoteBanking Commissioner Charles Austin in 1925

Most of these failures were banks knownto this department to be absolutely insol-vent for a long time, but which appear tohave been kept open with the hope thatthey might work out their own salvation.Bank failures do not develop overnight,but are usually the result of conditionswell known and fully recognized for along time before the crisis develops. Noneof the failures which have occurred so farthis year should have been any surprise toany person having access to the bank ex-aminer’s reports, and most of them shouldhave been no surprise to the public atlarge for the reason that the banks havebeen notoriously insolvent and generallydiscussed in the communities where theyexisted for many months prior to theirclosing.

These words highlight the important dis-tinction between bank insolvency and bank failure or closure. Banks are insolvent when themarket value of their assets is less than the market value of their liabilities. Here, failure or closure refers to a decision by the charteringauthority that effectively terminates the opera-tions of the bank. We do not have a definitiveanswer as to what ultimately triggered closuresof Texas banks during the 1920s. It appears likely, however, that regulatory assessments ofcapital strength were not the primary factor indetermining when a failing bank was ultimatelyclosed. Warburton (1959, Table 10, 46) pro-vides evidence that incompetent managementand associated heavy depositor withdrawalswere the primary causes of Texas bank failuresover the period 1921–30. Liquidity pressures

on banks, then, may have forced the regula-tors’ hand in triggering the decision to close afailing bank.32

However, even though our limited data,represented by examination reports on thirty-two state bank failures, do not allow us to deter-mine definitively whether or to what extentforbearance was widespread, we can use theavailable examination reports to at least shedsome light on this issue. Most of the recordscontain a question that asks if the examinerconsiders the bank to be solvent. Of the thirty-two failures, only four examination reports indi-cated that the bank was not solvent when closeto failure. Several more banks were classified aseither questionably solvent or “barely” solventwhen approaching failure. In only one case wasa bank classified as insolvent more than oneyear prior to failure. However, as stated earlier,the data on the increasing assessment burdensand ultimate depletion of the deposit insurancefund are consistent with the view that the con-dition of troubled banks was substantially worsethan examiners’ assessments would indicate. In this sense, forbearance was still practiced,though unintentionally.

In almost every case, however, while themajority of banks were classified as solvent at orbefore failure, most of the descriptions of theirfinancial condition indicated serious financialdifficulties occurring several years prior to fail-ure. The extent of banks’ financial weakness inthe years prior to failure depends on the par-ticular measure of capital difficulties, as shownin Table 2. The least restrictive measure is thereported, or book-value capital ratio. By thismeasure, only 12 percent of the banks in oursample did not meet the minimum capital-to-deposit requirement when close to failure, com-pared with 53 percent of banks when using theadjusted-capital measure. Almost three-fourthsof the banks that failed were classified as capi-tal impaired one year or less prior to failure. Theproportion of banks operating in financial diffi-culty trails off as the failure date recedes. Inevery case, the least restrictive measure is thereported book-value capital ratio, while the designation of capital impairment captures thelargest percentage of subsequent failures.33 Ifrestraints had been set in place for capital-impaired banks similar to those legislated forbanks not meeting the formal capital require-ments, then perhaps the regulatory apparatuswould have been more successful in limitinglosses to the insurance fund.

Assessment. Overall, the available evi-dence indicates that this early attempt to imple-

31 Warburton (1959) also cites forbear-ance as a problem in Texas during the1920s.

32 Weaver (1926, 62) points out that thecommissioner had the authority to closeinstitutions where “the interests of thedepositors are jeopardized” by fraud orheavy deposit withdrawals. Friedmanand Schwartz (1963, Chapter 7) alsonote the importance of liquidity pres-sures on bank failures during the1930s.

33 Data for all thirty-two bank failures arenot available for up to four years priorto failure. The number of banks forwhich we have examination data in theyears before failure are as follows:twenty-seven banks one to two yearsbefore failure, twenty-six banks two tothree years before failure, and twenty-two banks three to four years beforefailure.

Table 2Measures of Bank Financial Distress Prior to Failure

Percentage of banks that

Years Are Do not meet capital Do not meet capitalbefore capital requirement requirementfailure impaired (adjusted) (book value)

0 – 1 72 53 121 – 2 59 19 72 – 3 46 15 43 – 4 27 9 4

SOURCE: Examiner’s Report of Condition, Federal Reserve Bank of Dallas.

FEDERAL RESERVE BANK OF DALLAS 9 FINANCIAL INDUSTRY STUDIES AUGUST 1997

ment capital standards on banks producedmixed results, at best. Unlike seven other statedeposit insurance systems in place at the time,no insured depositor suffered any losses underthe Texas system.34 However, though reportedcapital levels were quite high, even at the timeof failure, they were largely fictitious, and thedepositors guaranty fund suffered such heavylosses that its cost proved too great for its members. We surmise that a primary reason for the failure of capital regulation in pre-Depression Texas was that the capital require-ments were based on book-value measures that did not adequately reflect the true finan-cial strength of individual banks. As a result,capital-impaired banks were allowed to con-tinue operating without fully compensatingrestrictions on their behavior, since specificrestrictions were legislated only for banks thatfailed the capital requirement on a book-valuebasis.35 The lesson we take from this earlyattempt at capital regulation is: Without mean-ingful measures of capitalization that provide at least a close approximation of the net eco-nomic value of an institution relative to the size and risk of its activities, together withappropriate compensating restrictions on thefinancial behavior of capital-impaired institu-tions, capital regulation will tend not to providean adequate level of protection to the depositinsurance fund.

What about today?Capital regulation is probably more impor-

tant today than it was in pre-Depression Texas.This reflects the aftermath of the bank and thriftdifficulties of the 1980s and early 1990s. Thecollapse of many savings and loans has beenestimated to have cost taxpayers between $150billion and $175 billion.36 Banking difficulties ofthe 1980s and early 1990s were not as severe,but did result in the Federal Deposit InsuranceCorporation reporting a negative balance of $7billion for the first time in 1991.37 Despite a sys-tem of formal capital requirements in place,bank failures generated substantial losses, asindicated in Table 3. The more recent Basle risk-based capital requirements, along with the prompt corrective action and early closure features of FDICIA, were implemented to avoida replay of these difficulties. And the promptcorrective action and early closure provisions ofFDICIA are specifically designed to preventbanks from pursuing any risk-taking or lootingincentives that may arise if their capital shouldbecome impaired.

However, immediately related to the

above analysis is whether the current regula-tory accounting standards supporting Basle and FDICIA provide an adequate measure of bankcapital relative to the size and riskiness of bankactivities. Several recent analyses have ques-tioned whether the current capital-based super-visory system in place will make a difference,should banking difficulties develop in thefuture. Jones and King (1995) estimate that mostof the banks exhibiting a high risk of insolvencyfrom 1984 to 1989 would not have been classi-fied as undercapitalized had the FDICIA guide-lines been in place. In fact, more than one-thirdof the banks would have been classified as wellcapitalized. Interestingly, Jones and King sug-gest a modification to the calculation of loanloss reserves based on examination results asone possible adjustment to the risk-based capi-tal requirements. It was also an adjustment tocapital based primarily on estimated loan lossesthat Texas examiners in the 1920s used in anapparent attempt to derive more meaningfulcapital measures. Along these same lines, Peekand Rosengren (1996) question whether theprompt corrective action provisions of FDICIAwould have made a difference during recentbanking difficulties in New England.

This evidence from the past decade, plusour analysis of the Texas banking difficulties ofthe 1920s, highlights how problematic book-value accounting measures can be. Market-valueaccounting was part of the original proposalthat eventually became FDICIA, but the objec-tions of regulators and bankers caused it to bedropped from the final bill.38 It should be em-phasized that the estimation of the economic ormarket value of banks’ assets and liabilities,many of which are not publicly traded, presentsa number of formidable difficulties.39 As a result,the same types of measurement problems thathave plagued traditional forms of capital regula-

34 Seven other states experimented withdeposit insurance systems around thistime, including Oklahoma, Kansas,Nebraska, Mississippi, South Dakota,North Dakota, and Washington. All ofthese state-run plans experienced finan-cial difficulties and ultimately collapsed.See FDIC (1956), American BankersAssociation (1933), and Calomiris(1989) for descriptions of these variousdeposit insurance systems. While noinsured depositors suffered any lossesin Texas, in the other states losses oninsured deposits ranged from $1.2 million to $33.7 million.

35 Grant and Crum (1978, 158–59) reportthat the actions available to regulatorsincluded: (1) stockholder assessmentsto restore capital, which were largelyunsuccessful due to the fact that theprinciple stockholders were often officers of the bank who had depletedtheir own resources in an attempt tosave their business, (2) “overnight reorganizations” with new stockholdersand a new charter quickly arranged, and (3) liquidations.

36 See National Commission on FinancialInstitution Reform, Recovery andEnforcement (1993).

37 See FDIC Annual Report (1991). Onelikely reason why the thrift industry’sdifficulties were much more severe wasthe less stringent supervision and regu-lation of savings and loans. While thriftswere subject to capital requirements inthe 1980s administered by the FederalHome Loan Bank Board, the amountsrequired were substantially reducedthroughout the 1980s, while the defini-tion of regulatory capital was watereddown. See Barth (1991) and Kane(1989).

38 See Benston and Kaufman (1994).

39 See Berger, King, and O’Brien (1991).

Table 3Losses to Bank Insurance Fund from Bank Failures, 1985–94

Estimated lossYear Number of failed banks (in millions)

1985 120 $1,0081986 145 1,7251987 203 2,0211988 221 6,8721989 207 6,1231990 169 2,8131991 127 6,2691992 122 3,9601993 41 5841994 13 139

SOURCE: United States General Accounting Office (1996).

10

tion since the Texas banking crisis of the 1920sare likely to remain a source of concern, even ifattempts were made to implement market-valueaccounting.

In fact, the measurement problems associ-ated with capital regulation may have increasedas banks have moved beyond their traditionalrole of taking deposits and making loans intonew forms of financial intermediation. In this re-gard, many have argued that traditional methodsof assessing capital adequacy are not applicableto these new lines of business, particularly thetrading of financial derivatives, because meas-urement of the size and risk potential of deriva-tives portfolios is exceedingly difficult.

In response to these developments, severalnew approaches to capital regulation have beenadvanced.40 A prominent example is the so-called precommitment approach, whereby banksdetermine their maximum precommitted tradinglosses and set aside sufficient capital based onthese estimates.41 Under this approach, penaltieswould be assessed on banks if their trading losses exceeded their maximum expected losses.This approach addresses the measurementproblem squarely, since regulators are requiredto know only the actual losses, leaving the measurement of the size and riskiness of tradingactivities to the banks themselves. Unfortu-nately, to the extent that this approach hasmerit, its primary application is to portions of abank’s overall activities, rather than the bank asa whole.

As the banking industry continues toevolve in response to market pressures, theimplementation of capital regulation most likelywill continue to adapt as well. However, despiteall this change, measurement problems akin tothe inflated book values that plagued an earlyattempt at capital regulation that occurred someeighty years ago in the state of Texas are likelyto remain as a significant challenge to effectivecapital regulation.

ReferencesAkerlof, George A., and Paul M. Romer (1993), “Looting:

The Economic Underworld of Bankruptcy for Profit,”

Brookings Papers on Economic Activity, no. 2: 1–73.

American Bankers Association (1933), The Guaranty ofBank Deposits (New York: American Bankers Association).

Barth, James R. (1991), The Great Savings and LoanDebacle (Washington, D.C.: The AEI Press).

Benston, George J., and George G. Kaufman (1994),

“The Intellectual History of the Federal Deposit Insurance

Corporation Improvement Act of 1991,” in ReformingFinancial Institutions and Markets in the United States,ed. George G. Kaufman (Boston: Kluwer Academic

Publishers), 1–17.

Berger, Alan N., Kathleen Kuester King, and James M.

O’Brien (1991), “The Limitations of Market Value

Accounting and a More Realistic Alternative,” Journal ofBanking and Finance 15 (September): 753–83.

Bliss, Robert R. (1995), “Risk-based Capital: Issues and

Solutions,” Federal Reserve Bank of Atlanta EconomicReview, September/October, 32–40.

Board of Governors of the Federal Reserve System

(1959), All-Bank Statistics, United States 1896–1955(Washington, D.C.: Board of Governors of the Federal

Reserve System).

Calomiris, Charles W. (1989), “Deposit Insurance:

Lessons from the Record,” Federal Reserve Bank of

Chicago Economic Perspectives, May/June, 10 – 30.

Examiner’s Report of Condition, Federal Reserve Bank

of Dallas, unpublished examination reports (Dallas,

1919–1928).

Federal Deposit Insurance Corporation (1991), AnnualReport (Washington, D.C.: Federal Deposit Insurance

Corporation).

——— (1956), Annual Report (Washington, D.C.: Federal

Deposit Insurance Corporation).

Forty-Seventh Annual Report of the Commissioner ofInsurance and Banking (Austin: Von Boeckmann-Jones).

Forty-Sixth Annual Report of the Commissioner ofInsurance and Banking (Austin: Von Boeckmann-Jones).

Friedman, Milton, and Anna Jacobson Schwartz (1963),

A Monetary History of the United States 1867–1960

(Princeton: Princeton University Press).

General Laws of Texas, Thirty-first Legislature, Regular,First, and Second Called Sessions, 1909 (Austin: Von

Boeckmann-Jones), Chapter 15, 406 –29.

Grant, Joseph M., and Lawrence L. Crum (1978), TheDevelopment of State-Chartered Banking in Texas(Austin: Bureau of Business Research, University of

Texas at Austin).

Hooks, Linda M., and Kenneth J. Robinson (1996),

“Moral Hazard and Texas Banking in the 1920s,” Federal

Reserve Bank of Dallas, Financial Industry Studies

Working Paper no. 1–96 (Dallas, October).

40 For more on these approaches to determining capital, as well as someshortcomings, see Bliss (1995).

41 See Kupiec and O’Brien (1995).

FEDERAL RESERVE BANK OF DALLAS 11 FINANCIAL INDUSTRY STUDIES AUGUST 1997

Jones, David S., and Kathleen Kuester King (1995),

“Implementation of Prompt Corrective Action: An

Assessment,” Journal of Banking and Finance 19

(June): 491–510.

Kane, Edward J. (1989), The S & L Insurance Mess: How Did It Happen? (Washington, D.C.: The Urban

Institute Press).

Keeley, Michael C. (1988), “Bank Capital Regulation in

the Early 1980s,” Federal Reserve Bank of San Francisco

Weekly Letter (San Francisco, January 22).

Kupiec, Paul, and James O’Brien (1995), “A Pre-commit-

ment Approach to Capital Requirements for Market Risk,”

Federal Reserve Board of Governors Working Paper no.

95–36 (Washington, D.C., July).

Marcus, Alan J. (1984), “Deregulation and Bank Financial

Policy,” Journal of Banking and Finance 8: 557–65.

Merton, Robert C. (1977), “An Analytical Derivation of

the Cost of Deposit Insurance and Loan Guarantees:

An Application of Modern Option Pricing Theory,” Journalof Banking and Finance 3: 3–11.

National Commission on Financial Institution Reform,

Recovery and Enforcement (1993), Origins and Causesof the S&L Debacle: A Blueprint for Reform (Washington

D.C.: U.S. Government Printing Office, July).

Peek, Joe, and Eric S. Rosengren (1996), “Will Legis-

lated Early Intervention Prevent the Next Banking Crisis?”

Federal Reserve Bank of Boston Working Paper no. 96–5

(Boston, September).

Robb, Thomas Bruce (1921), The Guaranty of BankDeposits (Boston: Houghton and Mifflin Company).

Short, Genie D., and Jeffery W. Gunther (1988), “The

Texas Thrift Situation: Implications for the Texas Financial

Industry,” Federal Reserve Bank of Dallas FinancialIndustry Studies (Dallas, September).

State of Kansas (1909), Session Laws of Kansas(Kansas: Official State Paper, March 10, 1909).

U.S. General Accounting Office (1996), “Bank and Thrift

Regulation: Implementation of FDICIA’s Prompt Regula-

tory Action Provisions,” Report to Congressional Com-mittees (Washington, D.C.: U.S. Government Printing

Office, November).

Warburton, Clark (1959), Deposit Insurance in EightStates During the Period 1908 –1930 (Washington, D.C.:

Federal Deposit Insurance Corporation).

Weaver, Findley (1926), “State Banking In Texas,”

unpublished Master of Arts Thesis, University of Texas,

Austin (June).

12

Most of my understanding of theMexican payments systems is based oninterviews and written correspondencewith the major players in the Mexicanfinancial markets. I benefited from con-versations with Juan Antonia, GilbertoCalvillo, Abdón Sánchez-Arroyo, andFrancisco Solís at the Bank of Mexico,Héctor Pérez Galindo and Carlos H.Garza at INDEVAL, and individuals atother institutions that are major playersin the payments systems described. Inaddition, I would like to thank AltonGilbert, Jeff Gunther, Genie Short, EdStevens, Bruce Summers, and JimThomson for comments on previousdrafts.

1 For a discussion of Mexican financialsystem reforms, see Welch and Gruben(1993) and Gruben, Welch, and Gunther(1993).

2 For a description of these systems, seeChakravorti (1997).

3 For a comparison of large-value pay-ments systems, see Horii and Summers(1994) and Bank for InternationalSettlements (1997).

4 For an overview of payments systemissues in developing countries, seeListfield and Montes-Negret (1994) andSato and Humphrey (1995).

Since the late 1980s, Mexico has engagedin a series of far-reaching financial and eco-nomic reforms. These reforms include the pri-vatization of its banks and other state-ownedenterprises, interest rate deregulation, an easingof reserve requirements, reductions in restric-tions on trade and foreign bank entry, and anoverhaul of the payments systems.1 This articleinvestigates the ongoing payments systemreforms that Mexico began in 1994.

The safe and efficient transfer of monetaryvalue in exchange for goods, services, andfinancial assets is vital to any market economy.The apparatus used to transfer monetary valueis the payments system. For the purpose ofanalysis, the payments system as a whole canbe divided into large-value and small-value systems. Large-value, or wholesale, paymentssystems are primarily used to transfer fundsbetween banks, and the average value of eachtransfer is relatively large. Folkerts-Landau,Garber, and Lane (1994) list the important func-tions of large-value systems: to provide the necessary infrastructure for the intermediationof household and business payments, to enablemore efficient liquidity management by banks,to assist the development of security markets,and to allow for more effective implementationof monetary policy. The primary thrust of pay-ments system reform in Mexico thus far hasconcentrated on large-value systems.

Small-value, or retail, payments systemsprocess relatively small payments among con-sumers and businesses. Retail payment in-struments include cash, checks, automatedclearinghouse payments, credit and debit cards,and, more recently, electronic money.2 Pingitzerand Summers (1994) state that “the efficientoperation of a market economy depends on the availability of a smoothly functioning small-value transfer system that connects all economicagents.”

Although financial analysts agree thatlarge-value payments systems should be safeand efficient, there is little consensus on theiroptimal design and operation. Major differencesexist in the types of large-value payments sys-tems employed in developed countries.3 As aresult, developing countries seeking to enhancetheir integration with international capital mar-kets face difficulty in identifying the mostappropriate blueprint for strengthening theirown large-value systems.4 Mexico provides aninteresting example of the recent push to en-hance the safety and efficiency of large-valuepayments systems in emerging financial markets.

In early 1994, the Bank of Mexico, the

Mexican PaymentsSystem Reforms

Sujit “Bob” ChakravortiSenior Economist

Financial Industry Studies DepartmentFederal Reserve Bank of Dallas

A s with the other major

financial reforms initiated

since the late 1980s, Mexico’s

recent efforts to enhance the role

of market-based decisions in

the payments system represent

a significant step in the

right direction.

FEDERAL RESERVE BANK OF DALLAS 13 FINANCIAL INDUSTRY STUDIES AUGUST 1997

5 I discuss these circumstances below.

6 This conversion is based on the pre-vailing exchange rate of 7.81 HK dollar–U.S. dollar in October 1987. For a description of this event, seeFolkerts-Landau et al. (1995).

7 When discussing large-value paymentssystems, good funds are usuallyreserves held at the central bank byfinancial institutions.

8 Bank for International Settlements(1992), A2–7. In the broader bankingliterature, systemic risk is often definedas the failure of a financial institutionleading to the failure of one or morefinancial institutions, with adverse con-sequences to both the financial systemand the economy as a whole.

central bank of Mexico, proposed reforming itspayments system. The goals of these reformsare to decrease the amount of unsecured intra-day credit it extends to banks over the large-value interbank payments system, to promotemarket discipline in the determination of creditexposures related to the payments system, andto move large-value transactions away fromchecks to electronic systems. The first two goals are designed to reduce payments systemrisk, while the last one is aimed at increasingefficiency.

Progress has been made on each of thesefronts. Except under certain circumstances, theBank of Mexico no longer extends uncollateral-ized intraday credit to settle payments on itslarge-value payments system.5 Instead, to main-tain adequate liquidity while imposing marketdiscipline, the Bank of Mexico implemented anet large-value payments system, in which participants send payments based on intradaylines of credit they extend to one another. In addition, the Bank of Mexico has success-fully promoted this system as an alternative to high-value checks. However, some challengesremain for the Bank of Mexico in implementingthe desired market discipline in the intradaycredit market.

Trade-offs in payments system designThe details of the recent payments system

reforms in Mexico are best understood in thecontext of the policy alternatives facing pay-ments system operators in general. To provide aframework for an analysis of Mexico’s reforms,this section discusses some of the major issuesassociated with the operation of large-valuepayments systems.

A safe payments system minimizes therisks involved in the transfer of monetary value.From a public policy perspective, a safe pay-ments system can prevent the costly distur-bances that result from the stoppage of clearingand settlement caused by a failure of one ormore participants to settle. An example of sucha stoppage occurred in the Hong Kong futuresmarket during October 1987, when the marketwas closed for four days to sort out its settle-ment problems. The Hong Kong government,along with leading banks and brokerage firms,helped the various parties meet their obliga-tions by extending credit totaling HK$2 billion(US$256 million).6

The risks that safe payments systemsattempt to minimize are often collectivelyreferred to as payments system risk. Paymentssystem risk includes liquidity risk, settlement

risk, and systemic risk. (See the box entitled“Glossary of Terms.”) Liquidity risk is the riskthat a participant does not have good funds atthe time of settlement but can provide them ata later time. Settlement risk is the risk that oneparty to a transaction does not deliver theunderlying asset in its entirety at the specifiedsettlement time. This asset could be good funds,another financial asset, or a physical asset.7

Systemic risk, as defined by the Bank forInternational Settlements (1992), is “the risk thatthe inability of one institution to meet its obli-gations when due will cause other institutions tobe unable to meet their obligations when due.” 8

However, safety is not the only factorinfluencing the economic benefits provided bya payments system. An efficient payments sys-tem also promotes an efficient allocation offinancial resources. At the level of individualmarket participants, this efficiency results inlower transactions costs. For example, a com-parison of the cost differentials between securi-ties clearing and settlement systems used inemerging markets and the United States illus-

Glossary of Terms

Clearing/Clearance “Clearing is the process of transmitting, reconciling and insome cases confirming payment orders or security transfer instructions prior to settlement, possibly including netting of instructions and the establishment of finalpositions for settlement” (Bank for International Settlements 1993).

Clearing House for Interbank Payments System (CHIPS) CHIPS is the primaryelectronic large-value funds transfer system for the dollar component of foreign ex-change and cross-border transactions. CHIPS, established in 1970, is operated bythe New York Clearing House Association.

Delivery Versus Payment (DVP) DVP describes transactions in which delivery ofan asset occurs if and only if payment occurs. Participants need not deliver goodfunds but only a payment instrument with the underlying value.

Fedwire Fedwire, the U.S. large-value gross settlement system operated by theFederal Reserve, is used for the transfer of funds and government securities.

Liquidity Risk The risk that a participant does not have good funds at the time ofsettlement, but can provide them at a later time.

Settlement “An act that discharges obligations in respect of funds or securitiestransfers between two or more parties” (Bank for International Settlements 1993).

Settlement Risk The risk that one party to a transaction does not deliver the under-lying asset in its entirety at the specified settlement time. This asset could be goodfunds, another financial asset, or a physical asset.

Sistema de Atención a Cuentahabientes de Banco de México (SIAC) SIAC isthe large-value gross settlement system that transfers funds between reserveaccounts at the Bank of Mexico.

Sistema de Información de Depósito de Valores (SIDV) SIDV is the large-valuesecurities transfer system. SIDV is operated by the Instituto de Depósito de Valores(INDEVAL).

Sistema de Pagos Electrónico de Uso Ampliado (SPEUA) SPEUA is the large-value funds transfer system that nets payments and settles over SIAC. SPEUA isoperated by the Bank of Mexico.

Systemic Risk “The risk that the inability of one institution to meet its obligationswhen due will cause other institutions to be unable to meet their obligations whendue” (Bank for International Settlements 1992). This definition applies in the contextof payments systems.

14

9 Financial institutions usually do not earn interest on these reserves, sothey attempt to minimize their reserveholdings.

10 For a description of Swiss InterbankClearing, see Vital and Mengle (1988).

11 For a discussion of these issues, seeHancock and Wilcox (1996) andRichards (1995).

12 The risk and efficiency trade-offs ofpayments systems that net are modeledin Chakravorti (1996).

trates the potential cost savings to participants.According to Stehm (1996), the average cost toprocess and settle a securities trade in emergingmarkets is probably between ten and one hun-dred times greater than in the United States.

When designing payments systems, oper-ators can choose between gross settlement systems and net settlement systems. In gross settlement systems, each transaction is settledindividually; in net settlement systems, partici-pants settle the net of their incoming and out-going payments at the end of a specified periodof time, usually a day. (For a comparison ofgross and net settlement systems, see the boxentitled “Gross Versus Net Settlement.”) Grosssettlement offers participants the immediacy ofusing the underlying funds and reduced settle-ment risk because each transaction is settledwith good funds.9 However, these systems aremore expensive for participants to use than net-ting systems because of the need for greaterquantities of good funds to settle. Operators ofpayments systems must weigh the safer grosssettlement system against the more efficient netsettlement system.

Payments system operators often adoptpolicies to increase the efficiency of gross settlement systems or decrease the settlementrisk of netting systems. To decrease cost to participants, gross settlement system operatorsmay extend free intraday credit. To decreasesettlement risk to participants, net settlementsystem operators may impose market-baseddebit caps and/or loss-sharing arrangementsamong participants. Market-based debit capsrestrict the amount a participant can owe at any time during the day. Loss-sharing rules dis-tribute the losses associated with the failure ofone or more participants to settle among theremaining participants.

One area in which central banks’ grosssettlement systems differ is the quantity of intra-day credit they extend. In this context, intradaycredit is used to facilitate payment flows duringthe day. Payments system participants areexpected to end the day with a zero balance. Atone extreme is Swiss Interbank Clearing, wherethe Swiss National Bank, the central bank ofSwitzerland, extends no intraday credit.10 At theother extreme, until recently some central banksextended unlimited daylight credit. Until themid-1980s, the Federal Reserve extended nearlyunlimited daylight credit to Fedwire partici-pants. Since then, the Federal Reserve hasimposed limits on intraday credit and alsocharges fees based on the quantity of creditextended.11 The Bank of Mexico also used to

extend unlimited and uncollateralized intradaycredit to its banks to make payment over itslarge-value gross settlement system. As part ofthe recent reforms, however, the Bank ofMexico has replaced most unsecured intradaycredit with fully collateralized credit.

Central banks are faced with a trade-offwhen deciding to extend intraday credit. By notproviding intraday credit, they eliminate creditrisks associated with direct intraday lending.However, such a policy may result in paymentgridlock, especially in financial systems withoutwell-developed interbank funds markets. Pay-ment gridlock occurs when the flow of pay-ments stops because participants are waiting to receive payments before sending them. Byproviding intraday credit, central banks increaseintraday liquidity and prevent payment gridlock.A central bank’s major concern about such apolicy is the credit risk associated with intradaylending, especially if this credit is not properlypriced. In addition, the reliance on central bankcredit by payments system participants may dis-tort the market allocation of intraday credit.

Another way to reduce payment gridlockthat does not rely on central bank credit is tonet payments instead of settling each paymentindividually. In net settlement systems, par-ticipants extend each other credit during theday and settle their positions with good funds at the end of the day. An example of such a system is the Clearing House Interbank Payments System (CHIPS), the large-value payments system used primarily to settle inter-national dollar payments and dollar com-ponents of foreign exchange transactions.12

However, such systems do not usually guar-antee the immediacy of funds. In other words,good funds are not usually available until theend of the day for further transactions.

In its reform of the payments systems, theBank of Mexico reduced its direct exposure tointraday credit risk, while maintaining sufficientliquidity. The Bank of Mexico implemented parallel gross and net settlement systems. Thetwo net settlement systems settle over the grosssettlement system at the end of the day. Todecrease its exposure to credit risk, the Bank ofMexico eliminated the extension of daylightcredit to banks except under certain limited circumstances. Although the Bank of Mexicoeliminated direct unsecured intraday credit, itguaranteed payment of end-of-day net clearingbalances arising from the two parallel nettingsystems. However, these guarantees may stillexpose the Bank of Mexico to undesired levelsof credit risks.

FEDERAL RESERVE BANK OF DALLAS 15 FINANCIAL INDUSTRY STUDIES AUGUST 1997

To illustrate the difference between grossand net settlement systems, consider the followingsix individual payments between institutions A, B,and C:

A → B $400A → C $400B → A $300B → C $600C → A $500C → B $400

In gross settlement systems, each institution settles each payment individually (Chart 1). If weassume that the central bank does not grant intra-day credit, each participant would either have towait until it is paid, borrow funds in the interbankfunds market, or hold assets in the form of centralbank reserves to make payment. If each bank waited until it was paid, there is a possibility that

no one would send a payment, resulting in pay-ment gridlock. Let us assume that institutions donot wait for incoming payments before sending anoutgoing payment. In such a system, one cost ofparticipation is the cost of holding or borrowingcentral bank reserves. For illustrative purposes,assume there is a 1 percent cost for holding or borrowing reserves. In this example, institution Auses $800, institution B uses $900, and institutionC uses $900. The cost for A would be $8; for B, $9;and for C, $9.

Alternatively, these participants could bilat-erally net payments during the day and settle at theend of the day. By bilaterally netting, an institutionnets payments between itself and each of the otherinstitutions, resulting in only one transaction witheach of the other participants (Chart 2 ). In such asystem, institution A only needs $100 of centralbank reserves, reduced from $800; institution Bonly needs $200, reduced from $900; and insti-tution C only needs $100, reduced from $900.The cost for A is $1, reduced from $8; for B, $2,reduced from $9; and for C, $1, reduced from $9.

Multilateral netting would further reduceholdings of central bank reserves by the institutions

Gross Versus Net Settlement

(Chart 3 ). In this case, institution A reduces itsholdings of central bank reserves from $100 in thebilateral netting system to $0 in the multilateral netting system; institution B reduces its holdings of central bank reserves from $200 to $100; andinstitution C reduces its holdings of good fundsfrom $100 to $0. The cost of holding reserves to Aand C is zero and to B is $1. Multilateral settlementsystems require the least amount of central bankreserves to settle and are also the least costly toparticipants.

However, settlement of payments is not finalin net settlement systems. Settlement only becomesfinal at the end of the day when good funds aretransferred. One way to increase the efficiency ofthe payments system and to ensure settlement at the time of payment is for the central bank toextend free intraday credit and guarantee payment.In such systems, participants enjoy the benefits ofnetting, since they settle at the end of the day andalso benefit from immediacy of funds due to theguarantee. But by extending free intraday credit,central banks are exposed to settlement risk, andthe guarantee distorts the payments system par-ticipant’s credit assessments of other participants.

Chart 1Gross Settlement

A

C B

$400

$600

$400

$500

$300

$400

Chart 2Bilateral Net Settlement

A

C B

$200

$100

$100

Chart 3Multilateral Net Settlement

A

C B

$0

Clearinghouse

$100

$100

16

13 For a description of the Mexican pay-ments systems prior to 1994, seeSánchez-Arroyo (1996).

14 In addition to SPEUA and SIDV end-of-day positions, SIAC also settles posi-tions from the check clearinghouses.

15 The Bank of Mexico does not regulateor supervise brokers.

16 A bank that has defaulted on settlementhas three days to meet its shortage offunds and faces penalties for the lengthof time it takes to settle. If the bankcannot settle at the end of three days,the loss-sharing arrangements are used.At this point, the banks that granted thedefaulting bank credit share in the loss.This loss-sharing arrangement isdescribed in the box entitled “SPEUALoss-Sharing Arrangements.” Failure tomeet its overdraft within the three-daytime frame is not sufficient for the Bankof Mexico to close the bank.

17 Díaz (1996).

18 A full assessment of the Bank ofMexico’s success in reducing the levelof intraday credit would require a comparison of the aggregate bilateralSPEUA credit granted as a percentageof the total value of payments to theaggregate overdrafts on SIAC as a percentage of total value of paymentsbefore the reforms.

The Mexican payments system reformsBefore the proposed reforms in 1994,

the Bank of Mexico operated the Sistema deAtención a Cuentahabientes de Banco deMéxico (SIAC-BANXICO, or SIAC), which wasMexico’s only electronic large-value interbankpayments system.13 SIAC was introduced in1986, replacing the electronic system known asSistema de Información Contable. Participantsused SIAC to transfer Mexican pesos, U.S. dol-lars, and government securities. Each participanthad three SIAC accounts: a peso account, a U.S. dollar account, and a securities account.The Bank of Mexico guaranteed every paymentand granted free unlimited and unsecured day-light peso overdrafts to banks. However, theBank of Mexico did charge penalty rates forovernight borrowing resulting from daylightoverdraft positions.

As part of the reforms, the Bank of Mexicoreorganized SIAC into three linked paymentssystems: a new SIAC, the Sistema de PagosElectrónico de Uso Ampliado (SPEUA), and theSistema de Información de Depósito de Valores(SIDV). SIAC, still operated by the Bank ofMexico, is now used primarily to settle positionsfrom the other two systems. The Bank ofMexico replaced unlimited and unsecured over-drafts with 100 percent collateralized overdrafts.In addition, the Bank of Mexico placed limits onthe size of the fully collateralized overdraftsbased on bank size. SPEUA, also operated bythe Bank of Mexico, is another electronic large-value funds transfer system. Unlike SIAC,SPEUA participants use uncollateralized intradaycredit to make payment. However, participantsface credit limits based on the credit lines theyextend to one another. SIDV, operated byInstituto de Depósito de Valores (INDEVAL), aprivate firm, clears and settles government- andbank-issued securities and equities. Each ofthese systems is discussed in more detail below.

SIAC. Currently, SIAC participants holdonly peso accounts at the Bank of Mexico, andpayments are irrevocable. As mentioned above,most payments over SIAC must be collateralizedor made with good funds, which limits the Bankof Mexico’s exposure to unlimited and uncollat-eralized intraday credit. Most analysts agree thatthis policy has reduced the Bank of Mexico’srisk because the value of unsecured intradaycredit extended by the Bank of Mexico hasdecreased. SIAC’s major function is to settlepayments resulting from end-of-day positionsfrom the other systems.14 These types of SIACpayments are called nonrejectable payments. Inaddition, some individual payments continue to

be processed over SIAC. Brokers, for example,send payments via SIAC since they are notallowed to participate in SPEUA directly.15

If a SIAC participant does not have ade-quate collateral for a payment and the paymentis used to settle an end-of-day clearing obliga-tion from another system, such as SPEUA, theBank of Mexico will extend unsecured credit tothe bank to allow the payment to be made.Although the Bank of Mexico thus extendsunsecured credit, it charges penalty rates onsuch overdrafts and strongly encourages par-ticipants to avoid them. In addition to penaltiesfor each unsecured overdraft, the Bank ofMexico imposes sanctions based on a partici-pant’s unsecured overdrafts during a givenmonth. The Bank of Mexico may also increasecollateral requirements for a participant thatsends uncollateralized nonrejectable paymentstoo often. Because the Bank of Mexico is will-ing to allow such payments, the receiving par-ticipant of a SPEUA payment bears no same-dayliquidity risk. However, that participant doesface credit risk based on its share of the loss-sharing arrangement should the sending partic-ipant be unable to meet its obligation after threedays.16

After these changes were implemented inMarch 1995, SIAC participants learned to man-age their SIAC accounts better and reduce theirreliance on Bank of Mexico unsecured intradayand overnight credit. SIAC participants sig-nificantly reduced their reliance on Bank ofMexico unsecured credit after the first threemonths following the adoption of these policies.Of the penalties imposed for SIAC unsecuredoverdrafts in the first nine months after theadoption of these policies, 92 percent occurredin the first three months, whereas only 8 percentof the penalties occurred in the next sixmonths.17 Thus, early indications suggest thatthe Bank of Mexico has been successful inreducing the amount of uncollateralized credit itgrants to SIAC participants.18

SPEUA. SPEUA was developed to increaseintraday liquidity and to decrease the riskabsorbed by the Bank of Mexico. Unlike SIAC,SPEUA participants are limited to banks. InSPEUA, the participants determine the levels ofintraday credit through bilateral credit lines thatthey extend to each other. Further, each bankhas an aggregate credit limit that is the sum ofthe bilateral credit limits. Like SIAC payments,SPEUA payments are irrevocable, except if pay-ments are queued.

For example, if a sending bank exceeds itscredit limit, payment messages are placed in a

FEDERAL RESERVE BANK OF DALLAS 17 FINANCIAL INDUSTRY STUDIES AUGUST 1997

19 The Bank of Mexico has consideredallowing participants to decrease theircredit lines during the day. However,such a change would further complicatethe loss-sharing rules. The Bank ofMexico is considering the adoption ofless complex loss-sharing arrange-ments in conjunction with the introduc-tion of collateral requirements. If suchchanges are adopted, the Bank ofMexico may consider allowing partici-pants to decrease their credit lines.

20 The effectiveness of the loss-sharingprovision is critically dependent on howthe Bank of Mexico settles insolventbanks. If banks are not allowed to fail,or if interbank placements are not sub-ject to loss even in the event of failure,SPEUA participants extending creditwould discount the costs associatedwith the loss-sharing provision in theirinterbank lending decisions.

21 See Robinson and Flatraaker (1995) andHumphrey and Berger (1990) for costcomparisons of electronic forms andchecks or paper giros. Giro paymentsare credit transfers between the payorand the payee that may be used forrecurring or nonrecurring payments.The payor instructs the Giro, an organi-zational structure that receives andmakes payment, to debit his or heraccount and credit the payee’s account.Giro payments are a dominant form ofpayment in many European countries.Giro payments can be either electronicor paper based.

queue. Payments that are queued can be can-celed before they are sent. When the participantis again sufficiently below the credit limit, thequeued payment message is sent if it has notpreviously been canceled. However, due to thehigh credit lines extended to participants, fewpayments are queued. In addition, participantsusually stop sending payments when their credit limit is reached. The Bank of Mexicorestricts them from reducing credit lines duringthe day.19 At the end of the day, each bank mustmeet any debit positions and send payments viaSIAC. As part of the reforms, the Bank ofMexico also established loss-sharing rules to dis-tribute losses in the event of the failure of aSPEUA participant. (For a description of this

arrangement, see the box entitled “SPEUA Loss-Sharing Arrangement.”) According to theserules, SPEUA participants that grant intradaycredit to a failed participant share in the lossbased on a loss-sharing formula.20 In addition,the Bank of Mexico plans to impose collateralrequirements in the future.

In reforming its payments systems, theBank of Mexico also wanted to move high-valuepayments away from checks to electronic form.Several studies have shown that electronic alter-natives are significantly less costly to processand use than checks.21 This savings increases theefficiency of a country’s payments system. Toprovide an incentive to use SPEUA instead ofchecks, the Bank of Mexico changed the value

SPEUA Loss-Sharing Arrangement

These loss-sharing arrangements are used after a bank has failed to settle its SPEUA obligations forthree consecutive days. The additional settlement obligation (obligación adicional de liquidación — OAL) forinstitutions that grant credit to the defaulting institution is1

where:

OALijt = The additional settlement obligation of participant i as a result of the default of participant j onday t. Equation B.1 calculates the additional settlement obligation for day 1, equation B.2 for day2, and equation B.3 for day 3.

Cjt = The overdraft of defaulting bank j at day t. For days 2 and 3, Cjt measures the difference betweenthe overdraft position on day t and the overdraft position from the preceding day, or day t – 1.If the difference is negative, the overdraft position for the preceding day will be recalculated.

LERijt = The amount of the credit line extended to participant j by participant i at day t.

LERkjt = The amount of the credit line extended to participant j by participant k at day t.

i = The bank for which the additional settlement obligation is being calculated.

j = The overdraft bank.

k = All banks except overdraft bank j.

n = The total number of SPEUA participants.

In period 1, the additional settlement obligation of a participant is equal to the product of the partici-pant’s share of the total credit extended to the overdraft participant and the total amount of the overdraft.2

In periods 2 and 3, the calculation of the additional settlement obligation is similar, except that it is based onany additional credit extended to the overdraft bank. The total additional settlement obligation of a partici-pant is equal to the sum of the obligations in days 1 through 3. If there is a shortfall between the defaultingparticipant’s overdraft and the sum of the additional settlement obligations of the remaining participants, theBank of Mexico absorbs the loss.

1 Bank of Mexico (1997). The description of loss-sharing arrangement did not appear in the original version but appeared as an update.2 Loss-sharing rules are often based on the credit line and not the actual credit extended. For example, CHIPS’ loss-sharing arrangements are

also based on credit lines extended (see New York Clearing House Association 1996).

( ) ,( )

B.2 2 2

2 1

2 11

OAL CLER OAL

LER OALij j

ij ij

kj kjk

n=−

−

=

∑

( ) ,B.1 1 1

1

11

OAL CLER

LERij j

ij

kjk

n=

=

∑

( )( )

B.3 3 3

3 2 1

3 2 11

OAL CLER OAL OAL

LER OAL OALij j

ij ij ij

kj kj kjk

n=− −

− −

=

∑

18

22 In May 1996, the Bank of Mexico furtherreduced the minimum value per trans-action to 100,000 pesos. However, therewas no significant change in the valueor volume of check payments.

23 The figures for check value and volumeare from Díaz (1996) and correspon-dence with Bank of Mexico staff.

24 Named after the German bank that wasclosed in 1974 before it could makepayment on its dollar obligations,Herstatt risk is the risk in a foreignexchange transaction that one partydelivers one currency but the counter-party does not deliver the other. In thecase of Herstatt, the time zone differ-ence between Germany and New Yorkwas largely to blame for the dollardefaults. Although this type of settle-ment risk is named Herstatt risk, it isnot necessarily eliminated if there is little or no time zone difference betweenthe currencies being settled. The differ-ence in settlement times of the twounderlying currencies in a foreign exchange transaction leads to Herstattrisk. For a discussion of foreign ex-change settlement risk, see Chakravorti(1995) and Bank for InternationalSettlements (1996).

25 For a general overview of clearance andsettlement of securities in emergingmarkets, see Stehm (1996).

26 Based on a survey of various DVP systems, the Bank for InternationalSettlements (1992) categorized DVPsystems into three models. Model 2DVP systems settle the securities partof the transaction on a gross basis during the day and settle the funds sideon a net basis at the end of the day.Model 1 DVP systems settle both thesecurities and funds side on a grossbasis. Although model 1 systems haveless settlement risk than model 2 systems, such systems require greateramounts of good funds to settle and asa result are more expensive for partici-pants to use. Model 3 DVP systems settle both the securities and funds sideon a net basis.

27 The adoption of delivery-versus-paymentarrangements does not eliminate pay-ments system risk completely. Therecould still be a failure to settle the pay-ment. In non-DVP transactions, there is the potential for one party to neverdeliver its asset after receiving thecounterparty’s asset.

28 Most stock transactions occur on thestock exchange because of tax benefitsassociated with exchange traded stocks.

29 For SIAC transfers, nonbank partici-pants may transfer funds themselves ifthey are SIAC participants. However, forSPEUA transfers, nonbank SIDV partici-pants must have correspondent rela-tionships with a SPEUA participant.

date on checks to next day from same day inJanuary 1996. The Bank of Mexico also reducedthe minimum value for SPEUA transactions from500,000 pesos to 150,000 pesos in December1995.22 The Bank of Mexico believes that thesepolicies were responsible for the reduction ofthe average daily value of checks from 55 bil-lion pesos in 1995 to 6 billion pesos in 1996.23

Although the value of check transactionsdecreased significantly, the number of checksprocessed did not decrease significantly be-cause the number of checks with values of150,000 pesos and above was and continues tobe fairly small. The average daily number ofchecks decreased from 782,000 in 1995 to684,000 in 1996.

Most Mexican peso components of large-value foreign exchange transactions are settledvia SPEUA. Most foreign exchange peso trans-actions are for U.S. dollars, and the dollar com-ponents of each transaction are settled primarilyvia New York-based CHIPS. In most cases, if anondollar–peso foreign exchange transaction isrequested by a client, the trader would firstmake a dollar trade and then trade dollars forthe desired currency. Herstatt risk exists forpeso–dollar transactions using SPEUA andCHIPS, since the settlement of the dollar andpeso components of the transaction may notoccur simultaneously, even though SPEUA andCHIPS operate roughly during the same time.24

For large-bank–to –large-bank transactions, thepeso and dollar transactions are not settled inany specific order. However, for transactionsinvolving a small participant and a large bank,the large bank will often require the delivery ofone currency before releasing the other.

SIDV. Operated by INDEVAL, SIDV is usedto clear and settle bank and government securi-ties, and equities.25 SIDV participants are re-quired to have two types of SIDV accounts—afunds account and a securities account. All SIDVtransactions follow the Bank for InternationalSettlements’ Delivery Versus Payment (DVP)model 2.26 In a DVP transaction, the underlyingsecurity and the payment for that security areexchanged at the same time, thereby reducingsettlement risk.27 In October 1994, DVP wasimplemented for bank securities transactions. InJuly 1996, the DVP process was extended togovernment securities, and in April 1997, theDVP process was extended to equities.

For a DVP SIDV settlement to occur, thebuyer must have a positive balance in its SIDVfunds account or have access to overdraft facilities, and the seller must have the security inits securities account. Once INDEVAL has con-

firmed the seller’s possession of the security andthat the buyer has adequate funds or overdraftfacilities, the transaction cannot be reversed. Ifthe seller does not have the underlying securityor the buyer does not have the funds or suffi-cient overdraft facilities, the transaction isplaced in a queue and settled when each partyhas the necessary funds and securities to settle.If the queued transaction involves governmentor bank securities, the trade can be canceled.However, if the transaction is an equity trans-action, it cannot be canceled while in the queue because of stock exchange rules regard-ing trades.28

The buyer’s overdraft facility is the lesserof the fully collateralized credit line or buyer’sbank credit line. Collateral can be in the form ofbank or government securities. When used forcollateral, government securities receive a lesserdiscount than bank securities.

In addition to the collateralized creditlines, participants are granted credit lines thatare a component of their overdraft facility frombanks. Every morning, the Bank of Mexicoextends credit lines to banks for the purpose ofmaking SIDV payments. In turn, banks allocatethese credit lines to SIDV participants. Althoughthere is not a set policy for the amount of credit each bank is granted by the Bank ofMexico, in most cases banks receive credit linesof around 60 percent of their aggregate SPEUAcredit line to allocate to SIDV participants. Inaddition to the collateralized and bank creditlines, buyers can transfer funds from SPEUA or SIAC to use for payment.29 However, par-ticipants use the overdraft facility most of the time.

SIDV is linked to SPEUA and SIAC. Theselinks enable participants to transfer funds in realtime between these systems either directly, ifthey are banks, or through their correspondentbank. A participant, for example, can sell asecurity using SIDV and transfer the funds toSIAC and then use the funds to offset someother obligation, all within minutes.

The Group of Thirty in 1989 made recom-mendations for the clearance and settlement ofsecurities.30 (See the box entitled “Group ofThirty Recommendations for Securities Clearingand Settlement” for a complete list.) These recommendations have been accepted as a standard that securities markets around theworld should strive to meet, and, in 1992, theGroup of Thirty produced status reports on var-ious countries, including Mexico. At that time,Mexico did not satisfy two of the recommen-dations. First, Mexico did not satisfy recom-

FEDERAL RESERVE BANK OF DALLAS 19 FINANCIAL INDUSTRY STUDIES AUGUST 1997

30 The Group of Thirty, established in1978, is a private-sector nonprofit organization concerned with the work-ing of international financial markets. In 1989, the Group of Thirty publishedits recommendations to reduce risk and improve efficiency of securitiesmarkets around the world.

31 S. D. INDEVAL (1995).

mendation 5, which states that each countryshould use DVP to settle all securities trans-actions. Today, all transactions cleared and settled by INDEVAL use DVP.

Second, Mexico did not meet recommen-dation 9, which states that each country shouldadopt the international message standard devel-oped by the International Organisation forStandardisation (ISO Standard 7775). Use of onestandard for numbering and identifying securi-ties facilitates greater ease in cross-border trans-actions. In 1993, INDEVAL implemented theInternational Securities Identification Number(ISIN) code.31

Some market participants are concernedabout the funds-netting component of SIDV’sDVP system, especially if participants areallowed in the future to sell securities short oract as market makers. They fear that the risk ofopen positions taken by participants may affectend-of-day settlement. In a netting system, thedefault of one participant may affect others,even if they did not deal directly with thedefaulting participant. However, other partici-pants argue that, by allowing participants tomake the market or sell short, the liquidity ofthese markets should improve. Greater liquidityin the market should enable participants to havegreater ease in buying and selling securities,thereby reducing the settlement risk associatedwith open positions in general.

Remaining challengesAs part of its payments system reforms,

the Bank of Mexico attempted to implement amarket-based allocation of intraday credit. Byeliminating unsecured daylight overdrafts onSIAC and simultaneously developing SPEUA,the Bank of Mexico attempted to shift most ofthe credit risk associated with the extension ofintraday credit from itself to payments systemparticipants. Furthermore, by implementingexplicit loss-sharing rules for SPEUA settlementfailures, the Bank of Mexico attempted toincrease market discipline by imposing losseson creditors.

However, concerned about maintainingadequate liquidity to avoid payment gridlockand keeping the cost to participants relativelylow, the Bank of Mexico has implemented poli-cies that may have the unintended effect of dis-torting the market-based allocation of intradaycredit. For example, the Bank of Mexico’srestriction on participants’ decreasing their credit lines during the day may also increase theavailability of interbank funds to troubled par-ticipants, which, in turn, could increase the risk

of financial loss from a settlement default. Inaddition, the previously mentioned practice ofgranting immediacy of payment for SIAC trans-actions used to meet end-of-day settlement may distort the credit assessments made amonginterbank participants.

Banks evaluate their daily credit lineextensions knowing that the Bank of Mexicowill make payment at the end of the day. Aslong as the Bank of Mexico allows non-rejectable payments to exist on SIAC, market-based risk assessments may be distorted by theguaranteed end-of-day extension of liquidity by the central bank. Although the SPEUA loss-sharing rules can allocate losses to SPEUA creditors, even if the defaulting bank is not

Group of Thirty Recommendations for Securities Clearing and Settlement

The recommendations made by the Group of Thirty (1989) are:

Recommendation 1: Trade ComparisonBy 1990, all comparisons of trades between direct market participants (that is,brokers, broker/dealers, and other exchange members) should be accomplishedby T + 1.

Recommendation 2: Trade AffirmationIndirect market participants (such as institutional investors, or any trading coun-terparties which are not broker/dealers) should, by 1992, be members of a tradecomparison system which achieves positive affirmation of trade details.

Recommendation 3: Central Securities DepositoryEach country should have an effective and fully developed central securitiesdepository, organized and managed to encourage the broadest possible industryparticipation (directly and indirectly), in place by 1992.

Recommendation 4: Trade Netting SystemEach country should study its market volumes and participation to determinewhether a trade netting system would be beneficial in terms of reducing risk andpromoting efficiency. If a netting system would be appropriate, it should be imple-mented by 1992.

Recommendation 5: Delivery Versus PaymentDelivery versus payment (DVP) should be employed as the method for settling allsecurities transactions. A DVP system should be in place by 1992.

Recommendation 6: Same Day FundsPayments associated with the settlement of securities transactions and the servicing of securities portfolios should be made consistent across all instru-ments and markets by adopting the “same day” funds convention.

Recommendation 7: T + 3 SettlementA “Rolling Settlement” system should be adopted by all markets. Final settlementshould occur on T + 3 by 1992. As an interim target, final settlement should occur on T + 5 by 1990 at the latest, except where it hinders the achievement of T + 3 by 1992.

Recommendation 8: Securities LendingSecurities lending and borrowing should be encouraged as a method of expe-diting the settlement of securities transactions. Existing regulatory and taxationbarriers that inhibit the practice of lending securities should be removed by 1990.

Recommendation 9: Common Message StandardEach country should adopt the standard for securities messages developed by the International Organisation for Standardisation [ISO Standard 7775].In particular, countries should adopt the ISIN [International Securities Identifica-tion Number] numbering system for securities issues as defined in the ISOStandard 6166, at least for cross-border transactions. These standards should be universally applied by 1992.

20

declared failed, other considerations mightmake the Bank of Mexico reluctant to imposethese loss-sharing rules as a first step. To theextent that banks feel that the Bank of Mexicowill deal with insolvent banks without creditorsabsorbing losses, its efforts to induce partici-pants to monitor risk and allocate their expo-sures based on the establishment of interbankSPEUA credit lines with loss-sharing provisionswill not impose the degree of market disciplinethat would exist if market participants antici-pated the potential for interbank losses.

In this regard, the Bank of Mexico hasrevised its policies governing the liquidation of failed banks, with the purpose of promot-ing market discipline. In 1995, the Bank ofMexico established—through an amendment to Fondo Bancario de Protección al Ahorro(FOBAPROA)32 —explicit rules identifying cate-gories of bank liabilities that it will not guar-antee. The 1995 FOBAPROA amendment statesthat

FOBAPROA shall guarantee all liabilitiescontracted by participating financial in-stitutions, as long as said liabilities stemfrom their normal business operations,excluding:

1) subordinated debentures they mightissue,

2) liabilities resulting from illicit oranomalous acts or acts of bad faith,and

3) liabilities stemming from credit con-tracted between banks in order toguarantee liabilities payable in favorof the Bank of Mexico, provided thesaid banks participate in the fundtransfer systems administered by thecentral bank.33

Exception three of this amendment specificallyaddresses SPEUA credit lines. However, thisamendment applies only when the bank is inthe process of being liquidated. To date, noneof these exclusions has been imposed.

ConclusionThe 1994 large-value payments system

reforms implemented by the Bank of Mexico,including the introduction of parallel intradaylarge-value payments systems for funds andsecurities, have reduced payments system riskwhile keeping transaction costs relatively low.By eliminating free and unsecured daylightoverdrafts, the Bank of Mexico has reduced itscredit risk associated with direct intraday lend-

ing. In addition, with the implementation ofDVP for all securities transactions and 100 percent collateral requirements for the fundscomponent of securities transactions, credit risk in securities transactions has been signifi-cantly reduced. However, to maintain adequateliquidity at a relatively low cost for participants,the Bank of Mexico established the large-valueinterbank funds payments system, SPEUA, andthe securities clearing and settlement system,SIDV. Participants do not use good funds to settle each SPEUA or SIDV transaction but mustsettle their net positions at the end of the day.In addition, the Bank of Mexico extends unse-cured credit to allow banks lacking reserves to settle over SIAC their end-of-day clearing balances from the other systems, although theBank of Mexico strongly discourages banksfrom relying on such credit.

Payments system reforms are still beingimplemented, and further changes may be necessary for the Bank of Mexico to meet itsstated objectives. Market participants seem gen-erally pleased with the payments systemreforms implemented to date. However, theBank of Mexico may find that some of the policies designed to increase liquidity in thepayments system, such as the inability of SPEUAparticipants to decrease their credit lines duringthe day and the guarantee of payment for non-rejectable SIAC payments, unintentionally workagainst the goal of promoting market-basedintraday credit decisions. The need for some ofthese policies in Mexico may diminish over timewith renewed strength in the banking systemand a continued deepening of financial markets.

Finally, the SPEUA loss-sharing rules de-signed to help promote market discipline in thepayments system will not be fully effective ifparticipants feel that the Bank of Mexico willresolve insolvent banks without imposing losses. However, this struggle to offset the pos-sibility that government guarantees may weakenmarket discipline in the payments system is notunique to Mexico; it is, in fact, common todeveloping countries in general and even todeveloped countries. As with the other majorfinancial reforms initiated since the late 1980s,Mexico’s recent efforts to enhance the role ofmarket-based decisions in the payments systemrepresent a significant step in the right direction.

ReferencesBank for International Settlements (1997), Real-TimeGross Settlement Systems (Basle, Switzerland: Bank for

International Settlements).

32 FOBAPROA is the Mexican depositinsurance fund.

33 Bank of Mexico (1996).

FEDERAL RESERVE BANK OF DALLAS 21 FINANCIAL INDUSTRY STUDIES AUGUST 1997

——— (1996), Settlement Risk in Foreign ExchangeTransactions (Basle, Switzerland: Bank for International

Settlements).

——— (1993), Payment Systems in the Group of TenCountries (Basle, Switzerland: Bank for International

Settlements).

——— (1992), Delivery Versus Payment in SecuritiesSettlement Systems (Basle, Switzerland: Bank for

International Settlements).

Bank of Mexico (1996), The Mexican Economy 1996(Mexico City: Bank of Mexico, June 3).

——— (1995), “Disposiciones Aplicables a las Opera-

ciones Activas, Pasivas y de Servicios de la Banca

Múltiple” (Circular 2019/95), September 20 (date origi-

nally published but has been updated several times);

updated document can be found at <http://www.banxico.

org.mx/public_html/circulares/2019.html> (visited April 2,

1997).

Chakravorti, Sujit (1997), “How Do We Pay?” Federal

Reserve Bank of Dallas Financial Industry Issues, First

Quarter, 1– 6, 8.

——— (1996), “Analysis of Systemic Risk in the

Payments System,” Federal Reserve Bank of Dallas

Financial Industry Studies Working Paper no. 2–96

(Dallas, December).

——— (1995), “The Tradeoff Between Risk and Efficiency

in Foreign Exchange Settlement Systems” (Ph.D. disser-

tation, Brown University).

Díaz, Francisco G. (1996), “The Payments System in

Mexico: The Challenge of a Reform,” presentation at

Risk Reduction in Payments, Clearance and Settle-

ment Systems Symposium sponsored by Goldman,

Sachs & Co. (January).

Folkerts-Landau, David, Peter M. Garber, and Timothy

D. Lane (1994), “Payment System Reform in Formerly

Centrally Planned Economies,” in Building SoundFinance in Emerging Market Economies, ed. Gerard

Caprio, David Folkerts-Landau, and Timothy D. Lane

(Washington, D.C.: International Monetary Fund), 91–112.

———, Garry J. Schinasi, Marcel Cassard, Victor K. Ng,

Carment M. Reinhart, and Michael G. Spencer (1995),

“Effect of Capital Flows on the Domestic Financial

Sectors in APEC Developing Countries,” in Capital Flowsin the APEC Region, ed. Mohsin Kahn and Carment

Reinhart (Washington, D.C.: International Monetary

Fund), 31–74.

Group of Thirty (1992), Clearance and SettlementSystems Status Reports: Autumn 1992 (New York:

Group of Thirty).

——— (1989), Clearance and Settlement Systems in theWorld’s Securities Markets (New York: Group of Thirty).

Gruben, William C., John H. Welch, and Jeffery W.

Gunther (1993), “U.S. Banks, Competition, and the

Mexican Banking System: How Much Will NAFTA

Matter?” Federal Reserve Bank of Dallas FinancialIndustry Studies, October, 11–25.

Hancock, Diana, and James A. Wilcox (1996), “Intraday

Management of Bank Reserves: The Effects of Caps and

Fees on Daylight Overdrafts,” Journal of Money, Creditand Banking 28 (November, pt. 2): 870 – 908.

Horii, Akinari, and Bruce J. Summers (1994), “Large-

Value Transfer Systems,” in The Payment System: Design,Management, and Supervision, ed. Bruce J. Summers

(Washington D.C.: International Monetary Fund), 73–88.

Humphrey, David, and Allen N. Berger (1990), “Market

Failure and Resource Use: Economic Incentives to Use

Different Payment Instruments,” in The U.S. PaymentsSystem: Efficiency, Risk, and the Role of the FederalReserve, ed. David Humphrey (Norwell, Mass.: Kluwer

Academic Publishing), 45 – 86.

Listfield, Robert, and Fernando Montes-Negret (1994),

“Modernizing Payment Systems in Emerging Economies,”

World Bank Policy Research Working Paper no. 1336

(Washington, D.C.: World Bank, August).

New York Clearing House Organization (1996), CHIPS:Rules and Administrative Procedures 1996 (New York:

New York Clearing House Organization).

Pingitzer, Jürgen C., and Bruce J. Summers (1994),

“Small-Value Transfer Systems,” in The Payment System:Design, Management, and Supervision, ed. Bruce J.

Summers (Washington D.C.: International Monetary Fund),

106 –15.

Richards, Heidi W. (1995), “Daylight Overdraft Fees and

the Federal Reserve’s Payments System Risk Policy,”

Federal Reserve Bulletin, December, 1065–77.

Robinson, Pal E., and Dag-Inge Flatraaker (1995), “Costs

in the Payments System,” Norges Bank EconomicBulletin 66 (June): 207–16.

S. D. INDEVAL (1995), Annual Report (Mexico City:

INDEVAL).

22

Sánchez-Arroyo, Abdón (1996), “The Mexican Payment

and Settlement System: The Quality and Quantity Leaps

Ahead,” The North American Journal of Economics andFinance 7(2): 171–79.

Sato, Setsuya, and David B. Humphrey (1995), “Trans-

forming Payment Systems: Meeting the Needs of

Emerging Market Economies,” World Bank Discussion

Paper no. 291 (Washington, D.C.: World Bank, July).

Stehm, Jeff (1996), “Clearance and Settlement Systems

for Securities: Critical Design Choices in Emerging

Market Economies,” World Bank Discussion Paper no.

321 (Washington, D.C.: World Bank, April).

Vital, Christian, and David L. Mengle (1988), “SIC:

Switzerland’s New Electronic Interbank Payment System,”

Federal Reserve Bank of Richmond Economic Review,

November/December, 12–27.

Welch, John H., and William C. Gruben, “A Brief

Modern History of the Mexican Financial System,”

Federal Reserve Bank of Dallas Financial IndustryStudies, October, 1–10.

FEDERAL RESERVE BANK OF DALLASP.O. BOX 655906

DALLAS, TEXAS 75265–5906

BULK RATE

U.S. POSTAGE

P A I DDALLAS, TEXAS

PERMIT NO. 151