Printer Friendly

Determinants of the rate of return on S&L assets: 1970-97.

Abstract

This study identifies determinants of the rate of return on U.S. Saving and Loan (S&L) assets during the 1970-97 period. The Instrumental Variables (IV) estimation reveals that this rate of return is an increasing function of the spread between the S&L mortgage interest rate and the S&L cost of funds, the regulatory S&L capital-asset ratio, and the percentage growth rate of real GDP. It is negatively affected by the Tax Reform Act of 1986 and positively affected by the Federal Deposit Insurance Corporation Improvement Act of 1991. Based on these findings, certain policy implications and general conclusions are suggested. (JEL G20, G21)

Introduction

In the U.S., Savings and Loans (S&Ls) have historically been the principal institution supplying home mortgages to the public. The S&L crisis in the U.S. that began in the early 1980s and involved an enormous increase in the S&L failure rate has received considerable attention in the news media and research literature. (1) The media has focused largely on the number of S&L failures, the pecuniary cost of those failures to taxpayers, and allegations of fraudulent behavior on the part of S&L directors and officers. The research literature has focused more on the resolution costs of the S&L failures and on the apparent causes of these S&L failures, at least in part with the objective of providing insights to help prevent such failures in the future.

A considerable proportion of this research literature focuses on the role of the federal deposit insurance coverage system in S&L failures. A well-known study of the S&L crisis, Barth [1991, p. 101], convincingly argues that, "... federal deposit insurance was the unifying cause of the savings and loan disaster." Barth further argues that federal deposit insurance encouraged the S&Ls to take on additional risk and, in so doing, significantly contributed to the rate of S&L failures. He alleges that, "... the very availability of such insurance enabled many inadequately capitalized savings and loans to engage in high-risk activities and to gamble for resurrection." Other studies [Barth and Bartholomew, 1992; Barth and Brumbaugh, 1992; Brumbaugh, 1988; Cebula and Hung, 1992A; and Kane, 1982] argue similarly, although the study by Saltz [1995] seemingly provides empirical evidence to the contrary.

Federal deposit insurance coverage is by no means viewed as the only significant cause of S&L failures over time. Factors such as the rising cost of deposits, increased interest rate volatility, declining capital-asset ratios, declining crude oil prices, the 1981-82 recession, Regulation Q, and provisions of the Tax Reform Act of 1986 are also viewed as potentially having influenced the S&L failure rate.

Another dimension of the S&L crisis is the geographic variation in the S&L failure rate. Amos [1992] investigates determinants of interstate differentials in the commercial bank closing rate, focusing principally on interstate differences in the growth rate of gross state product, the volatility of gross state product, the percentage of gross state product deriving from manufacturing, from agriculture, or from oil and natural gas extraction, and other factors. Using Amos [1992] as a point of departure, Cebula [1994] expands the scope of inquiry into interstate differentials in commercial bank closings to include a variety of money market (and other) factors, such as the cost of funds, capital-asset ratios, and the extent to which interstate banking is permitted. (2) Finally, depending to some degree on Amos [1992], Barth [1991], and Cebula [1994], Chou and Cebula [1996] use the heteroskedastic-TOBIT model to investigate determinants of geographic (interstate) differentials in the S&L failure rate.

The present study seeks to extend the inquiry into the economic health of the S&L industry by empirically identifying key determinants of the rate of return on S&L assets over the period 1970-97. Clearly, if the rate of return on S&L assets declines sharply and becomes negative, then a prolonged experience of such negative rates of return would ultimately doom the S&Ls to eventual insolvency. In turn, the failing S&Ls may create significant financial burdens for taxpayers as well as for depositors. In identifying key factors systematically determining the rate of return on S&L assets, the study hopes to provide further insights into the factors that influence the economic and financial health of S&Ls so that poor rates of return can perhaps be prevented or at least moderated in the future.

Using semiannual data, the study period runs from the first half of 1970 (1970.1) through the second half of 1997 (1997.2). Unfortunately, quarterly data for several of the variables in the analysis, including the rate of return on S&L assets, the S&L capital to assets ratio, and the S&L cost of funds, are unavailable prior to 1984. In any event, the period 1970.1-1997.2 is an especially interesting one for the S&L industry because the S&Ls experienced a variety of significant challenges over this time span. These include two major oil-price shocks in the 1970s that sharply raised crude oil prices, followed by sharp oil-price declines in the latter part of the 1980s, two major federal statutes that acted to deregulate the financial services industry, major tax reform legislation enacted in 1986, major legislation to reform the financial services industry in 1991, and sharply rising interest rates, especially in late 1979 and the early 1980s, followed by lower interest rates later in the study period.

The Study Period: 1970.1-1997.2

Events involving an O.P.E.C. oil embargo in 1973 created a major oil-price shock that nearly doubled the price per barrel of imported crude oil for the U.S. For example, the average price per barrel of imported crude oil rose from $3.89 in 1973 to $6.87 in 1974. Another oil-price shock occurred in 1979. In this case, the price per barrel of imported crude oil rose from an average of $12.64 in 1979 to $21.59 in 1980 and to $31.77 in 1981 (the peak current-dollar average annual price). After 1981 and 1982, oil prices began to drop somewhat. However, after 1985, the price of imported oil dropped sharply. For instance, the average price per barrel of imported crude oil was $24.05 in 1985, whereas the average price had fallen to $12.51 per barrel by 1986.

Barth [1991], Barth and Bartholomew [1992], and others have argued that although the rising price of oil in the 1970s was beneficial to those parts of the country that engaged in extensive oil exploration and production (particularly the Southwest), the reverse effect was true when oil prices dropped during the 1980s. The declining oil prices resulted in lost jobs, lost income, and a declining real estate market. Potentially, this situation could be particularly problematic for S&Ls given the importance of the real estate industry to their financial viability. Moreover, recession during 1981-82 and in the early 1990s contributed further to these kinds of problems.

Deregulation of the financial service industries effectively began in 1980 under provisions of the Depository Institutions Deregulation and Monetary Control Act (DIDMCA). The deregulation process progressed further under the Garn-St. Germain Depository Institutions Act (GSDIA) of 1982. Under deregulation, as Barth [1990, p. 45] observed, there was, "... greater competition among financial service firms ...." The S&L industry faced increased competition, not only from commercial banks but also from other types of financial service firms. Indeed, Barth [1991, p. 45] observed that the S&L industry, "... experienced a substantially declining share of the assets of all financial service firms as new services and products came into existence and generated expanded opportunities for noninsured and less regulated firms in the 1980s." Barth and Bartholomew [1992, p. 39] observe further that, "... greater overall competition among financial service firms in the 1980s significantly narrowed the net interest margins for thrifts." Barth [1991, p. 45] asserted that, "... this increased competition contributed to the disappearance of many savings and loans ...."

It should also be emphasized that, under the deregulation statutes, the S&Ls were subject to declining capital requirements. For example, the DIDMCA of 1980 eliminated the 5 percent minimum statutory capital requirement and replaced it with a new statutory range of 3 to 6 percent. The Federal Home Loan Bank Board (FHLBB) lowered the capital requirement from 5 to 4 percent as of November 1980. By January 1982, the capital requirement had been lowered by the FHLBB to only 3 percent. Later in 1982, under the GSDIA of 1982, the 3 to 6 percent statutory capital requirement was replaced with a mandate that the FHLBB simply require S&Ls to maintain "adequate capital" [Barth, 1991, pp. 48-9]. The analysis in Taggart and Jennings [1934] would suggest that, in the context of federal government deposit insurance, financial institutions tend to pursue more (less) prudent and responsible business practices when their capital is higher (lower) because they have more (less) owners' capital to protect. Given the declining levels of S&L owners' capital in the 1980s and early 1990s, one would expect a diminution in the prudence and responsibility with which the thrifts conducted business.

Other challenges of the 1980s and 1990s for S&Ls include the Tax Reform Act of 1986, which came to be characterized as real estate unfriendly [Barth, 1990, p. 45; Barth and Bartholomew, 1992, p. 39; Sanger, Sirmans and Turnbull, 1990]. For example, Barth and Bartholomew [1992, p. 39] argued that provisions in the Tax Reform Act of 1986, "... adversely affected real estate values and thereby thrift institutions."

Next, Barth and Bartholomew [1992, p. 39] characterized the sharply rising interest rates, especially in late 1979 and in the early 1980s, as contributing, "... to the thrift crisis of the 1980s." Elaborating further on this issue, Barth and Brumbaugh [1992, xiii] observed,
      "The fixed-rate home mortgage from the savings and loan became an
      important part of Americana. But when interest rates unexpectedly
      soared in the late 1970s and early 1980s, the savings and loan
      house of cards built on lending long and borrowing short
      collapsed."


Given that S&Ls were extensively holding long-term, fixed-rate mortgages funded by shorter-term variable-rate deposits, Barth [1990, p. 38] argued that, "When interest rates skyrocketed in late 1979 and the early 1980s, net (S&L) operating income plummeted."

Since the challenges described above were experienced in varying degrees during the 1970s, 80s, and 90s, this study endeavors to formally identify the key determinants of S&L profits during this period. Indeed, while the study begins with 1970.1, it ends with 1997.2, the most recent period for which all of the data necessary to the study are currently available in semiannual form. By including the period following 1991, the potential impact of the Federal Deposit Insurance Corporation Improvement Act of 1991 (FDICIA) on S&L profitability can be allowed. Under FDICIA, numerous statutory provisions were implemented beginning in 1992, including risk-related deposit insurance premiums and a prompt corrective action mandate. These actions made a significant impact on the performance and operations of the S&L industry. As Madura and Bartunek [1995, p. 191] observed, "The FDIC Improvement Act mandated key provisions that affect the potential performance and risk of financial institutions." Given this potentially profound impact of FDICIA on the S&L industry, a variable to reflect the implementation of FDICIA is included in the system. (3)

The next section provides a rudimentary model of S&L profit-maximizing behavior. The empirical analysis in the subsequent section of the study provides concrete evidence, based on an Instrumental Variables (IV) estimation, as to the identity of key determinants of the rate of return on S&L assets over the 1970.1-1997.2 period. The conclusion provides a summary of the results and suggests certain policy and other implications of same.

A Rudimentary Model of S&L Profit-Maximizing Behavior

The S&L is viewed as a profit-maximizing, price-taking firm. The S&L generates revenues primarily through the issuance of mortgage loans to the public, (4) although, over time, non-mortgage loans, government securities, and certain other activities have increasingly become sources of revenues to S&Ls. Nevertheless, mortgage loans averaged nearly 70 percent of the assets of the thrift industry over the 1970-97 study period at the industry level [OTS, 1989, Table A3; FDIC, 1998]. While this average declined from a high of 85.2 percent in 1970 and fell gradually to a 50 percent range by 1997 as S&Ls diversified, it remained, even in 1997, the principal, single asset category (on average) for the S&Ls. The S&L obtains funds to support its loans principally through the deposit markets, although some funds are obtained from outside borrowings. The S&L's total costs are essentially the sum of its total payments for deposits (which principally include explicit interest costs), net interest payments for outside borrowed funds, and operating costs. Each S&L is constrained so that its total volume of outstanding loans cannot exceed the sum of its excess reserves plus net outside funds borrowed plus net worth. Theoretically, its capital(net worth)-asset ratio may not fall below a certain regulatory value.

Since mortgage loans comprise (on average) the principal form of S&L assets and the single largest revenue source for S&Ls over the study period, mortgage loan income is treated as the identified source of revenue for S&Ls. In addition, since deposits and outside borrowed funds are the predominant source of funds for S&Ls, the focus is on the cost of funds as the main cost consideration for S&Ls, ignoring the other S&L cost categories.

Clearly, S&L revenues depend significantly on the mortgage interest rate charged (rM) and the proportion of S&L mortgage loans that is not performing (PML). For a given value of PML, a higher mortgage rate (rM) implies higher revenues and hence a higher rate of return, ceteris paribus. The factors that influence PML are those that reflect risk dimensions of the S&L mortgage portfolio. There are a number of factors that have been argued in the literature to influence S&L mortgage portfolio risk. Based on Barth [1991], Barth and Bartholomew [1992], Barth and Brumbaugh [1992], Brumbaugh [1988], and Chou and Cebula [1996], PML is expected to be a function of (1) the propensity for S&L directors and officers to pursue low-risk or high-risk lending strategies and (2) the economic viability of the real estate market.

Based in principle on the arguments in Taggart and Jennings [1934], the greater the ratio of S&L net worth-to-S&L assets (RAP/ASSET), the greater the likelihood (given the existence of federal deposit insurance) that the S&L will adopt prudent and responsible lending and other business practices. This is because the higher the (RAP/ASSET) ratio, the greater the incentive for such behavior in order to protect owners' capital. Thus, the higher the (RAP/ASSET) ratio, the greater the likelihood that mortgage loans will perform, and the lower will be PML.

The economic health of the real estate market and S&L assets and revenues have been argued by Barth [1991], Barth and Bartholomew [1992], and others to have been influenced negatively by the Tax Reform Act of 1986, declining crude oil prices during the 1980s, and recession. As Barth [1991, 45] observes, the Tax Reform Act of 1986, "... adversely affected real estate values, thereby weakening the financial conditions of savings and loan institutions." According to Barth [1991, 45], the Tax Reform Act of 1986 (TRA) contributed to the industry's strains because it:
      "... reduced the depreciation benefits from investing in
      commercial and residential property, limited the offsetting losses
      on passive investments that affect limited partnership
      syndications, and eliminated the favorable capital gains
      treatment."


Arguments consistent with these are found in the study by Sanger, Sirmans and Turnbull [1990, p. 421], who conclude that, "For the 1986 Tax Reform Act, ... the market assessed the changes to be detrimental to the owners of real estate assets." Barth and Bartholomew [1992, p. 39] conclude that, "These changes in tax laws adversely affected real estate values and thereby thrift institutions." Arguably, the TRA (1) may have reduced S&L revenues and revenue growth by reducing real estate demand and (2) may have lowered S&L asset values and net worth, inducing further risk-taking behaviors by S&Ls.

Barth [1991, p. 40] also observes that S&Ls experienced a sharp decline in asset quality and revenues that was attributable to "... sharply falling energy prices (especially crude oil and natural gas) in the 1980s ...." This fall in energy prices (ENERGY PR) led to reductions in the exploration and the extraction of domestic crude oil and natural gas which further led to reductions in employment and income, especially in the Southwest. This circumstance led to increased mortgage loan delinquency and to foreclosure rates (which damaged S&L revenues), as well as to a sharp fall in housing prices, resulting from declining housing demand. As a consequence of such effects, Barth [1991], Barth and Bartholomew [1992], Barth and Brumbaugh [1992], and others, argue that S&L revenues as well as S&L assets, were adversely affected.

Furthermore, the financial well-being of the thrift industry has been argued by Barth [1990, p. 40], Cebula and Hung [1992A, p. 307], and others to also have been damaged by periods of declining real GDP growth rates, such as a falling Y. The latter presumably contributed to increased mortgage payment delinquencies and foreclosures. In turn, S&L revenues (as well as asset quality) suffered.

Regarding S&L costs, Barth [1991, p. 38], Barth and Bartholomew [1992, pp. 38-9], and others observe that the cost of funds at S&Ls (rE) rose substantially, especially during late 1979 and the early 1980s. As Barth and Bartholomew [1992, p. 39] state, "The higher rates in the early 1980s drove up deposit costs ...." Moreover, since deposits are overwhelmingly the primary source of funds to S&Ls, it is especially important that the cost of such funds to S&Ls be accounted for. The rate of return on S&L assets should be a decreasing function of rE based on the studies by Barth [1991], Barth and Bartholomew [1992], and Saltz [1995], as well as the microeconomics theory of the firm.

Barth and Brumbaugh [1992, xiii] observe that when interest rates soared in late 1979 and the early 1980s, S&L profitability was compromised. This occurred because "The higher [interest] rates ... drove up deposit costs without a corresponding increase in revenues from mortgage loans," [Barth and Bartholomew, 1992, p. 39]. Thus, it appears that a reasonable way to investigate the implications of changing interest rates on S&L earnings performance is to look at the spread between mortgage rates and deposit rates. (5) Given the presence of both the S&L mortgage interest rate (rM) and the S&L cost of funds (rE) in the model, the net interest income spread in terms of the difference between the mortgage rate and the cost of funds is approximately measured, such as (rM - rE).

Finally, there is the potential impact of FDICIA. As Cebula [1997A, p. 696] observed, "This statute contains numerous provisions aimed at diminishing the bank and S&L failure rates and at diminishing the numbers of problem banks and problem S&Ls ...." Cebula [1997A] discusses a variety of such provisions, which range from risk-related deposit insurance premiums to new real estate lending guidelines to a prompt corrective action mandate for institutions whose capital falls below certain prescribed levels. Given the number, complexity, and impracticality of trying to actually measure such provisions, this study will allow for the FDICIA by using a dummy variable.

Accordingly, based on the arguments expressed above, the rate of return on S&L assets (RET), which is used in this study as the proxy for the S&L profit rate, can be proximately modeled as follows: (6)

RET = f(rM - rE, RAP/ASSET, TRA, ENERGY PR, Y, FDICIA), (1)

where it is hypothesized that:

frM-rE > 0, fRAP/ASSET > 0, fTRA < 0, fENERGY PR > 0, fY > 0, fFDICIA > 0 (2)

Empirical Findings

Based on the model expressed in (1) and (2), the following reduced-form equation is estimated:

RE[T.sub.t] = [a.sub.0] + [a.sub.1](rM - rE[).sub.t] + [a.sub.2](RAP/ASSET[).sub.t-1] + [a.sub.3]TR[A.sub.t] + [a.sub.4]ENERGY P[R.sub.t-1] + [a.sub.5][Y.sub.t-1] + [a.sub.6]FDICI[A.sub.t] + [a.sub.7]TREND + [mu], (3)

where RE[T.sub.t] = the average rate of return on S&L assets in period t, as a percent per annum; RET is average S&L net income after taxes for period t divided by average assets for period t [OTS, 1989, E-6]; [a.sub.0] = constant term; (rM - rE[).sub.t] = the spread between the nominal average mortgage interest rate yield at S&L institutions in period t and the nominal average S&L cost of funds in period t, expressed as a percent per annum; (RAP/ASSET[).sub.t-1] = the ratio of regulatory capital to assets at S&Ls in period t - 1, as a percent; TR[A.sub.t] = a binary (dummy) variable indicating whether the Tax Reform Act of 1986 was in effect in period t; TR[A.sub.t] = 1 for those periods (t) when the Tax Reform Act was in effect and TR[A.sub.t] = 0 otherwise; ENERGYP[R.sub.t-1] = the price per barrel of imported crude oil in period t - 1, expressed in 1987 dollars; [Y.sub.t-1] = the percentage growth rate of real GDP in period t - 1; FDICI[A.sub.t] = a binary (dummy) variable indicating whether provisions of FDICIA were implemented in period t; FDICI[A.sub.t] = 1 for such periods, and FDICI[A.sub.t] = 0 otherwise; TREND = a simple linear trend variable; and [mu] = stochastic error term.

Over the study period, many S&Ls failed. They were either closed outright or forced to merge with other institutions. In addition, numerous mergers occurred on a voluntary basis and they were not based simply on an institution's being insolvent. In any event, there were numerous mergers and acquisitions of S&Ls, especially during the 1980s and 1990s. Potentially, this causes comparable RET data to be imperfect over time. Nevertheless, since the FDIC continues to treat the RET figures as comparable over time [FDIC, 1998, 7], the author follows the lead of the FDIC and uses the RET data that are available.

The variable (rM - rE[).sub.t] is contemporaneous with variable RE[T.sub.t]. Precisely how this situation is addressed is described below. Given the impracticality of attempting to create a variable that quantifies all the basic provisions of either the Tax Reform Act of 1986 or the Federal Deposit Insurance Corporation Improvement Act of 1991, the latter are each represented by a binary dummy variable, TR[A.sub.t] and FDICI[A.sub.t], respectively. Focusing on regulatory capital reflects provisions in the deregulation statutes of 1980 and 1982. The real price per barrel of imported crude oil, lagged one period, is used to measure energy prices. To measure the overall health of the economy and identify rapid growth and slow growth periods, the one-period lag of the percentage growth rate of real GNP ([Y.sub.t-1]) is adopted.

Since the variables RE[T.sub.t] and (rM - rE[).sub.t] are contemporaneous, the possibility of simultaneity bias exists. Accordingly, equation (3) is estimated using an Instrumental Variables (IV) technique, with the instrument being the two-period lag of the actual inflation rate of the consumer price index ([P.sub.t-2]). While variable [P.sub.t-2] was found to correlate highly with (rM - rE[).sub.t], it was not the same with the error terms in the system. The choice of the instrument is actually suggested in Barth [1991, p. 38] when he speaks of factors contributing to the rising interest rates of late 1979 and the early 1980s as including inflationary forces.

The Augmented Dickey-Fuller and Phillips-Perron tests both reveal that the variables RET (with a trend), Y, and P are all stationary in levels for the 1970.1-1997.2 study period. However, the variables (rM - rE), ENERGYPR, and (RAP/ASSET) were found to be stationary only in first differences for the period. (7) Therefore, in the estimate provided below, those variables stationary in levels are expressed appropriately in levels, whereas those stationary solely in first differences are expressed in first-differences form. The data sources include: Office of Thrift Supervision [1989, Tables A-3, A-16; 1990; 1991; 1998]; the Federal Home Loan Bank Board [1990; 1991]; Barth [1990; 1991]; the FDIC [1993; 1994; 1995; 1996; 1997; 1998]; the Council of Economic Advisors [1996; Table B-2; 1999, Tables B-2, B-64]; Cebula and Hung [1992B, Tables 3.7, 3.9]; and the U.S. Department of Commerce [1997, Table 933].

The results of the IV estimation of equation (3) for 1970.1-1997.2 are provided in column (1) of Table 1, where terms in parentheses are t-values. In Table 1, ignoring the TREND variable, five of the six estimated coefficients (those for the (rM - rE), (RAP/ASSET), TRA, Y, and FDICIA variables) exhibit the expected signs and are statistically significant at the 5 percent level or beyond. The sixth of these estimated coefficients (that for the variable ENERGYPR) does not exhibit the expected sign and is not significant even at the 10 percent level. In addition, the F-statistic is significant at the 1 percent level.

In column (2) of Table 1, a sensitivity measure is included, which provides a gauge of the economic significance of the explanatory variables. Ceteris paribus, for each explanatory variable, the figure in column (2) reports the change in the rate of return on S&L assets (RET) for a one standard deviation change in the explanatory variable. Based on this sensitivity measure, some significant impacts on the rate of return on S&L assets are exercised by FDICIA, the net interest rate spread (rM-rE), and the Tax Reform Act of 1986, respectively. The S&L capital-asset ratio and the real GDP growth rate impacts are somewhat more modest in magnitude, with the ENERGPR variable shown to exercise a negligible influence.

Based on the findings provided in Table 1, for the 1970.1-1997.2 period, it appears that the rate of return on S&L assets is an increasing function of the spread between the S&L mortgage rate and the S&L cost of funds. Thus, the greater this spread, the higher is the rate of return on S&L assets. It also appears that the higher the regulatory capital-asset ratio, the higher the rate of return on S&L assets. It may well be that when such capital-asset ratios are higher, more prudent business practices are pursued by S&Ls, practices that act to protect S&L rates of return as well as owners' capital. The Tax Reform Act of 1986 has been linked with a reduced rate of return on S&L assets. This negative impact on the rate of return on S&L assets presumably reflects the adverse impact of this tax legislation on the real estate market, which has been important to the economic health of the S&Ls [Barth, 1991; Barth and Bartholomew, 1992; Sanger, Sirmans and Turnbull, 1990]. It can also be seen that, for the overall 1970.1-1997.2 study period, the rate of return on S&L assets has not been significantly influenced by energy prices. This appears to be inconsistent with arguments by Barth [1991] and Barth and Bartholomew [1992]. Since the latter two studies focus on the 1980s and do not deal with the 1990s, their arguments might well be valid for the 1980s per se, although apparently not applicable to the longer period considered in this study.

Next, the findings in Table 1 reveal that the rate of return on S&L assets is an increasing function of the percentage growth rate of real GDP. This is consistent with arguments found in studies dealing with earlier periods, including Barth [1991] and Cebula and Hung [1992B]. In addition, the estimated coefficient on the variable used to reflect the effects of implementing provisions of the Federal Deposit Insurance Corporation Improvement Act of 1991 is statistically significant at the 1 percent level. Thus, it appears that the various provisions of this statute have acted jointly to increase the rate of return on S&L assets, which would be consistent with the objectives of this statute [Benston and Kaufman, 1997; Cebula, 1997A]. This finding is also consistent with the claim by Benston and Kaufman [1997, 140] that because of the various FDICIA provisions, "The thrift industry ... rebounded."

Finally, Table 2 provides a description of variables in the analysis. In particular, Table 2 provides the mean, standard deviation, and range of values for RET and each of the explanatory variables included in the IV estimation for the 1970.1-1997.2 study period.

Concluding Observations

Using semiannual data, this study empirically investigates determinants of the rate of return on S&L assets in the U.S. using an IV estimation for the 1970.1-1997.2 time period. The results revealed that the rate of return on S&L assets was an increasing function of the spread between the S&L mortgage interest rate and the S&L cost of funds, the regulatory S&L capital-asset ratio, and the percentage growth rate of real GDP. It was (1) negatively affected by the Tax Reform Act of 1986, which has been commonly characterized as unfriendly to the real estate markets, and (2) positively affected by the Federal Deposit Insurance Corporation Improvement Act of 1991, which was intended in part to improve the health of the S&L industry.

Clearly, in the context of federal deposit insurance, the S&L's capital-asset ratio is a regulatory tool that can be used to help protect, if not elevate, the rate of return on S&L assets. By statutorily requiring an adequately high and reasonably stringent capital-asset ratio (which is based on market-value accounting (8) and which omits goodwill from the calculation of the capital requirement), S&L activities are likely to be more prudent and responsible since S&L decision-makers will have incentives to protect the owners' capital. Interestingly, the passage and implementation of FDICIA in 1991 imposed such reforms on the financial services industry. This reflected in part by the fact that, in 1997, the FDIC found 98.8 percent of all operating S&Ls to be either well capitalized or at least adequately capitalized [FDIC, 1998, p. 17]. In fact, while S&L industry earnings rose to a record level in 1997 ($8.8 billion), the industry's average "... capital to assets ratio at the end of 1997 rose to 8.71 percent ... the highest level since 1943" [FDIC, 1998, p. 9].

In addition, it would seem that monetary and fiscal policies that limit inflation and thereby limit upward pressure on the S&L cost of funds also benefit the rate of return on S&L assets. Indeed, since 1992, the spread between the S&L mortgage rate (as well as other S&L lending rates) and the S&L cost of funds has become and remains quite favorable to the S&L industry. The return on assets for S&Ls in 1997 (0.93 percent) was the highest it had been since 1946 [FDIC, 1998, p. 8]. Furthermore, as one might expect, policies that improve the health and stability of real estate markets per se should also benefit S&Ls.

The future of the S&L industry remains uncertain despite factors such as the favorable spread that continues between the S&L mortgage rate (as well as other S&L lending rates) and the S&L cost of funds; the expansion of the U.S. economy; the greatly improved capital to asset ratios for most of the industry; and the continued diligent application of FDICIA. Part of the uncertainty is whether the S&L institutions will prove to be a viable industry in the long run. S&Ls are limited in the degree to which they can compete with much larger, more diversified, and generally more efficient global financial institutions. Thus, the recent success experienced by the S&Ls may contribute to their disappearance because of continued and increased mergers and acquisitions. For example, during 1997 alone, the commercial banking industry absorbed 116 S&Ls with assets in excess of $75 billion. As the FDIC [1998, p. 8] observed, "This is the largest number of institutions and the largest amount of assets ever transferred in a single year between the two industries." While circumstances such as those identified in this study may be favorable for the S&L industry's continued prosperity for the foreseeable future, they may serve ultimately only to determine the manner in which the S&Ls eventually become extinct. Rather than failing, the S&Ls may be absorbed by larger banking institutions because of their success.
TABLE 1

IV Estimates

Variable\Column (a) (b)           (1)            (2)
                               70.1-97.2   Sensitivity (c)

Constant                       +0.713

[delta](rM - rE[).sub.t]       +0.444 **   +0.19655
                              (+2.37)

[delta](RAP/ASSET[).sub.t-1]   +0.26 *     +0.11866
                              (+2.29)

TR[A.sub.t]                    -0.54 **    -0.2681
                              (-2.63)

[delta]ENERGYP[R.sub.t-1]      -0.003      -0.008
                              (-0.18)

[Y.sub.t-1]                    +0.05 **    +0.1112
                              (+2.35)

FDICI[A.sub.t]                 +1.568 ***  +0.64921
                              (+7.67)

TREND                          -0.027 ***
                              (-5.20)

F-Statistic                    14.45 ***

(a.) Terms in parentheses are t-values, which have been corrected for
heteroskedasticity, using the White [1980] procedure.
(b.) [delta] is the first-differences operator.
(c.) The sensitivity measure indicates the change in the average rate of
return on S&L assets (RET) for a one standard deviation change in the
explanatory variable.
*** Indicates statistically significant at a 1.0 percent level.
** Indicates statistically significant at a 2.5 percent level.
* Indicates statistically significant at a 5.0 percent level.

TABLE 2

Summary Statistics

     Variable        Mean    Standard Deviation  Minimum  Maximum

RET                 0.21875       0.67477         -1.55   0.93
[delta](rM - rE)    0.04464       0.44268         -1.07   1.56
[delta](RAP/ASSET)  0.02909       0.45637         -1.22   1.18
TRA                 0.41071       0.49642          0.0    1.0
[delta]ENERGY PR    0.03218       2.66589        -10.74   9.11
Y                   2.77321       2.22404         -2.1    5.8
FDICIA              0.21429       0.41404          0.0    1.0


* Armstrong Atlantic State University--U.S.A.

Footnotes

(1) A failed S&L is one that was closed outright or forced to merge with another institution. Voluntary mergers are not failures.

(2) See also the well-done study by Loucks [1994].

(3) Regarding the impact of FDICIA on bank failures, see Cebula [1997A].

(4) It should be noted that a number of the better operated S&Ls made more concerted efforts to move out of the mortgage business. However, separate profit data at the aggregate level for these S&Ls are not readily available.

(5) Along these same lines, it has been alleged by the FDIC [1994, p. 6] that in the early 1990s, "Higher [financial institution] earnings ... benefitted from rising net interest income."

(6) Unlike Saltz [1994], who focuses on bank failures, the author does not consider the inclusion of a variable to reflect Regulation Q. This is because this variable is statistically insignificant over a variety of specifications. Moreover, this statistical insignificance may reflect the fact that while Regulation Q may have at times protected S&Ls from rising deposit costs, it very likely contributed to disintermediation. Regulation Q was phased out under provisions of the Depository Institutions Deregulation and the Monetary Control Act of 1980.

(7) For the 1965-91 period, the variables RET and ENERGY PR have been found to be non-stationary [Cebula, 1998] or even cointegrated [Cebula, 1997B].

(8) Barth and Brumbaugh [1992, xvi] state that empirical evidence suggests that "... even very generic forms of market valuation are superior to the traditional accounting principles" for evaluating S&L capital and asset ratios.

References

Amos, O. M. "The Regional Distribution of Bank Closings in the United States from 1982 to 1988," Southern Economic Journal, pp. 805-15, January, 1992.

Barth, J. R. The Great Savings and Loan Debacle, Washington, D.C.: AEI, 1991.

--. Statement Before the House Committee on Banking, Finance, and Urban Affairs, 101st Congress, April 11, 1990.

Barth, J. R. and Bartholomew, P. "The Thrift Industry Crisis: Revealed Weaknesses in the Federal Deposit Insurance System," Their Chapter in a Cooperative Work, in The Reform of Federal Deposit Insurance, edited by J. R. Barth and R. D. Brumbaugh, New York: HarperBusiness, pp. 36-116, 1992.

Barth, J. R. and Brumbaugh, R. D., eds. The Reform of Federal Deposit Insurance, New York: HarperBusiness, 1992.

Benston, G. J. and Kaufman, G. G. "FDICIA After Five Years," Journal of Economic Perspectives, pp. 139-58, Summer, 1997.

Brumbaugh, D. B. Thrifts Under Siege, Cambridge, MA: Ballinger Publishing Co., 1988.

Cebula, R. J. "The Regional Distribution of Bank Failures in the United States: An Extension of the Amos Analysis," Southern Economic Journal, pp. 202-8, July, 1994.

--. "An Exploratory Empirical Analysis of the Impact of the Federal Deposit Insurance Corporation Improvement Act of 1991 on Bank Failures in the United States," Applied Financial Economics, pp. 695-702, December, 1997A.

--. "The Rate of Return on Savings and Loan Assets: Evidence from Cointegration, 1965-1991," Studies in Economics and Finance, pp. 3-24, Spring, 1997B.

--. "Thrift Institution Profit Rates, 1965-1991," Applied Economics Letters, pp. 635-8, October, 1998.

Cebula, R. J. and Hung, C. S. "Barth's Analysis of the Savings and Loan Debacle," Southern Economic Journal, pp. 305-9, July, 1992A.

--. The Savings and Loan Crisis. Dubuque, Iowa: Kendall/Hunt, 1992B.

Chou, R. Y. and Cebula, R. J. "Determinants of Geographic Differentials in the Savings and Loan Failure Rate," Journal of Financial Services Research, pp. 5-26, March, 1996.

Council of Economic Advisors. Economic Report of the President, 1996, Washington, D.C.: U.S. Government Printing Office, 1996.

--. Economic Report of the President, 1999, Washington, D.C.: U.S. Government Printing Office, 1999.

Federal Deposit Insurance Corporation. 1992 Annual Report, Washington, D.C.: FDIC, 1993.

--. 1993 Annual Report, Washington, D.C.: FDIC, 1994.

--. 1994 Annual Report, Washington, D.C.: FDIC, 1995

--. 1995 Annual Report, Washington, D.C.: FDIC, 1996.

--. 1996 Annual Report, Washington, D.C.: FDIC, 1997.

--. 1997 Annual Report, Washington, D.C.: FDIC. 1998.

Federal Home Loan Bank Board, Mortgage Interest Rate Survey, Washington, D.C.: FHLBB, September, 1990.

--. Mortgage Interest Rate Survey. Washington, D.C.: FHLBB, December, 1990.

Kane, E. J. "S&L's and Interest-Rate Reregulation: The FSLIC as an In-Place Bailout System," Housing Finance Review, pp. 219-43, May, 1982.

Loucks, C. "The Regional Distribution of Bank Closings in the United States from 1982 to 1988: A Brief Note," Southern Economic Journal, pp. 191-4, July, 1994.

Madura, J. and Bartunek, K. "Valuation Effects of the FDIC Improvement Act," Applied Financial Economics, pp. 191-8, October, 1995.

Office of Thrift Supervision. Savings and Home Financing Source Book, Washington, D.C.: U.S. Government Printing Office, 1989.

--. News. Washington, D.C.: U.S. Government Printing Office, October 15, 1990.

--. News. Washington, D.C.: U.S. Government Printing Office, December 15, 1991.

--. News. Washington, D.C.: U.S. Government Printing Office, December, 15, 1998.

Saltz, I. "Federal Deposit Insurance Coverage and Bank Failures: A Cointegration Analysis with SemiAnnual Data, 1963-1991," Journal of Economics and Finance, pp. 3-10, Spring, 1997.

--. "The Impact of Federal Deposit Insurance on Savings and Loan Failures," Southern Economic Journal, pp. 253-5, July, 1995.

--. "On Regulation Q and Bank Closings in the U.S.," International Review of Economics and Business, pp. 185-91, February, 1994.

Sanger, G. C., Sirmans, C. F., and Turnbull, G. K. "The Effects of Tax Reform on Real Estate: Some Empirical Estimates," Land Economics, pp. 409-24, November, 1990.

Taggart, G. H. and Jennings, L. D. "The Insurance of Bank Deposits," Quarterly Journal of Economics, pp. 508-16, May, 1934.

U.S. Department of Commerce. Statistical Abstract of the United States, 1997, Washington, D.C.: U.S. Government Printing Office, 1997.

White, H. "A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity," Econometrica, pp. 817-38, May, 1980.

RICHARD J. CEBULA *
COPYRIGHT 2004 Atlantic Economic Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2004 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Cebula, Richard J.
Publication:International Advances in Economic Research
Geographic Code:1USA
Date:Feb 1, 2004
Words:6705
Previous Article:Cost structure, technological change, and productivity growth in the Greek banking sector.
Next Article:Ownership structure, corporate governance, and enterprise performance: empirical results for Ukraine.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters