Assessing Systemic Risk Exposure Under Alternative Approaches for Capital Adequacy
Paul Kupiec Deputy Division Chief, Banking Regulation Division, Monetary and Exchange Affairs Department The International Monetary Fund
David Nickerson Associate Professor, Economics and Finance Departments, Colorado State University
Date of this Draft: May 5, 2001
Prepared for the Bank of England Conference, “Banks and Systemic Risk,” London, UK 25 May 2001
The opinions expressed here are those of the authors and do not necessarily reflect the views of the International Monetary Fund or those of FreddieMac. Contact information: Paul Kupiec at
[email protected], phone: 202-623-9733; fax: 202-623-8964; or David Nickerson at
[email protected], phone: 970-491-5249; fax: 970-491-2925.
Abstract A key function of capital regulation is maintaining public confidence in the ability of market participants to honor financial obligations in times of market stress. Non-bank institutions are increasingly important participants in wholesale risk-sharing markets. Consequently, ensuring public confidence may require a solvency assessment of all key institutions that interact in wholesale risk-transfer markets. After establishing the importance of non-bank counterparties in global financial markets, the paper reviews three alternative regulatory measures of capital adequacy: the Basle Banking Standards, the U.S. NAIC Model Insurance Solvency Regulation Act, and the approaches used by U.S. housing GSEs. These approaches differ markedly in the techniques and formulas that are used to measure capital adequacy. A formal equilibrium model is used to motivate an explanation for the significant differences in these regulatory solvency measures. The model includes asymmetric information that precludes trading of some financial contracts because of uncertainty about counterparty solvency. In the context of the model, important measurement issues arise when the solvency of counterparties are assessed using different regulatory capital standards. Significant bias is introduced if the functional regulatory capital requirements for a bank are used to measure the solvency of insurers or intermediaries that primarily offer contingent claim liabilities. In the formal model, the asymmetric information imposed limits on contract trading are identified as a form of systemic risk. Regulatory capital requirements cannot be used to attenuate systematic risk unless safety net subsidies are increased. While systemic risk can be eliminated by requiring complete transparency in the model economy, firms offering standard debt and equity claims have no incentives or mechanism to credibly signal information to market investors. Absent a private market signaling mechanism it is demonstrated that a pre-commitment approach to regulation---which requires lower regulatory capital requirements and ex post penalties---can remove systemic risk and allow trade in contracts that, while still subject to limited restrictions, are close to the set of contracts traded in the complete transparency solution.
1
I. Introduction Systemic risk can be defined as the potential for a relatively modest economic shock to induce disproportionate volatility in the prices of financial assets, significant reductions in corporate liquidity, substantial mark-to-market losses, and perhaps even unanticipated bankruptcies among financial institutions. A key feature in the propagation of such a systemic shock is acute uncertainty regarding an institution’s ability to satisfy its immediate payment obligations and a simultaneous inability of counterparties to hedge such risk. If a financial institution of sufficient importance experiences a “crisis of confidence” in the institution’s ability to make payments, this in turn may raise questions about the liquidity and solvency of its major counterparties and potentially lead to a loss of market liquidity and the disruption of financial trade. Owing to market incompleteness, these informational externalities constrain market liquidity and, potentially, real economic activity, as market participants are uncertain about the financial stability of their counterparties. Uniform regulatory capital requirements may reduce systemic risk if credible enforcement introduces a measure of confidence in the solvency of financial counterparties. This fundamental objective of minimum solvency standards is clearly articulated in the U.S. Federal Deposit Insurance Corporation’s Division of Supervision Manual of Exam Policies:1 “Capital provides a measure of assurance to the public that an institution will continue to provide financial services even when losses have been incurred, thereby helping to maintain confidence in the banking system and minimize liquidity concerns.” The 1988 Basle Accord and its subsequent and proposed amendments have established a recommended uniform minimum regulatory capital standard for internationally active banking institutions. Banks, however, are not the only intermediaries that are active in international risk-exchange markets. Insurance and reinsurance companies, energy firms, corporate conglomerates, and other non-bank financial intermediaries are active participants in global risk-sharing markets. Linkages between insurance products and equity and credit derivative markets have, for example, increased insurance institutions’ participations in wholesale risk-transfer markets. Deregulation and the associated need to manage energy price risks have expanded the derivative market activities of energyrelated firms. The potential importance of these firms has been demonstrated recently as shortcomings in deregulation strategies and improper hedging by California electric utilities lead to bankruptcy fears and serious concerns about credit risk in the U.S. commercial paper markets.2 While counterparties to banks, these non-bank firms are subject to widely differing regulatory capital standards or indeed no regulation at all. 1
February 2000. See Capital, Section 2.1, available at www.fdic.gov/regulations/safety/manual/99CAPITA_main.htm. 2 After PG&E Corporation postponed, on 12 January, release of their fourth-quarter earnings data and Southern California Edison announced on 16 January the suspension of payments for all outstanding debt obligations, Standard and Poors downgraded credit ratings for both utilities. On 6 February, S&P issued a
2
While systemic risk historically has been almost exclusively the concern of central bankers and banking regulatory agencies,3 modern financial engineering technology and the increasing use of wholesale risk-sharing markets by non-bank institutions mandate a more inclusive approach toward financial stability assessment. An erosion of confidence in a key non-bank financial intermediary or corporation could have significant consequences for systemic risk. Wider concerns about the potential systemic importance of non-bank institutions are clearly reflected in the official reports published by international coordination groups. For example, in a recent report, the Group of Ten concludes that, “… non-bank financial institutions, not just banks, have the potential to be sources of systemic risk…even a medium-sized foreign bank (or perhaps a non-bank financial institution) from a large nation would be a potential source of instability to a relatively small host country.”4 The Group of Ten’s findings echo those of the influential “Tietmeyer Report”:“Systemic threats can also arise from unsupervised financial service providers, notably major highly leveraged institutions…Additionally, spill-over effects could arise from difficulties at non-bank financial institutions and large insurance companies.”5 The Tietmeyer Report ultimately lead to the formation of The Financial Stability Forum (FSF), a group composed of the Finance Ministers and Central Bank Governors of the G7 countries. The FSF was chartered to: (1) to assess vulnerabilities affecting the international financial system; (2) to identify and oversee action needed to address these vulnerabilities; and (3) to improve co-ordination and information exchange among the various authorities responsible for financial stability.6 The FSF charter clearly recognizes the importance of monitoring non-bank financial intermediaries and institutions in assessments of financial system stability. Notwithstanding the fact that, in some circumstances, financial system stability may hinge on the solvency of non-bank market participants, there has been relatively little discussion of how to gauge the solvency of non-bank market participants relative to the regulatory requirements for internationally active banks. Several important non-bank market participants are regulated and subject to supervisory minimum capital regulation, but the capital regulations that establish minimum solvency standards for these types of firms often differ markedly from those used for internationally active banks.
public statement criticizing the California Public Utilities Commission, the California State Legislature and the Gray Davis, the California Governor, for failing to reassure debt holders that the utilities would honor their credit and off-balance sheet obligations. See Kelleher, Orange County Register, 12 January, 2001; City News Service, 16 January 2001; Coleman, Associated Press Newswire, 9 March 2001; and Associated Press State and Local Wire 7 February 2001 for additional details. 3 For example, the European Central Bank, and the Central Banks of Finland, Sweden, and Norway conduct (and publish) a financial system stability reviews with emphasis on banking sector developments. 4 “Report on Consolidation in the Financial Sector,” Group of Ten, January 2001, p.4. 5 Report by Hans Tietmeyer, President of the Deutsche Bundesbank, to the Finance Ministers and Central Bank Governors of the G-7, February 11, 1999. 6 These objectives are taken verbatim from the Financial Stability Forum’s web page, www.fsforum.org.
3
In the United States, insurance companies are subject to minimum solvency requirements set by State Insurance Commissions in each state in which these companies operate, while Government Sponsored Enterprises (GSEs), including FannieMae and FreddieMac in the secondary mortgage market, are subject to regulatory capital requirements imposed by a national regulatory agency (OFHEO).7 The supervisory capital guidelines that apply to the GSEs require that the institution remain solvent under specific economic stress scenarios, while virtually all US State Insurance Commissions require that both life and property-casualty insurance firms conduct stress tests, under multiple interest rate scenarios, for their long-lived assets, using procedures established by the National Association of Insurance Commissioners (NAIC). In the United Kingdom, the Financial Services Authority (FSA), acting on behalf of HM Treasury, conducts direct supervision of insurance companies, buildings societies and friendly societies operating in the United Kingdom, using a combination of risk-based capital rules and dynamic ruin-probability scenarios.8 Based on guidelines issued by the International Association of Insurance Supervisors (IAIS), the European Union is adopting regulations of both life and non-life insurance firms, as well as regulatory guidelines for financial conglomerates, that combine stress-testing with traditional riskbased capital ratios. Finally, even those corporations, such as General Motors and British Telecom, which are active in international financial markets but not under the direct authority of financial regulators, use a variety of risk-measurement procedures to assess their capital adequacy.9 The use of alternative regulatory approaches for setting solvency standards need not engender systemic risks. Issues of regulatory arbitrage or so-called “level playing field” concerns, however, often lead to policy debates that require measures of the comparative rigor of alternative capital regulations. Given the historical importance of banks in systemic crisis management, such comparison are often facilitated by measuring the capital adequacy of non-bank financial institutions using banking regulatory standards.10 While it has not been widely recognized, such a comparison can produce highly 7
Other US GSEs include the Farm Credit Banks, the Student Loan Marketing Association, the Federal Home Loan Banks and the Federal Agricultural Mortgage Corporation, all supervised by different agencies and subject to different capital adequacy regulations, as described in General Accounting Office Report to Congress, Government Sponsored Enterprises: A Framework for Limiting the Government’s Exposure to Risks (1991). 8 Although in Parliament for more than two years, the Financial Services and Markets Act 2000, to be fully implemented in the second half of 2001, will merge the traditional regulatory functions of the Building Societies Commission (BSC), the Friendly Societies Commission (FSC) the Registry of Friendly Societies (RSC), the Insurance Directorate (ID) of the Department of Trade and Industry, and other regulators, under the auspices of the Financial Services Authority. While conforming to general EU guidelines, the FSA will utilize both risk-based capital standards and stress-testing methods in their assessment of the capital adequacy of insurance firms, building societies and credit unions. See Financial Services Authority, Insurance Draft Interim Prudential Sourcebook (2000) and Interim Prudential Sourcebook for Building Societies (2000) for details. 9 Such firms include DaimlerChrysler, Dell and Motorola, for example, each of which uses one or more VAR models and various econometric forecasting procedures to assess their solvency risk and volatility of their cash-flow. See Business Wire, 17 January, 2001. 10 See for example, the General Accounting Office Report to Congress, Government Sponsored Enterprises: A Framework for Limiting the Government’s Exposure to Risks (1991).
4
misleading assessments of solvency risk and may create substantial allocational inefficiencies if counterparties or supervisors are mislead by conflicting measures of solvency. There are many alternative approaches that are used to assess capital adequacy. Perhaps the most well-known approaches are the Basle Bank Supervisors approach and standards that focus on stress testing. The Basle approach focuses on risk-weighted asset ratios and value-at-risk measures while the GSEs capital requirements and those widely used in the insurance industry focus on minimum ratios and a stress test survivorship requirement. To the extent that key market participants are evaluated under different approaches for determining capital adequacy, any systemic risk assessment procedure must be capable of reconciling differences in capital adequacy standards. This compatibility requirement is problematic, as these regulatory measures of capital adequacy are fundamentally not comparable.11 The incompatibility arises because, contrary to basic portfolio theory, the Basle Bank Supervisors approach for assessing capital adequacy focuses only on the assets and offbalance sheet positions of a bank, and ignores the stochastic characteristics of the bank’s debt liabilities. This may not be a problematic feature for determining bank solvency as bank balance sheet liabilities typically do not include embedded contingent claims. The stress tests that characterize the insurance industry and GSE regulatory capital standards, however, typically account for the risks inherent in the market values or cash flows associated with a firm’s liabilities. The consequence is that the capital adequacy measures produced by the Basle standards are not comparable to those produced by stress test approaches. While international bank capital regulations may provide an accurate assessment of many banking institution’s risk profiles, the failure to recognize information on the characteristics of debt liabilities may be problematic if the Basle approach is applied in a non-bank setting. If the structure of an institution’s debt liabilities are managed so that the net market value of a firm’s assets and liabilities is relatively insensitive to changes in the value of macro-financial factors, its risk of insolvency could be minimal and yet it may appear to be under-capitalized under the Basle capital standards. Many non-bank financial institutions use structured debt issues as a form of risk management. Examples include the significant callable debt issues of the GSE’s and in the insurance industry, catastrophe bonds, reinsurance contracts, and indeed even the insurance contracts themselves. This paper considers the issues associated with capital adequacy based measures of systemic risk in markets in which key players follow different approaches for assessing capital adequacy. Section II documents the participation of non-bank intermediaries in global financial markets. Section III reviews capital regulations for internationally active banks and for selected key non-bank financial institutions, including capital regulations for life and property-casualty insurers in the United States, and for the GSEs in the
11
See Kupiec (2000), “Stress Tests and Risk Capital,” The Journal of Risk, Summer, pp. 22-46.
5
United States.12 Section IV, introduces a formal equilibrium model that provides basis for explaining the observed differences between regulatory measures of capital adequacy. We consider the optimal functional regulatory capital requirements for a bank and an insurance intermediary with identical asset portfolios and identify significant differences in these alternative regulatory capital standards. While these measures are optimal given the underlying regulatory objectives, they set solvency standards that are inherently noncomparable. For example, these differences imply significant solvency measurement bias if a bank regulatory standard is applied to measure the solvency of an intermediary that funds itself with a significant amount of liabilities that included explicit or embedded contingent claims. Section V investigates links between minimum regulatory capital standards and systemic risk in the context of the formal model developed in Section IV. Here we formally demonstrate that regulatory capital regulation alone may not be able to mitigate the systemic risk that arises from information asymmetry. Regulation may, however, play a role in limiting systemic risk. It is demonstrated that a pre-commitment approach for minimum capital regulation can remove systemic risks by creating a credible signaling mechanism that removes the moral hazard issues that limited financial contract trading. A final section concludes the paper.
II. Bank and Nonbank Participation in Financial Markets: Evidence from Derivatives Markets Firms in many different industries have become participants in international financial markets, owing to increased globalization of trade, the liberalization of regulations on direct foreign investment, and rapid technological advances in communications and financial engineering technology.. The growth in derivatives transactions has greatly expanded the means to manage risk available to nonbank firms. Multinational firms and firms engaged in real international trade are increasingly using financial markets, and especially derivatives transactions, to manage their asset and liability risk. International markets for derivatives have grown rapidly over the last two decades and offer a wide variety of contracts, ranging from standardized derivatives traded on organized exchanges to bilateral trade of specialized, over-the-counter (OTC) contracts, created for buyers by derivative dealers.13 Table 1 presents data on the notional stock, in US dollars, of all derivatives contracts open at year-end 1999 by those twelve of the largest twenty US firms which reported
12
This section, while abbreviated, still includes a mind-numbing set or regulatory details. Notwithstanding the practical importance of details, a reader pressed for time can skim this section and still follow our more general arguments. 13 Since many derivatives transactions, including futures and forward contracts (such as those traded to hedge foreign currency risk in international commodity trade) generate contingent cash flows rather than acting as a liability or asset on a corporate balance sheet, derivatives are often classified as off-balancesheet (OBS) contracts, and are only indirectly accounted for in capital regulations for banks and other types of financial intermediaries.
6
derivatives positions in SEC filings at this date.14 .The average notional value of all derivatives contracts per firm open at this date was 3,447,700,833,333 USD. The data in Table 1 show that, while Chase Manhattan, whose banking activities are regulated by the Federal Reserve Board, was engaged in the largest volume of derivatives transactions, financial firms such as J.P. Morgan, Morgan Stanley and Merrill Lynch, whose primary regulator is the SEC, comprised four of the largest six derivatives users in the sample. FannieMae and FreddieMac, regulated by OFHEO, Metropolitan Life, regulated by U.S. state insurance commissioners, and General Motors, regulated by the SEC, also reported substantial volumes of derivatives contracts.15
Table 1 FIRM
ASSETS
DERIVATIVES CONTRACTS
FannieMae FreddieMac Citigroup Chase Manhattan Morgan Stanley Merrill Lynch J.P. Morgan and Company First Union Corporation Goldman Sachs General Motors Metropolitan Life Insurance Wells Fargo Lehman Brothers Holdings
$575,167,000,000 $386,684,000,000 $716,937,000,000 $406,105,000,000 $366,967,000,000 $328,071,000,000 $260,898,000,000 $253,024,000,000 $250,491,000,000 $274,730,000,000 $225,232,000,000 $218,102,000,000 $192,244,000,000
$274,200,000,000 $424,244,000,000 $7,373,2000,000,000 $12,936,600,000,000 $3,404,000,000,000 $3,939,000,000,000 $8,876,300,000,000 $190,280,000,000 $809,205,000,000 $6,300,000,000 $28,834,000,000 $220,897,000,000 $2,895,650,000,000
Table 2 presents data on the disclosed notional stock, in pounds sterling, of derivatives contracts for year-end 1999 by eight of the largest firms, again measured by asset size, on the FTSE index. Although Barclays’s usage of derivatives is several orders of magnitude higher than all other reporting firms in this sample, several nonbank financial conglomerates, including Old Mutual and the Halifax Group, reported derivatives usage ranging from 11% to 26% of the sample mean notional contract value of 283,440,957,642 pounds. Nonfinancial firms, such as British Telecom, Glaxo Wellcome, Unilever and Royal Dutch Shell, also reported derivatives usage of substantial magnitude, in absolute terms, ranging respectively from one hundred million pounds to over fifteen billion pounds. Table 3 presents data on the disclosed notional stock, in US dollars, of derivatives contracts open at year-end 1999 by those six of the eleven largest international insurance and reinsurance firms that report derivatives activity in 1999. Although the mean notional contract value for these insurance firms, 9,272,886, 337 USD, was modest relative to that for the banking and nonbank firms in the sample of U.S. firms reported in 14
Firm size is measured by year-end 1999 book value of assets. Specifically, the derivatives stock is measured by the book-value of the sum of all derivatives contracts written and purchased, as of year-end 1999, in USD, in accord with FASB accounting convention. 15 Derivatives usage by several nonfinancial firms in the sample, including Ford Motors, are unavailable from SEC filings or corporate financial statements.
7
Table 1, it compares, in order of magnitude, to the mean for nonbank financial and nonfinancial firms reported in both Tables 1 and 2.
Table 2 FIRM
ASSETS
DERIVATIVES CONTRACTS
Royal Dutch Shell British Telecom Barclays PLC Unilever Abbey National Group Halifax Group PLC Old Mutual Glaxo Wellcome
70,449,484,077 27,962,000,000 254,793,000,000 1,136,3745,320 180,744,000,000 162,078,000,000 38,956,000,000 6,080,000,000
15,248,806,077 100,000,000 1,806,304,000,000 6,390,855,058 328,804,000,000 75,529,000,000 32,577,000,000 2,574,000,000
FIRM
ASSETS
DERIVATIVES CONTRACTS
Allianz Allied Zurich PLC Munich RE Swiss RE Marsh-McLennan AON
38463581268 9562778240 180739826400 68211244678 13021000000 21132000000
8679992986 10399036080 2993239620 26671049336 387000000 6507000000
Table 3
Owing to the recent implementation of FAS 133 derivatives reporting requirements for SEC filings in the United States, derivatives usage data for nonbank firms is rare and available only for limited samples.16 Firms operating in the European Union have to date also been reluctant to provide such data in public filings. Insurance firms in the United States, however, are required to report derivatives transactions to the state insurance commission in each state in which they conduct life or property/casualty books of business. This data, compiled by the National Association of Insurance Commissioners in the U.S., offers evidence on the role of banks as counterparties in derivatives trading with insurance firms. Tables 4, 5 and 6 report the extent of usage of derivative securities, of all varieties, by a sample of 1,207 life insurance and 2,063 property-casualty insurance firms operating in the United States during 1994, as well as all domestic US and foreign banks that acted as counterparties to insurers in this year in the exchange of OTC contracts.17 Table 4 reports summary statistics for alternative types of derivatives contracts opened by both life and 16
Annual SEC filings for 2000, the first date for which FAS133 will be in effect, are not yet available at the time of this writing. 17 This sample consists of the universe of all 1,760 life and 2,707 property-casualty insurance firms that filed regulatory annual statements with the NAIC for 1994, either directly or through state insurance commissioners, less firms identified as having zero assets or equity values, as discussed in Cummins, Phillips and Smith (1996). The authors of the current paper have procured analogous data for fiscal year 2000, released this quarter by the NAIC, and will incorporate these current values in future drafts. The extent and pattern of derivatives usage by insurance firms in 2000, however, is even more illustrative of the extent of such trading than the comparable data for 1994.
8
property-casualty insurers during 1994, in terms of the number of users and the size of these contracts. Buying puts and writing calls account for the largest amounts of exchange, in terms of participants, during the year, with calls also exhibiting the largest mean contract value.18 TABLE 4 CONTRACT TYPE
NUMBER CONTRACT OF VALUE: USERS MEAN
CONTRACT VALUE: STANDARD DEVIATION
50 62 19 1 8
173,408,546 664,872,903 585,091,711 89,000,000 832,156,550
404,317,751 3,432,286,046 1,450,341,545
121 62 4 1
1,337,880,976 357,705,270 570,665,200 141,352,155
10,687,000,591 1,148,541,536 787,748,330
161,883,598,051 11,446,568,625 2,282,660,800 141,352,155
71 4 39
412,856,106 402,500,123 831,412,118
960,521,239 706,780,951 960,521,239
29,312,783,517 1,610,000,492 29,312,783,517
54 62
817,476,057 1,072,196,368
3,195,160,149 4,389,527,625
44,143,707,077 66,476,174,791
268
1,557,751,717
10,100,039,739
417,477,460,263
TOTALS
Options Purchased Calls Puts Caps Corridors Floors
1,222,401,871
8,670,427,298 41,222,119,955 11,116,742,511 89,000,000 6,657,252,403
Options Written Calls Puts Caps Floors
Swaps, Collars and Forwards Opened Swaps Collars Forwards
Futures Opened Long Position Short Position
All Contracts
Table 5 illustrates characteristics of the distribution of derivative contract counterparties for the insurance firms in the sample. The mean number of counterparties for the 138 insurance firms was 4.72 and the concentration of exposure, as measured by a standard Herfindahl-Hirschman index for counterparty transactions, was 62%.19 Table 6 lists banks which are counterparties to the insurance firms in our sample, ranked by the value of contracts outstanding during 1994. The 1994 data indicate that the magnitude of derivatives exposures between insurance firms and banks is large. 18
Although not reported here, the mean number of open positions significantly exceeds the median for most contract types (including calls and puts), indicating that a greater than proportional share of activity is concentrated in a relatively small number of (large) insurance firms. 19 The Herfindahl-Hirschman index is, for each insurance firm transacting in OTC contracts, the sum of the squared values of the ratio of the total notional principal with counterparty i, relative to total notional principal with all counterparties for this firm, across all counterparties with which the firm dealt in 1994. A high percentage indicates that the firm deals with relatively few counterparties, with a maximal value of one indicating that all OTC contracts were with a single counterparty.
9
TABLE 5 NUMBER OF INSURERS 138 138
Counterparties per Insurer Counterparty Concentration
MEAN
STANDARD DEVIATION
MIN
MAX
4.72 62%
5.55 .353
1 6.7%
27 100%
TABLE 6 COUNTERPARTY
TOTAL NOTIONAL INSURERS USING CONTRACTS COUNTERPARTY OUTSTANDING
PERCENT OF TOTAL INDUSTRY OTC CONTRACTS
Credit Suisse Citibank Deutsche Bank Republic National Bank First Chicago Chase Manhattan Swiss Bank Barclays Bank Chemical Bank ABN-Amro Bank Bank of America Bank of Montreal Royal Bank of Canada Bank of Tokyo First Boston Nomura International Bank of New York Bank of Nova Scotia CIBC Sumitomo Bank NationsBank Lloyds Bank First National Bank of Chicago Toronto Dominion Bank
27 18 14 2 8 15 12 20 21 3 10 6 11 2 7 6 6 8 8 2 4 4 7 5
2.1% 2.1% 1.8% 1.7% 1.6% 1.6% 1.3% 1.2% .9% .9% .7% .7% .6% .4% .4% .3% .3% .3% .2% .2% .2% .1% .1% .1%
2,037799,372 2,018,544,392 1,724,412,763 1,680,187,168 1,558,846,084 1,540,110,495 1,245,353,905 1,165,776,187 850,294,905 850,000,000 707,078,653 637,483,190 538,732,814 422,000,000 357,420,000 337,144,300 327,133,959 258,340,286 218,152,334 205,000,000 161,600,000 132,950,000 125,797,383 104,910,747
While data limitations are significant, available data suggests that non-bank firms have substantial exposures in derivative markets and are important counterparties to banks. While derivatives usage offers many benefits to these non-financial firms, it may create complications in regard to the ability of investors to assess the creditworthiness of counterparties to such transactions. When transactions involve firms facing different capital regulations, or the absence of such regulations, the costs of assessing counterparty risk may increase and result in restrictions on efficient risk trading.20 We return to these issues in Section IV.
20
Financial reporting requirements, such as FASB 115, which requires mark-to-market accounting for fixed income securities held by insurance firms, and FASB 119 and 133, which require public reporting of derivatives usage by insurance companies and other nonbanking firms, have increased transparency in this context, but counterparty risk, especially in international transactions, remains substantial.
10
III. Regulatory Approaches for Capital Adequacy A.
US Bank/Thrift Capital Standards and the Basle Framework
The regulatory capital requirements that apply to bank and thrift institutions in the U.S. are a ratio-based standard that is consistent with the 1988 Basel Capital Accord and its amendments.21 Minimum capital requirements are defined with respect to an institution’s risk-based capital ratio that is calculated by dividing its qualifying capital (the numerator of the ratio) by its risk-weighted assets (the denominator). Total risk-weighted assets (the denominator) include a capital requirement for credit risk (its risk-weighted assets) plus a capital requirement for market risk, which is calculated using either a standardized approach or an internal models-based approach. An institution’s Tier 1 capital (so-called core capital) includes the book value of an institution’s issued and fully paid ordinary shares, perpetual non-cumulative preferred shares, disclosed reserves, and minority interests in less than wholly-owned subsidiaries. Tier 2 capital includes undisclosed reserves, revaluation reserves, general loan loss reserves (up to 1.25 percentage points of 8 percent of total risk-weighted assets), unsecured subordinated fully paid up participating debt, cumulative preferred shares, mandatory convertible debt, and term subordinated debt with original maturity greater than 5 years. Tier 2 capital is limited to the amount of a bank’s Tier 1 capital. Tier 3 capital includes qualified short-term subordinated debt.22 Tier 3 capital is limited to the magnitude of the bank’s market risk and cannot be used to support credit risk capital requirements. Minimum capital requirements are two-fold: the bank’s Tier 1 capital must be at least 4 percent of total risk weighted assets and the bank’s total risk-weighted capital ratio must be at least 8 percent of its total risk-weighted assets. In the U.S., the Basel minimum capital requirements establish an absolute regulatory minimum and expanded bank powers engender higher regulatory capital minimums that are typically negotiated at the time that extended powers are granted.23 Although recognized by regulatory agencies, interest rate risk is not explicitly addressed under U.S. bank capital standards nor is it captured under these standards for traditional held-to-maturity assets. Bank examiners do have discretion to recommend increased capital for excessive interest rate risk exposures, but this is not formally measured nor are such recommendations often made in practice. For example, while the U.S. Office of Thrift Supervision (OTS) has adopted a final rule that requires an explicit interest rate risk 21
U. S. banks and thrifts operate under slightly different capital requirements, in particular, with regard to interest rate risk for held-to-maturity assets. Moreover, the U.S. bank/thrift capital regulations differ is some details from Guidelines recommended by the Basle Accord. 22 Original maturity greater than 2 years, unsecured, and fully paid, non-redeemable before maturity, nonredeemable at maturity if bank has a capital shortfall. 23 In the U.S., these extended powers include underwriting securities, selling insurance, and expanded trade in derivatives.
11
capital requirement for thrifts with excessive interest rate risk, enforcement of this rule has been suspended for competitive purposes until U.S. banking regulators adopt an explicit interest rate risk capital rule. Pending this, the OTS relies on examiner discretion in recommending any additional capital requirement for interest rate risk on a thrifts held-tomaturity portfolio.24 Capital requirements for the credit risk of traditional held-to-maturity assets are assigned by risk-weighting gross balance sheet carrying values. Although there are proposed changes to the credit risk weighting scheme, the current set of risk weights established by the 1988 Capital Accord are detailed in Table 7. Credit risk for off-balance sheet activities are calculated as follows. Derivatives and other off-balance sheet exposures are translated into balance sheet equivalent values using a specific set of rules. The balance sheet equivalent amounts are risk-weighted according to the underlying counterparty credit risk category. The conversion factors are: 100 percent for traditional off-balance sheet banking guarantees; the sum of the mark-to-market value (so-called current exposure) plus an add on for potential future exposure for derivatives that is a percentage of the notional contract amount. The derivative add-on factors are detailed in Table 8. Capital relief is granted for contracts that are subject to legally recognized netting agreements. Market risk capital requirements apply to positions in a bank’s trading book. These assets are marked-to-market for accounting purposes. Market risk capital requirements apply to all U.S. banks with total assets exceeding $5 billion and either: trading activity exceeding 3 percent of total notional asset value or derivative contracts with notional value in excess of $5 billion. Alternatively, any institution with total assets of $5 billion or less and trading activity that represents at least 10 percent of total assets must comply with the market risk capital requirement. The market risk capital requirement has a number of qualitative features including: a bank’s risk management function must be independent of trading operation; a bank’s internal value-at-risk model must be approved by supervisor annually; a bank’s risk model must be fully integrated in operations; and banks must conduct stress testing exercises and back testing exercises in addition to approved value-at-risk model procedures. If a bank meets these qualitative standards, its market risk capital charge is given by a supervisory determined multiple (between 3 and 4) of a specific internal model value-at-risk estimate (99 percent, 10 day horizon, 1 year sample for estimation) plus and add-on factor for specific risk.25 The specific risk add-on rules generate a capital requirement for risks that are not adequately measured in the bank’s value-at-risk model. For many banks, the specific risk capital charge reportedly is a large component of its market risk capital requirement.
24
O.T.S. thrift examination staff have, however, been encouraged not to make such a recommendation, for example, owing to the competitive disadvantage this would place on thrifts relative to banks. See Nickerson (1998), OTS internal memorandum. 25 See Kupiec and O’Brien (1997) for more details.
12
The market risk capital charge applies to all positions in the trading account and all foreign exchange positions and commodity positions regardless of whether they are in the bank trading account. Positions in the trading account, except foreign exchange and OTC derivatives, are excluded from the credit risk capital requirement.
Table 7: Current Credit Risk Weights Risk Weight
Assets
0%
Cash, claims on central governments and central banks in domestic currency, claims on OECD central governments of central banks, claims collateralized by cash of OECD central government securities or guaranteed by OECD central governments. Claims on multilateral development banks, claims on banks incorporated in the OECD and loans guaranteed by such entities, claims on or guaranteed by other banks with residual maturity less than 1 year, claims on nondomestic OECD public sector entities, excluding central governments and loans guaranteed by such entities. Claims on the private sector, claims on banks incorporated outside the OECD with residual maturity greater than 1 year, claims on non-OECD central governments, claims on commercial companies owned by the public sector, real estate and other investments, fixed assets (plant and equipment), capital instruments issued by other banks, all other assets. Claims on domestic public-sector entities, excluding the central government, and loans guaranteed by such entities.
20%
100%
Other (0,20,50, or 100% at national discretion)
Table 8: Potential Future Exposure Conversion Factors Residual Maturity 1 year or less 1 to 5 years Over 5 years
Interest Rate 0.0% 0.5% 1.5%
Exchange Rate and Gold 1.0% 5.0% 7.5%
Equity Metals Except Gold 6.0% 7.0% 8.0% 7.0% 10.0% 8.0%
Other Commodities 10.0% 12.0% 15.0%
B. US Capital Standards for Life and Property/Casualty Insurers Both life and property-casualty insurance firms adhere to risk-based capital regulations, which differ from those applying to banks and thrifts in three ways. First, they explicitly include interest rate risk and are applied to liabilities as well as assets. Second, they explicitly include provisions for the effects of correlation between securities held in the
13
insurer’s portfolio. Finally, they are designed and administered at a state, not a national, level. Unlike other financial services industries in the United States, capital requirements for both life and property/casualty insurance firms are, since passage of the McCarranFerguson Act in 1945, the responsibility of individual states in which the firms operate. Each state implements capital requirements and operates insurance guarantee funds through its Office of the Insurance Commissioner. Regulatory initiatives and actual implementation of regulations are, however, coordinated by the states through the National Association of Insurance Commissioners (NAIC), which published its Risk Based Capital for Insurers Model Act, Model #312-1, in 1995. Forty-seven of fifty US states have adopted versions of this Model Act, the provisions of which we describe below.26 Insurance firm, as intermediaries finance the purchase of assets through issuing specialized types of state-contingent debt contracts (insurance policies). Assets held by both life and property/casualty insurers include corporate stock, government and tradable corporate debt, and, primarily for life insurers, private bond issues and mortgages and mortgage-backed securities (MBSs). The acquisition of assets is financed by the issue of liabilities which involve the purchase of certain types of specified risks, posed by random losses to the values of life, health, property and certain forms of legal liability and borne by individuals, business firms and some public institutions, in exchange for the receipt of premia payments on a fixed basis, over either a fixed or variable term to maturity.27 Although insurance policies are the predominant type of debt contract written by insurance firms, such firms also issue a variety of other types of debt obligations, including annuities, private pension funds and guaranteed investment contracts (GICs).
26
Based on reviews performed as part of the NAIC Accreditation Program, 47 of the U.S. insurance jurisdictions had, by 1999, adopted laws, regulations or bulletins that are considered to be substantially similar to the NAIC’s Model Act #312-1. In addition to the risk-based capital requirements, the NAIC proposal also describes model “prompt corrective action” legislation that grants automatic authority to the state insurance commissioner to take specific actions against an insurance firm, contingent on the degree of noncompliance exhibited by that firm. See NAIC Model Act #312-1, Section 4, pp. 312-6-8. Similar regulations have been proposed by the International Association of Insurance Supervisors, through the Conference of Insurance Supervisory Authorities of the Member States of the European Union; see IAIS Issues Paper (2000), “On Solvency, Solvency Assessments and Actuarial Issues.” 27 Although insurers diversify (pool) a portion of so-called “underwriting” risk by writing such contracts for large numbers of policyholders whose individual losses are relatively uncorrelated, diversification offers only an incomplete hedge against insurance/underwriting risk. NAIC RBC regulations are intended to account for the remaining portion of the solvency risk posed by such liabilities, although these regulations are limited by implicit distributional assumptions on actual losses. In addition, although reinsurance is still the dominant means of hedging insurance risk, a nascent market in insurance derivatives has developed in the last decade. The first exchange-traded insurance derivatives are the “catastrophe” insurance futures and options, introduced by the Chicago Board of Trade in 1993, joined in 1995 by catastrophe bonds. See Cummins and Geman (1995) for a description of these securities and Hogan and Nickerson (2000) for an analysis of their hedging properties.
14
Insurance policies exhibit distinct patterns of cash flows from other types of debt obligations.28 Owing to the potentially disparate pattern of the cash flows generated by these liabilities, relative to their assets, a primary task of both life and property-casualty insurers is to manage the duration and convexity of their portfolios, as well as the default risk on assets. Relative to the capital regulations applicable to banks, thrifts and comparable intermediaries, insurance regulations explicitly recognize that solvency risk is a function of the entire portfolio of an insurance firm, inclusive of assets and liabilities. The NAIC Model Act regulations, consequently, address the stochastic properties of both assets and liabilities appearing on the balance sheets of such a firm. The NAIC provides for separate, but analogous, risk-based capital models for life insurance and property/casualty insurance firms; we explicitly describe only the former here.29 Two distinct features of the NAIC RBC regulations are the classification of both asset and liability risks, and the formalized usage of explicit stress-testing to evaluate interest rate risk faced by life insurers. The common classes of risks identified in both life and property/casualty regulations include “Asset Risk” (C1); “Insurance Risk” (underwriting risk) (C2), “Interest Rate/Credit Risk” (C3), and “Business Risk” (C4).30 Asset risk (C1) is analogous to Basle risk-adjusted asset values for banks. Relative asset risk weights for life and property-casualty firms appears below, in Table 9.31 Insurance risk (C2) is simply underwriting risk. Distinct risk-based capital factors are used for both life insurers, to calculate the surplus needed to provide for excess claims, owing to random fluctuations as well as an allowance for measurement error in pricing future claims, and for each line of business for property-casualty insurers, with specified adjustments for firm experience in each line.32 Property-casualty insurers also calculate 28
Debt obligations of insurers with unusual cash flow patterns include universal life policies, where policyholders hold options on the magnitude and timing of premium payments, variable life policies and annuities, the cash flows of which are often linked to equity or other indices, and single-premium deferred annuities and GICs. 29 The NAIC also provides a third model for health organizations, but this model is similar to that for property/casualty firms and we do not describe it here. 30 More specifically, risk calculations for life insurance companies include C0 – Asset Risk – Affiliates; C1 Asset Risk – Other, Plus Reinsurance; C-2 Insurance Risk; C-3a Interest Rate Risk; C-3b Health Provider Credit Risk; C-4a Business Risk, Guaranty Fund Assessment Risk; and C-4b Business Risk, Health Administration Expense Risk, while the formula for property/casualty insurance firms expand to six the number of distinct risk types, including three subcategories for asset risk. 31 Asset risk is further subdivided, in the NAIC Model Act, into two subcategories. Asset risk – affiliate (CO) is the risk of default of assets for affiliated investments. This represents the risk-based capital requirement of the downstream insurance subsidiaries owned by the insurer and applies factors to other subsidiaries. The parent is required to hold an equivalent amount of risk-based capital to protect against financial downturns of affiliates. For life companies, off-balance sheet items are included in this risk component and these include non-controlled assets, derivative instruments, guarantees for affiliates and contingent liabilities. Asset risk – other (C1) represents the risk of default for debt assets and loss in market value for equity assets. Fixed income assets include bonds, mortgages, and short-term investments. Equity assets include common and preferred stock, real estate, and other long-term assets. All insurance companies are also subject to an asset concentration factor that reflects the additional risk of high concentrations in single exposures. 32 See NAIC, Risk Based Capital (RBC) for Insurers Model Act ,#312-1 (1994) for specific risk weights.
15
risk-based capital for so-called “excessive growth,” based on historical averages calculated by the NAIC.33 Interest rate risk (C3) is the risk of losses due to changes in interest rate levels and their effect on differentially timed cash flows from assets and liabilities.34 Insurers divide liabilities into alternative risk classes; for life insurers, these are low (.005 capital charge), moderate (.01 capital charge) and high (.02 capital charge), with liability types for each class specified by state insurance commissioners and based on contractual withdrawal provisions.35 Life insurance firms, which exhibit the greatest disparity in maturities, can, under certain conditions, adjust this value for (C3) by filing an approved report with the NAIC on the performance of their portfolios under a stress test involving a choice of either twelve or fifty alternative interest-rate scenarios.36 Business Risk (C4) is measured, for the purposes of capital regulations, by the maximum potential assessment on the firm by state insurance guaranty funds. Once calculated, values for C1-C4 are used to compute the total value of risk-based capital (RBC) for the representative life insurance firm, through the expression, RBC = [ (C1 + C3)2 + C22 ]1/2 + C4. This measure of RBC is adjusted for the effects of covariance among different classes of assets and liabilities, as represented by the addition of the values of risk categories that are believed to be correlated. The covariance adjustment then squares these resulting groups, adds the resulting squares together and takes the square root of the sum of the squares.37 This value of RBC is the minimum required capital for the representative life insurance firm. The regulatory risk of this firm is then measured with respect to the ratio, Total Adjusted Capital RBC
33
This calculation, as described in NAIC Model Act #312-1, pp. 312-1, takes into account the positive relationship between reserve deficiencies and rapid liabilities growth. 34 This risk category is not (yet) calculated for property-casualty insurers. 35 Since the impact of interest rate change is greatest on those contracts where the policyholder is most likely to respond to changes in interest rates by withdrawing funds from the insurer. See NAIC (2000) “Reserves on Certain Annuities and Single Premium Life Insurance Policies Cash-Flow Tested for Reserve Adequacy-Factor Based RBC,” for suggested classifications. 36 See NAIC (2000), “Reserves on Certain Annuities and Single Premium Life Insurance Policies CashFlow Tested for Reserve Adequacy-Factor Based RBC,” and especially Technical Appendix 1. 37 Usage of linear correlation in the NAIC RBC expression rests implicitly on an assumption that the value of each security held in portfolio has a marginal distribution in the class of normal distributions and that the joint value distribution is multivariate normal, an assumption which may be inappropriate for lossdistributions with “heavy” tails. See Doherty (1997), Embrechts, MacNeil and Straumann (1999), Klugman and Parsa (1999), Rantala (1982), and Wang (1997), among others, for a discussion of such distributions in insurance.
16
Table 9: NAIC Asset Risk (C1) Weights Risk Weight: Life 0% .3%
Risk Weight: P-C 0% .3%
Assets
1.0% 4.0% 9.0% 20.0% 30.0% .5% 3.0% 30.0% 2.0% +
1.0% 2.0% 4.5% 10.0% 30.0% 5.0% 5.0% 15.0% 2.0% +
NAIC BBB rated bonds NAIC BB rated bonds NAIC B rated bonds NAIC CCC rated bonds Bonds “near default” Whole residential mortgages Commercial mortgages Common stock Preferred stock: weight is corresponding NAIC bond weight plus two percent.
Cash and US government bonds. NAIC AAA-A rated bonds, including GSE debt and most collateralized mortgage obligations.
where Total Adjusted Capital includes capital and surplus (reserves) as determined by each state’s statutory accounting requirement for the annual financial statement filed by the representative life insurance firm. Values of this ratio then trigger a regulatory response according to a schedule of actions specified for a partition of values for this ration between zero and two.38
38
Deviations from compliance with the RBC requirements are measured, for both life and propertycasualty firms, by the ratio of Total Adjusted Capital (total surplus, including reserves and capital) to the value, under the respective expressions above, for Risk-Based Capital. Deviations can result in the following actions, contingent on this ratio. A company reporting total adjusted capital of 200% or more of minimum risk-based capital is a “no action” level company, specifying no action to be taken by the state regulator. Total Adjusted Capital of 150% to 200% of minimum risk-based capital requires the submission of a report to the regulator outlining corrective actions planned by the firm. The firm must submit to the state insurance commissioner a comprehensive financial plan that identifies the conditions that contribute to the company’s financial condition. This plan must contain proposals to correct the company’s financial problems and provide projections of the company’s financial condition, both with and without the proposed corrections. The plan also must list the key assumptions underlying the projections and identify the quality of, and the problems associated with, the insurer’s business. If a company fails to file this comprehensive financial plan, this failure to respond triggers the following action. Total Adjusted Capital of 70 to 100% of the minimum risk-based capital triggers an Authorized Control Level. This is the first point that the law authorizes the regulator to take control of the insurer. This authorization is in addition to the remedies available at the higher action levels. It is important to note that the law grants the insurance commissioner this power automatically. This action level occurs at a point where the insurer may still be technically solvent according to traditional standards – that is, the company’s assets may still be greater than it liabilities. Finally Total Adjusted Capital of less than 70% triggers a “Mandatory Control Level” that requires the regulator to take steps to place the insurer under control. This situation can occur while the insurer still has a positive level of capital and surplus, although a number of the companies that trigger this action level are technically insolvent (liabilities exceed assets).
17
C. U.S. Housing Government-Sponsored Enterprise Capital Standards A third class of intermediaries participating in international financial markets consists of U.S. Housing Government-Sponsored Enterprises (GSEs). The GSEs are the Federal National Mortgage Association (FannieMae), the Federal National Home Loan Mortgage Corporation (FreddieMac), and the twelve Federal Home Loan Banks (FHFBs). These institutions were originally government agencies, created to increase liquidity in residential mortgage markets. FannieMae and FreddieMac are now private corporations with publicly traded shares, while the FHFBs are owned by their member institutions (primarily savings banks) and linked in terms of their credit liability through a mutual system.39 All three GSEs intermediate lending by creating a secondary market for residential mortgages. FannieMae, originally created in 1938, and FreddieMac, created in 1970, create mortgage-backed securities (MBSs) by purchasing certain types of residential mortgage loans from primary lenders, financing such purchases by selling MBSs to other investors.40 The MBSs combine the cash flows from such pools with a credit guarantee respectively issued by FannieMae or FreddieMac, leaving the investors exposed only to interest rate risk. The primary role of the FHFBs has been to issue collateralised loans, termed “FHFB advances,” to member institutions at subsidized yields. These advances are financed by sales of debt securities, termed “collateralised obligations,” and include discount notes and bonds with fixed rates and maturities as well as variable rate bonds. These bonds may be callable or putable and are issued, on behalf of the FHFBs, by a single agent, the FHLB Office of Finance. The bonds are marketed to foreign as well as domestic investors.41 Current capital standards for FannieMae and FreddieMac have been determined by legislation and regulation over the last ten years. Capital standards for the two largest housing GSEs, Freddie Mac and Fannie Mae, were established by the Federal Housing Enterprises Financial Safety and Soundness Act of 1992. The Office of Federal Housing Enterprise Oversight (OFHEO) is responsible for promulgating regulations to implement the capital standards, as well as for designing the stress tests that form a component of the capital standards . 39
Although private ownership distinguishes FreddieMac and FannieMae from Crown Corporations found in many Commonwealth nations, a vestige of their public origins persists in the view, held by some investors, academics and members of the business press, that each firm enjoys some form of implicit guarantee on its corporate debt from the U.S. Treasury. Although some have argued that each firm possesses a line of credit with the U.S. Treasury, the actual provisions of each firm’s charter act to give the U.S. Treasury a limited call option on this debt, rather than the put inherent in a line of credit or as explicitly possessed by banks and thrifts in the U.S. insured by the FDIC. 40 These other investors include both foreign and domestic life insurance firms, pension funds and other corporations, as well as primary lenders, who swap a portion of newly-originated mortgages for MBSs. Both FannieMae and FreddieMac, in addition, hold a portion of purchased mortgages in their own portfolios. Residential mortgages eligible for purchase, under U.S. regulations, include mortgages carrying direct U.S. government guarantees (“FHA/VA” mortgages) as well as so-called conventional mortgages which have loan to value ratios below 80% or some form of private credit insurance. 41 See “Federal Home Loan Bank System: An Overview,” at http://www.fhlb-of.com.
18
The recent Gramm-Leach-Bliley Act (GLBA), passed by the U.S. Congress in 1999, established the outline of capital standards for the Federal Home Loan Banks (FHLBs). The Federal Housing Finance Board is responsible for designing and implementing specific capital regulations for the FHLBs, as well as supervising the issue of FHFB collateralised obligations by the FHLB Office of Finance. C.1 Capital standards for Freddie Mac and Fannie Mae Capital regulations for Freddie Mac and Fannie Mae contain both a traditional minimum ratio-based capital requirement and, in addition, a risk-based capital standard that employs a stress test methodology to measure whether the GSEs can withstand a severe economic downturn. Minimum capital requirement Under the minimum capital standard, Tier 1 or core capital must exceed 2.5% of onbalance sheet assets, plus 0.45% of off-balance sheet obligations, primarily off-balance sheet guaranteed mortgage backed securities; plus a capital charge for potential future exposure on interest-rate contracts such as swaps. This ratio based approach for setting capital requirements is similar to the approach used for bank capital standards. In addition to these regulatory minimum requirements, Freddie Mac and Fannie Mae have committed to issue subordinated debt such that the sum of core capital and subordinated debt exceed 4.0% of on-balance sheet assets. The commitment is to reach this target by October 2003. Stress tests and the risk-based capital requirement The risk-based capital requirement employs a stress test to set capital levels and includes specific requirements based on both credit and interest rate risk. The enterprise must have sufficient capital to survive ten years under severe economic conditions. The stress test has both a credit risk and an interest rate risk component. There is an additional 30% add-on for management and operations risk, i.e., the capital required for management and operations risk is 30% of the capital required for credit and interest rate risk. The credit risk component of the stress test requires sufficient capital to withstand severe mortgage credit losses. The mortgage credit losses, which are based on historical data, assume that approximately one-quarter of those mortgages with original loan-to-value ratios of 95% experience defaults and that loss to creditors, after foreclosure costs, exceed 50% of collateral.42 Mortgage default models are employed to predict how the enterprises’ current book of mortgages would perform under these economic conditions. Non-mortgage assets are assumed to experience default losses consistent with the mortgage defaults.
42
Specifically, this data is for the losses recorded for mortgages in the US South “Oil Patch” (primarily Arkansas, Louisiana, Mississippi, Oklahoma and Texas) for the years 1982-83, the period following the decline in world oil prices.
19
The interest rate risk component of the stress test requires sufficient capital to withstand severe changes in interest rates. The ten-year US Treasury rate is assumed to increase by 75% or decrease by 50% (whichever would require more capital) over a one-year period and remain at these levels for the next nine years. Other interest rates are assumed to change consistent with the change in the ten-year US Treasury rate. Mortgage prepayment models are employed to predict how the enterprises’ current book of mortgages would perform under these economic conditions. Based on these models, the income from the mortgage assets can be calculated. Similarly, the expense on the enterprises’ debt and other liabilities are calculated based on both the contractual terms of the debt and other liabilities and the interest rates during the ten-year stress test. Net income during the stress period is calculated as the income on the assets minus the expense on the liabilities minus credit losses, operating expenses, and taxes. The enterprises’ capital must be sufficient to absorb any net losses arising from the credit risk and interest rate risk components along with the 30% add-on for management and operations risk. Capital standards for these GSEs has, for the last several years, been largely set through self-regulatory measures.43 While OFHEO has pledged to issue final risk-based capital regulations by the end of 2001, Freddie Mac and Fannie Mae have, pending the publication of these final regulations, implemented stress tests consistent with the statutory requirements prescribed by the U.S. Congress.44
C.2 Capital standards for the Federal Home Loan Banks The FHLBs have both minimum leverage ratios and risk-based capital requirements and hold two forms of capital: FHLBank stock purchased by member financial institutions and retained earnings. FHLBank stock provides the principal capitalization for the System.45 Historically, member financial institutions were required to purchase stock valued at 5% of their total level of FHLBank advances or 1% of their total mortgage assets, whichever
43
These measures can be explained as signaling to both capital markets and the U.S. Congress, bolstering the arguments for the efficacy of the incentives for transparency, as discussed in the context of insurance regulation above. 44 These requirements are described in the Federal Housing Enterprises Financial Safety and Soundness Act of 1992, a subsection of the Housing and Community Development Act of 1992, Pub. L. No. 102-550. 45 Retained earnings in the FHLB bank System are a nearly-negligible source of capital: for example, there was $30.6 billion of FHLBank stock, which equals 4.8% of the total assets of the FHLBanks, on 30.09.01, while, on the same date, there was $0.7 billion of retained earnings, which equals 0.11% of total assets. See FHLBank Office of Finance, http://www.fhlb-of.com/financial/. This primary role of FHLB equity raises measurement difficulties in comparing the capital of FHLBs with other GSEs, as well as with banks and other intermediaries. Since the Federal Home Loan Bank System has a mutual form, the member financial institutions own the stock of the FHLBs and the capital stock of the FHLBs is an asset on the books of the member financial institutions. FHLB equity is, consequently, double counted, once as capital for the FHLBs and once as capital for the member financial institutions.
20
was greater.46 The GLBA of 1999 eliminated these “subscription” requirements for FHLBank stock, requiring the Federal Housing Finance Board to establish regulations mandating both minimum leverage and risk-based capital requirements for FHLBanks. The Federal Housing Finance Board specified these new capital requirements in their January 30, 2001 regulation, “Capital Requirements for Federal Home Loan Banks: Final Rule.” The leverage-based capital requirements are, broadly, similar in form to current regulation of most financial institutions, requiring that the ratio of each FHLB’s capital to assets cannot fall below specified levels, although the nature of the capital used to meet this leverage requirement requires a somewhat complex overall measure of compliance. The risk-based capital requirements are explicitly based on stress tests for both credit and interest rate risk. The minimum leverage requirement for FHLBs is based on measures termed “permanent capital” and “total capital,” and these measures of capital, in turn, depend on different classes of equity shares issued by the FHLBs. Under GLBA, more specifically, FHLBs can issue two classes of shares, Class A and Class B, to replace shares currently held by their members. Class A shares are redeemable in cash at par upon 6 months written notice, while Class B shares are redeemable in cash at par at 5 years written notice. “Permanent capital” is comprised of the sum of the amounts paid-in for Class B stock plus retained earnings of the FHLB. “Total capital” consists of permanent capital plus amounts paid-in for Class A stock, plus general loss allowances. Under the Finance Board regulations, the leverage-based capital requirements are calculated by two alternative methods. The first method requires that “weighted” permanent capital (permanent capital (the amount paid in for Class B stock plus retained earnings) multiplied by a weighting factor of 1.5) equal a minimum of 5% of total assets. The second method requires that total capital (permanent capital plus amounts paid in for Class A stock plus general loss allowances) equal at least 4% of total assets. Although the Finance Board will explicitly rely on stress testing for capital requirements for both credit and interest rate risk, at the time of this paper the actual design and implementation of these tests are the responsibility of each individual FHFB, which will then submit the design to the Finance Board for final approval.47 These tests must satisfy the legislative requirement that each FHLB must “...maintain ‘permanent capital’ ... sufficient to meet the credit and interest rate risk to which the (FHFB) is subject, with the (measure of ) risk being based on a stress test established by the Finance Board that tests for changes in certain specified market variables.”48 Issues of transparency, with 46
Although this requirement resulted in high capitalization rates for the FHLBanks, in combination with mandated payments to the REFCorp, AHP, and CIP programs of the Federal Home Loan System, it reduced the market value of FHLBank stock.
47
After the issuance of the Final rule on January 30, 2001, individual FHFBs have 270 days to submit their capital plans to the FHFB for approval, and three additional years to satisfy compliance requirements.
48
“Capital Requirements for Federal Home Loan Banks: Final Rule,” Federal Register 66 FR 8262 (Jan. 30, 2001) at 8264. This rule does not, however, mandate transparency, by requiring either the FHFB or the Finance Board to make public the structure of the FHFB’s internal test, nor does it address the potential for
21
respect to public information about the stress test criterion employed by individual FHFBs, the issues created by the ability of FHFB members to redeem the equity shares that comprise the measures of permanent and total capital, are not as yet explicitly addressed in Finance Board regulations.
IV. Functional Regulation and Minimum Capital Requirements A. Overview In the section we develop a simple model that is used to establishes one of our primary results.. Under a functional regulatory system in which separate supervisory authorities are charged with protecting the interests of different specific classes of a financial intermediary’s stakeholders, the optimal set of regulatory capital requirements will differ according to the stakeholder interest that is being protected. Thus a financial institution with an identical asset portfolio will face a different set of regulatory capital requirements depending on which functional regulator has supervisory responsibility. This result may in part provide an explanation as to why the minimum regulatory capital regimes discussed earlier differ markedly among bank, insurance, and GSE regulatory authorities.49 A secondary result relates to the institutional form of existing functional regulatory capital requirements. While bank capital requirements have traditionally focused on the riskiness of a bank’s asset portfolio, other types of financial intermediaries make use of liability contracts to hedge their risk exposures. When the structure of liabilities are important component of an intermediary’s overall risk management strategy (they are not in our stylized model of a bank), capital requirements that focus on assets alone can give rise to a seriously biased assessment of an intermediaries risk of insolvency. B. The Model Consider a two-date economy in which all contingent claims are traded at date 1 and claims payoff at date 2. We assume that there exist no arbitrage opportunities at date 1 and that all traded securities and contingent claims can be valued as if agents are risk neutral. 50 The risk free rate is arbitrarily normalized to zero and the units traded are arbitrarily labeled dollars. Under the simplifying assumptions of the model, the date 1
systemic risk created by the incentives of individual FHFB members to redeem their stock, withdrawing capital from the FHFB and from the System as a whole. The option of members to redeem their stock at par is a feature of the FHLB System that differentiates it from the other GSEs and from banks. 49
It also may help to explain why ratings firms appear –at least to industry participants--to value capital differently depending on the type of financial firm they are rating. This might be expected if the rating is reflecting the implicit credit risk on a different set of liabilities in, for example, banks versus insurance firms. 50 We do not assume that agents are risk neutral, only that assets can be priced as if agents are risk neutral after an appropriate change of measure. Harrison and Pliska (1981) establish that the absence of arbitrage in a discrete time and discrete state space model implies the existence of a risk-neutral pseudo probability measure and the equivalent martingale valuation condition. The probability measure is unique if markets are complete.
22
value of contingent claims and securities are equal to the expected value of the cash flows generated by the claim on date 2. There are two production process in this economy and these determine the output (the date two asset values) of two alternative types of firms: type A and type B. We arbitrarily fix the total number of firms at 100 and assume that there are 50 identical type A firms, and 50 identical type B firms. We assume, moreover, that absent receiving a costly signal that is perhaps gained in the context of intimate banking or insurance business relationship, it is impossible for investors, regulators or other firms to differentiate between type A and type B firms at date 1. Aggregate output and the output of individual firm types is stochastic can be characterized by an 11 state process described in Table 10.51 Table 10: Exchange Economy Assumptions state probability 1 0.02 2 0.05 3 0.09 4 0.12 5 0.14 6 0.16 7 0.14 8 0.12 9 0.09 10 0.05 11 0.02 aggregate market value individual firm value
individual type A output 0.5 1 2 3 4 10 12 9 6 3 1.5
individual type B output 1.5 3 6 9 12 10 4 3 2 1 0.5
6.24
6.24
aggregate output 100 200 400 600 800 1000 800 600 400 200 100 624
The cash flows for type A and type B firms are perfectly symmetric with respect to the aggregate output of the economy: type B firms produce a greater share of output in states 1 through 5; type A firms produce the largest output shares in states 6 through 11. For either type firm, the minimum date 2 asset value recorded is 0.50. This value occurs in state 1 for a type A firm and in state 11 for a type B firm. Under the symmetry and risk neutrality assumptions, type A and B firms have identical market values at date 1 before uncertainty is resolved. Consider the introduction of financial intermediaries, for example, a bank or an insurance company. Assume that financial intermediaries extend loans to type A and B firms in the first period. A bank funds its loan portfolio with deposits and equity. An insurance company funds its loan portfolio with equity and by offering insurance contracts whose payoffs are contingent on a particular observed value of aggregate output. 51
The probabilities in Table X are risk neutral probabilities.
23
Financial intermediaries are assumed to make loans to type A or type B firms at fair market terms under the risk neutral valuation assumption. As type A and type B firms’ date 2 asset values and probabilities are identical with respect to the valuation of a standard discount loan contract, financial institutions offer identical loan terms. A loan to either firm with a given par value will have an identical probability of default, loss given default, and default option value. An example of a loan valuation calculation for a 2 dollar par value discount loan for type A and B firms appears in Table 11. Table 11: Loan Pricing and Default Characteristics promised promised actual actual payment payment payment payment state probability type A type B type A type B 1 0.02 2 2 0.5 1.5 2 0.05 2 2 1 2 3 0.09 2 2 2 2 4 0.12 2 2 2 2 5 0.14 2 2 2 2 6 0.16 2 2 2 2 7 0.14 2 2 2 2 8 0.12 2 2 2 2 9 0.09 2 2 2 2 10 0.05 2 2 2 1 11 0.02 2 2 1.5 0.5 market value of loan 1.91 1.91 default option value as % market value 4.71% 4.71% probability of default 0.09 0.09 Loan default characteristics and pricing for loans with a maturity value of 2 to either a type A or a type B firm.
C. Functional Regulation, Capital Requirements and Asymmetric Information In this section we consider the optimal regulatory capital requirements that apply under different functional approaches for regulating an intermediary with an identical asset portfolio. It will be shown that the optimal set of regulatory capital requirements depends on the objective of the functional regulator. C.1. Bank Regulation: Ensuring the Safety of Depositors We adopt the traditional objective function and assume that bank regulators attempt to ensure the safety of bank depositors and limit the potential for any explicit insurance cost under a formal government deposit guarantee scheme or any implicit benefit that may be garnered by banks under an implicit guarantee system. Accordingly, it is assumed that it is socially important that the deposits of the bank be default risk free and that a regulator is charged with the duty of ensuring the safety of deposits by regulating bank leverage (regulatory capital requirement). Moreover, initially at least, it is assumed that banks are restricted to making loans and are prohibited from issuing any liabilities other than debt. 24
If a regulator were entrusted with the knowledge in Table 10 and the charge of ensuring the safety of a bank’s deposits, it cold do so by setting a the regulatory capital requirement on each loan equal to Min[.50, par ] where par is the maturity value of the loan, Additional par loan value in excess of 0.50 must be financed dollar for dollar with bank equity. Bank loans with par values smaller than .50 can be total financed with deposits. This maximum deposit-to-loan capital requirement ensures the safety of deposits without requiring regulatory knowledge of the type of firms represented in bank loan portfolios. No matter what degree of leverage banks offered to type A or type B firms, the proceeds from the bank loans would ensure that deposits are risk free under the minimum capital rule. If a bank were limited to 0.50 deposit funding , a bank purchasing a 2 dollar par value loan would for example be required to have equity capital of 1.41 unit per loan or a 73.82% capital requirement. The simple regulatory capital scheme ignores the possibility of bank loan diversification benefits. If a bank were to diversify its loan portfolio and make only simultaneous paired loans of equal par value to type A and type B firms, then a pair of such loans could at most support Min[2, par ] units of risk free deposits.52 The diversification from the paired loan strategy would allow the bank to support a larger share of its assets with riskless bank deposits. An example of the potential diversification benefits for the provision of risk free deposits appears in Table 12. A problem arises when a bank has the ability to differentiate between type A and B firms at date 1 but the regulator does not. In this instance, if the regulator were to grant the regulatory capital diversification benefit, and allow 2 dollars of deposits for every pair of 2 dollar par value (or greater) loans to a type A and B firms, a moral hazard problem ensues. It is easily shown that the call option nature of bank equity returns will create an incentive for a bank shareholder-managers to concentrate the bank’s loans to a single firm type but represent itself has holding a diversified loan portfolio. If bank depositors relied on an explicit guarantee provided in conjunction with the regulatory oversight, depositors would not limit their exposures to banks or price in the credit risk that would arise form the moral hazard, but guarantee losses would arise in some states of nature. In the example in Table 12, an ex ante gain of .02 would accrue to a bank’s equity holders if it chose to hold a concentrated portfolio. In this instance, the deposit insurer would be required to pay out 1 if state 1 were realized. Recognizing this moral hazard problem associated with asymmetric information, in order to limit deposit insurance losses, regulators must disallow diversification benefits and enforce Min[.50, par ] as the maximum amount of deposit funding per loan. It should be noted that the single loan regulatory capital requirement for risk free deposits is identical to the regulatory capital requirement that would be applied by the market absent regulations. It can be shown that, in the absence of credible information about the 52
The combination of one type A firm and one type B firm will produce a total output of 2 in states 1 and 11, and larger output in all other states.
25
type of firms and diversification benefits in a bank’s loan portfolio, the market would impose a moral hazard discount on any “risk free” liability claims that a bank tried to offer in excess of the regulatory maximum deposit funding requirement. Since investors would price these claims as if they were risky regardless of whether or not a bank had a diversified portfolio, the bank would never hold a diversified loan portfolio if it wanted to issue liabilities in excess of the regulatory limit. Table 12: Deposit Value and Loan Diversification Concentrated Loan Portfolio Diversified Loan Portfolio two 2 dollar payoff 2 dollar payoff par loans to on 2 dollar par loan on 2 dollar separate bank to a type A bank state type A firms deposit and typeB deposit 1 1 1 2 2 2 2 2 3 2 3 4 2 4 2 4 4 2 4 2 5 4 2 4 2 6 4 2 4 2 7 4 2 4 2 8 4 2 4 2 9 4 2 4 2 10 4 2 3 2 11 3 2 2 2 1.98 market value 2 market value 0.02 default option value 0 default option value 0.02 probability of default 0 probability of default Example of the value of bank deposits and deposit default option values for banks with diversified and non-diversified bank loan portfolios.
In the context of this model, with or without a regulator, in the absence of safety net subsidies, a bank is limited to issuing Min[.50, par ] of riskfree deposits. The bank could, however, issue risky liabilities subject only to fair market pricing restrictions or explicit regulatory restrictions (whichever is binding). If a regulator could ensure strict priority of depositors’ claims in the case of bankruptcy, the bank could fund itself entirely with subordinated debt, but because of the moral hazard discount, the market will assume that the bank holds---and thus the bank will be forced to hold---a concentrated loan portfolio. Despite any bank claims to the contrary, because the market cannot determine ex ante whether or not a bank holds a diversified loan portfolio and bank shareholders will profit if investors are deceived into believing the portfolio is diversified, investors will assume banks hold concentrated loan portfolios and price subordinated debt accordingly. Notwithstanding the theoretical possibility of extending bank leverage using subordinated debt, in practice bank capital regulations generally are formulated to discourage all but a
26
de minimus probability of default for the entire bank.53 Thus while banks could take on substantial amounts of leverage (up to 100%) without incurring any significant market pricing problems from asymmetric information, the nature of their liability claims—an identical payment promised in every state of nature--- requires that greater leverage be associated with greater probability bank default. Table 13 reports the market value of deposits, subordinated debt, equity, and the probability of default for a bank with a single 2 dollar par value loan to a type A or B firm, funded with .50 deposits and alternative par values of its own subordinated funding debt. Table 13: Subordinated Debt, Bank Leverage and Bank Default Probability Par Value of Bank Market Value Market Probability of Bank Bank Deposits of Bank Value Default on Bank Equity-toSubordinated Subordinated of Bank Subordinated Asset Debt Debt Equity Debt Ratio (%) 0 .5 0 1.41 0 73.82 .5 .5 .49 .92 .02 48.17 1 .5 .955 .455 .07 23.82 1.5 .5 1.41 0 .09 0 Trade-off between the probability of default on bank subordinated debt and bank leverage for a bank with .50 in deposits and a single type A or B 2 dollar par value loan.
C.2 Insurance Regulation: Ensuring the Integrity of Insurance Contracts Since aggregate output is observable, it is feasible to offer contingent claims that provide payoffs that are keyed to the realized value of aggregate output. Because it is impossible to differentiate between states other than the mode of the output distribution (state 6) other individual state contingent insurance policies cannot be offered in equilibrium. The inability to offer a complete set of state contingent claims is a significant source of systemic risk in the case of the provision of insurance services. Another source of systemic risk are moral hazard generated limits on the capacity of intermediaries to provide insurance based upon aggregate output realizations. Assume that society has deemed it beneficial to establish a regulatory function to ensure that insurance intermediaries are capable of paying their policy claims. Associated with this function maybe an explicit insurance guarantee system, but assume that regulatory objectives are to minimize explicit guarantee claims and any implicit benefits that are conveyed by the guarantee program. Assume that the insurance regulatory agent carries out its appointed task by mandating minimum regulatory capital requirements for the insurance industry. Further, assume that insurance companies are limited to investing in firm loans and can only issue liabilities in the form of insurance contracts.
53
A maximum probability of bank default, for example, is the objective function that appears in recent Basel Bank Supervisors discussions of credit risk modeling. See for example, “Credit Risk Modelling: Current Practices and Applications,” The Basle Committee on Banking Supervision, April 1999, p. 13.
27
By assumption, insurance intermediaries’ assets are standard discount loans to type A or type B firms that are made at fair market value. As the only insurable set of events are the 6 separate aggregate output levels reported in Table 14, assume that the firms offers insurance (contingent claims contracts) in the form that a contract pays off a single unit if the specific insured aggregate output state materializes and pays nothing otherwise. Table 14: Regulatory Insurance Liability Limits Aggregate States of Maximum Notional Insurable Output Event Nature Amount per Bank Loan 100 1,11 Min[loan par value, 0.5] 200 2,10 Min[loan par value,1] 400 3,9 Min[loan par value, 2] 600 4,8 Min[loan par value, 3] 800 5,7 Min[loan par value, 4] 100 6 Min[loan par value,10] Regulatory maximum insurance liability funding schedule for each bank loan.
Because it is impossible for an insurance regulator to identify the type of firm in the insurance intermediary’s loan portfolio, to ensure the claims paying ability of the insurer, the regulator must enforce a regulatory requirement that takes the form of a schedule of the maximum number (or the maximum total notional insured value) of specific insurance contracts that can be used to fund a bank loan. The optimal regulatory schedule is given in Table 14. Similar to bank regulations, it is also true that the maximum permitted regulatory insurance sales are also the maximum possible insurance sales in the absence of regulation. Insurers would be limited in their abilities to offer contracts to these level because, absent information on the type of firm’s in the insurer’s loan portfolio, investors would be unable to confirm claimed loan portfolio diversification, and a moral hazard discount (a credit risk discount from investors viewpoint) would be applied to any contracts offered in excess of the regulatory maximum levels. The regulatory (or competitive market) maximum limits on an intermediaries ability to offer insurance policies implicitly translates into an equity capital requirement. The equity capital requirement that would apply for an insurer with a 2 dollar par value a type A or B loan are calculated in Table 15. It is worth noting that a portion of the insurance intermediaries’ liabilities could be packaged and sold as a catastrophe bond in place of offering individual insurance claims. Taking the example in Table 15, if a insurance intermediary with a $2 par loan to a type A or B firm were to offer a liability that paid out $1 in all states of nature except when aggregate output is 100, it would in effect be offering a catastrophe bond (in place of the equivalent bundle of individual insurance claims). This “cat bond” would be fairly priced by the market because the bond terms eliminate any uncertainty as to the firm’s ability to honor its liabilities given the unknown type of firm exposure in its loan portfolio. The cat
28
bond is an example in which a firm can tailor the contract terms on its liabilities to avoid positive (or increasing) default probabilities and moral hazard discounts. In risk management terms, the cat bond liability is, in part, a hedge of the cash flows of the insurer’s asset portfolio. The use of a funding liability with embedded contingent claims reduces the probability of default over a traditional debt contract. Table 15: Implied Regulatory Equity Capital for Insurers payment on $2 par value loan state probability to type A 1 0.02 0.5 2 0.05 1 3 0.09 2 4 0.12 2 5 0.14 2 6 0.16 2 7 0.14 2 8 0.12 2 9 0.09 2 10 0.05 2 11 0.02 1.5 Loan market value 1.91
insurer's insurer's payment maximum residual residual required required on $2 par regulatory insurer's insurer's equity equity value loan insurable equity equity value value to type B value type A type B A B 1.5 0.5 0 1 0 0.02 2 1 0 1 0 0.05 2 2 0 0 0 0 2 2 0 0 0 0 2 2 0 0 0 0 2 2 0 0 0 0 2 2 0 0 0 0 2 2 0 0 0 0 2 2 0 0 0 0 1 1 1 0 0.05 0 0.5 0.5 1 0 0.02 0 1.91 Equity capital requirement 0.07 0.07 Equity capital requirement as % loan value 3.66 3.66
Minimum regulatory capital requirement for insurance an intermediaries’ equity.
C.3 Prudential Capital Requirements Under Functional Regulation: Voyez la difference A comparison of the minimum regulatory capital calculations for a bank and an insurance company that hold an identical asset portfolio demonstrate that functional prudential regulations (an indeed competitive market forces) may imply strikingly different regulatory capital requirements for banks and insurance companies. In the case of a bank intermediary, the provision of risk free deposits requires that the regulator impose a 73.82 percent minimum regulatory capital requirement on a bank whose loan portfolio consists of a single 2 dollar par value fairly priced loan to a type A or B firm. This regulatory capital requirement can be relaxed if the regulator allows for a positive probability of bankruptcy on a bank’s subordinated liabilities. Indeed if the regulator is willing to tolerate a 9 percent probability of bank default, the bank could use 100 percent debt to financed in the $2 par debt example.54
54
As the par value of bank loans to type A and B firms increase, 100 percent bank debt finance will be associated with a greater probability of bank default.
29
In contrast to the case of bank liabilities, a completely rigorous prudential standard for an insurance intermediary with an identical asset portfolio will require a minimum capital requirement of only 3.66 percent. With only a 3.66 equity to asset ratio, an insurance companies entire set of liabilities can be completely risk free. Again, for a bank’s liabilities to be risk free, the bank’s equity to asset ratio must be at least 73.82 percent. This enormous difference in regulatory capital requirements will only increase as the par value of the intermediary’s asset loan increases. The intuition that underlies these strikingly large differences in prudential equity capital requirements for banks and insurers is that the ability to provide a riskfree deposit is limited by the magnitude of aggregate economic output (or an intermediary’s loan portfolio’s performance) in the worst possible state of nature, and may be limited further by asymmetric information constraints. The provision of sound insurance contracts is far less limited as output is significantly larger in all other states of natures. While the model is highly stylized to remove unnecessary details, the intuition of this result is more general. A riskfree claim is by its very nature the most restrictive of all claims that can be offered as the solvency of a unit claim must span all possible states of nature. While bank leverage can be increased by allowing the issuance of subordinated debt (that is costless in this model), bank regulators seemingly focus attention on the probability of bank solvency (rather than the solvency of deposits) and so bank leverage is limited in practice by limits on the issuance of defaultable debt. While such limits are often justified as a control for systemic risk, the results in the next section that they do not fulfill that role in this model.
V. Asymmetric Information and Systemic Risk A. Overview This section uses the model of Section IV to investigate the relationship between functional regulatory capital requirements and systemic risk. In the context of this simple model, systemic risk is identified with the dead weight loss in welfare that occurs when asymmetric information problems preclude intermediaries from offering a complete class of debt and contingent claims contracts.55 We interpret the losses associated with market incompleteness in this static model as an analog (admittedly a rough analog) to the dead weight losses that are engendered in a dynamic market setting when intermediaries refuse to trade financial claims under stressful market conditions. The analysis will show that, in the context of this model at least, it is impossible to costlessly remove this form of systemic risk with regulatory capital requirements alone. While one might conjecture that regulatory capital requirements could play a role in alleviating this problem, absent an increase in investor information, it is shown that capital requirements alone cannot induce trade in the missing claims markets unless the
55
While we do not model agents utility directly and so do not measure directly the welfare losses associated with incomplete contingent claims markets, we are confident that it is possible to formally establish this claim and have omitted the consumer utility calculus merely to simplify the presentation.
30
regulatory safety net is extended and investors are offered an explicit guarantee that has positive value to banks. Complete transparency will removes the information asymmetry and systemic risk in this exchange economy. The complication is that firms offering standard debt and equity claims have no incentive or a credible means of signaling their private information to market investors. In this setting we show that regulatory capital requirements can play a role in improving transparency. It is demonstrated that if regulators lower capital requirements to a level that reflects the true diversification benefits available to banks and adopt an appropriate pre-commitment style penalty schedule for banks that violate their date 1 commitment to hold a diversified loan portfolio, not only are markets completed and systemic risks reduced, but regulatory capital requirements can be significantly lowered without any effect on the value of government guarantees. We note in passing that Pillar 3 of the New Basle Capital Accord proposal reflects an increase in the regulatory focus on transparency . While the optimality of the transparency solution is clear in the context of this model, achieving transparency requires that firms have the ability to credibly signal their “type” to investors or banks can credibly signal their true asset diversification strategy. Pillar 3 of the New Basel Capital Accord is an attempt to increase bank transparency by mandating additional accounting disclose requirements for banks. Notwithstanding detailed disclose requirements, such provisions will provide a credible signaling mechanism for banks and non-financial firms only if legal/regulatory penalties are set commensurate with the potential gains from moral hazard---thus the importance of penalties in a pre-commitment approach to regulation. In practice, it remains unclear whether or not disclosure requirements alone can provide a credible signaling mechanism to achieve the necessary transparency and mitigate system risks. While the analysis focuses on issues related to functional regulation, there are deeper issues regarding the organization and objectives of regulation that are not addressed in this paper. For example, make we no general claims regarding the optimality of a functional approach to regulation outside the limited context of this model. Moreover, the simple expansion of the types of liability contracts that are offer by intermediaries cannot solve the underlying systemic problem as the insurance companies liabilities are span the largest set of claims that can be traded without additional information or regulatory guarantees with safety net subsidies. Thus, allowing banks to offer insurance policies in addition to deposits does not in any way reduce systemic risk in this setting. B. Limits to Trade The asymmetric information problem has no effect on the pricing of loans to firms. Asymmetric information problems will, however, significantly limit intermediaries’ abilities to exploit diversification benefits and the value of financial claims that they can offer will be significantly limited relative to the potential value of claims that could be traded.
31
Asymmetric information limits the abilities of financial intermediaries to offer risk free and risky claims on diversified loan portfolios. Firms themselves can issue risk free claims, but the par value of a risk free claims that could be offered by any single firm is 0.50 and thus the aggregate amount of risk free claims offered in the securities market is limited to 50, despite the fact that the minimum aggregate output is 100. While the introduction of a financial intermediary would allow for pooling of financial claims on type A and type B firms, and through diversification benefits, allow intermediaries to offer 100 units of risk free claims, or insurance contracts with aggregate values equal to the total aggregate output, asymmetric information prevents intermediaries from performing these functions. Table 16 summaries the asymmetric information imposed limitations on the trading of financial claims. The limits on the potential trade of financial asset claims could be a significant source of dead weight loss in this economy as market incompleteness limits investors’ abilities to optimize their date 2 consumption bundles. We equate the potential dead weight loss in utility with the so-called systemic risk phenomenon that arises in dynamic situations when claims trading among financial institutions is significantly depressed by imperfect information about counterparty solvency. Table 16: Asymmetric Information Limits on Trade total aggregate total individual individual tradeable minimum risk free type A type B aggregate insurance riskfree deposits state output output output claims output offered 1 0.5 1.5 100 50 100 50 2 1 3 200 100 100 50 3 2 6 400 200 100 50 4 3 9 600 300 100 50 5 4 12 800 400 100 50 6 10 10 1000 1000 100 50 7 12 4 800 400 100 50 8 9 3 600 300 100 50 9 6 2 400 200 100 50 10 3 1 200 100 100 50 11 1.5 0.5 100 50 100 50 Asymmetric information limitations on the trade of financial claims.
C. Risky Debt, Regulatory Signals, and the Extension of the Safety Net While asymmetric information limits the aggregate supply of risk free debt to 50 percent of the potential supply of risk free securities, there are no such restrictions on firm’s ability to issue risky discount debt. While firms can fund themselves with any level of risky debt they desire at a price that is independent of the firm type, the asymmetric information problem does creates issues for the valuation of the risky debt and equity claims of intermediaries.
32
Absent intermediary transparency and a credible signaling mechanism at the individual firm level, secondary market participants will always favor an assumption that moral hazard will lead intermediaries to hold concentrated loan portfolios. Indeed if market participants impose any moral hazard related discount on risky debt or equity securities issued by an intermediary, the intermediary will be required to hold concentrated loan portfolios if they are to issue fairly priced claims. The discount applied in the secondary market for bank equity and insurance equity claims makes external equity issuance costly for a diversified intermediary (a claim often made by bankers) and favors an equilibrium where intermediaries hold concentrated asset portfolios or shareholders internalize information by also being the intermediary’s managers and thereby avoid the asymmetric information problem. Consider now the application of regulatory capital requirements with an objective of reducing systemic risk in the economy. An increase in the issuance of risk free claims would become possible if the bank regulator would, for example. recognize the maximum possible diversification benefit and allow the maximum deposit funding support of Min( par value,2) for every paired loan, where par value is the value of each type A and type B loan pair. Notwithstanding lower regulation capital limits, absent some form of insurance guarantee or increased information on the characteristics of a banks loan portfolio, investors would continue to impose the competitive market deposit limit of Min( par value,.5) as they would recognize the moral hazard incentives facing bank manager-shareholders. Absent new information or some sort of enforcement scheme, the only way the regulator can reduce systemic risk and promote the provision of a greater amount of riskfree claims is to lower regulatory capital requirements and accept a positive probability of insurance losses. Moreover, in this situation, if banks were given the ability to continue to offer subordinated debt contracts, banks would in effect be required by the market to hold a concentrated exposure if they wanted to issue fairly priced subordinated debt claims as investors would still apply a moral hazard (default risk) discount to these liabilities. In the prior example, absent any new information, systemic risk can only be reduced by lowering bank capital requirements and providing an explicit deposit insurance guarantee. The increase in supply of risk free claims will, however, create expected deposit insurance costs as deposit insurance is no longer fairly priced ex ante. While it may be possible to use a bank ex post bank taxation scheme to fund the costly deposit insurance scheme, industry specific income tax levies may be illegal in practice. A general taxation scheme that charges all investors or all firms might fund the guarantee, but may still leave banks with a net gain from moral hazard behavior. D. Transparency and a Pre-Commitment Approach for Capital Regulation As an alternative to a general tax levy deposit insurance funding scheme, it is possible to impose an ex post penalty scheme similar in nature to the pre-commitment approach for market risk capital regulation to remove the moral hazard problem and establish fairly
33
priced deposit insurance at the expanded level.56 Conditional bank fines can be set in a manner that induces firms to avoid moral hazard behavior and thereby make the expanded deposit insurance guarantee a fairly priced. In addition, an effective precommitment regulation program will establish a credible signaling mechanism that will enable intermediaries to offer a more complete set of risky contingent claims. Under a portfolio specific bank pre-commitment fine schedule, a bank will be induced to adopt, ex ante, a diversified loan portfolio an eschew moral hazard behavior. The precommitment approach will require limits on the use of bank subordinated debt, as, absent a pre-commitment fine, a bank must have sufficient equity gains from concentration in some states of nature for the application of the pre-commitment find to have the desired effect. Table 17 illustrates the design of a pre-commitment penalty schedule for a bank that is representing itself as taking $2 in deposits, and investing in a $2 par value loan to a type A firm and a $2 par value loan to a type B firm. In this example, the banks is precluded by regulation form issuing any subordinated debt.57 This paired loan strategy offers the greatest amount of diversification benefit to bank net loan income. Absent a precommitment penalty fine, Table 17 shows that bank management would maximize shareholder value by concentrating its loan portfolio into type A or type B loans.58 The shareholder benefit from loan concentration can be removed by imposing an ex post penalty on the bank that requires the bank to pay regulatory fines in states of nature in which its equity income exceeds the equity income it would earn on a perfectly diversified loan portfolio---the portfolio to which it must pre-commit to quality for regulatory permission to accept greater deposit funding. In the example in Table 17, at a penalty rate set at 2 percent of income in excess of the bank’s pre-committed portfolio income, the bank shareholders no longer accrue gains from concentrating the bank’s loan portfolio. Any pre-commitment penalty rate higher than 2 percent will, in this example, be sufficient to ensure that banks select a fully diversified portfolio. If a pre-commitment approach were used to enforce a truth revelation mechanism, bank fine schedules would need to be tailored to their representation of their loan portfolio. A secondary requirement is limits on the use of bank subordinated debt. This requirement is necessary to ensure that the pre-commitment penalty imposes adequate costs on bank shareholders. An interesting feature of the pre-commitment approach is that, not only does it allow for a larger share of loan value to be funded with deposits ($2 of deposits to support every diversified loan pair of $2 par value loans) and allow for the issuance of a socially optimal volume of risk free claims without the potential for any deposit insurance losses, 56
For further discussion of the Pre-Commitment Approach, see Kupiec and O’Brien (1996,1997,1998). The bank in this example could be allowed to issue limited amounts of subordinated debt provided that the penalty schedule is adjusted appropriately. 58 We assume that the regulator can observe the loan par values and the magnitude of a bank’s deposits so the only unobserved component is the type of firms to which the bank extends loans. 57
34
but the increased share of deposit funding will provide an accurate signal the secondary market for risky bank claims. Under an optimal (i.e., fairly priced deposit insurance) precommitment penalty schedule, banks will be able to issue risky claims that reflect their diversified loan portfolios at fair market terms. In other words, the issuance of equity or subordinated debt by a diversified bank is no longer costly as the moral hazard pricing discount is removed by truth-revealing pre-commitment penalty schedule. While we have illustrated the pre-commitment approach in the context of bank capital regulation, similar results apply in the case of insurance regulation. The pre-commitment penalty structure (with minimum equity limits) can be used to create a credible signaling mechanism that enables investors to identify insurance intermediaries that hold diversified asset portfolios. Table 17: Pre-Commitment Penalty Rate Calculation Assumes banks funds with $2 in deposits income equity value equity value equity value excess excess from $2 par from portfolio from portfolio from portfolio equity value equity value loan to of a $2 par of two $2 par of two $2 par from A from B 1 type A loan to type A loans loans concentration concentration probability & 1 type B and a type B to type A to type B strategy strategy 0.02 2 0 0 1 0 0 0.05 3 1 0 2 1 0 0.09 4 2 2 2 0 0 0.12 4 2 2 2 0 0 0.14 4 2 2 2 0 0 0.16 4 2 2 2 0 0 0.14 4 2 2 2 0 0 0.12 4 2 2 2 0 0 0.09 4 2 2 2 0 0 0.05 3 1 2 0 0 1 0.02 2 0 1 0 0 0 Equity Value 1.82 1.84 1.84 Ex ante gains from concentration 0.02 0.02 Minimum required pre-commitment fine rate 0.02 0.02
VI. Implications and Concluding Remarks The paper has established a number of important results regarding functional regulation, minimum capital requirements, and systemic risk. After establishing that nonbank intermediaries are important participants in international risk transfer markets, it bored to tears any reader that actually read all the details that document three (four?) different approaches to setting minimum regulatory capital requirements. After the mandatory discussions of institutional details and reality, the paper proceeded into a comfortable theoretical world with far fewer pesky details where somewhat clever mathematical constructs can be used to illustrate an underlying truth or structure. In this theoretical setting with market failures that arise from fundamental information asymmetries, it establishes that, if regulations are designed to protect specific classes of
35
intermediary liability holders, minimum regulatory capital requirements will differ depending on the characteristics of the claims that are being protected. This sound like a trivial result, but the fact that some (many?) attempt to assess the prudential adequacy of non-bank financial intermediaries by comparing the equity of these institutions to those that would be required if the institution were a bank, suggests that our seemingly trivial result is not well known.59 Following this we show that minimum regulatory capital requirements alone may not control systemic risk if the source of this risk is asymmetric information. In our model, systemic risk can be attenuated by improving transparency. Here we argue that the pre-commitment approach to minimum capital regulation has merit in that it provides useful information to investors.
References Doherty, N. (1997) “Financial innovations in the management of catastrophe risk,” Journal of Applied Corporate Finance 10(3):84-95. Embrechts, P., A. MacNeil, and D. Straumann, 1999, “Correlation: Pitfalls and Alternatives,” preprint ETII Zurich, Switzerland Federal Deposit Insurance Corporation (2000), Division of Supervision Manual of Exam Policies. Federal Register (2001) “Capital Requirements for Federal Home Loan Banks: Final Rule,” 66 FR 8262, Government Printing Office, Washington DC. Financial Services Authority (2000) Insurance Draft Interim Prudential Sourcebook, London. Financial Services Authority (2000) Interim Prudential Sourcebook for Building Societies, London. General Accounting Office (1991), Government Sponsored Enterprises: A Framework for Limiting Government Exposure to Risks, Report to Congress, Washington DC. Harrison, M. and S. Pliska (1981) “Martingales and Stochastic Integrals in the Theory of Continuous Trading,” Stochastic Processes and their Applications, v11, pp. 313-316. Hogan, A. and D. Nickerson (2000) “Attracting Financing for Catastrophic Risks: Does Zero-beta Accurately Measure Catastrophic Bond Risk?” Working Paper, Economics Department, Colorado State University.
59
Indeed this seems to be the thrust of the regulatory capital approach to supervising financial holding companies and complex conglomerate---but that is a topic for another paper.
36
International Association of Insurance Supervisors (2000) “On Solvency, Solvency Assessments and Actuarial Issues,” Issues Paper, Bank for International Settlements, Basel, Switzerland (http://www.iaisweb.org/framesets/map1.html ) Klugman, S. and R. Parsa (1999) “Fitting bivariate loss distributions with copulas,” Insurance Mathematics and Economics 24(2):139-148. Kupiec, P. (2000) Stress Tests and Risk Capital,” Journal of Risk, Summer, pp. 22-46. Kupiec, P. and James O’Brien (1997). “Recent Developments in Bank Capital Regulation of Market Risks,” in Advances in Finance, Investment and Banking: Derivatives Regulation and Banking, Barry Schachter editor, Amsterdam: North Holland. Kupiec, Paul, and James O’Brien, (1996). “Commitment is the Key,” Risk, September. Kupiec, Paul, and James O’Brien, (1996). “Pre-Commitment,” The Financial Regulator, Vol. 1, No. 3, pp. 41-46. “ Kupiec, Paul, and James O’Brien, (1998). “Deposit Insurance, Bank Incentives, and the Design of Regulatory Policy,” The Federal Reserve Bank of New York Economic Policy Review, Vol. 4, No. 3. National Association of Insurance Commissioners (1994) “Risk Based Capital for Insurers Model Act,” Model Act #312-1, Kansas City, Missouri. National Association of Insurance Commissioners (2000) “Reserves on Certain Annuities and Single Premium Life Insurance Policies Cash-Flow Tested for ReserveAdequacy Factor Based RBC,” Kansas City, Missouri Nickerson, D. (1998) “Governance, Capital Regulations and Efficiency,” Memorandum for Capital Markets Group, Office of Thrift Supervision. Rantala, J., 1982, Solvency of Insurers and Equalization Reserves (Vol. II), Insurance Publishing Co. Helsinki. Tietmeyer, H. (1999), Report to the Finance Ministers and Central Bank Governors of the G-7, Deutsche Bundesbank. Wang, S., 1997, Aggregation of correlated risk portfolios: models and algorithms. Casualty Actuarial Society.
37
38
