An application of unit root tests with a structural break to risk-based capital and bank portfolio composition.

• Author: Jacques, Kevin T.

1. Introduction

   During the 1990s, considerable attention focused on the composition of commercial bank portfolios in general and the relationship between bank holdings of business loans and government securities in particular. Prior to 1992, U.S. commercial banks routinely held commercial and industrial loans in excess of U.S. government securities. But in the three-year period ending December 1993, commercial bank holdings of U.S. government securities increased by 60.1%, from $456.0 billion to $730.1 billion, while holdings of business loans fell by 8.6%, from $641.2 billion to $586.4 billion. In fact, from May 1992 through June 1995, the dollar volume of U.S. government securities held by commercial banks actually exceeded the dollar volume of business loans, a virtually unprecedented event.

   The unusual behavior of the relationship between bank holdings of business loans and government securities during the 1990s generated considerable interest among economists, policymakers, and bank regulators. Early attention concerning this issue was devoted to assessing what factors were responsible for the shift, including demand factors (Bemanke and Lown 1991) and more stringent bank examination standards (Peek and Rosengren 1995). Alternatively, according to some observers, another factor influencing the changes in bank portfolio composition was the implementation by U.S. bank regulators of the risk-based capital standards. Because the risk-based capital standards account primarily for credit risk, they require banks to hold greater capital, at the margin, for assets with potentially high levels of credit risk, such as business loans, than for assets deemed to have no credit risk, such as U.S. government securities. Thus, as Berger and Udell (1994) note, the risk-based standards may function as a regulato ry tax, one that reduces the profitability of business loans relative to government securities, thereby creating an incentive for banks to alter the composition of their portfolios.

   This study examines the time-series behavior of aggregate business lending and government security holdings in the U.S. commercial banking system by considering whether a trend specification with structural break model where the structural break coincides with the implementation of the risk-based capital standards is consistent with the recent history of bank portfolio composition. While almost all existing empirical studies of the impact of the risk-based capital standards on portfolio composition are performed on individual bank data, the relationship over time between business loans and government security holdings for the banking system in the aggregate is important for a number of reasons. First, if the credit view of monetary policy is correct, then the composition of bank portfolios in the aggregate has important implications for the future level of economic activity. (1) This is because many businesses, particularly small ones, rely solely on banks for credit, and if aggregate bank lending is reduced, then the disruption in financial intermediation may impair economic activity and growth. Second, Silber (1969) argues that a change in monetary policy may be more quickly transmitted to the economy by a change in bank loans than by a change in bank holdings of securities. This occurs because inventory investment is very responsive to changes in loan rates, while investment spending is less responsive to changes in interest rates on securities. Under these conditions, an exogenous shock to bank portfolio composition may have an impact on the speed and effectiveness of monetary policy. Third, from a regulatory perspective, the shift in portfolio composition may have important implications for the safety and soundness of the banking system because a relative increase in security holdings, if not properly immunized, may lead to an increase in interest rate risk for the banking system. And research by Allen, Jagtiani, and Landskroner (1996) finds that after implementation of the risk-based capital standards, bank s substituted interest rate risk for credit risk.

   Finally, Greenspan (1998), McDonough (1998), and Hawke (1999) note that efforts are currently under way by both U.S. and foreign bank regulators to revise the risk-based capital standards. Given the limited understanding of the impact of the risk-based capital standards, as noted by Dowd (1998), a time-series examination of the impact of risk-based capital on the aggregate composition of bank portfolios may provide useful insights for regulators as they revise the risk-based capital standards.

2. Risk-Based Capital Standards

   In July 1988, the Basle Committee on Banking Regulation and Supervisory Practices, comprised of representatives from 12 major industrialized countries, approved adoption of the risk-based capital standard for banks in their respective countries. (2) The primary purpose of the risk-based standards was to require banks to hold capital in accordance with the perceived credit risk in their portfolio of assets as well as the risk arising from their off-balance sheet activities. To achieve this objective, the risk-based standards classify bank assets into one of four broad risk categories: 0%, 20%, 50%, and 100%. Certain assets, such as U.S. government securities, are considered to have no default risk and are assigned a risk weight of 0%, while commercial and industrial loans are assumed to have considerable credit risk and are assigned to the 100% risk-weight category. Having assigned assets to the appropriate risk-weight category, a bank computes its total risk-weighted assets by summing the dollar value of each asset times its corresponding risk weight. As a final step, banks are required to keep a certain minimum percentage of their total risk-weighted assets in the form of capital. (3) Effective December 31, 1990, the risk-based capital standards required banks to hold a minimum of 7.25% of their total risk-weighted assets in the form of capital. (4)

   While the idea behind the risk-based capital standards was to get banks to hold capital commensurate with the level of primarily credit risk in their portfolio of assets, previous studies by Avery and Berger (1991) and Baer and McElravey (1993) recognize the many limitations of the risk-based capital standards. One problem is that by substituting assets with low risk weights, such as government securities, for assets with high risk weights, such as business loans, a bank could lower its minimum regulatory capital requirement yet not necessarily reduce the overall level of risk in their portfolio. Thus, Berger and Udell (1994) note, the risk-based capital standards may function as a regulatory tax, one that beginning December 31, 1990, places a higher marginal tax rate on commercial and industrial loans (7.25%) than on U.S. government securities (0%). This situation is further compounded because, as Avery and Berger (1991) and Keeton (1994) observe, if the risk weights used in the risk-based capital standards do not accurately reflect the true risk of an asset, then banks have an incentive to arbitrage assets both within and across risk-weight categories. Thus, for capital-constrained banks, the risk-based capital standards create an incentive to reallocate the assets in their portfolios since compliance can be achieved by shifting a bank's portfolio toward lower risk-weighted assets, such as government securities. Former Securities and Exchange Commission Chairman Richard Breeden and former Federal Deposit Insurance Corporation Chairman William Isaac (1992, p. A2) note the incentive structure created by risk-based capital when they state,

   Say what you may about bankers, they tend to be rational economic beings. Tell them they have to maintain 8% capital against business and consumer loans--and no capital or materially less capital against government bonds or single-family mortgage loans--and most bankers will put much of their money in the assets that require little or no capital.

   Recent work by Haubrich and Wachtel (1993) confirms this point finding that banks shifted their existing portfolios away from high risk-weighted assets, such as business loans, to low risk-weighted assets, such as government securities, thereby reducing their risk-based capital requirements. They found that these changes occurred after implementation of the standards because the composition of bank portfolios can be quickly changed, thereby making portfolio changes before implementation of risk-based capital unnecessary.

   In addition, banks that are not explicitly capital constrained also have an incentive to reallocate their portfolios toward low credit risk assets. As Hancock and Wilcox (1994), Jacques and Nigro (1997), and...