Printer Friendly

An analysis of the foreign and domestic balance sheet strategies of the U.S. banks and their association to profitability performance.

Abstract

* This study analyzed the 1987 data of 176 relatively large U.S. banks that have both foreign and domestic offices. Canonical analysis and the interpretive framework of asset/liability management were used to identify and interpret their foreign and domestic balance sheet strategies in the context of the "crisis in lending to LDCs."

* The analysis found a consistent dichotomy in foreign and domestic asset/liability matching strategies, the former being more generally conservative with respect to interest-rate and liquidity risks. Among the 44 very large banks, those that were found to follow a predominant or consistent foreign strategy are more profitable than those that follow a mixed or, especially, domestic strategy. Further, these banks that follow a consistent foreign (domestic) matching strategy have the smallest (largest) mean proportions of all foreign asset and liability variables.

Key Words

* In conclusion, the least profitable very large banks have the largest proportions of foreign loans, yet they emphasize domestic balance sheet (asset/liability) matching strategies. Conversely, the most profitable very large banks have the smallest proportions of foreign loans, but, nonetheless, they emphasize foreign balance sheet matching strategies.

Introduction

The purposes of this study are (1) to determine the nature and implications of the foreign and domestic strategies reflected on both sides of the balance sheets of U.S. commercial banks and (2) to determine how the types of strategies followed by individual banks relate to relative profitability performance. The analysis uses 1987 balance sheet ratio data computed from the call reports of 176 relatively large U.S. banks that have both foreign and domestic banking offices.

The major impetus for this study is the need for an improved understanding and performance evaluation of the balance sheet strategies followed by U.S. banks within the context of the crisis in lending to less-developed countries (LDCs). The rising trend in these loans began in the early 1970's and peaked in 1983.(2) The high inflation of the 1970's had the effect of reducing the real debt service burden of the LDCs. However, this situation changed in the 1980's as inflation dropped significantly, real interest rates rose, the value of the dollar increased against LDC currencies, and a worldwide recession reduced demand for the exports of LDCs, especially oil. These factors made it extremely expensive for LDCs to service their international debts. The resulting debt crisis has been characterized by loan defaults, rollovers of maturing loans, maturity rescheduling of other loans, nonpayment of loan interest, and new loans to cover contractual interest payments. The subsequent reduction in the market value of U.S. bank portfolios of LDC loans led the banks to increase their capital and to curtail growth in general and loans to LDCs in particular.

The year 1987 was selected for study because it was the first year that banks gave major recognition to the reality of the "crisis in lending to LDCs."(3) The crisis intensified then when two LDCs each declared a moratorium on interest payments. Their creditor banks were then obliged to place large proportions of this debt on "nonperforming" status. This resulted in huge additions to loan-loss reserves and losses for these banks.

Further, 1987 was also a generally bad year for banks. It was characterized by banks earning the lowest return on assets since World War II. This poor performance was attributed to declining net interest margins, increased overhead expenses, and deteriorating credit quality. Thus, from a financial strategy standpoint, 1987 provides a background of likely sharp contrasts for examining the nature and performance implications of bank foreign and domestic balance sheet strategies.

Bank asset/liability management provides a framework for interpreting the general nature and implications of the identified foreign and domestic balance sheet strategies.(4) The essence of this systems approach is the management of discretionary assets and liabilities and the coordination of their interrelationships consistent with short term (usually) financial objectives.(5) This coordination is made more complex when both foreign and domestic operations are involved.

To the extent that foreign and domestic assets and liabilities are not matched consistently, asset/liability management provides a means for generally interpreting whether the identified strategies are relatively "conservative" or "aggressive" in terms of interest-rate and liquidity risks when compared to one another.

The use of canonical correlation analysis (discussed below) makes it possible to obtain a more specific and complete understanding of the nature of the relationships between bank asset and liabilities that are correlated both within and across their respective sides of the balance sheet. Further, canonical analysis is not limited to relationships with a single independent and multiple explanatory variables. It provides an analysis of the magnitude and nature of the relationships between sets of both "criterion" variables and "predictor" variables.

Related Bank Research

Previous research by Simonson et al. (1983) on U.S. bank balance sheet relationships focused on overall asset and liability variables. The results of the analysis supported the interdependence between asset and liability portfolio choice. There was strong evidence of systematic asset/liability hedging to manage interest-rate risk. Several other important links were also identified between bank asset and liability structures. Their empirical results provide support for research examining the nature of foreign and domestic strategies reflected in bank asset/liability structures.

Two other studies analyzed the relative effectiveness of the management of foreign and domestic banking activities. The earlier study (Haslem, et al. 1983) analyzed the nature of the association between selected foreign activity ratios and the relative profitability of 99 U.S. banks for each of the years 1978-1980. In part, the analysis found that 19 percent of the 48 selected ratios of foreign banking activity had either a consistently positive or negative association with bank profitability. In fact, 75 percent of the foreign activity ratios of the most profitable banks reflected either the largest or smallest degree of foreign activity relative to its domestic counterpart. This suggests that U.S. banks should focus on improved asset/liability management of their foreign activities if they wish to benefit from the potential for improved overall profitability.

The latter study (Haslem, et al. 1986) used factor analysis on the same basic data to determine the "factors" that explained most of the total variance in foreign banking activities relative to their domestic counterparts that was explained by all of the factors. The most important types of factors identified were "loan assets" and "cash assets." The significant factors were then related to overall bank profitability. Three general conclusions were reached. First, the banks managed some asset-related foreign banking activities, especially various loan activities, less consistently than their liability-related foreign activities. This finding may have reflected shortcomings in the asset/liability management of balance sheet sources and uses of funds. Second, some of the various foreign loan assets were significantly less profitable than their domestic counterparts. This finding was consistent with the widespread occurrence of problem loans to LDCs. Third, the results of the 1978 - 1980 analyses were generally consistent in the factors that explained most of the variance in the behavior of foreign banking activities relative to their domestic counterparts and in the nature of their relationship to overall bank profitability. While these two related studies separately analyzed foreign banking activities relative to their domestic counterparts, they did not analyze and identify the interdependent and simultaneous nature of the relationships between the foreign and domestic assets and liabilities of U.S. banks.

Sample Selection and Construction

All 256 banks that in 1987 were required to file Federal Financial Institutions Examination Council Form 031 (FFIEC 031) are the source of data for this study. This call-report form is required of all U.S. banks that have foreign branches or subsidiaries in U.S. territories or possessions, Edge Act or "agreement" corporation subsidiaries, foreign branches, consolidated foreign subsidiaries, or International Banking Facilities.(6)

After initially selecting all banks that filed Form 031, those with any of the following data attributes were eliminated from the sample: (1) no foreign office assets reported; or, (2) no foreign office liabilities reported; or, (3) missing data in any of the data items (accounts) used in the analysis. This resulted in 177 banks, but an additional bank was deleted because it had both negative income and capital, which would have distorted its profitability ratios. Thus, the final sample includes 176 banks with both foreign and domestic offices.

The total sample of 176 banks was then divided into "large bank" and "very large bank" samples. This was done because, as expected, the sample balance sheet variables were sensitive to total asset size. The total sample was dichotomized at a bank total asset size of $8.0 billion, the point at which the variable means provided the first general evidence of statistically significant differences. The mean values of these variables for the two resulting samples are shown in Appendix A. Eight of the thirteen variables for the two resulting samples are statistically different at the 0.05 level and the ninth at the 0.10 level.

The result is a "large bank" sample of 132 banks each with less that $8.00 billion in total assets and a "very large bank" sample of 44 banks each with $8.0 billion or more in total assets. The large banks have mean total assets of $3.0 billion, with the smallest having total assets of $45.8 million. Further, the very large banks have mean total assets of $26.4 billion, with the largest having total assets of $154.1 billion.

Variable Derivation and Definitions

As mentioned above, the variables analyzed in this study were computed from Form 031 data items. All of the balance sheet (report of conditions) variables were computed as proportions of total assets. The net income variable from the report of income was computed both as a proportion of total assets and of total equity capital. These variables are defined in Appendix B.

The counterpart foreign and domestic variables (e.g., foreign cash and domestic cash) derived from the call report data items are more general than some of the quite specific variables (e.g., interest-sensitive loans) available for asset/liability management studies of overall bank balance sheet relationships. This tradeoff in variable specificity is necessary if foreign and domestic balance sheet strategies are to be separately analyzed, identified, and interpreted. This tradeoff dictates that variables such as net domestic loans and foreign investment securities include both rate- sensitive and fixed-rate assets.

Method of Analysis

Canonical correlation analysis was used to analyze the balance sheet relationships for the large and very large bank samples.(7) This method was selected because, as discussed above, it is a general multivariate statistical model that facilitates the study of interdependent relationships among sets of multiple criterion and predictor variables, where causality need not be assumed.

In the context of this study, canonical correlation relates the asset and liability variables from both within and across the variable sets. The lack of a causality assumption is important here because an asset or liability account "is not simply determined but is jointly determined with balances of other accounts on the same side of the balance sheet, as well as on the opposite side" (Crum, et al. 1987). Canonical correlation provides the ability the analyze these complex interactions simultaneously and produce summary measures of the underlying relationships. This is necessary to gain an understanding of the complex nature of balance sheet strategies.

The model derives a linear combination/composite of asset and liability variables from each of the two variables sets such that the correlation between the pair of linear composites is maximized. Each successive pair of composites derived are orthogonally independent of the prior pair and maximizes the residual correlation between the two variable sets. The number of pairs of linear composites (canonical variates) derived from this analysis necessarily equal the number (six) of variables in the set (liability) with the fewest variables.

To interpret the results further, three statistical tests were performed. F-test were used to determined the number of significant variates - four in the large bank sample and three in the very large bank sample. Canonical roots were computed to indicate the amount of shared variance between the optimally weighted linear composites of the variable sets. Redundancy analysis was used to measure the ability of a set of variables to explain variability in the other set taken one variable at a time. The larger the redundancy index (proportion of total redundancy), the larger the shared variance (dependence) of the variable sets.

Further, because the asset and liability variables are measured as proportions of total assets and of total liabilities and capital, respectively, totalling 100 percent, the potential singularity condition was overcome by dropping one variable from each side of the balance sheet. Nonetheless, the procedure used maintains the information content of all the variables in the asset and liability sets, with the canonical correlation and loading being invariant with respect to the omitted variable in each set.

Analysis of the canonical loadings of the variables that define each variable set identifies the variables that are most useful for explaining the observed covariances for each pair of canonical variates. A canonical loading measures the correlation between a variable and its respective linear composite. The larger its loading the greater the contribution of a variable to its composite. Variables with loadings of the same (different) sign are directly (inversely) related to one another.

Canonical analysis was also used to generate canonical variate scores for each of the sample banks. A bank's canonical score for each of its set of variables is obtained by multiplying the standardized values for its variables by the respective canonical weights derived for the variable set.(8) The sign of a bank's canonical score for each of its variable sets is determined by the variables with the largest influence in the construction of the canonical variate. This influence is directly proportional to the size of the canonical loadings for these variables. Thus, the predominant pattern of algebraic loading signs for particular foreign or domestic variables determines the signs of a bank's canonical scores. The canonical scores make it possible to divide the banks into categories representing the type of balance sheet matching strategy (e.g., predominant foreign strategy) followed by the bank. These categories facilitate the determination of the relative profitability performance associated with banks following particular balance sheet matching strategies.

Overall Results and Interpretation of the Analysis

Table 1 presents the test statistics for the most significant canonical variates identified by the canonical analysis in both the large bank and very large bank samples. It is apparent from their probabilities that variates 1 - 4 and 1 - 3 are very highly significant for the large bank and very large bank samples, respectively.
Table 1. Test Statistics for Canonical Variates Large and Very
Large Banks - 1987
Variate Large Banks
 Likelihood Ratio F Approximation Probability
First 0.1435 10.1195 0.0001
Second 0.4370 5.7493 0.0001
Third 0.6182 5.4106 0.0001
Fourth 0.8291 4.0591 0.0007
P>0.05 for fifth canonical function.
Variate Very Large Banks
 Likelihood Ratio F Approximation Probability
First 0.0089 10.0901 0.0001
Second 0.1432 4.5303 0.0001
Third 0.3786 3.4340 0.0003
P>0.05 for fourth and fifth canonical functions.


Table 2 presents the canonical roots of the linear composites in the most significant variates and the redundancy analysis results of the linear composites and their opposite variable sets in both the large bank and very large bank samples. The linear composites in the first variate in the large bank and very large bank samples have canonical roots of 67.2 percent and 93.8 percent, respectively. These represent the largest amount of shared variance between the linear composites of the asset and liability sets. As is to be expected, the roots for the linear composites in both samples get smaller within each succeeding variate. The redundancy analysis of the large bank sample reveals that the linear composite of asset (liability) variables explains 16.5 (18.4) percent of the variance in the liability (asset) set of variables. The redundancy analysis of the very large bank sample reveals that the linear composite of asset (liability) variables explains 48.8 (37.9) percent of the variance in the liability (asset) set of variables. As is to be expected, the proportion of total redundancy in both samples gets smaller within each succeeding variate.

[TABULAR DATA OMITTED]

Tables 3 and 4 present the rotated canonical loadings of the variables in the significant canonical variates in the large and very large bank samples, respectively. The varimax rotation criterion was employed to perform the canonical rotation analysis (Cliff and Krus 1976, Kaiser 1958, Perreault and Spiro 1978). Each loading represents the simple correlation between a given variable in the particular asset or liability variable set and that set's canonical variate. The asset and liability variables with significant loadings and consistent signs (both positive or both negative) are positively related, and may be considered as representing a "matched balance sheet strategy."(9)

[TABULAR DATA OMITTED]

The matched balance sheet relationships in each variate represent one set of characteristics of balance sheet strategies across all sample banks. For example, it may be seen in Table 3 that the large banks in the first variate matched domestic loans (-0.7443 loading) and domestic interest-bearing deposits (- 0.6411); also, they matched foreign cash (0.9597) and foreign interest-bearing deposits (0.9080). In this case, asset/liability management separately coordinated (matched) foreign and domestic balance sheet variables. Further, within the limits of the general definitions of the variables, the asset/liability management of the foreign strategy is more conservative than that of the domestic strategy. The large banks are primarily funding their rate-sensitive and fixed-rate domestic loans with rate-sensitive domestic deposits. No focus on use of the more stable core deposits is indicated.(10)

To simplify the discussion of the interpretation of the significant foreign and domestic asset and liability interrelationships, Table 5 was prepared. This table provides a summary of the significant variates and closely matched foreign and domestic asset and liability strategies for the large and very large bank samples.

[TABULAR DATA OMITTED]

The most striking overall result of the canonical analysis of the large and very large bank samples is the consistent dichotomy in the asset/liability management funds matching strategies. In all cases where the asset and liability variables are strictly foreign or domestic, the consistent signs and large sizes of the canonical loadings match only foreign liabilities with foreign assets and/or only domestic liabilities with domestic assets. In no case are there any significant mixed-sign correlations which match foreign assets (liabilities) with domestic liabilities (assets).

This dichotomy strongly suggests that asset/liability management maintained a very significant degree of independence between foreign and domestic strategies in both samples. This dichotomy appears inconsistent with the normative practice of asset/liability management as a global, systems approach to balance sheet management. Whether or not this dichotomous behavior is consistent with high or low profitability performance was determined by an analysis of individual bank funds matching strategies (discussed below).

Results of the Analysis of the Large Bank Sample

Table 5 presents the summary results (details in Table 3) of the canonical analysis of the large bank sample. The first variate reveals a significant foreign relationship that correlates interest-bearing deposits with cash. It also reveals a significant domestic relationship that correlates interest-bearing deposits with loans. Thus, as discussed in the previous example, the foreign strategy is more conservative than the domestic strategy which has interest-rate and liquidity risks.

The second variate does not reveal a significant, strictly foreign relationship. However, it does indicate a significant partial foreign relationship that correlates total "other" assets and total equity capital. Total other assets is a residually computed variable that includes foreign and domestic assets. These include rate-sensitive assets such as Federal funds sold, securities purchased under agreements to resell, and bankers acceptances outstanding; assets in trading accounts; and premises and fixed assets. Total equity capital includes undivided profits and capital reserves from both foreign and domestic operations and the other accounts that provide a bank's "permanent" shareholder base. Because these variables reflect a variety of relationships, both foreign and domestic, they provide no matching of strictly foreign funds. The variate also includes a significant domestic relationship that correlates interest-bearing deposits and loans. The domestic strategy of primarily funding rate-sensitive and fixed-rate loans with rate-sensitive deposits has interest-rate and liquidity risks. This is not a particularly conservative strategy. Nevertheless, the domestic strategy cannot be compared with a strictly foreign funds matching strategy.

The third variate does not reveal a significant domestic relationship. Neither does it reveal a significant, strictly foreign relationship. However, it does indicate significant, partial foreign relationships that correlate foreign noninterest-bearing deposits and total "other" liabilities with loans. Total other liabilities is a residually computed variable that includes foreign and domestic liabilities. These include rate-sensitive liabilities such as Federal funds purchased, securities sold unter agreements to repurchase, and bankers acceptances outstanding; mortgage indebtedness; and other borrowed funds. The partial foreign strategy of primarily funding rate-sensitive and fixed-rate loans with core deposits and both rate- sensitive and long-term liabilities may reveal relatively conservative "matched risk-class" funds. Nevertheless, no comparison of strictly foreign and domestic funds matching strategies can be made.

The fourth variate does not reveal a significant, strictly foreign relationship. However, it does indicate a significant, partial foreign relationship that correlates total other liabilities with foreign securities. The variate includes a domestic relationship that correlates noninterest-bearing deposits with loans. The partial foreign strategy of primarily funding rate-sensitive and longer-term securities with both rate- sensitive and long-term liabilities may reveal relatively conservative matched risk-class funds. The domestic strategy of primarily funding rate-sensitive and fixed-rate loans with core deposits is a conservative one. Nevertheless, this conservative domestic strategy cannot be compared with a strictly foreign funds matching strategy.

In sum, based on the larger total redundancy (amount of variance in one variable set explained by a linear composite of the other variable set) of variates one plus four compared to variates two plus three, it may be concluded that large banks follow generally more conservative strategies in matching their canonically significant foreign assets and liabilities than they do in the strategies matching their significant domestic assets and liabilities. The first variate is given additional weight in this conclusion because the other three variates do not reflect strategies that match strictly foreign and domestic funds. These findings carry with them implicit, but traditional, assumptions about the relative riskiness of cash, securities, and loans in meeting the borrowing needs of current and potential customers.(11)

Results of the Analysis of the Very Large Bank Sample

Table 5 also presents the summary results (details in Table 4) of the canonical analysis of the very large bank sample. The first variate does not reveal a significant, strictly foreign relationship. However, it does indicate significant, partial foreign relationships that correlate interest- bearing deposits and total equity capital with cash, loans, and total other assets. The total other assets and total equity capital variables include both foreign and domestic relationships. The variate includes a significant domestic relationship that correlates interesting bearing deposits with loans. The partial foreign strategy of funding rate-sensitive cash assets, rate- sensitive and fixed-rate loans, rate-sensitive and long-term other assets with rate-sensitive deposits and shareholder capital may reveal a relatively conservative strategy of matched risk-class funds. The domestic strategy of funding rate-sensitive and fixed-rate loans with rate-sensitive deposits has interest-rate and liquidity risks. Nevertheless, the domestic strategy cannot be compared with a strictly foreign funds matching strategy.

The second variate reveals significant foreign relationships that correlate interest-bearing deposits with cash and securities. It also indicates a significant domestic relationship that correlates noninterest-bearing deposits with very liquid cash assets. The foreign strategy that primarily funds rate-sensitive cash assets and securities as well as fixed-rate securities with rate-sensitive deposits has interest-rate and liquidity risks. The domestic strategy primarily funds rate-sensitive cash assets with core deposits. Thus, the domestic strategy is more conservative than the foreign funds matching strategy.

The third variate reveals significant foreign relationships that correlate noninterest-bearing deposits and interest-bearing deposits with securities and loans. It also indicates a significant domestic relationship that correlates interest-bearing deposits with loans. The foreign strategy that primarily funds both rate-sensitive and fixed-rate securities and loans with rate-sensitive and core deposits may reveal a relatively conservative strategy of matched risk-class funds. The domestic strategy that primarily funds rate-sensitive and fixed-rate loans with rate-sensitive deposits has interest-rate and liquidity risks. Thus, the foreign strategy is more conservative than the domestic funds matching strategy.

In sum, based on the larger total redundancy of variates one plus three compared to that of variate two, it may be concluded that very large banks follow more conservative strategies in matching their canonically significant foreign assets and liabilities than they do in the strategies matching their significant domestic assets and liabilities. These findings also carry with them implicit assumptions about the relative riskiness of bank assets.

Analysis of Balance Sheet Strategy and Performance

Once the bank balance sheet matching strategies for foreign and domestic assets and liabilities were analyzed for the two samples, the types of strategies followed by the individual banks were related to relative profitability performance (measured by net income to total assets and by net income to total equity capital). These two performance measures are highly correlated; nevertheless, two measures are used to show the stability of the results. As discussed above, individual bank strategies were related to profitability performance by using the canonical scores for each bank in the paired variates. The particular combination of signs of the canonical scores identified the nature of a bank's predominant strategy in matching its foreign and domestic assets and liabilities. Consistent signs, (+/+ or -/-) of both canonical scores (i.e., of both asset and liability variable sets) indicate a "predominant strategy" of matching either domestic assets and domestic liabilities or foreign assets and foreign liabilities. Mixed (+/- or -/+) signs of both canonical scores indicate a cross matching of foreign (domestic) assets and domestic (foreign) liabilities. The combination of signs of a bank's canonical scores in a given pair of variates that represent a particular balance sheet strategy is determined, as previously discussed, by the predominant size of the canonical loadings of the variables that have the largest influence on the canonical variates. This discussion of strategy and performance is limited to the findings for the very large bank sample. This is the most relevant sample because it includes the banks most involved in lending to LDCs, and, not surprisingly, it demonstrates major differences in the number of banks that follow the various types of matching strategies.

As shown in Table 6, a pair of consistent positive (negative) canonical scores in the first canonical variate indicates a predominant foreign (domestic) asset and liability matching strategy. A pair of consistent positive (negative) scores in the second variate indicates a predominant domestic (foreign) matching strategy. And a pair of consistent positive (negative) canonical scores in the third variate indicates a predominant foreign (domestic) matching strategy. A pair of mixed canonical scores (either +/- or -/+) in any variate indicates a mixed strategy of matching either domestic assets with foreign liabilities or foreign assets with domestic liabilities. The nature of the mixed strategies, as shown by the particular order of the signs of the canonical scores in a given variate, does not matter for purposes of the analysis.

[TABULAR DATA OMITTED]

Table 6 also records the profitability performance of the banks which follow the various types of matching strategies. Those banks in each variate that follow the predominant foreign strategy of matching foreign assets and liabilities have the largest ROA (net income to total assets) and ROE (net income to total equity capital); those banks that follow the predominant mixed strategy have the second largest ROA and ROE; and those banks that follow the predominant domestic strategy have the smallest ROA and ROE.(12)

The more significant the variate, the larger the number of banks that follow the predominant foreign matching strategy. The predominant foreign strategy is followed by over 61 percent of the banks in the first variate, by over 36 percent of those in the second variate, and by over 34 percent of the banks in the third variate. The predominant mixed strategy is followed by over 9 percent of the banks in the first variate, by over 31 percent of those in the second variate, and by over 29 percent of the banks in the third variate. The predominant domestic strategy is followed by nearly 30 percent of the banks in the first variate, by nearly 32 percent of those in the second variate, and by over 36 percent of the banks in the third variate. Thus, the most profitable predominant foreign strategy is followed by a wide majority of the banks in the first variate but only by about one-third of those in the less important second and third variates.

Next, the performance of the individual very large banks that follow "consistent strategies" - foreign, domestic or mixed - simultaneously across all three canonical variates were analyzed. Consistent strategy behavior was analyzed because it has the potential for revealing the sharpest evidence of the performance implications of a particular strategy.

Table 6 indicates that banks that follow a consistent foreign strategy of matching foreign assets and liabilities have the largest ROA and ROE; those banks that follow a consistent mixed strategy have the second largest ROA and ROE; and those banks that follow a consistent domestic strategy have the smallest ROA and ROE. Unfortunately, only about 7 percent of the banks follow a consistent foreign strategy. Not unexpectedly, over 86 percent of the banks follow a consistent mixed strategy. The remaining nearly 7 percent of banks follow a consistent domestic strategy. These percentages suggest a great deal of opportunity for very large banks to improve their profitability performance.

Finally, the mean values of the balance sheet variables for the very large banks that have consistent foreign, domestic or mixed strategies (across all variates) were calculated. As also shown in Table 7, the evidence is very striking and consistent.(13) Banks that follow a consistent foreign strategy have the smallest mean values for all five of the strictly foreign asset and liability variables. They also have the largest mean values for two of the five strictly domestic asset and liability variables (securities and interest-bearing deposits). On the other hand, banks that follow a consistent domestic strategy have the largest mean values for all five of the strictly foreign asset and liability variables. They also have the smallest mean values for four of the five strictly domestic asset and liability variables.

[TABULAR DATA OMITTED]

Thus, the evidence points to the highest (lowest) performing very large banks following a balance sheet matching strategy that involves a high degree of association between specific foreign (domestic) asset and liability variables, while they maintain the foreign variables at relatively low (high) levels. For the highest performing banks, this evidence reflects a close matching of foreign assets and liabilities that are maintained at relatively low levels -careful asset/ liability management of foreign activities "at the margin" in conjunction with conservative "mean values" of foreign assets and liabilities. This is not a surprising finding given the revealed high risk of lending to LDCs. The high-performance banks evidently anticipated this problem and practice relatively conservative credit and asset/liability management strategies with their foreign assets and liabilities.

Summary

This study analyzed the 1987 data of 176 relatively large U.S. banks that have both foreign and domestic banking offices. Canonical analysis and the interpretive framework of asset/liability management were used to identify and interpret their foreign and domestic balance sheet strategies in the context of the "crisis in lending to LDCs." The analysis generated the following results: (1) a consistent dichotomy in strictly foreign and domestic balance sheet matching strategies, that appears inconsistent with a normative systems approach to asset/ liability management; (2) strategies involving the matching of foreign assets and liabilities are generally more conservative (less interest-rate and liquidity risks) than those matching domestic assets and liabilities; (3) very large banks that follow a "predominant" or "consistent" foreign strategy of matching foreign assets and liabilities are more profitable than those that follow a mixed or, especially, domestic matching strategy; and (4) very large banks that follow a consistent foreign (domestic) matching strategy have the smallest (largest) mean balance sheet proportions of all foreign asset and liability variables. Thus, the least profitable very large banks have the largest proportions of foreign loans, yet they emphasize domestic balance sheet matching strategies. Conversely, the most profitable very large banks have the smallest proportions of foreign loans, but, nonetheless, they emphasize foreign balance sheet matching strategies.

Appendix A. Mean Values of Balance Sheet Proportions and Returns Large vs. Very Large Banks

- 1987

[TABULAR DATA OMITTED] * P<0.10; ** <0.05; *** P<0.01.

Appendix B. Variable Definitions

Domestic Cash

This item is RCON 0010 ("Cash and balances due from depository institutions" for domestic offices).

Foreign Cash

This item is derived by the Federal Reserve by deducting RCON 0010 ("Cash and balances due from depository institutions" for domestic offices) from RCFD 0010 ("Cash and balances due from depository institutions" for the consolidated bank) to arrive at RCFN 0010 ("Cash and balances due from depository institutions" for foreign offices).

Domestic Investment Securities This is RCON 0390 ("Total investment securities-book value" for domestic offices).

Foreign Investment Securities

This item is derived by the Federal Reserve by deducting RCON 0390 ("Total investment securities-book value" for domestic offices) from RCFD 0390 ("Total investment securities-book value" for the consolidated bank) to arrive at RCFN 0390 ("Total securities-book value" for foreign offices).

Net Domestic Loans

This is RCON 2122 ("Total loans and leases-net of unearned income" for domestic offices) reduced by an allocated portion of RCFD 3123 ("Allowance for possible loan losses") and RCFD 3128 ("Allocated transfer risk reserve"); the allocated portion is the percentage that RCON 2122 constitutes of RCFD 2122 ("Total loans and leases-net of unearned income" for the consolidated bank).

Net Foreign Loans

This is RCFD 2125 ("Net loans" for the consolidated bank) reduced by Net Domestic Loans computed above.

Total Other Assets

This is RCFD 2170 ("Total assets" for the consolidated bank) reduced by the sum of 1) Domestic Cash; 2) Foreign Cash; 3) Domestic Investment Securities; 4) Foreign Investment Securities; 5) Domestic Net Loans; and 6) Foreign Net Loans.

Total Domestic Noninterest-bearing Deposits

This is RCON 6631 ("Noninterest-bearing deposits" in domestic offices).

Total Foreign Noninterest-bearing Deposits

This is RCFN 6631 ("Noninterest-bearing deposits" in foreign offices).

Total Domestic Interest-bearing Deposits

This is RCON 6636("Interest-bearing deposits" in domestic offices).

Total Foreign Interest-bearing Deposits

This is RCFN 6636 ("Interest-bearing deposits" in foreign offices).

Total Other Liabilities

This is RCFD 2948 ("Total liabilities" for the consolidated bank) less: 1) Total Domestic Noninterest-bearing Deposits; 2) Total Foreign Noninterest-bearing Deposits; 3) Total Domestic Interest-bearing Deposits; and 4) Total Foreign Interest-bearing Deposits; plus RCFD 3282 ("Limited-life preferred stock" for the consolidated bank).

Total Equity Capital

This is item RCFD 3210 ("Total equity capital" for the consolidated bank).

Net Income

This is item RIAD 4340 ("Net income" for the consolidated bank).

Total Assets

This is item RCFD 2170 ("Total assets" for the consolidated bank). It was used as the denominator in all of the ratios except where Net Income was related to Total Equity Capital. For the liability variables it is equivalent to item RCFD 3300 ("Total liabilities, limited-life preferred stock and equity capital" for the consolidated bank).

Notes (1) The support of the University of Maryland Computer Science Center and the College of Business and Management Office of Computing Services are acknowledged. The helpful comments of Richard M. Durand and Thomas M. Corsi, University of Maryland are also gratefully acknowledged. A special thanks is due David Scholtzhauer of the SAS Institute for his kind assistance. (2) For a more complete discussion of the history and developments in lending to LDCs, see Bennett and Zimmerman [1988]; also, Todd [1988]. (3) See Gregorash and Ford [1988] for a succinct discussion of general and LDC banking developments in 1987. (4) For a discussion of asset/liability management, see Haslem [1984] and Sinkey [1989]. The major operating task of asset/liability management is the control of interest-rate and liquidity risks by avoiding mismatches in the interest-rate sensitivities (and often maturities) of assets and liabilities. A rate-sensitivity mismatch increases interest-rate and liquidity risks by, for example, funding fixed-rate term loans with rate- sensitive short-term deposits. (5) The implications of these asset and liability interrelationships are normally measured by the impact on a bank's net interest margin return and variability, which is a function of the interest-rate sensitivity, volume, and mix of its earning assets and its liabilities. (6) The data wer computed from the Report of Condition and Income tapes [Board of Governors 1987]. [7] The analysis was performed using the SAS Institute (1985) statistical package. For useful background discussions of canonical analysis, se Hair et al. [1987]; also Green et al. [1966] and Kuylen and Verhallen [1981]. For early applications in international finance, finance, transportation and geography, see Schmidt [1976], Stowe et al. [1980], Corsi and Scheraga [1989], and Corsi and Harvey [1975], respectively. [8] For each original variable, [x.sub.ij] (value of the ith variable in the jth case), the standard score, [z.sub.ij], is obtained from [z.sub.ij] = [x.sub.ij] - [x.sub.j]/[s.sub.i] where [x.sub.j] is the mean of [x.sub.i] across all observations and [s.sub.i] is the standard deviation of [x.sub.i]. These z-values are then substituted in the canonical variate to obtain the canonical score. [9] Previous studies have used values of significant loadings ranging from 0.20 to 0.40, with 0.30 being most common. This study used a reasonably conservative value of 0.35 for significant loadings. See Lambert and Durand [1975] for a discussion of significance levels. [10] This situation has potential risks when the amount of risk-sensitive assets is less than the amount of rate-sensitive liabilities (and the yield curve slopes upward), an increase in interest rates causes a squeeze in profits as interest expenses increase more rapidly than the returns on earning assets ("interest-rate risk"). Also, the increase in interest rates may cause disintermediation of the rate-sensitive liabilities that are funding the rate-sensitive and fixed-rate assets ("liquidity risk"). Further, these banks are also primarily funding their rate-sensitive foreign cash assets with rate-sensitive foreign deposits. Thus, the domestic strategy reveals greater interest-rate and liquidity risks than the foreign strategy. [11] These asset risks are derived from the risks of default, declines in market values, interest rate increases, and inflationary price increases. The assumptions about the relative risks of bank assets are probably consistent within domestic markets and perhaps within foreign markets. But they may well be at different absolute levels between domestic and foreign markets due to their different environments. Thus, in addition to the above risks, foreign banking activities are also subject to country risk (e.g., adverse laws, nationalization of assets, expropriation of assets, and prohibition of foreign loan payments) and foreign exchange risk (e.g., adverse exchange rates, lack of foreign exchange markets, currency restrictions, and foreign exchange controls). The loan exposure to LDCs is the major source of country risk for U.S. creditor banks. [12] Due to the large differences in the number of banks in the strategy groups, multiple comparisons of means using such techniques as Duncan's multiple range or Waller-Duncan k-ratio tests could not be used. [13] As would be expected, this evidence is generally more clearcut than it is for the banks that follow a consistent strategy in a given variate, especially the second and third variates.

References Bennett, Barbara A., and Gary C. Zimmerman (1988): U.S. banks' exposure to developing countries: An examination of recent trends. Economic Review, Federal Reserve Bank of San Francisco, Spring: 14-29. Cliff, Norman, and David J. Krus (1976): Interpretation of canonical analysis: Rotated vs. unrotated solutions. Psychometrika, March 41: 35-42. Corsi, T.M., and M.E. Harvey (1975): The socio-economic determinants of crime in the city of Cleveland: The application of canonical scores to geographic processes. Journal of Economic and Social Geography, June 66: 323-336. Corsi, Thomas M., and Carl A. Scheraga (1989): Pre- and post-deregulation financial strategies and linkages to performance in the motor carrier industry: An application of canonical scores. Transportation Research, 2, 23A: 161-171. Crum, Michael R., Daulatram B. Lund, and Howard E. Van Auken (1987): A canonical correlation analysis of carrier financial strategies: The case of airline deregulation. Transportation Research, 3, 21 A: 179-190. Green, Paul E., Michael H. Halbert, and Patrick J. Robinson (1966): Canonical analysis: An exposition and illustrative application. Journal of Marketing Research, February 3: 32-39. Gregorash, George, and Theresa Ford (1988): Banking 1987: A year of reckoning. Economic Perspectives, Federal Reserve Bank of Chicago, July/August: 3-13. Hair, Jr., F. Joseph, Ralph E. Anderson, and Ronald L. Tatham (1987): Multivariate data analysis with readings, 2nd ed., New York: Macmillan Publishing Co. Haslem, John A. (1984): Bank funds management. Reston, VA: Reston Publishing Co. Haslem, John A., James P. Bedingfield, and A.J. Stagliano (1983): An analysis of international banking measures and relative profitability. Management International Review, 3, 23: 48-60. Haslem, John A., Andreas Christofi, James P. Bedingfield, and A.J. Stagliano (1986): A statistical analysis of international banking measures and relative profitability. Management International Review, 2, 26: 5-13 Kaiser, Henry F. (1958): The varimax criterion for analytic rotation in factor analysis. Psychometrika, September 23: 187-200. Kuylen, Anton A.A., and Theo M.M. Verhallen (1981): The use of canonical analysis. Journal of Economic Psychology, 1: 217-237. Lambert, Zarrel V., and Richard M. Durand (1975): Some precautions in using canonical analysis. Journal of Marketing Research, November 12: 468-475. Perreault, Jr., William D., and Rosann L. Spiro (1978): An approach for improved interpretation of multivariate analysis. Decision Sciences, July 9: 402-413. Report of condition and income for commercial banks and selected other finanical institutions (1987): Washington, D.C.: Board of Governors of the Federal Reserve System, December 31, 1987 data. SAS user's guide: Statistics, ver 5 ed. (1985): Carey, North Carolina: SAS Institute. Schmidt, Reinhart (1976): Determinants of corporate debt ratios in Germany. 1975 Proceedings of the European Finance Association, ed. Richard Brealey and Graeme Rankine, Amsterdam: North-Holland Publishing Co. Simonson, Donald G., John D. Stowe, and Collin J. Watson (1983): A canonical correlation analysis of commercial bank asset/liability structures. Journal of Financial and Quantitative Analysis, March 18: 125 -140. Sinkey, Jr., F. Joseph (1989): Commercial bank financial management, 3rd ed., New York: Macmillan Publishing Co. Stowe, John D., Collin J. Watson, and Terry D. Robertson (1980): Relationships between the two sides of the balance sheet: A canonical correlation analysis. Journal of Finance, September 35: 973-980. Todd, Walker (1988) Developing country lending and current banking conditions. Economic Review, Federal Reserve Bank of Cleveland, February 24: 27-36.
COPYRIGHT 1992 Gabler Verlag
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:includes appendices
Author:Haslem, John A.; Scheraga, Carl A.; Bedingfield, James P.
Publication:Management International Review
Date:Jan 1, 1992
Words:7170
Previous Article:Under what conditions can exports exert a positive influence on profitability?
Next Article:Cultural baggage and the adaptation of expatriate American and Japanese managers.
Topics:


Related Articles
Recent developments affecting the profitability and practices of commercial banks.
An analysis of the impact of international activity on the domestic balance sheet of U.S. banks.
Profits and Balance Sheet Developments at U.S. Commercial Banks in 2000.
Steinway net sales declined in third quarter. (Items of Interest).
Fitch Ratings Affirms UMB Financial at 'A+'; Outlook Remains Negative.
Profits and balance sheet developments at U.S. Commercial Banks in 2003.
Profits and balance sheet developments at U.S. commercial banks in 2004.
The Chinese Banking Industry is Expected to Witness an Annual Growth Rate of 10%.
Steady Economic Growth in Norway Has Driven Bank Lending, Thereby, Sustaining the Net Interest Income - Norway Retail Banking Industry: An Analysis.
Fitch Affirms National Commercial Bank Jamaica Ratings; Outlook Stable.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters