Abstract

Central to this research is two interrelated studies concerning the measurements of international bank capital profile and return performance in the contexts of prudential and functional regulation.  The first study argues that the conventional measures of capital and return, such as the Basel Accord's capital standard (Basel) and the accounting-based return on capital employed (ROCE), are not adequate in capturing the overall riskiness of the banks operating and competing in the new global financial environment.  The second study contends that, despite a high prospect of regulatory convergence among countries due to increasing international cooperation and multilateral reciprocity, differences in banking regulations among those countries still exist.  Two alternative capital and return measures have been proposed, namely the capital-at-risk (CAR) and the risk-adjusted return on capital (RAROC), to supplement and enhance the effectiveness of conventional measures.  These new measures help alleviate the problems of asymmetric information in signaling or disclosing more information about the bank's risky portfolios from financial reports based on the distribution of its loss reserves.  In order to test the effectiveness of the new �capital measure,� a ratio between regulatory capital and economic capital is used to indicate the degree of willingness of the bank to provide capital cushion against total risks.  Likewise, the �return measure� expressed as the ratio between economic return and accounting return represents the level of ability of the bank to generate adequate returns that can compensate for total risks.  These two ratios are specified as the dependent variables in order to find any association with the independent variables given by functional and prudential regulations.  The first hypothesis is that functional determinants such as dynamic asset-liability gaps and opacity ratio can explain the willingness and the ability of international banks better than prudential determinants can through such standard ratios as liquidity and solvency.  The second hypothesis follows that functional determinants can differentiate the willingness and ability among international banks of various countries better than prudential determinants can.  Based on the pooled regression of panel data of one hundred banks from nineteen countries categorized into five regions, it is found that both functional and prudential determinants perform equally well in explaining both capital and return measures, which refutes the first hypothesis.  However, only functional determinants can differentiate the willingness and the ability of the banks among five regions.  It is concluded that the new measures can effectively supplement the conventional ones in light of the second study.

1.1          Statement of the Problem

The research problem for this study centers on the measurements of bank capital profile and return performance.  Conventionally, the bank regulatory agencies rely on prudential measures such as CAMEL (capital, assets, management, earnings, and liquidity) index and the Basel Accord's capital adequacy ratio, which is based on forfeit risk-weighted assets to standardize the bank's capital cushion for credit risk.  Several improvements have been made on prudential capital regulation to extend such measures to cover the market risk, most notably the value-at-risk (VAR) measure developed during the first half of 1990s (Jorion, 1997; Best, 1998; Dowd, 1998).  However, as the banking communities around the world have gradually undergone structural changes from being institutionally based financial intermediary to functionally based financial supermarket due to competitive pressure and technological impetus (Merton, 1992; Merton and Bodie, 1995), the conventional measures of banks� capital cushion for credit and market risks are deemed inadequate, especially in preventing the more subtle bank-own risks like derivative and operational risks.  Therefore, without an up-to-date or more effective means to measure banks� opaque operations, the banking industry in the eyes of the regulatory agencies as well as the general public would still be relatively fragile and subject to unpredictable failure, which undermines the systemic stability of the global economy.

The proposals for and development of functional regulation during the past decade as a complementary framework to prudential regulation have yielded a substantial impact on the banking industry as a whole in which major banks� activities are regulated separately to ensure their compliance with the overall objectives of safety and soundness as well as competitive efficiency improvement.  Despite its potential benefits, functional regulation is still in an early, controversial stage and quite difficult to enforce due to the lack of appropriate measures of functional impacts on the bank's capital profile and return performance.  Two new measures have been theorized in the context of functional regulation namely, the capital-at-risk (CAR) measure and the risk-adjusted return on capital (RAROC) measure (Bessis, 1998).  Yet, their applicability is limited to the fact that they have not been empirically tested.  Therefore, it is in the interest of this research that these new measures are evaluated based on the banks' past performance in terms of their congruence with regulatory objectives as well as international regulatory convergence.

1.2          Purpose of the Study

The purpose of this study is to evaluate the effectiveness of the new measures for bank capital profile and return performance in two folds.  These measures can primarily be used to estimate a bank's willingness to provide capital cushion against foreseeable risks and its ability to generate future returns that absorb those risks.  In the first study, the two measures will be tested against the objectives of both functional and prudential regulations using historical data obtained from international banks.  Also, how well these measures perform in determining the degree of regulatory convergence among different countries is the subject of the second study.

The evaluation of international banks' financial reports both on the cross-sectional and intertemporal dimensions will provide an enriched analysis of how well the new measures perform vis-�-vis the conventional measures.  This can be achieved through the use of Panel Data Analysis of Pooled Regression (Hsiao, 1986; Griffith et al., 1993), which will be outlined and discussed later in research methodology section.

1.3          Structure of the Study

The structure of this research is divided into five parts.  The statement of the problem and the purpose of the study as well as its contribution are outlined in Chapter One.  Chapter Two lays the conceptual foundations for the economics of banking to date.  In this chapter, theories of bank failure, empirical studies on signaling/disclosure behavior of the firm�s financial health, various banking regulatory frameworks, and the measurements of bank capital profile and return performance are discussed and reviewed.  Chapter Three involves the establishment of two-stage research hypotheses and includes the quantitative methodology used in empirical testing of these research hypotheses.  The research methodology section contains three sub-sections: variables determination, variables derivation, and statistical method.  Chapter Four gives the summary and discussion of empirical findings.  Finally, implications of the empirical findings for the banking industry and on the public policy, research limitations, and future research directions conclude this research in Chapter Five.

Four appendices that follow are provided to substantiate the overall study.  They comprise the definition of terms to be used throughout the research, the name list of 100 international banks that are participating in this study, the derivation of capital and return measures on an individual bank basis, and the input data for statistical analysis.

1.4          Contribution of the Study

It is expected that this study will provide a better justification for the implementation of the new measures of bank capital profile and return performance in light of the dynamics in the international banking industry.  In this connection, the bank regulatory agencies and the market participants will be able to utilize these measures for their own purposes.  In addition, the findings of this research will provide additional insights to the advances in both financial economic and public policy disciplines with respect to the ongoing issues in international banks' risk management and financial engineering activities.  Specifically, it utilizes a positive approach, rather than a normative argument, to refute conventional wisdom that prudential regulation is most effective in lessening bank fragility and thereby systemic instability.  More importantly, its results will shed light on how policy makers in industrialized and emerging countries can modify their current systems of banking regulation and supervision to improve both performance and safety within their domestic banking industries.

The study will also pave the way for further investigation into the area of regulatory replications, coined by Crockett (1997), the general manager of the Bank for International Settlements (BIS), whereby bank regulatory and supervisory policies are objectively formulated, implemented, and adapted by all parties concerned to make banks function more efficiently and effectively without unnecessary and costly government intervention.  The contributions of this research are to enhance not only the theoretical concepts of financial intermediation, information, and regulation and their empirical implications in the academic sphere, but also the practicality in policy-making and professional worlds.

1.5          Acknowledgements

The author wishes to express his deepest gratitude to all academic advisors, colleagues, and institutional providers of primary data as well as personal inspiration without whom this research will not accomplish its intended objectives. Despite best efforts, all unintentional errors and omissions shall be the sole responsibility of the author.

Literature Review

Four building blocks pertaining to this research consist of 1) the potential causes of bank failure, 2) various indicators of bank fragility, 3) the economics and frameworks of banking regulation, and 4) the measurements of bank's capital profile and return performance.  The first deals with two premises of bank's operating fragility, i.e., external misfortune and internal mismanagement, which lead to the eventual institutional failure.  It also covers the theories and models developed in light of bankruptcy classification and prediction.  The second block is related to the first one in that these two potential causes of bank failure can be detected beforehand through some objective signaling/disclosure indicators.  Researchers in financial economics have attempted to find these indicators that are closely representative of the behavior of the problem banks, or of the troubled firms in general.  A summary of empirical research on signaling/disclosure behavior of the firms is given in this block.  Third, with the combined effects of institutional and functional dynamism and the persistence of agency problems, three frameworks of regulatory economics will be discussed, namely the mandate-based, the market-based, and the model-based frameworks of banking regulation.  And finally, the third block introduces both conventional and alternative measurements of bank capital profile and return performance from the perspectives of prudential and functional regulation.  These measurements will serve as the basis for empirical investigation as to whether the alternative ways of evaluating bank capital profile and return performance is more effective than the conventional ones.

2.1          Potential Causes of Bank Failure

The fragility and eventual failure of banks operating domestically and internationally can be predicated on either mismanagement or misfortune in their increasingly opaque and illiquid asset-liability profiles based on sophisticated risk management strategies and financial engineering techniques.  Mismanagement in the banks' asset and liability portfolios involves imprudent lending and investment decisions and excessive risk-taking without adequate internal risk control.  Most banks have functional motives and regulatory incentives to conceal their risk management activities from outsiders because of the existence of asymmetric and imperfect information.  Economics of differential information is thus one of the prime concerns of international banking regulators to optimize the trade-off between global financial stability and the promotion of efficient and competitive systems of financial innovation.

Mishkin (1986) and Edwards (1996) list a series of historical events that indicate the clues to potential causes of bank failures: 1) Bankaus Herstatt for time zone-induced settlement risk and Franklin National Bank for foreign-exchange trading risk, both in 1974; 2) Continental Illinois Bank for default risk from foreign loans in 1984; 3) the Bank of Credit and Commerce International (BCCI) for money-laundering activity in 1991; 4) Credit Lyonnais Banque for its investments and derivatives mismanagement in 1994; and 5) Barings PLC and Daiwa Bank for poor internal controls and operational risk in 1995.  These events triggered either bank runs or depositor runs, depending on how transparent and sensitive the issues were to the markets and how well the regulatory agencies coped with them.  Basically, bank fragility and subsequent failure can stem from two sources: 1) external misfortune, and 2) internal mismanagement.

External Misfortune

Perceivably, the general cause of bank failures is deteriorating quality of banks' loan and investment portfolios due to market prices fluctuation and/or returns volatility.  Large banks, on the one hand, fail when loan losses so erode earnings that uninsured creditors become frightened and refuse to roll over deposit and non-deposit funds. On the other hand, small banks fail when rising loan losses cause regulators to declare them insolvent and bankrupt.  According to Rose (1989,1990), the risk in a portfolio of bank loans depends on three components: 1) the risk of each individual loan, 2) the share of the portfolio devoted to each loan, and 3) the correlation among loan assets, i.e., the extent to which borrower�s portfolios tend to move together.  Practically, banks have spent much time assessing individual loan risks through analysis of borrower�s balance sheets but usually devoted little attention to the other two components of loan risk.  Other alternative explanations that lead to banks' portfolio losses are based on competitive pressure from non-bank financial services firms disintermediating traditional bank loans and crowding out commercial banking operations.  As a consequence, banks have no alternative but to take higher risk in lending to speculative projects in order to survive the competition.

To this regard, the U.S. Federal Deposit Insurance Corporation (FDIC) has expressed its concern over the banks' inferior loan quality (Hove, 1998).  By observing the lending practices at the 1,212 banks during the six months ended March 31, 1998, FDIC found that of the 387 banks, about 31 percent frequently funded speculative construction projects, compared with 18 percent of the 593 active construction lenders in the report a year earlier.  Those speculative real estate projects have been known to be susceptible to market conditions whereby banks are forced to bear the ultimate risk of loan losses. While weakened underwriting standards are still most evident in commercial real estate and construction lending, such an indication of adverse trends of risk-taking behavior has triggered the FDIC to maintain its close monitoring of underwriting practices for all major types of bank loans.

Internal Mismanagement

The U.S. Office of the Comptroller of the Currency (OCC) determined that the primary cause of bank failure is mismanagement by imprudent managers who are either inexperienced in banking or lax in their supervision of a bank's affairs (Clarke, 1992).  The OCC found that the most common characteristic of failed banks was that they were poorly managed.  Most banks failed to establish and adhere to policies by condoning the development of managerial direction and routine that would guide them through volatile business and economic cycles.

In the study, the OCC examined 171 failed national banks between 1979 and 1987, of which 51 had come close to failure but recovered and 38 healthy ones of similar size and geographic location had faced the same economic conditions.  The study found that half of the failed banks had boards that either lacked necessary banking knowledge or were uninformed or passive in their supervision of the bank's affairs.  And more than half of the rehabilitated banks had similar inexperience among their directors during their declining days. It also found that most failed banks had no standard loan policies, poor documentation and inadequate systems to insure compliance with lending rules.  In addition to the laxity by directors, the lack of an experienced chief executive contributed most frequently to failures. Two-thirds of failed banks had chief executives who clearly lacked the capability, experience or integrity necessary to make their banks successful.   Finally, insiders' abuse and fraud had led to the demise of banks in about one-third of the failures.

Bankruptcy Classification and Prediction Models

The study of the causes and consequences of corporate fragility and failure has been a domain of theory of bankruptcy since the 1960s.  At that time, researchers were interested in explaining the conditions experienced by, and the characteristics of, the firms that lead to business demise.  In other words, they have been in search of effective indicators and predictors of corporate bankruptcy using several analytical and statistical techniques.  Traditionally, most practitioners prefer the arbitrary rule-of-thumb-based ratio analysis.  Prior to the development of statistics-based quantitative analysis, the measurement and evaluation of company past performance had been derived from accounting information.

Among the works in the area of ratio analysis and bankruptcy classification was the one conducted by Beaver (1967) utilizing univariate factors.  More specifically, his "paired sample method" is based on a use of selected sample of industry and asset size to find the relationship between financial ratios and business failure.  He argued that these two factors should be controlled because the same ratios may imply a different probability of bankruptcy across industries.  Generally, ratios measuring profitability, liquidity, and solvency prevail as the most significant indicators.

The accounting data selected by Beaver were based on three criteria: 1) popularity in the literature, 2) performance in previous studies, and 3) definition of the ratio in terms of cash flow concept.  Six variables are considered as having the most robust explanatory power for bankruptcy classification and prediction:

1)       Current ratio

2)       No-credit interval

3)       Cash flow to total debt

4)       Net income to total assets

5)       Total liabilities to total assets

6)       Working capital to total assets

However, the shortfall of Beaver's method is that the controlled variables of industry and asset size may be important predictors of failure yet remain undetected because their predictive power is masked by the paired sample technique.  He also provided caveats about the use of univariate ratio analysis for bankruptcy prediction.  First, ratios have greater success predicting non-failure than failure.  Second, not all ratios predict with the same degree of accuracy.  And third, for decision-making purposes, frequency distributions and likelihood ratios should complement financial ratios.  Despite self-imposed and others' criticisms, he concluded that financial ratios have a desirable ability to predict failure for at least five years before the firms actually go bankrupt.

With several shortcomings of the univariate model in mind, particularly the third caveat, Altman (1968) attempted to improve upon Beaver's approach by integrating a so-called "multiple discriminant analysis" (MDA) into the model.  MDA is a statistical technique used to classify an observation into one of several a priori groupings depending upon the observed individual characteristics.  It is also used to derive a linear combination of such characteristics which best discriminates between the groups, the same way the multiple regression technique, such as ordinary least squares (OLS), is used to derive a linear combination that offers the least random error among variables.  The result of MDA is in the form of discriminant function, which according to Altman is called the "Z-score model," utilizing multivariate factors.

The Z-score multivariate factors are selected based on the following procedures: 1) observation of the statistical significant of various alternative functions including determination of the relative contributions of each independent variable; 2) evaluation of inter-correlations among the relevant variables; 3) observation of the predictive accuracy of the various ratio profiles; and 4) judgment of the analyst.  As a result, five independent variables are pinpointed:

1)       Sales to total assets

2)       Working capital to total assets

3)       Retained earnings to total assets

4)       Earnings before interest and taxes to total assets

5)       Market value of equity to book value of total liabilities

While the Z-score method produces much improvement over the paired sample method, the MDA model does not recognize and scrutinize very large or very small firms, the latter group comprising the largest number of business failures.  Thus about nine years later, Altman, Haldeman, and Narayanan (1977) came up with an improved method called the "ZETA model," which researchers and practitioners alike have found more effective.  There are five considerations that make the ZETA model superior to the original Z-score model:

• The ZETA model utilizes a bankrupt firm sample where the average adjusted asset size two annual reporting periods prior to failure was approximately $100 million as opposed to the Z-score model where the largest firm had assets of less than $25 million.

• The ZETA model can be made as current as possible with respect to the temporal nature of the data.

• With analytical adjustments, a particularly vulnerable group of firms can be analyzed on an equivalent basis with groups of firms in other industries.

• The ZETA model can accommodate the most recent changes in financial reporting standards and accepted accounting practices, e.g., lease and R&D capitalization, reserves, minority interests, and intangible assets.

In terms of empirical efficacy, Altman et al. concluded that the ZETA model appears to be quite accurate for up to five years prior to failure, with successful classification of well over 90 percent one year prior and 70 percent accuracy up to five years.  As a result, several large banks have utilized the ZETA model in their operations because of the model's satisfactory robustness.

2.2          Indicators of Bank Fragility

Empirically, two types of indicators can be used to signify financial health of the firm: ex ante signaling indicators and ex post disclosure indicators, both of which can be derived or observed from the firm�s financial statements.  Considering the previous empirical studies conducted on the firms� financial indicators, it is unfortunate that there is no pattern that matches the purpose of this research by studying the reserves for potential losses as an indicator for the bank fragility.  Those studies using capital structure as indicator are quite close to becoming a principal theoretical basis for banks' signaling/disclosure behavior.  As addressed earlier, the main contribution of this research is on how to verify the validity and effectiveness of the newly proposed measures of bank capital profile and return performance amid the changes and shifts in the current banking operations.  Since it is difficult to measure opaque activities of the bank, the only measures we can rely on is its ex post disclosure of capital profile and return performance based on reserves for potential losses.  And since the previous studies on signaling behavior have not yet focused on this issue, it is worthwhile exploring into such new indicators.

Following the past empirical studies on signaling/disclosure behavior of the firm, five patterns are identified: 1) capital structure, 2) investment, 3) equity-related, 4) debt-related, and 5) dividends payout.  First, capital structure has been studied in at least two aspects in terms of optimal capital structure in the presence of asymmetric information and the use of debt financing as a means to indicate the firm�s financial strength.  Second, the observed investment patterns in relations to debt capacity and dividend payouts can indicate how prudent the firm has been.  Third, the ways in which the firm issue, repurchase, and split common stocks reveal certain information about its investment or portfolio quality as well as its risk exposures.  Fourth, the decisions to issue various types of debt instruments indicate the firm�s earnings ability and the management�s willingness to be monitored by the markets.  And fifth, dividend-payout patterns are regarded as important indicator of management�s concealed behavior.  Table 1 summarizes the rationale, evidence, and finding of these studies:

Table 1

Summary of Past Empirical Studies

2.3          Economics of Banking Regulation

Following the collapse of banks in various countries during the last two decades, many questions have been raised as to how effective the current regulatory systems are in reducing the banking industry�s fragility and in preventing systemic instability.  Until recently, many regulatory agencies at both national and international levels have resorted to prudential regulation which mandates banks to maintain a minimum level of capital to cushion their credit and market risks.  The emphasis of prudential regulation is on the "fiduciary" relationships between the banks' shareholders and the general depositors in terms of potential problems of conflict of interest within the banking industry.  Yet, prudential regulation has become more costly and less effective than regulation through market forces, in which the focus is on the "agency" relationships between the banks' management and the shareholders.  Some market-based private regulation and model-based functional regulation have been proposed as the alternatives to prudential regulation.

This section provides a review of theory of banking regulation and supervision based upon various economic frameworks of financial regulation, as well as a discussion of it in relation to the problems of differential information.  Banking regulation arises mainly for two reasons: systemic stability and contractual reliability.  Systemic stability deals with the issues of institutional safety and soundness in general in which banks are expected to perform �competent,� �prudent,� and �honest� economic functions.  This regulatory objective is explored more in the sub-section of regulatory theory of public interest.  Contractual reliability, on the other hand, is concerned with the relationship between the banks and their clients that they are required to perform intermediation functions with �reliability� and �accountability.�  Under the sub-section of supervisory theory of private interest, the regulatory objective to monitor and enforce contractual obligations will also be discussed.  While the first two sections emphasize regulatory objectives, the last section will discuss three alternative regulatory frameworks in connection to the economics of information and the regulatory players.

Regulatory Theory of Public Interest

As banks can incur significant social costs and negative externalities via bank runs, it is required that they be subject to strong scrutiny both from within the financial markets and the public authority (Miller, 1994).  The main objective of public scrutiny and oversight is to maximize welfare and avoid market failure.  Economic efficiency and social welfare can be achieved and derived from four areas that the banks are expected to facilitate: allocation, distribution, transaction, and information.  The avoidance of market failure is to ascertain that banks will not operate sub-optimally and be the primary cause of systemic crisis.  The criteria for welfare maximization are high performance and considerate prudence, while the criteria for market failure minimization are high liquidity (safety) and adequate solvency (soundness).

Systemic Stability Objective

Due to the potential competitive inefficiency in the banking industry resulting from specific rent-seeking behavior or market failure in general, which can lead to a system-wide banking and financial crisis, banks are very tightly regulated.  Three kinds of behavioral characteristics are the foci for banking regulation: competence, prudence, and honesty (Kane, 1987).  In order to guarantee a minimum level of banks' competence, an entry regulation such as bank registration or fit and proper standard is required.  However, just because the banks could attain the minimum competency level does not mean that they will execute their best efforts with prudence to represent the public and maximize the uninformed depositors' interests.  Thus, the prudential regulations such as activity firewalls and geographical restriction are usually mandated.  To elicit honest behavior from the banks that they contemplate the institutional safety and soundness as their first priority, many ex ante safeguarding measures, e.g., capital adequacy and required reserve ratios and ex post safety nets, e.g., discount window, deposit insurance, and lender-of-last-resort, are imperative.

The negative consequences of systemic instability can be put in sequence as follows: 1) depositor runs due to short-run illiquidity of the banks, 2) bank runs and insolvency problems as a result of mismanagement or misfortune, 3) systemic collapse of the overall banking system from herd-like withdrawing panics due to the public inability to distinguish between healthy banks and distressed or failed banks, and 4) the contagion effects of systemic crisis that spill over to other financial markets in domestic and other countries and result in negative externalities to the real sectors of the economy.  Therefore, it is of the utmost importance that the government intervene to regulate banking industry to prevent a series of undesirable consequences.

Competitive Efficiency Objective

Two perspectives of regulation for competitive efficiency are used to explain why banks are subject to government regulation: 1) the rent-seeking behavior of the banks, i.e., self-interest maximization, and 2) the negative externalities the banks generate toward the economy, i.e., market failure.  In the rent-seeking perspective, Stigler (1971) argues that market participants will demand regulation when the expected marginal benefits of such regulation exceed the expected marginal costs.  Regulation is demanded as a means by which a bank can raise the costs of its competitors or reduce its costs while leaving the costs of its competitors unchanged.  When banks demand regulation that is designed to increase their market share at the expense of their competitors, they are said to be engaged in rent-seeking behavior.  Economic rent is the abnormal profit a firm can earn when its market power is increased.  Thus, rent seeking is the process by which banks allocate real resources to lobbying and the use of pressure groups to generate rents from regulation.

Market failures refer to conditions under which the market-guiding invisible hand of voluntary exchange would fail to direct society's resources to an allocative pattern capable of achieving optimum social welfare.  Three types of market failure are: 1) sub-optimal Pareto allocation, 2) natural or informational monopolies, and 3) negative externalities and social costs.  A market failure is presumed to occur when market participants do not take into account all the costs of their decisions.  A bank's private cost, or the cost that governs its private behavior, is presumed to diverge from the true social cost of the firm's decisions, thus resulting in an externality and a failure of the market to allocate resources most efficiently.  Regulation is an appropriate means of correcting this inefficiency and other market failures.

Supervisory Theory of Private Interest

From the standpoint of the fiduciary nature of the banking industry and its relation to the markets, efficiency in contractual relationship with respect to the reliability and accountability of the banks is the primary concern for banking supervision (Miller, 1994).  Supervisory theory of private interest searches for low-cost mechanisms by which to reconcile conflict between regulators' private and societal goals.  Resource costs are associated with the markets' need to control its banking industry.  It is desirable for the markets that this control be produced at minimum cost.  To achieve this objective, banking supervision, therefore, has to engender contractual efficiency and bank-specific or idiosyncratic transparency.

Contractual Efficiency Objective

Private and public fiduciary relationships may be viewed as contracts that delegate obligations to the counterparties.  Each obligation establishes a principal-agent relationship.  Every agent has an objective function that differs from that of the principal.  Every agent faces a temptation to evade some of its obligations whenever it can hide self-serving actions or other relevant information from the principal.  Contracting efficiency looks past market failures to focus on society's need to monitor and police at minimum resource cost.  Jensen and Meckling (1976, 1994) describes the central proposition of agency theory as follows: rational self-interested people involved in cooperative endeavors always have incentive to reduce or control conflicts of interest so as to reduce the losses resulting from them.  A reliable outside supervisor can improve the fairness, efficiency, and enforceability of financial agreements by offering to mediate transactions in which the interests of bank's management, shareholders, debtholders, and depositors diverge.  Agency theory recognizes the value of incentive-based compensation and ex post settlement as ways for society to constrain the tradeoffs between public and private benefits that are made by banks' management and by the government and private parties that supervise them.  Agents are willing to accept tighter control only if they are properly compensated for the benefits that tighter controls take away from them.

Idiosyncratic Transparency Objective

One major incentive for the banks to pursue their self-interest and make their financial contracts with outsiders ineffectual is the presence of differential information.  Several supervisory options can be used to resolve this idiosyncratic opaqueness of the banks' behavior including internal control, industry oversight, and public oversight (Smith and Walter, 1997).  First, internal control refers to resistance to financial failure, abuse of fiduciary responsibilities, and other lapses in management.  Management problems should be reflected in the reputation of the bank, which in turn will be reflected in its value.  Thus, an important responsibility of bank's management is to maximize its franchise value, and put in place the necessary safeguards against its erosion.  If the bank is subject to a high level of transparency, the pressure on management to enforce safe and sound practices and high standards of conduct may obviate external supervision that could entail greater regulatory burdens.  Second, industry oversight refers to supervision by industry association and self-regulatory organizations (SROs) to encourage and enforce high standards on their members and to discipline them for non-compliance.  It may be that SROs, together with external audits by public accounting firms and the threat of lawsuits by disgruntled private parties, can alleviate shortcomings in self-regulation without incurring some of the regulatory costs associated with external supervision.  Third, public oversight refers to supervision by agencies set up for a specific purpose whose mandates are anchored in regulatory statutes.  This requires an infrastructure of qualified, motivated banking examiners who have at their disposal both civil and criminal penalties.  Most financial systems have found that a structure of external supervision is necessary no matter how effective internal control and SRO supervision become, since there are always gaps that could lead to instability and breaches of appropriate conduct in financial markets.

Alternative Regulatory Frameworks

To alleviate asymmetric information-induced problems in the banking industry, three alternative regulatory frameworks are categorized along the lines of regulatory players: 1) publicly-mandated prudential regulation, 2) market-based private regulation, and 3) model-based functional or self-regulation (Culp, 1995).

Mandate-Based Regulation

As discussed before in the regulatory theory of public interest, prudential regulation incorporates both preventive (safeguarding) and protective (safety-net) measures in mandating the banking industry to behave in the interest of the public at large according to the safety and soundness doctrine as well as to perform financial intermediation functions in the manner that meet the market expectations for competitive and contractual efficiency (Walter, 1996; White, 1996).  The preventive measures include various activity restrictions, such as the fit and proper entry requirements and the separation between commercial banking and investment banking functions, as well as disclosure requirements and monitoring such as the SEC�s mandate on off-balance-sheet financial reporting and early warning signals relating to liquidity and solvency positions of the banks.

The protective measures, on the other hand, include capital adequacy ratios, deposit insurance, and lending of last resort from the central bank.  Capital requirement such as the Basel Accord capital adequacy ratio is imposed on banks as a self-insurance against risk of losses from their credit and investment operations.  Deposit insurance is used to protect depositors� funds thereby building strong public confidence in the banking systems in the event that some banks are faced with temporary liquidity problem.  Finally, lender-of-last-resort scheme is deployed as the ultimate safety net to rescue the banks that have serious liquidity problem but still possess a sound solvency position.

Market-Based Regulation

Market-based regulations deal with the efforts by the banking industry to self-regulate and impose market discipline on the conduct of banking activities.  They involve practices and organizations like industry watchdog by bankers associations or societies, performance monitoring by independent credit-rating agencies, and accounting-disclosure compliance by auditing firms.  Despite these engulfed self-regulatory features and strong market discipline, market-based regulations still suffer from the lack of enforcement mechanisms that can otherwise be imposed upon problem banks that are on the verge of bankruptcy.

In connection to the supervisory theory of private interest, agency theory has given rise to various private regulatory applications in both corporate finance and financial institutions.  Costly monitoring, free riding, corporate governance, and shareholder activism are among the most prominent areas of research in the field of market-based regulation, which build on the frameworks of securities and incentive systems design.  In essence, the theory addresses the two-stage conflictual relationships between principal and agent.  The first stage involves shareholders as being the agent for debtholders (depositors) in monitoring the bank management�s decisions and activities.  The second stage is concerned with the bank management as being the agent for shareholders in monitoring the performance of the bank�s loan, investment, and derivative portfolios.  However, since behavioral monitoring is costly the market-based regulation encounters an inevitable free-riding problem whereby debtholders free ride on the shareholders� effort to monitor the bank�s risk-prone activities.  The tensions and interactions between debtholders and shareholders of the banks are manifested in the optimal design of debt contracts and covenants.  Despite optimal securities design, the tradeoff between information monitoring and free-riding costs still represents a deadweight burden, which the banking industry as a whole has to shoulder.

The inherent conflict of interests between shareholders and the bank management can potentially be resolved through optimal incentive systems design (Anderson and Sundaresan, 1996).  Theoretically, the effectiveness of incentive systems rests upon how active the incumbent shareholders are in influencing the behavior of the bank management through the board of directors.  The so-called �corporate governance� specifically deals with the issues like markets for corporate control, executive compensations, and risk-adjusted performance measurements to allow shareholder activism to play a more important role in less transparent decision-making (Giammarino and Neave, 1982; Berkovitch and Israle, 1996).

Model-Based Regulation

The concept of model-based regulation stems from an increasing exposure of corporations and banks to credit and market risks.  In corporate finance, credit and market risks can be shared or shifted to counterparties through hedging and risk management activities.  Dynamic trading of and assuming hedged positions in underlying and derivative financial instruments are not uncommon to corporate risk management practices (Walmsley, 1998).  In the banking industry, however, the management of credit and market risks entails much more sophistication than a mere hedging activity in that banks, as a prime provider of those financial instruments, have to be engaged in the ongoing financial innovation and opaque engineering activities which involve complex mathematical programming and stochastic models.  Such activities collectively increase the operational risk of the banks.

Due to the opaqueness of the banks� financial engineering activities, it is extremely difficult for the outsiders to observe or measure the risk of their derivative portfolios, let alone to measure their intrinsic values.  The only way in which banks can disclose or report their portfolio positions is through the use of option-pricing model, or more generally, contingent-claim models.  For example, many banks currently incorporate the VAR model to measure their portfolios� daily exposure to market risk (Jorion, 1997).  This has made banking regulation even more difficult for public regulators, depositors, and shareholders alike.  The proposed model-based regulation focuses on CAR and RAROC as the main indicators of banks� risk and potential fragility.  One limitation of these measures is that they are derived internally by the banks in question, which in turn are based upon the historical profiles of their portfolios in connection with the current asset-liability management and the expected rates of return from those portfolios.

International Banking Regulation Environments

The surveys on the differences in international banking regulations are considered the important aspect of the study on the trends toward regulatory convergence and international banking cooperation.  Baker (1978) is among the first to survey and discuss the different systems of bank regulation and supervision as well as bank deposit insurance programs among various countries including the United States, Western Europe, Middle Eastern countries, and Japan.  Along with this comparative study, the rationale for and the causes of growth of international banks� presence and operations are provided as the primer to comparative regulatory regimes.  With the growth of internationalization of the banking industry during the last three decades, issue like regulatory arbitrage in which international banks from one country reap the benefits of lax regulation in another thereby increasing systemic risk has been a major concern.  Hence, the need for bilateral and multilateral reciprocity as well as the harmonization of bank regulatory systems to remove the incentives for regulatory arbitrage and the causes of fragility of the international banking system has become a mandate for increasing cooperation at the international level.

Dale (1985,1992) also conducted similar comparative survey focusing on different aspects of banking regulation, which encompass entry restriction, activity restriction, capital adequacy, liquidity control, foreign currency exposure, loan concentration and country risk, and bank inspection and examination.  Countries under survey include the Benelux nations, Canada, France, Hong Kong, Italy, Japan, Singapore, Switzerland, United Kingdom, and United States.  It is concluded that different countries put different weights to their regulatory aspects depending upon the concentration of domestic banking industry.  Some countries exercise tighter control over the banks� management and operations when the banking industry is less concentrated.  The survey also indicates the shared regulatory practices and common framework that have been increasingly adopted as international cooperation began to develop during the mid-1970s after the establishment of the BIS.  In spite of this favorable trend toward regulatory convergence, the issues of supervisory jurisdictions and enforcement have yet to be settled at both national and international levels.

A more in-depth comparative study on banking regulation and supervision was conducted by Hall (1994) by scrutinizing among the three industrialized countries, i.e., United Kingdom, United States, and Japan, whose banking industries always remain top ranked in terms of size and operations.  Differences in regulatory aspects are mainly attributed to different interpretation and adoption of rules and standards by the three national governments.  Interestingly enough, banking industries from these three countries are competing effectively against each other both in their respective domestic markets and on the global arena although their home-country regulatory regimes are highly distinct.  This aspect has been ascribed to the ways the national governments employ regulatory discretion and practices to enhance the strength and increase competitiveness of their international banks.  However, with a global perspective on systemic stability and competitive efficiency, it is necessary that further development and certain reforms in international banking regulation and supervision still be promoted.

The most recent global survey on regulatory and market developments in banking, securities, and insurance industries was conducted by the Institute of International Bankers (IIB) in 1997, which covers 50 countries and the European Union.  This comprehensive survey was jointly prepared with the cooperation of the bankers� associations and societies in respective countries.  The interesting feature of this survey is the report on regulatory convergence in terms of the implementation of market-risk capital requirements (standard or internal VAR approaches).  Table 2 shows the list of countries that have adopted either approach for market-risk capital.

Table 2

Countries Adopting Market-Risk Capital Requirements

Source: Global Survey 1997, Institute of International Bankers.

2.4          Measurements of Bank Capital Profile and Return Performance

To successfully operationalize and enforce all three alternative regulatory frameworks to the banking industry, efficient as well as effective measurements of bank's capital profile and return performance have to be established and practiced.  The following six sub-sections discuss the development of regulatory measures that both have already been standardized and are being proposed in the literature.  The first is called the capital adequacy measure, following the Cooke Committee's mandate under the Basel Accord of 1988.  The purpose of this measure is to require banks to provide adequate capital cushion against risk of losses from default loan portfolios, i.e., credit risk.  The second measure, the standard value at risk (VAR) measure, was proposed in 1993 within the Basel Accord framework to address the issue of market risk -- the risk of losses from market value fluctuation of investment portfolios.  However, the international banking community found that this standard VAR measure was not practical to follow since many banks' functional structures were not standardized.  Therefore in 1995, the BIS proposed an alternative measure called the internal VAR to remedy the non-standardization issues as discussed in sub-section three.  Despite these attempts by the BIS, the problems arising from the banks� illiquid instruments and opaque operations have not yet been fully addressed.    An overview of VAR derivation and the various approaches to VAR measure are provided in sub-section four.  In the last two sub-sections, the newly proposed CAR and RAROC measures are theoretically presented in comparison with the conventional BIS regulatory frameworks.

The Basel Accord of 1988 and Capital Adequacy Measure

In January 1987, the Bank of England and the U.S. federal banking regulatory authorities, consisting of the Federal Reserve Board, the FDIC, and the OCC, announced that they had reached agreement on proposals for a common measure of capital adequacy for banks.  The proposals were for a risk-related approach similar in many respects to that already in use in the U.K. and that proposed in papers released by the U.S. regulatory authorities in January 1986.  The proposals also drew on work of the Cooke Committee.

The Cooke Committee's approach was to seek a convergence of the various regulatory methods to form a package that could be used by banking regulators from all of the G-10 plus Switzerland and Luxembourg.  The issues involved were controversial, and the goal of the Committee was ambitious.  The 1987 British-American joint proposals were circulated for comment then adopted by the Basel Committee as a whole in July 1988 and subsequently ratified by each respective country.  The Federal Reserve Board announced its final version of the guidelines in January 1989.  In its announcement of the new guidelines, the Federal Reserve noted that they had been designed to achieve certain important goals:

• Establishment of a uniform capital framework, applicable to all federally supervised banking organizations;

• Encouragement of international banking organizations to strengthen their capital positions;

• Reduction of competitive inequality arising from differences in supervisory requirements among countries.

The guidelines were intended to 1) establish a systematic analytical frame work that make regulatory capital requirements more sensitive to differences in risk profiles among banking organizations, 2) take off-balance-sheet exposures into explicit account in assessing capital adequacy, and 3) minimize disincentives to holding liquid, low-risk assets.  In principle, the Basel Accord's capital adequacy measure is simple although its actual structure is rather complex.  The basic idea is to assign each asset owned by a bank to one of four risk categories.  Each risk category is assigned a risk weight, which is used to multiply the amounts in each risk category to determine the amount of risk-based regulatory capital required by the bank.

The Basel rules included a schedule for implementing the new system worldwide, with a ratio of 8 percent, of which at least 4 percent must be in the form of tier 1 capital.  They now provide a common standard for safe and prudent banking capitalization.  Once countries have agreed on the same minimum base, there is no advantage in being undercapitalized or for countries to unduly subsidize banking institutions by setting interest-rate controls to allow banks to accumulate excess profits as a cushion against future losses, at the expense of economic growth and efficiency.  Riskier instruments have become more costly to hold, lessening the chances of excessive exposure and the prospect that regulators will have to step in to provide support in a crisis.  There is less incentive to underprice off-balance-sheet commitments.  Given the rate of financial innovation and the deluge of new instruments, it was necessary that the guidelines provide regulators with a coherent framework into which to slot new types of exposures as they evolve, instead of always lagging events by as much as several years.  Each new type of instrument is assigned to the highest risk category until such time as the regulators rule otherwise.

There had been difficulties and problem issues concerning the implementation of the Basel rules.  First, the risk-based capital requirements increased pressure on banks to charge higher spreads or fees for financial transactions in which they participate, in order to recover the incremental cost of the additional capital needed to support specific loans and advances, or simply to recover the higher overall cost of capital.  This condition was seen by many banks to place them at a substantial disadvantage relative to securities firms, with which they are increasingly in direct competition.  Securities firms, not being regulated by banking authorities that must look after the deposits they are guaranteeing, are not subject to the new rules or to any similar constraints.  Coordination with authorities regulating the securities industry is essential if competitive rules under which firms in the two sectors (banking and securities) of the industry operate and are not to serve as further distortions to competitive conditions in the case of financial services performed by both bank and non-bank institutions.

Second, the provisions of the Basel Accord are not uniform between banks of different countries.  Wide international differences exist in the availability of information on bank performance, which may influence their relative competitive positioning and certainly affect the ability to determine whether the international competitive playing field is in fact relatively level.  Transparency in U.S. accounting for banks is assured by the regulatory structure, and any disclosure problems are relatively quickly remedies - including cross-border exposures and off-balance-sheet exposures in such transactions as swaps.  In other countries disclosure is far less extensive, and in some cases relatively meaningless.  Disclosure of off-balance-sheet exposures in many cases is absent altogether, and many home countries of multinational banks fail to disclose their worldwide operations on a consolidated basis.

The third issue is whether banks may emerge among the nonparticipating countries that would challenge banks from the participating countries for business and possess a competitive edge over the participating banks through less rigorous regulatory standards.  Conceivably, banks could migrate to unregulated areas for the purpose of competing with the banks from the major countries.  However, since only a small percentage of the world's international banking assets are booked outside the participating BIS countries, the impact of such a migration would not appear to be large.

Finally, it has been argued that coordinated risk-based capital adequacy requirements can actually be counterproductive, since assets categorized in the same risk class may have vastly different risk profiles.  Moreover, since different assets and off-balance-sheet exposures require different levels of capital, the result may well be distortions in banking decisions - for example, loading up on highly interest-rate-sensitive U.S. government securities that require less capital backing than perhaps less volatile asset deployments - decisions that ultimately may lead to increased, rather than decreased, vulnerability of individual institutions.  It could also reduce financial innovation and, as noted, place banks at a competitive disadvantage against non-financial institutions operating in the securities markets that are not subject to similar requirements.

In general, by the mid-1990s banks in most of the advanced countries had attained the BIS guidelines with greater or less difficulty.  Banks in each country faced more or less unique difficulties associated with, for example, loan losses related to real estate and country lending in the U.S. and various European countries, stock-market collapse and the end of the bubble economy in Japan, and the simultaneous creation of a single market in financial services under universal banking conditions in the European Union.

The Revised Accord of 1993 and Standard VAR Measure

The Basel Accord of 1988 has been criticized on many aspects.  First, it did not account for the portfolio risk of the bank.  Correlations between components of the portfolio would significantly alter total portfolio risk.  Credit risk would be offset by diversification across issuers, industries, and geographical locations.  The Accord actually raised the banks' capital requirements from their hedging operations.  Second, it did not account for netting of risk exposures.  If a bank matches lenders and borrowers, its net exposure might be small.  If the counterparty fails, the loss could be small if the amount lent was matched by the amount borrowed.  Netting is an important driving force behind swap contracts.  In the event of default, banks are exposed to only the net exposure, not the notional amount.  Third, the Accord of 1988 poorly accounted for market risk, such as interest-rate risk and foreign-exchange risk.  Assets were recorded at book values, which might substantially differ from their current market values.  As a result, accounting lags could create a situation where a healthy balance sheet with acceptable capital hides losses in market values.  This omission was particularly glaring for the trading portfolio of banks with positions in derivatives.  Recognizing these shortcomings, the Basel Committee moved toward measuring market risk with the value-at-risk (VAR) approach.

The revised Accord proposal, issued in April 1993, is based on a building block approach.  VAR is first computed for portfolios exposed to interest-rate risk, exchange-rate risk, equity risk, and commodity-price risk, using specific guidelines.  The bank's total VAR is then obtained from the summation of VARs across the four risk categories.  Because the construction of VAR follows a highly structured and standardized process, this approach is called the Standard VAR Model.

In terms of interest-rate risk, the proposal defines a set of maturity bands, within which net positions are identified across all on- and off-balance-sheet items.  A duration weight is assigned to each of the 13 bands, varying from 0.2 percent for positions under three months to 12.5 percent for positions over 20 years.  The sum of all weighted net positions then yields an overall interest-rate-risk indicator.  The market-risk capital charge for exchange-rate risk and equity risk is 8 percent of the net position and 15 percent for commodity-price risk.

Although Standard VAR Model aims at identifying banks with unusual market exposures, it is still beset by problems.  The duration of some instruments cannot be easily identified.  Another problem is that the Standard VAR Model does not account for diversification across risks.  Low correlations imply that the risk of a portfolio can be much less than the sum of individual component risks.  This diversification effect applies across market risks or across different types of financial risks.  Correlations across different types of risks are more difficult to deal with.  Default risk may be related to interest-rate risk.  This is true for most floating-rate instruments where borrowers may default if interest rates increase to insufferable amounts.

The New Accord of 1995 and Internal VAR Measure

The second revision of the Accord took place in April 1995 with a major extension of Standard VAR Model.  For the first time, the Basel Committee would allow banks the option of using their own risk measurement models to determine their capital charge, thus the Internal VAR Model.  This decision stemmed from the recognition that many banks have already developed sophisticated risk management system, which, in many cases, are more complex than that of the Standard VAR Model.  As for banks lagging behind the times, this new proposal provides a further impetus to create sound risk management systems.

In this Internal VAR Model, banks have to satisfy various qualitative requirements, including regular review by various management levels within the bank and by regulators.  More specifically, the Internal VAR Model is based on the following approach:

• The computation of VAR shall be based on a set of uniform quantitative inputs: 1) a horizon of 10 trading days; 2) a 99 percent confidence interval; and 3) an observation period based on at least a year of historical data and updated at least once a quarter.

• Correlations can be recognized in broad categories as well as across categories.

• The capital charge shall be set at the higher of the previous day's VAR or the average VAR over the last 60 business days, times a factor called "hysteria" factor.  The exact value of this factor is to be determined by local regulators, subject to an absolute floor factor of 3.0.  This factor is intended to provide additional protection against environments that are much less stable than historical data would lead to believe.

• A penalty factor shall be added to the hysteria factor if back-testing reveals that the bank's Internal VAR Model incorrectly forecasts risks.  The purpose of this factor is to give incentives to banks to improve the predictive accuracy of their models and to avoid overly optimistic projections of profits and losses due to model fitting.  As the penalty factor may depend on the quality of internal controls at the bank, this system is designed to reward truthful internal monitoring, as well as develop sound risk management systems.

To obtain total capital-adequacy requirements, banks will add their credit-risk charge to their market-risk charge applied to trading operations.  In exchange for having to allocate additional capital, banks will be allowed to use a new class of capital, i.e., tier 3 capital, which consists of short-term subordinated debt.  The amount of tier 3 capital is limited to 250 percent of tier 1 capital allocated to support market risks.

Various Approaches to VAR Measure

Dowd (1998) provides a formal definition of the VAR measure as the maximum expected loss over a given horizon period at a given level of confidence.  This definition involves two arbitrary parameters: the horizon period, which can be daily, weekly, monthly, quarterly, or annually; and the level of confidence, which might be 0.90, 0.95, 0.99, or any chosen probability.  The VAR concept has been extensively used in developing the internal models for measuring and controlling market-related risks of financial institutions.  The VAR measure can be derived in terms of absolute dollar loss, or in terms of loss relative to the mean revenue.  Absolute VAR is simply the maximum amount the bank is expected to lose with a given level of confidence, measured from the change in the level of wealth during the holding period, e.g., current loss.  Relative VAR is the maximum expected loss with a given confidence level but measured relative to what the bank expect its wealth to be at the end of the holding period, e.g., expected loss.

Absolute VAR     =      Current Maximum Loss        =       W₀ � W^*         =           rW₀

Relative VAR       =     Expected Maximum Loss       =     E(W) � W^*       =    -rW₀ + mW₀    =     (m - r)W₀

where:                    W₀          =              Initial wealth

                                W^*          =              Minimum level of wealth during the holding period

                                E(W)      =              Expected wealth at the end of the holding period

                                r              =              Minimum realized rate of return during the holding period

                                m              =              Mean rate of return over the holding period

Three major approaches to VAR measure are identified as appropriate for specific types of portfolios: 1) numerical and parametric, 2) delta-normal and delta-gamma, 3) historical and Monte Carlo simulation.

Numerical and Parametric VAR

Numerical VAR is derived from the histogram of return�s actual distribution of a given portfolio (P):

VAR = [E(r) - min(r)] P

If the probability distribution of the portfolio returns is known or estimable, then the parametric VAR can be derived based upon the standard normal distribution of return of stock portfolio (S):

VAR = as S

where:    a   =   Alpha parameter representing 1.96 standard deviation associated with 95% confidence level

                s   =   Sigma parameter capturing the returns volatility of stock portfolio

Delta-Normal and Delta-Gamma VAR

The Delta-Normal VAR is derived from non-linear normal distribution of bond or fixed-income portfolio (B) involving duration and convexity of returns:

VAR = asd B

where:        d   =   Delta parameter capturing the price-yield relationships of bond or fixed-income portfolio

The Delta-Gamma VAR is derived from non-normal distribution of derivative instruments portfolio (D) involving embedded options:

VAR = as(d - gda/2) D

where:        g   =   Gamma parameter capturing the changes in option price embedded in derivative portfolio

Historical and Monte Carlo Simulation VAR

The historical simulation VAR is approximated from a past series of portfolio value changes over a given period.  Monte Carlo simulation VAR is calculated from a stochastic process using geometric Brownian motion of portfolio return series, implied volatilities, and correlation matrices.

VAR = S f (%D value, a)

Capital-At-Risk (CAR) Measure

CAR is a measure of the bank's economic capital based on objective risk measures.  It is distinguished from the Basel Accord's capital adequacy ratio or from accounting measure of capital.  Regulatory capital is valued with simple standardized formulae, or forfeits, that do not measure the actual risks.  The accounting-based capital does not coincide with the economic capital as measured by CAR.  Therefore, using regulatory or accounting-based capital as a surrogate for economic capital generates distortions because of the divergence between real risks and the forfeited risks of regulatory capital.

Bessis (1998) offers some arguments against regulatory capital measures as follows.  With regulatory forfeit, credit risk is identical for short-term and long-term commitments, even though the former has a higher risk than the latter.  The credit risk measure is also the same for private counterparties that have different ratings.  This signifies that the AAA and Baa ratings are assigned the same amount of capital despite the large gap in default riskiness between these two rating categories.  Moreover, the regulation measures risk over a portfolio by a simple aggregation of individual risks for credit risk; it ignores the diversification effects and results in the same measure of risk for well-diversified portfolios and for highly concentrated portfolios.

These shortfalls of standardized capital measures have implications for the entire risk management practice in the banking industry.  The target return performance is also set according to forfeit measures of capital when in fact it should not be.  The allocation of this capital across business units does not depend upon their true risks.  Any risk-based policy for measuring risk-adjusted performance suffers from such distortions.  CAR is, therefore, intended to correct such distortions.

Yet, the derivation of CAR relies on the same principle and method that is used to derive the VAR.  When using this method to value economic capital, the following assumptions apply:

• Capital is intended to provide a protection against the deviation of losses beyond the average.  It does not provide protection to predictable average losses, which should be taken care of through reserves and provision policies.

• The deviations from the average are the unexpected losses, which are defined subject to a tolerance level. Thus, CAR is associated with the tolerance of, say, 2.5 percent based on the 95-percent confidence level.

• The losses that exceed the upper bound set by the tolerance level are exceptional or unexpected losses.  By definition, unexpected losses cannot be absorbed by CAR.  If they occur, the bank defaults.

• The tolerance level (a) represents the probability that actual losses deviate by more than the upper bound.  Therefore, the tolerance level is the default probability of the bank.

• It is not equivalent to measure unexpected loss at the business unit level in terms of VAR and those unexpected losses that put the solvency of the entire bank at stake.

Using a 2.5 percent tolerance level of market risk is acceptable for the purpose of risk monitoring and enforcing limits at the level of a business unit.  If the aggregate loss of the bank goes beyond capital, this event does trigger default.  CAR requires stricter rules to be defined than VAR used for day-to-day operations.

The CAR measure is more comprehensive than the VAR measure as it covers all bank-own risks and not just the market risk exposures. By bank-own risks, we mean the whole range of external and internal uncertainties the bank exposes itself to. It includes 1) liquidity risk, 2) credit risk, 3) market risk, 4) derivative risks, both for counterparty risk and mark-to-market position risk which are comparable to credit and market risks for loans and investments, and 5) operational risk, i.e., exposure to inadequate internal control in trading and hedging financial instruments.

The CAR measure is obtained using the following formula:

CAR = k (UL - EL) V

where:    k     =   Parameter representing 2.33 standard deviation associated with the 1% tolerance level (a)

                UL =   Upper bound of unexpected losses in bank�s total portfolios

                EL  =   Mean of expected losses in bank�s total portfolios

                V     =   Notional value of portfolios exposed to bank�s total risks

The expected losses can be observed from the bank's financial reports in terms of reserve for loan and market losses in the balance sheet. The mean of expected losses can then be calculated from each year's data over the desired period under study.  The unexpected losses are simply the standard deviation of the expected losses we have derived based on the mean value.  However, this standard deviation figure has to be adjusted for time-varying factor, i.e., the square root of number of periods, to come up with each year's loss volatility.  The notional amounts of portfolios are obtained from the bank's balance sheet as well including the amount of loan portfolios outstanding, the value of investment securities and financial instruments the bank holds, and the value of off-balance-sheet instruments reported both in terms of guarantee commitments and contingent liabilities.

By putting all parameters together, the CAR measure for each year can be calculated.  The key to evaluate CAR is on how the bank behaves in reporting its loss reserves, which is quite subjective in the eyes of outsiders. But if the bank has incentive to reveal truthful inside information in order to differentiate its quality and low riskiness from other banks, then the market can view the level of loss reserves as a signal for the bank's risk profile.  Unlike the standard Basel risk-weighted-asset capital ratio, CAR reflects what the bank believes its true risk exposures would have been, or potentially, would be. Therefore, it has more advantage over the conventional Basel capital ratio, which I use as the benchmark in measuring the bank's risk-based capital.

Risk-Adjusted Return on Capital (RAROC) Measure

Return performance is traditionally measured with accounting data, such as profit margins or return on equity.  For market transactions, the profit and loss of the bank is the change in the mark-to-market value of financial instruments during a given period.  However, there is no performance level without a trade-off or sacrifice in terms of risk.  Thus, only the risk-return combination is meaningful in setting the target performance.  If risk is omitted, some important issues emerge as to 1) how the performance of individual or business units' transactions can be compared without adjustment for risk, 2) how the risk of counterparties can be priced, and 3) how the total risk can be allocated to individual or business units' transactions.

According to Bessis (1998), RAROC measure addresses these issues with three main functions: 1) the reporting and the comparison of profitability and risk across business units, products and customers, and with the profitability targets; 2) the aid for risk pricing; and 3) the allocation of risks and capital across business units, products and customers, based on the risk-return profile.

The RAROC adjusts the bank's earnings from portfolios to their risks based on CAR.  The earnings can be defined as the net interest margin, excluding fees, plus trading and investment income.  They can be calculated both before and after operating costs.  The minimum value of the required ratio depends upon the perimeter of the earnings calculation.  For the purpose of this research, the earnings are calculated before operating costs, which truly reflect various banks' management and exposures to portfolio risks cross-sectionally without imposing any comparative advantage on operating cost management among them.

The assumptions on which the derivation of RAROC is based are:

�         The measures of risks are within a specific time horizon, i.e., fiscal year.

�         The time profile of earnings over the same horizon.

�         The expected and unexpected credit losses are derived from default rates and recoveries.

�         The expected and unexpected losses from market risk exposure are derived from VAR measure.

The formula used to derive RAROC is as follow:

RAROC = (Earnings - Expected Losses)

            CAR

The earnings can be defined at various levels.  Earnings can be limited to the net interest margin, excluding fees, or they can by defined as the net interest margin plus fees.  They can be calculated both before and after operating costs.  The minimum value of the required ratio depends on the perimeter of the earnings calculation, e.g., the hurdle rate.

The earnings figures can be observed from the income statement of the bank.  The definition of earnings can either be limited to interest margins plus fees, trading, and investment incomes or be more expansive in terms of earnings before interest and taxes (EBIT), which come after all operating costs and expenses have been deducted.  Nonetheless, since the bank�s earning power is directly derived from or influenced by its main sources of earnings, the bank's operating costs do not matter for the calculation because different banks incur different activity expenses and.  For off-balance-sheet activities, the earnings should be taken up by the fee income. If hedging activities are involved, trading and investment incomes should reflect any earnings generated.

The benchmark with which the RAROC measure is compared is the accounting-based return on capital employed (ROCE).  In this study, the ROCE definition is as follow:

ROCE = Earnings

Capital

Table 3 provides a general comparison between the conventional measures of bank capital profile and return performance and the newly proposed ones.  Table 4 distinguishes between the VAR and CAR measures in terms of their developments, parametric variables, data characteristics and requirements, and usefulness.

Table 3

Comparison between Conventional and New Measures

Table 4

Distinctions between VAR and CAR Measures

Research Hypotheses and Methodology

3.1          Research Hypotheses

As stated in the statement of purpose of the study, both CAR and RAROC are subject to test for their quantitative effectiveness and statistical robustness and reliability in measuring the bank's capital profile and return performance.  Therefore, it is necessary to establish a testable hypothesis to confirm that they are empirically viable.  The CAR measure, on the one hand, represents the capital cushion of the bank indicating the level of self-insurance against the overall bank-own risk profile.  The RAROC measure, on the other hand, represents the bank's return performance indicating how well the banks could manage both their on- and off-balance-sheet portfolios given the overall bank-own risk exposures.

The null hypothesis to be tested in this study is that there is no relationship between these two measures and the two groups of regulatory determinants, i.e., prudential and functional, which represent the independent variables.  The distinction between functional and prudential determinants is given by the objectives the bank attaches for each determinant.  For instance, the bank's asset-liability management and off-balance-sheet activities aim to achieve the optimal levels of the bank's asset utilization thereby maximizing returns, whereas the management of liquidity and solvency positions are used to serve the bank in times of temporary financial difficulty and prolonged financial distress.  However, since the operations and functioning of the bank are in fact substantially restricted and governed by regulations, it follows that the null hypothesis is unlikely to be true.  Nonetheless, the test of null hypothesis will be implicitly embedded in the test of main hypotheses that follow.

In order to establish the alternative hypotheses, both CAR and RAROC will be compared with the benchmarked capital adequacy ratio (hereinafter referred to as the Basel ratio) and the ROCE, respectively, to form the dependent variables.  If these dependent variables are sensitive to the changes in either group of regulatory determinants, then it can be concluded that they are quite effective in measuring the bank's capital cushion and return performance.  And since there are differences in regulatory enforcement among groups of countries, the cross-sectional comparisons between geographical regions will be introduced as another independent variable dimension.

With the above rationale, the hypotheses for this study can be established as follows:

H₁:  Functional determinants can explain the willingness and the ability of the international banks better than prudential determinants can.

H₂:  Functional determinants can differentiate the willingness and ability among international banks from various regions better than prudential determinants can.

3.2          Research Methodology

Research methodology is employed to empirically test the above research hypotheses using historical data from both primary sources (financial statements of the banks) and secondary sources (S&P Global Vantage database).  Predicated on theoretical rationale and past empirical studies, both dependent and independent variables are determined and derived from these two sources of data.  These variables will be operationalized in the form of proper statistical measures in which meaningful inference can be made.  The research methodology is thus divided into three sub-sections: variables determination, variables derivation, and statistical specification.

Variables Determination

After the necessary measures are duly established, i.e., CAR, Basel, RAROC, and ROCE, the required dependent variables can be derived.  For capital profile disclosure, the ratio between the Basel capital ratio and CAR as the indicator for bank's willingness to cushion for risks can be used.  If the bank wanted to signal/disclose to the markets its true riskiness, it would then use CAR instead of Basel, otherwise it just complied with the minimum capital level set by the BIS and the central bank.  For return performance disclosure, the ratio between RAROC and ROCE is used as the indicator for the bank's ability to generate risk-adjusted returns.  If the RAROC was greater than the ROCE, it means that the bank performed satisfactorily.  Otherwise, it fell under the required rate of return expected by the markets.

Two main groups of independent variables can be determined based upon the objectives of functional and prudential regulations.  For functional determinants, the ALM gaps and opacity ratios are used to represent the bank�s effectiveness in its risk-management activities and investment in opaque portfolios.  For prudential determinants, the liquidity ratios and solvency ratios are employed to signify the bank�s effectiveness in the management of its balance-sheet position, which indicate its level of safety and soundness.

An overview of method for determining all required variables is summarized in Tables 4 and 5 below.

Table 4

Dependent Variables

Table 5

Independent Variables

Based upon the above research hypotheses and the bankruptcy classification and prediction frameworks, the dependent variables of interest are capital profile disclosure (capital measure) and return performance disclosure (return measure), each of which conveys different information about banks.  As for their underlying determinants, four types of independent variables are selected based upon the objectives that functional and prudential regulations aim to achieve: ALM gaps and opacity ratio for functional regulation, and liquidity and solvency ratios for prudential regulation.   These variables and determinants can all be observed and elicited from the banks' financial statements.  The following sub-sections provide detailed discussions as to why and how these variables are determined.

Dependent Variables

Two types of response variables, namely 1) capital profile disclosure and 2) return performance disclosure, are determined as follows.

1)      Capital Profile Disclosure

Capital profile disclosure can be represented by the ratio between the Basel ratio and the CAR plus bank reserves.  If this ratio is greater than 1, then it means that the bank was willing to provide capital to cushion its overall risks more than it required.  This implies that the bank intended to disclose to the general public about its safety and soundness conditions, which connotes low bank risks.   According to the Basel Accord of 1988, the minimum capital adequacy rate is set at 8 percent of the bank's total net worth to risk-weighted asset classes.  Asset classes are rated according to their credit quality starting from 0 percent for secured assets, 20 percent for loans to other banks and municipalities within OECD countries, 50 percent for mortgage-backed loans, and 100 percent for unsecured loans and investment in securities portfolios as well as holdings in subsidiaries.  Notional balances of off-balance-sheet commitments are weighted at 2.5 percent while contingent claims at 5 percent.  The pitfall of the Basel capital ratio is its subjectivity that may not reflect the actual risk exposures of the banks.

When equity is the risk-based capital with internal models, the various business units of the bank have to meet a target return on this economic capital.  This ensures that they actually charge risks, measured as economic capital, to customers.  The required reserves diverge from those that would be based on mandate-based capital to the extent that economic capital and regulatory capital are different. The internal capital reserve rate, therefore, is the ratio between the bank's reserves for expected portfolio risks plus its CAR over its total loan and investment portfolios.  The amount of bank reserves plus CAR properly reflects the actual risk exposures and can be used to compare against the total net worth to obtain the bank�s capital signaling behavior.

2)      Return Performance Disclosure

Return performance disclosure, which is associated with the behavior of bank's management in managing its portfolios, can be represented by the ratio between the RAROC and the conventional ROCE.  If a bank's RAROC exceeds its benchmark ROCE, then it means that the bank management is able to disclose to its shareholders about its internal capability to manage and control portfolio risks while meeting the required return expectations.

Traditionally, the profitability measures include return on assets (ROA), return on equity (ROE), and ROCE.  New risk-based performance measures have been induced by regulations.  Since banks are committed to obtain mandate-based capital, they have to provide a sufficient return on this capital and to monitor the cost of mandate-based capital used up by transactions and portfolios.  However as discussed earlier, mandate-based capital such as the Basel ratio has many drawbacks since it is subjectively determined.  The use of CAR allows banks to adjust profitability with economic measure of risks.

Independent Variables

Two main groups of independent variables are also derived from the banks' financial statements. The first group is called "functional determinants," the second group "prudential determinants."  An additional group of independent variables is used to represent the cross-sectional difference in regions where different banking regulations are compared and contrasted.

1)      Functional Determinants

Functional determinants are predicated on two important banking activities: 1) asset-liability management (ALM) to tackle the risks associated with liquidity gaps, and 2) off-balance-sheet management to address the risk-shifting issues.  The measures derived from them will be used to test how well they contribute to the capital signaling and return signaling behavior of the banks.

Liquidity gaps are defined by Bessis (1998) as follows.  When assets are greater than resources from operations, a "funding gap" occurs which should be financed externally from deposits and borrowings.  On the other hand, when liabilities are greater than assets, an "investment gap" appears, and the excess resources have to be deployed through loans, securities, and derivative investment vehicles.  In terms of ALM, banks want to ensure that their funding gaps can be financed under normal conditions without being subject to higher interest rates associated with the emergency funding of unexpected large deficits, while their investment gaps can be narrowed down without imposing higher exposures to credit and market risks on the asset side and interest rate risk on the liability side.

A measure used to represent the banks' ALM function and a proxy for the first explanatory variable is called an "ALM gap."  It is calculated as the difference between the annual percentage change of assets and of liabilities:

ALM Gap = %D Assets - %D Liabilities

A positive ALM gap means that the variation in assets exceeds that of liabilities, and vice versa for a negative ALM gap.  The banks that consistently maintain the positive ALM gaps have experienced more uncertainties in their uses of excess funds than in their sources of deficit funds.  Such an instance indicates their vulnerability to asset management and all associated risks.  On the contrary, those that have maintained the negative ALM gaps are more susceptible to liability-related risks than to asset-related risks, e.g., early withdrawals or competitive pressure to increase deposit interest rates.

Based on the four non-traditional banking functions in terms of credit enhancement, liquidity securitization, risk unbundling, and financial engineering, the banks are intensively engaged in off-balance-sheet activities with the primary purposes to hedge the down-side risks of unfavorable credit and market exposures while gaining on the up-side volatility through dynamic trading of underlying financial assets and derivatives instruments.  They also extensively utilize the off-balance-sheet products to manage risk for their clients through risk sharing and risk shifting.  Commitment loans and their derivatives are those that have the futures-contract features such as letters of credit, project-loan drawdowns, and securitized instruments.  Contingent claims are options-like contracts that banks originate, e.g., stand-by lines of credit, swap contracts, floating-rate notes, equity derivatives, and structured products.

To measure the degree of the banks' opaque activities, the so-called "opacity ratios" � the ratios between the off-balance-sheet portfolios and the on-balance-sheet portfolios � are used:

Opacity   =   Off-balance-sheet Portfolios

                     On-balance-sheet Portfolios

The higher the opacity ratio, the more intensive the risk management activities in which the banks are engaged.  Unless they have the appropriate internal risk control systems in place, high opacity ratios indicate that the banks are likely to be more vulnerable to operational risk than to the external credit and market risks they intend to reduce for themselves and their clients.

2)      Prudential Determinants

Prudential determinants are directly derived from the banks' liquidity and solvency policy.  The rationale for using these two measures is that in order to achieve the optimal capital structures and the desirable levels of internal financial cushion, the banks would maintain their liquidity and solvency positions that are commensurate with their expected rates of return.

The measure of the banks' liquidity position is called a "liquidity ratio."  The liquidity ratio is simply the ratio between the banks' secured or collateralized asset portfolios and their total demand and time deposits:

Liquidity = Collateralized Asset Portfolios

    Demand and Time Deposits

As their liquidity ratios increase, the banks are said to have strong and sound collateralized asset classes to cushion or self-insure for unexpected credit and investment losses without jeopardizing their creditworthiness.  Liquidity ratio also indicates the degree of a bank�s short-term vulnerability in the event of massive withdrawals or depositor runs.

The solvency position of the banks is given by the ratio between their total on-balance-sheet portfolios and total obligations.  The resultant "solvency ratio" measures both safety of the banks' total indebtedness as being protected by their total asset base and long-term viability to forced liquidation in the event of bank runs.

Solvency = Total Asset Portfolios

                     Total Debt Obligations

The higher the solvency ratios, the more safety cushions the banks� external depositors and creditors have.  However, no quality distinction is provided for banks� total asset portfolios in a solvency ratio.  Banks that have high solvency ratios may be considered operationally risky if majority of their portfolios consists of unsecured asset classes.

3)      Regional Determinants

Five groups of regions are distinguished from each other according to their trade ties, economic cooperation, geographical proximity, and cultural similarity.  These groups are arranged as follows:

• North America region includes the United States and Canada in light of their geographical proximity and the North America Free Trade Area (NAFTA) regional arrangement.

• British Commonwealth region includes the United Kingdom, Australia, and New Zealand based upon their allegiance to the British Crown and intra-group trade ties.

• Continental Europe region includes the Benelux nations, France, Germany, Switzerland, Italy, and Spain based upon the European Economic Cooperation (EEC), the precursor of the European Union.

• East Asia region includes Japan, South Korea, and Taiwan Republic of China in terms of their comparable rates of industrialization and trade ties.

• Southeast Asia region includes Indonesia, Malaysia, the Philippines, Singapore, and Thailand based on the regional arrangements for the Association of Southeast Asian Nations (ASEAN).

Variables Derivation

The analysis phase of the research methodology involves data manipulation and transformation after all variable proxies have been identified in the theoretical phase.  In this phase, accounting and financial information obtained from the banks' annual reports is analyzed using the spreadsheet program.  There are two steps involved: 1) data manipulation and 2) data transformation.

Data Manipulation

Data manipulation is a step involving classification of data inputs into three tables (see Credit Suisse Group�s data as illustration below or Appendix III for entirety).  Table 7, �Bank Portfolios,� is used to derive the bank's total assets, including secured assets, semi-secured (mortgaged) assets, loan and investment portfolios which are unsecured, off-balance-sheet instruments (commitments and contingents), and the Basel Accord's risk-weighted assets (RWA).  Table 8, "Bank Obligations," consisting of data inputs from the banks' current and long-term obligations and, in connection with Bank Portfolios, serves as a basis for establishing the bank's "Asset-Liability Management Profiles" given in Table 9 to be used as the proxies for explanatory variables.

Table 7

Bank Portfolios

(Figures in million U.S. dollars unless otherwise specified)

Table 8

Bank Obligations

(Figures in million U.S. dollars unless otherwise specified)

Table 9

Asset-Liability Management Profiles

Data Transformation

Based on the works of Philippe (1997) and Jo�l (1998), data transformation is an important and necessary step under Analytical Method in order to derive comparable measures for the required variables.  For dependent variables, capital and return disclosures are transformed in Table 10 �Bank Capital Profiles� and Table 11 �Bank Return Profiles,� respectively.  For independent variables, the banks' ALM gaps, opacity ratios, liquidity ratios, and solvency ratios are transformed in Table 9 above from Bank Portfolios in connection with Bank Obligations.  In Table 10, the banks' realized credit and market losses are manipulated as well as their reserves in order to arrive at the Reserve & CAR measure (internal capital).  The end result of this Capital Profile, on the one hand, is the dependent variable for the bank capital profile.  Table 11, on the other hand, represents the derivation of the dependent variable for bank return performance.

Table 10

Bank Capital Profiles

(Figures in million U.S. dollars unless otherwise specified)

Table 11

Bank Return Profiles

(Figures in million U.S. dollars unless otherwise specified)

Statistical Method

Within the statistical phase of research, four important steps are required: 1) method selection, 2) sampling and data collection, 3) model specification, and 4) statistical estimation and inference.

Method Selection

Since the characteristic of data involves both cross-sections and time-series, i.e. panel data, the most appropriate statistical method should be the pooled regression.  According to Hsiao (1986), the panel data sets have several major advantages over conventional cross-sectional or time-series data sets.  First, they provide a larger number of data points, increasing the degrees of freedom, and reducing the collinearity among explanatory variables, hence improving the efficiency of econometric estimates. 

Second, they allow a researcher to analyze a number of important economic questions that cannot be addressed using cross-sectional or time-series data sets alone.  A single cross-sectional data set usually cannot provide a direct choice between hypotheses, because the estimates are likely to reflect inter-individual differences inherent in comparisons of different firms.  However, if panel data set is used, one can construct a proper recursive structure to study the before/after effect.  Similarly, a single time-series data set usually cannot provide precise estimates of dynamic coefficients without specifying a priori that each of them is a function of only a very small number of parameters.  If panel data are available, one can utilize the individual differences in explanatory variables to reduce the problem of multicollinearity.

Third, the use of panel data set also provides a means of resolving or reducing the magnitude of a key econometric problem that often arises in empirical studies, e.g., omitted or unobserved variables that are correlated with explanatory variables.  By utilizing information on both the time-series and the individuality of the units of analysis, one is better able to control in a more natural way for the effects of missing variables.

Sampling and Data Collection

The data collection adopted in this research is non-random but predetermined from the available name list of 100 international banks from 19 countries based upon Fortune Global 500 ranking published for 1998, of which 12 belong to the Organization of Economic Cooperation and Development (OECD) and 7 to the less-developed countries (LDC), which were subject to the recent global financial crisis.  The main reason for using large international banks as sample is that they should be least prone to financial fragility, which lends the great advantage in terms of research�s robustness.  Smaller regional banks are excluded from the research because they are more susceptible to systemic risk than large banks, which might contaminate the robustness of statistical inference.  Further, many small banks do not provide complete financial data to cover the period of study.

The time span of this research is a 10-year period coverage from 1988 to 1997.  This period is most appropriate because it covers the banks' financial data from 1988 when the Basel capital ratio was first implemented until 1997 when all annual reports were completed and issued.  In terms of data collection, the primary data came from the bank's annual reports while the secondary data were retrieved from the Standard & Poors (S&P) Global Vantage database.

Model Specification

Four statistical models are specified according to the pooled regression using region variable j to represent the cross-section and t to represent the time-series.  The individual bank's observed data points being pooled for panel data are represented by i.

1)                   Capital measure_ijt    =       Sb_j(Region_j) + b₆(ALM Gap)_it + b₇(Opacity)_it

2)                   Capital measure_ijt    =       Sb_j(Region_j) + b₆(Liquidity)_it + b₇(Solvency)_it

3)                   Return measure_ijt    =       Sb_j(Region_j) + b₆(ALM Gap)_it + b₇(Opacity)_it

4)                   Return measure_ijt    =       Sb_j(Region_j) + b₆(Liquidity)_it + b₇(Solvency)_it

where:    t = 1988,�,1997 time series

i = 1,�,100 international banks

j = 1,�,5 regional blocs consisting of:

1 = North America (Canada and United States)

2 = British Commonwealth (Australia, New Zealand, and United Kingdom)

3 = Continental Europe (Benelux, France, Germany, Italy, Spain, and Switzerland)

4 = East Asia (Japan, South Korea, and Taiwan)

5 = Southeast Asia (Indonesia, Malaysia, the Philippines, Singapore, and Thailand)

The number of banks represented in each country for each region totaling 100 banks is as follows:

 The name list of these international banks and their 1998 Fortune Global 500 ranking is provided in Appendix II.

Statistical Estimation and Inference

Utilizing the SAS statistical program, parametric estimation can be conducted for 1) the estimation of country cross-sectional differences based on dummy parameters and 2) the estimation of time-series differences based on explanatory parameters.  The resultant statistical inference will include:

1)      Test of significant relationships (F)

Based on the analysis of variance (ANOVA), the F-values of each regression indicate how well the set of explanatory variables in the model is related to the response variable at a pre-specified level of significance (a).  If the regression's F-value is greater than the critical F-value with a = 0.05, then there is a significant relationship between the explanatory variables and the response variable.  The probability of the F-value for the regression is a guide to how important the independent variables are in explaining the behavior of the dependent variable.

2)      Test of significant parameters (t)

This test involves the estimation of each explanatory variable's parameter (b-coefficient) and how statistically significant it is in explaining the response variable.  The adequacy of each independent variable in the model can therefore be assessed.  The t-value is used to test the hypothesis that there is no linear relationship between the dependent variable and the independent variable.  The t-value takes into account the other variables in the regression model.  Similar to the F-value, the test of significant parameter is based on the t-value measured against the critical t-value with a = 0.05.

Empirical Results

Below are test statistics for hypothesis testing of the estimated parameters from the models, which infer that there are certain significant relationships between dependent variables and independent determinants.

4.1          Models Estimation and Inference

Using the significance level (a) of 0.05 as cut-off point, the F-values of all four pooled regressions are statistically significant at each corresponding p-value, i.e., less than 0.0001.  It can be inferred from the table that there are strong relationships between both capital and return measures and both regulatory determinants, with functional determinants being more able to explain the dependent variables than prudential determinants. 

Table 12

Test of Significant Relationships

4.2          Parameters Estimation and Inference

The test of significant parameters will reveal the most information from the pooled regression functions whereby each independent determinant is evaluated separately.  As stated earlier in the model specification, these pooled regressions consist of both cross-sectional and intertemporal dimensions of the panel data.  The individual bank's observation (i) and its corresponding time-series (t) are regressed in terms of main effects of functional determinants (ALM gap and opacity ratio) and prudential determinants (liquidity ratio and solvency ratio).  The parameters for five different regions are estimated via dummy variables represented by j to compartmentalize for each specific region.  At the cut-off level of significance of 0.05, the tables below provide test statistics, including the b coefficients, standard errors of the b coefficients, t-values, and their corresponding p-values, for making inferences.

Table 13

Tests of Significant Parameters

Capital Measure vs. Functional Determinants

 Capital Measure vs. Prudential Determinants

 Return Measure vs. Functional Determinants

Return Measure vs. Prudential Determinants

Regarding the main effect of each individual functional determinant, the p-value of opacity ratio (0.0726) signifies a tendency toward a high degree of association with the capital measure whereas the p-value of ALM gap (0.3660) is not statistically significant.  However, when considering the cross-section effect of regional variables, functional determinants are able to differentiate the capital measures among international banks from the five regions while prudential determinants are not. 

In terms of the main and cross-sectional effects of prudential determinants, capital measure is relatively effective in measuring the bank's solvency of asset portfolios (p-value of 0.0743) with no clear distinction among regions except for Southeast Asian countries whose p-value is as low as 0.0004.  The rationale behind this exception is that high level of government discretion in policy making plays an important role in differentiating among prudential regulation among those Southeast Asian nations.  This indicates how less stable the regulatory frameworks and the banking systems within those countries have experienced compared with the other four regions, which resulted in the recent financial crisis during 1997.

With respect to the return measure, the main parametric effect of both functional determinants does not yield any statistical significance even though their cross-sectional effects do exhibit differences among the five regions.  For prudential regulation, solvency ratio with the p-value of 0.0083 is the only determinant that shows strong association with the return measure.  This could be explained in terms of capital structure and the leverage effect whereby returns on equity are improved as banks gain higher debt capacity.  Cross-sectionally, only Southeast Asian region stands out in terms of regional difference with the corresponding p-value of 0.0225.  Again, the uneven paces in regulatory development as well as the inevitable governments� regulatory discretion within the region during the last decade could be attributed as the underlying causes of such a difference and the global financial crisis that had ensued from this region.

Research Conclusions

5.1          Major findings

Based on the above empirical results, it can be concluded for the first study that both functional and prudential determinants can equally explain the behavior of international banks in terms of their willingness to provide capital cushion and ability to generate returns.  This conclusion has refuted the first hypothesis that only functional determinants could explain such behavior better than prudential determinants could.  It follows that the implementation of functional regulation should not replace, but rather should complement, the conventional framework of prudential regulation, especially at the international level.

For the second study, it can be concluded that the cross-sectional parameters of functional regulation could distinguish among international banks from the five regions under study.  This conclusion thus supports the second hypothesis that functional determinants are more able than prudential determinants to point out the differences among banks that operate under the similar regulatory frameworks.  It also indicates that despite the high hope for regulatory convergence induced at the international level by such supra-national organization like the BIS, different patterns of bank capital profile and return performance still persisted.  Thus, the area in which more effort can be put to promote faster regulatory convergence might be at the regional or national level.

5.2          Policy Implications

In view of its research purposes and empirical findings, this research has made a major contribution towards public policy in at least two fronts.  The first deals with the role of these newly proposed capital and return measures in enhancing banking regulation.  The second is concerned with the potential reactions of bank regulators to the banks using these measures.

More specifically, the focus of the first area is on the impacts of CAR and RAROC measures on current regulatory frameworks by helping them to alleviate the problems of asymmetric information, which is the main goal of bank supervision for private interest (i.e., idiosyncratic transparency and contractual efficiency).  With respect to the hidden-knowledge problem, which engenders the non-contractual adverse selection and the conflict of interest between the bank and its depositors, the CAR measure enhances functional regulation by capturing the true profiles of illiquid and risky portfolios from the loss reserves the bank puts aside for each operating cycle.  The mapping of loss reserves onto CAR represents the signal that the bank intends to disseminate to the less-informed market participants.  Through its simplicity, the CAR formula enables the markets to estimate the riskiness of the bank's portfolios without having to resort to the standardized and less flexible risk-weighted asset.  The efficiency of the CAR measure can be confirmed from its ability to predict the bank's solvency ratio, which represents how well its overall assets are utilized to meet and exceed its fixed liability obligations.  With low level of CAR, the bank is confident that its asset base is strong and liquid enough to withstand massive withdrawals before maturities.

In terms of the hidden-action problem of asset opaqueness, which gives rise to the contractual moral hazard and the conflict of interest between the bank's shareholders and the management, the CAR and RAROC help elicit more information from the bank's off-balance-sheet portfolios and other financial innovation and engineering activities.  Again, they do so by means of mapping the perceived loss reserves onto the bank's economic capital (measured by CAR) and expected risk-adjusted rates of return (measured by RAROC).  Based on the empirical results, the association between capital measure and opacity ratio signifies that the bank's opaque operations can inherently be observed or estimated through CAR and RAROC.

On the second area, which focuses on the responses of bank regulators to the new measures, the functional regulation is proven to provide better incentive systems and monitoring and enforcement mechanisms to the banking industry than the institutionally oriented prudential regulation.  For the purpose of incentive systems, functional regulation based on CAR and RAROC allows banks to have more flexibility to set their own capital levels and return targets in the manner that best suits their own operations and reduces costs and expenses relating to regulatory compliance.  With regard to effective monitoring and enforcement, the use of CAR and RAROC measures will allow bank regulatory agencies to selectively and objectively focus on the most relevant and crucial bank's activities without imposing redundant restrictions or inducing potential abuses.  As a result of increased flexibility, reduced operating costs, and decreased restrictions, the overall benefits of functional regulation can be realized in terms of improved competitive efficiency among banks and more stable systems of financial intermediation, which are the ultimate goals of banking regulation for public interest.

5.3          Research Limitations

Four areas of limitation are identified in this research. The first is a critique on the nature of the CAR and RAROC measures.  As the CAR formula indicates, the constant parameter k is standardized with respect to the 2.5 percent level of loss tolerance based on the VAR method.   In practice, this constant parameter can differ across various banks when determining the appropriate level of CAR and RAROC.  The k value, in theory, should be set commensurate with the expected losses each bank believes it would incur.  Therefore, a correct identification and use of the k parameter to derive CAR is as important as an identification and classification of the bank's loss reserves.

Following the first limitation above, the second caveat is the subjectivity in the bank's loss reserves.  Although the risks of bank's portfolios can be objectively measured (through various complex mathematical and econometric models), the level at which the bank is willing to provide as loss reserves to reflect those risks is discretionary in nature based on the view and the interest of the bank management.  An important assumption must be made here under the incentive-compatible framework that if the bank management desires to signal the quality of its prudence, then the discretionary loss reserves can be deemed as a reliable indicator of the bank's overall risk profiles.  This assumption is not valid in the environment where banks are not competing based on reputation and credibility.  It is therefore the duty of regulatory agencies to promote competitive efficiency within the banking industry to ensure that the bank management has adequate incentive to set its loss reserves level that is truly objective.  In effect, these two research limitations open up an avenue for future research in how to improve or extend the CAR formula.

The third research limitation is concerned with the availability of primary data from some banks. Originally, it was intended that the periods of study cover from 1968 to 1997, since they would allow for the before-after study of the banks� signaling/disclosure behavior prior to and after the establishment of the BIS in 1987 and related implementation of international guidelines on bank capital standards thereafter.  However, only a few of them were able to provide a complete set of financial data as requested.   It was decided that the appropriate time span is ten years starting from 1988 to 1997, which covers the periods after the BIS�s inception.  Therefore, the use of secondary data from S&P Global Vantage database to supplement the primary source becomes more important in light of the lack of data and for numerical consistency purposes.

And finally, despite the use of S&P database, the classification between the types of off-balance-sheet items remains a clear restriction. The database does not differentiate between the commitment type and the contingent type of off-balance-sheet activities, as they possess varying degrees of risk exposure. In essence, the commitment transactions involve guarantees and stand-by credits and loans that have the futures contract characteristics whereas the contingent transactions are much like the options contracts in which off-balance-sheet instruments are transacted conditionally.

5.4          Future Research Directions

Based on several research limitations and impending issues outlined above, improvement or extension of the current CAR and RAROC measures seems to be the most promising future research direction.  The process of introducing the better and improved measures and testing them will be an everlasting task so long as banks and regulatory agencies continue to change.  Needless to say, changes in either player have profound impacts on one another.  Hence, the improvement in regulatory frameworks (regulatory economics) should go hand in hand with the improvement in financial measures (financial economics).

Mainly, there are two ways to approach the interdisciplinary research in regulatory and financial economics. From a qualitative angle, the impacts of alternative model-based regulations on systemic stability could be studied.  From a quantitative angle, the modeling of prudential regulation that is tied to competitive efficiency and performance of the banking industry vis-�-vis other non-bank financial services industries, e.g., securities and insurance, should be explored. These entail more elaborately designed empirical research and mathematical modeling both on the intra-industry and inter-industry cross-sectional scales. In addition, more robust non-linear econometric models will be employed to these future studies in terms of time-series analysis.

From the empirical-research standpoint, the next area of future study calls for an improvement in sampling and data collection by obtaining the non-public information from the banks through questionnaires and interviews.  Based on the efficient markets hypothesis, market participants take into account not only the publicly available information (semi-strong-form efficiency) but also privately available information (strong-form efficiency), including inside intelligence about how the bank management would report its past performance and future plans.  It is, therefore, expected that with a more enriched database, the empirical studies of this nature will reveal more rationales and conclusions in support of this current study. 

At present, there has been an increasing trend of advanced research in the areas of financial intermediation and regulation at both domestic and international levels. What has been theoretically conceptualized and empirically proven in the economics of monetary and capital markets as well as the economics of corporate finance, can be extended to the banking and financial services industries. It seems optimistic that the momentum of this rudimentary research will continue to become more important and gain more participation from diverse groups of practitioners in the twenty-first century.

References

Anderson, Ronald W., and Suresh Sundaresan. "Design and Valuation of Debt Contracts." The Review of Financial Studies 9 (Spring 1996): 37-68.

Berkovitch, Elazar, and Ronen Israel. "The Design of Internal Control and Capital Structure." The Review of Financial Studies 9 (Spring 1996): 209-240.

Bessis, Jo�l. Risk Management in Banking. New York: John Wiley & Sons, 1998.

Crockett, Andrew. "The Changing Role of Central Banks." The Money Macro and Finance Annual Conference. Durham, 11 Sep. 1997.

Culp, Christopher L. "Functional and Institutional Interaction, Regulatory Uncertainty, and the Economics of Derivatives Regulation." The Financier 2 (1995): 458-487.

Diamond, D. "Financial Intermediation and Delegated Monitoring." Review of Economic Studies 51 (1984): 393-414.

Diamond, D., and Phillip Dybvig. "Bank Runs, Deposit Insurance, and Liquidity." Journal of Political Economy 91 (1983): 401-419.

Dybvig, Phillip H., and Jaime F. Zender. "Capital Structure and Dividend Irrelevance with Asymmetric Information." Review of Financial Studies 4 (1991): 201-219.

Edwards, Franklin R. The New Finance: Regulation and Financial Stability. Washington, D.C.: American Enterprise Institute for Public Policy Research Press, 1996.

Giammarino, R.M., and E.H. Neave. "The Failure of Financial Contracts and the Relavance of Financial Policy." Working paper No. 82-3. Queen's University, Kingston, Ontario, 1982.

Greenspan, Alan. "The Role of Capital in Optimal Banking Supervision and Regulation." The Conference on Capital Regulation in the 21st Century. New York, 26 Feb. 1998.

Jensen, M.C., and W. Meckling. "Theory of the Firm: Managerial Behavior, Agency Costs and Capital Structure." Journal of Financial Economics 3 (1976): 305-360.

Jorion, Philippe. Value at Risk: The New Benchmark for Controlling Derivative Risk. New York: McGraw-Hill, 1997.

Kane, Edward J. "Competitive Financial Reregulation: An International Perspective." Threats to International Financial Stability. Ed. Richard Portes and Alexander K. Swoboda. Cambridge, UK: Cambridge University Press, 1987. 111-149.

__________. "How Market Forces Influence the Structure of Financial Regulation." Restructuring Banking and Financial Services in America. Ed. William S. Haraf and Rose Marie Kushmeider. Washington, D.C.: American Enterprise Institute for Public Policy Research Press, 1988. 343-382.

__________. "Ethical Foundations of Financial Regulation." Journal of Financial Services Research 12 (1997): 51-74.

Merton, Robert C. "On the Application of the Continuous-Time Theory of Finance to Financial Intermediation and Insurance." The Geneva Papers on Risk and Insurance 14, Jul. 1989: 225-262.

__________. "Operation and Regulation in Financial Intermediation: A Functional Perspective." Working paper No. 93-020. Harvard University, Graduate School of Business Administration, Boston, MA, 1992.

Merton, Robert C., and Zvi Bodie. "A Conceptual Framework for Analyzing the Financial Environment." The Global Financial System: A Functional Perspective. By Dwight B. Crane, et al. Boston, MA: Harvard University Press, 1995. 3-31.

__________. "Financial Infrastructure and Public Policy: A Functional Perspective." The Global Financial System: A Functional Perspective. By Dwight B. Crane, et al. Boston, MA: Harvard University Press, 1995. 263-282.

Mishkin, Frederic S. The Economics of Money, Banking, and Financial Markets. Boston, MA: Little, Brown and Co., 1986.

Modigliani, Franco, and Merton H. Miller. "The Cost of Capital, Corporation Finance, and the Theory of Investment." American Economic Review 48 (1958): 261-297.

__________. "Corporate Income Taxes and the Cost of Capital: A Correction." American Economic Review 53 (1963): 433-443.

Muller, Huib. "Supervisory Cooperation: A Condition for Fair Competition in International Banking." Competition and Cooperation in World Banking. London: The Institute of Bankers, 1985. 41-60.

Myers, Stewart C., and Nicholas S. Majluf. "Corporate Financing and Investment Decision When Firms Have Information that Investors Do Not Have." Journal of Financial Economics 13, (1984): 187-221.

Myers, Stewart C., and Raghuram G. Rajan. "The Paradox of Liquidity." Working Paper No.5143. National Bureau of Economic Research, 1995.

Perold, Andre F. "The Payment System and Derivative Instruments." The Global Financial System: A Functional Perspective. By Dwight B. Crane et al. Boston, MA: Harvard University Press, 1995. 33-79.

Ross, Stephen A. "The Determination of Financial Structure: The Incentive-Signalling Approach." Bell Journal of Economics 8 (1977): 23-40.

__________. "Institutional Markets, Financial Marketing, and Financial Innovation." Journal of Finance 44, Jul. 1989: 541-556.

Smith, Roy C., and Ingo Walter. Global Banking. Oxford: Oxford University Press, 1997.

Stigler, George J. "The Theory of Economic Regulation." Bell Journal of Economics and Management Science 2 (1971): 2-19.

Walmsley, Julian. New Financial Instruments. 2^nd Ed. New York: John Wiley & Sons, Inc., 1998.

Back to Top

About the Author

* Worapot Ongkrutaraksa is currently a lecturer of finance at Curtin University's School of Economics & Finance, Australia. He used to conduct his post-graduate research in financial economics at Kent State University and international political economy at Harvard University through the Fulbright sponsorship between 1995 and 1998.

E-mail: [email protected]

Home Page