An analysis of the impact of international activity on the domestic balance sheet of U.S. banks.
| Verfasser | Haslem, John A. |
| Posten | Includes appendix |
Introduction
The primary purpose of this study is to compare the domestic balance sheet strategy ("profiles") of international U.S. banks with those of domestic U.S. banks. The analysis was performed in the context of the banking events of the 1984 to 1989 period that included the "crisis in lending" to less-developed countries ("LDCs") and was motivated primarily by two related questions:
(1) Did the domestic office balance sheet profiles and performance of international banks in the period of the "LDC loan crisis" differ from those of domestic banks?
(2) Did these international banks generally alter their domestic office balance sheet profiles over this same crisis period?
These two questions arose from the need to understand better the balance sheet profiles and performance of international U.S. banks in light of the events of the difficult, dynamic, and more global banking environment that included the "LDC loan crisis." This crisis had its roots in the early 1970s when the increase in LDC loans began and culminated at its peak in 1983. As reported in Bennett and Zimmerman (1988), Todd (1988), and elsewhere, this crisis has been characterized by loan defaults, rollovers and reschedulings of maturing loans, nonpayments of loan interest, and new loans to cover contractual interest pay-merits. This ongoing crisis led banks to reduce their international banking activities, especially their loans to LDCs. Faced with the resulting decline in market value of these loans, banks raised additional capital through increased profit retention, sale of assets, and new issues of equity capital and subordinated debt. The banks also reduced pressure on their declining capital ratios by slowing the growth rates of their assets. Nonetheless, as reported in The Economist (1990), bank net loan chargeoffs relative to their declining net interest margins tripled during the 1980s, with total and LDC loan-loss reserves reaching alltime peaks in 1987.
It is important from the standpoints of both bank regulators and managers to understand specifically the impact of the period that included the "LDC loan crisis" on the domestic (but also foreign) balance sheet profiles and performance of international banks, given their degree of foreign office activity. By understanding the implications of this experience, they should be better prepared to avoid or at least to minimize any adverse impacts of future crises of an international nature.
Expectations in this regard might well be influenced by the extent to which international banks explicitly or implicitly practiced normative (systems oriented) asset/liability management (ALM). If they did it would be expected that their domestic office balance sheet profiles and performance would differ from those of domestic banks. This is because ALM would be expected to integrate domestic and foreign office balance sheet profiles.(1)
Related Literature
Empirically, the presumed difference in domestic office balance sheet profiles and performance due to foreign office activities is less certain. Haslem, Scheraga and Bedingfield (1992) studied the nature and implications of foreign and domestic strategy relationships reflected on both sides of U.S. bank balance sheets. They analyzed the 1987 data of 132 "large" banks (total assets less than $8 billion) and 44 "very large" banks (total assets $8 billion or more) with both foreign and domestic offices. First, a consistent dichotomy in balance sheet strategies was found. This dichotomy reflected either an intense matching of domestic assets/liabilities or a similar pattern with respect to foreign assets/liabilities. This apparent lack of integration of foreign and domestic balance sheet strategies would appear inconsistent with ALM. Second, foreign asset/liability relationships were generally found to have relatively less interest-rate and liquidity risks than domestic asset/liability relationships. Third, "very large" banks that focused on strictly matching foreign assets/liabilities were more profitable than those that focused on mixed-matching foreign and domestic assets/liabilities and, especially, those that focused on strictly matching domestic assets/liabilities. The smaller profitability of the mixed-matching strategy would also appear inconsistent with the performance potential of ALM. Fourth, banks that consistently pursued a strategy of matching foreign (domestic) assets/liabilities in their balance sheets were those that had the smallest (largest) mean balance sheet proportions of each foreign asset/liability. The results thus found that the most (least) profitable strategy focused on the matching of foreign (domestic) assets/liabilities, while maintaining the smallest (largest) proportion of foreign loans.
The greater the extent to which "very large" banks consistently followed mixed-matching of foreign and domestic assets/liabilities, the more likely it is that they practiced ALM. For these banks the percussions of events, including the "LDC loan crisis," would likely have affected their matched domestic balance sheet strategies. However, as discussed above, banks that consistently followed a mixed-matching strategy were less profitable than those that consistently followed a strategy of strictly matching foreign assets/liabilities. And, again, banks that consistently followed a strategy of strictly matching foreign (domestic) assets/liabilities had the smallest (largest) mean values for each of the foreign assets/liabilities. Thus, it remains to be determined here whether the foreign office activities of international banks impacted their domestic office balance sheet profiles and performance.
Previous research by Simonson, Stowe and Watson (1983) provides empirical support for research examining the nature of U.S. bank strategies, as reflected in the behavior of their asset/liability structures. Their study of balance sheet relationships focused on total (as opposed to foreign versus domestic) bank asset and liability variables. The results supported the concept of interdependence between asset and liability portfolio choice. Strong evidence was found of systematic asset/liability hedging to manage interest-rate risk, as well as other important links between specific asset/liability variables.
Later research by Amel and Rhoades (1988) provides both empirical and methodological support for studies of U.S. bank balance sheet strategies. They used cluster analysis (discussed below) to test for the existence of strategy groups in banking markets. No presumption was made concerning either the existence of or the number of these groups, if any, in these markets. The balance sheet composition of banks in different markets was analyzed for each of several years. They identified approximately six strategy groups and also found that bank membership in these groups was relatively stable over time.
More recently, research by Haslem, Scheraga and Stagliano (1993) also provides empirical and methodological support for studies of U.S. bank balance sheet strategies. This study utilized a time-series framework to analyze the foreign and domestic asset/liability matching strategies of large international banks. This was done in context of a changing banking environment using a 1984 and 1987 pooled sample.
Three significant asset/liability matching strategies were identified and use to characterize individual bank overall asset/liability strategies. These strategies were then classified into "strategy groups" using cluster analysis. The three major strategy groups that were identified may be characterized as "domestic focused" (emphasis on the matching of domestic assets/liabilities), "foreign focused" (emphasis on the matching of foreign assets/liabilities), and "mixed unfocused" (mixed-matching of foreign and domestic assets/liabilities). The general types of markets served in each strategy group were also identified. Further, the performance implications for banks that either did or did not change strategies were also identified.
Research Questions and Method of Analysis
The first question tested is whether the domestic balance sheet profiles of domestic U.S. banks differed from those of international U.S. banks. On an a priori basis, and as discussed above, these profiles would be expected to differ, especially in context of the banking events of the study period.
Traditionally, the test of this hypothesis would present an interesting methodological problem. As detailed in Haslem, Scheraga and Bedingfield (1992), this problem involves the need to a priori define classes of balance sheet strategies and then determine which banks meet the pre-defined criteria for inclusion in each class. There are also the problems associated with testing for overall significant differences in the pre-defined classes through use of pairwise t-tests of the individual variable means in each class.
To avoid these types of problems, cluster analysis was used to analyze the data.(2) In general, cluster analysis refers to a group of multivariate techniques that have a primary purpose of identifying similar variable constructs from the characteristics they have. It is a well documented, useful techniques that identifies and classifies pre-selected variables so that the values of each are similar to others in its cluster. The resulting clusters exhibit high inter-cluster heterogeneity and high intra-cluster monogeneity. Further, it explicitly identifies variable constructs on an ex post basis and thereby avoids the need to pre-define these constructs subjectively.
Cluster analysis was used in this study to identify the discrete classes of domestic office balance sheet profiles derived from the pooled sample of domestic and international banks. An iterative partitioning (i.e., a nonhierarchical rather than "tree like" hierarchical structure) algorithm was used to generate disjoint clusters. The cubic clustering criterion was used on an ex post basis to determine...
Um weiterzulesen
FORDERN SIE IHR PROBEABO ANCOPYRIGHT GALE, Cengage Learning. All rights reserved.
