The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
™?|Econoniic lisaiS Review FEDERAL RESERVE BANK O F ATLANTA MARCH/APRIL 1991 Supply Shocks and Household Demand for Motor Fuel Beyond Plain Vanilla: A Taxonomy of Swaps FYI: The Impact of Private-Sector Defense Cuts on Regions of the United States Book Review The Interstate Banking Revolution: Benefits, Risks, and Tradeoffs for Bankers and Consumers by Peter S. Rose Economic Review President Robert P. Forrestal Senior Vice President and Director of Research Sheila L. Tschinkel Vice President and Associate Director of Research B. Frank King Vice Presidents William Curt Hunter, Basic Research Mary Susan Rosenbaum, Macropolicy Research Officers William Roberds, Macropolicy Gene D. Sullivan, Regional Larry D. Wall, Financial David D. Whitehead, Regional Public Affairs Officer Bobbie H. McCrackin Publications Joycelyn T. Woolfolk, Publications Coordinator I.ynn H. Foley, Production Carole Starkey, Graphics Ellen Arth, Circulation The Economic Review seeks to inform the public about Federal Reserve policies and the e c o n o m i c environment and, in particular, to narrow the g a p between specialists and concerned laypersons Views expressed in the Economic Review are not n e c e s s a r i l y t h o s e of t h i s Bank or of the Federal Reserve System. Materia] may be reprinted or a b s t r a c t e d if the Review and author are credited. Please provide the Bank's Public Affairs Department with a copy of any publication containing reprinted material Free subscriptions and limited additional copies are a v a i l a b l e from the Public Affairs Department, Federal Reserve Bank of Atlanta, 104 Marietta Street, N.W., Atlanta, Georgia 30303-2713 (404/521-8020) Change-of-address notices and subscription canc e l l a t i o n s s h o u l d be s e n t d i r e c t l y t o t h e P u b l i c Affairs Department. Please include the current mailing label as well as any new information. ISSN 0732-1813 VOLUME 76, NO. 2, MARCH/APRIL 1991, ECONOMIC REVIEW Supply Shocks and Household Demand for Motor Fuel W i l l i a m C. H u n t e r a n d Mary S. R o s e n b a u m \2 Beyond Plain Vanilla: A Taxonomy of Swaps Research into the response of U.S. household demand for motor fuel to oil price shocks during the 1970-89 period lends perspective to the recent oil shock. The author surveys numerous varieties of interest rate, currency, commodity, and equity swaps. Peter A. A b k e n 30 R Y L David D. W h i t e h e a d 42 Book Review A r u n a Srinivasan FEDERAL RESERVE BANK O F ATLANTA The Impact of Private-Sector Defense Cuts on Regions of the United States The Interstate Banking Revolution: Benefits, Risks, and Tradeoffs for Bankers and Consumers by Peter S. Rose Supply Shocks and Household Demand for Motor Fuel William C. Hunter and Mary S. Rosenbaum Many economists believe that oil price shocks to the U.S. economy cause changes in demand for oil that persist long after the shocks'first effects. Casual observation suggests that consumption of motor fuel during the past two decades has not been very sensitive to motor fuel pnces. The authors' study of U.S. household demand for motor fuel from 1970 through 1989 supports an asymmetric response hypothesis, which holds that decreased demand for motorfuel immediately following an increase in its price will exceed the increased demand following a comparable price decrease. They find no support for an explanation that links household demand to conservation policies. T he unexpected reduction in the world supply of crude o i l — a n d subsequent increase in its price—following the August 2, 1990, Iraqi invasion of Kuwait is an example of what economists call a supply shock. The term is used to refer to a situation in which the supply of a key input into the economy's aggregate p r o d u c t i o n f u n c t i o n is u n e x p e c t e d l y c h a n g e d (interrupted or s u p p l e m e n t e d ) or in which the price of a key input changes substantially and quickly. Such shocks can affect the economy in numerous ways, leading to a number of potential economywide adjustments. The scope of these adjustments will depend on the nature and size of the shock and on the economy's current structure. Stated differently, a supply shock is a change in the economy's ability to produce a given level of output at a given price. A favorable supply shock, such as an increase in labor productivity or a technological breakthrough, raises the level of output produced by a fixed amount of input— that is, for any given price, more can be prod u c e d . A n adverse s u p p l y s h o c k , such as an increase in oil prices, lowers the output potential for existing resources. Capital, because it becomes more expensive to operate, and labor, because each unit of labor has, in essence, less capital with which to produce output, are both less productive. At any price, less can be produced with given inputs. For example, if a factory's fuel bill that had been $100 per unit of output increases to $125 per unit of output because of the rise in oil prices, the production ability of both the existing capital (plant and equipment) and labor are reduced. FEDERAL RESERVE BANK OF ATLANTA Adverse supply shocks like the infamous oil and food price shocks of the 1970s can thus be thought of as destroying some of the economy's existing capital stock or as reducing labor productivity. Either outcome leads to decreases in equilibrium real wages. The quadrupling of oil prices during the supply shock o f 1974 (see Chart 1) provides a good illustration of these points. In response to the oil shock, m a n y firms scrapped m u c h of their energy-inefficient plant and equipment. Supported by a smaller capital stock with which to work, labor productivity naturally fell in the short run. Economists have long speculated that certain shocks, especially adverse supply shocks, can lead to changes in behavioral relationships that persist long after a shock's initial impulse has passed. The U.S. economy has experienced several such shocks in the last twenty years, and these episodes provide data that can help determine how households and businesses respond. Some understanding of ways the economy has adjusted in the past may lend perspective on the recent oil shock and the likely efficacy of policies designed to offset its impact. An examination of U.S. household demand for motor fuel over the 1970-89 period, presented in this article, sheds some light on these issues of different sectoral responses to oil price shocks. The authors are, respectively, vice president and senior financial economist responsible for basic research and vice president and senior economist in charge of the macropolicy section of the Atlanta Fed's research department. They thank Mary Beth Walker for helpful comments. Excellent research assistance was provided by Jeff Watson and Danny Moses. Mark Rogers derived and calculated the motorfuel price index. 1 Chart 1 Oil Prices, 1968-90* 40 * Annual average crude oil refiners' acquisition cost, composite. Source: U.S. Department of Energy. Sensitivity of Oil Consumption to Oil Prices The ease or difficulty with which the economy responds to a supply shock depends on the nature of the commodity whose price changes. If there are few acceptable substitutes for the commodity, as for oil, then households and businesses find it difficult to switch to an alternative in the short run. In this circumstance, demand does not appear to be price sensitive; it does not have much price elasticity (to use economists' term for describing and measuring the sensitivity of demand to price changes). In the longer run, as substitutes for the commodity are developed or discovered, demand becomes more sensitive to price changes. In other words, the short-run price elasticity of demand is less than the longer-run price elasticity. The oil price shocks of the 1970s were large and disruptive. The first shock, between 1973 and 1974, sent oil prices from about $3-60 per barrel (1972 average) to more than $10 per barrel in 1974.1 Moreover, this episode included an embargo limiting U.S. 2 gasoline supplies and resulting temporarily in informal rationing and other nonprice restraints on gasoline purchases. The second shock, a series of price increases during 1979 and 1980, raised the per barrel cost from $12.46 in 1978 to more than $35.00 in 1981. Despite much early empirical evidence to the contrary, recent history lends casual support to the notion that U.S. oil consumption has not been very sensitive to oil prices, especially since 1975.2 As Stephen P.A. Brown and Keith R. Phillips (1989a) have noted, U.S. c o n s u m p t i o n of oil increased sharply between 1975 and 1979, despite the quadrupling of oil prices following the 1974 oil shock. Furthermore, although both oil prices and consumption fell from 1981 through 1985, consumption increased only modestly after prices plunged in 1986. Each of these observations is consistent with the assumption that the short-run price elasticity of demand for oil is quite low. Brown and Phillips observed that at least three theories compete to explain the apparent insensitivity of oil consumption to oil prices. The "nonprice conservation" theory is based on the notion that ECONOMIC REVIEW, MARCH/APRIL 1991 changes in U.S. government policies reduced the growth in demand for oil independently of changes in the price of oil. For example, the Energy Policy and Conservation Act of 1975 and the National Energy Act of 1977, which mandated a number of conservation measures, imposed m i n i m u m average gas-mileage standards for U.S. auto manufacturers and a 55-mile-per-hour national speed limit, as well as improvements in technology. A second theory is that U.S. oil consumption responds asymmetrically to changes in oil prices, falling more following a price rise than it rises after an oil price decline. Neither the nonprice conservation or asymmetric response theory stood up to empirical testing at the aggregate ( e c o n o m y w i d e ) level by Brown and Phillips (1989a). Using a quarterly econometric model of U.S. oil consumption applied to the period covering the first quarter of 1972 through the first quarter of 1988, the authors rejected both theories. A third theoretical explanation of the observed insensitivity of U.S. oil consumption to changes in oil prices is based on the notion that oil consumption patterns are persistent. According to this perspective, because substantial changes in the economywide ratio of oil consumption to output require new capital investment, consumption responds very slowly to oil price changes, giving the possibly false impression that oil consumption is not very sensitive to price. This observation is simply another way of saying that the long-run price elasticity of demand for oil is more than the short-run elasticity. Empirical support demonstrating the lag in the responsiveness of oil consumption to price changes appears in studies by W.W. Hogan (1989), Dermot Gately and Peter Rappoport (1988), and Brown and Phillips (1984). Another explanation, consistent with the consumption persistence theory, is that the household and business sectors of the U.S. economy have different sensitivities to oil shocks and different time patterns of adjustment to changes in oil prices. For example, businesses, especially manufacturing and transportation, may face very high fixed costs of converting to alternative energy sources, or they may have extremely limited opportunities to conserve. For such producers to decrease oil dependency may require investment in more fuel-efficient plant and equipment, so reducing demand might be difficult in the short run. At the same time, households may have more discretion in their conservation o p t i o n s because conservation at the household level involves trading off comfort (heat, air conditioning, private car use) for cost. Thus, the sensitivity of household motor fuel consumption to changes in fuel prices may differ significantly from that of aggregate or economywide oil consumption FEDERAL RESERVE BANK OF ATLANTA to changes in oil prices. Given these considerations, it is possible that broadly defined energy conservation policies (not targeted to business or household behavior) will fail to elicit the expected response. The specific question investigated is whether the adverse oil shocks of 1974 and 1979 led to significant changes in household motor fuel consumption patterns. Providing answers to this question as well as empirical evidence on the nonprice conservation and asymmetric response theories at the level of the household should prove valuable in any attempt to assess the impact of the 1990 oil shock on household demand for gasoline during the next several years. The same evidence could be useful in the design of energy conservation programs and contribute to the formation of future tax (fiscal) policy. The research published here is a first step toward understanding the various ways sectors respond to oil price shocks. In the study a simple quarterly econometric model of household motor fuel demand has been used to conduct tests of the nonprice c o n s e r v a t i o n a n d asymmetric response theories for household motor fuel consumption. The analysis was carried out using the linear expenditure system derived by Richard Stone (1954). A brief intuitive explanation of this approach to demand analysis appears in the next section. Readers interested in a rigorous derivation of the linear expenditure system from first principles are referred to the box on page 4. Household Demand: The Linear Expenditure System Defined The demand function of an individual or household for a product is simply a schedule of the amounts of the product the individual or household would buy at various possible prices in a particular period for a given level of income. Thus, the concept of a demand function describes a functional relationship between two variables: the price of a product and the quantity of the product demanded, holding income constant. However, it is generally true that the demand for any given product will also depend on a host of considerations or influences besides price. Depending on the issues being examined, various of these other influences—which include the prices of other goods or services (in particular, possible substitutes), the household's income and tastes, and expected future prices—should be taken into account explicitly in the household's demand 3 Deriving the Linear Expenditure System Utility Maximization and Demand Functions The classical consumer or household demand model is based on the assumption that the household chooses to consume that set of goods which yields the household the maximum utility or satisfaction. This choice is constrained by the size of the household's income. The total utility the household derives from the consumption of a set of goods can be represented by a utility function. In general notation, the household's utility function can be written as U- Utqv q2,..., qn\ second-order conditions for a maximum—that is, that the matrix of second-order partial derivatives (the Hessian matrix) is negative definite and symmetric—are assumed to hold. The solution of the n equations in relationship (B5) and the budget constraint in equation (B3) give the system of household demand equations. Depending on the exact functional form of the household's utility function, the demand equations may be written explicitly as qr q^pv . . . , pn, EX (B6) (Bl) The Linear Expenditure System where U is household utility and qv . . . , qn are the quantities of the n different (exhaustive) goods consumed by the household. It is assumed that the household's utility function is twice differentiable with respect to each of the goods contained in the function. In other words, the changes in utility associated with changes in the amounts of each good consumed can be computed. It is further assumed that marginal utility of each good (the extra utility or satisfaction derived from one additional unit of any one of the goods holding constant the levels of all other goods) is positive. The household's budget constraint can be written as n S M=E> i= 1 (B2) where the p(s are prices, the qfs are quantities, and E represents total household expenditures or income. For ease of presentation, boldface type denotes vectors, the entire array of prices and quantities. Using this notation equation (B2) can be rewritten compactly as p'q = E, (B4) where X is the Lagrange multiplier. The first-order conditions for the maximization of equation (B4) are dU/dq = \p, (B5) along with the budget constraint given in equation (B3)- This solution is a global maximum because the 4 n U - X ^ r t f ^ i— 1 (B7) where qi > qt° and q? is the minimum amount or subsistence level of the ith good that is consumed.1 If this particular functional form of the household's utility function is substituted into the Lagrangian equation (B4), the first-order conditions for utility maximization are n ft X (<7, - qpPj Vtq{ - qp = xp,., J- 1 (B8) which yield demand curves (expenditure functions) of the following form: (B3) where the prime denotes vector transposition. Using the differentiable calculus, the household's utility maximization problem can be expressed in terms of the Lagrangian function U(q)-Kp'q-E\ The Linear Expenditure System was first proposed and estimated by Richard Stone (1954) and represents the first formal treatment of demand analysis using a specific utility function. That utility function is n M r e r Pi tf+J^O?- 2 Pjip* j- i i= 1,..., n. CB9) The demand or linear expenditure system given in equation (B9) is estimated subject to the technical restriction that the /6?s be normalized so that they sum to unity and that 0 < /3(- < 1 for each of the equations. The jSj-s represent the marginal budget shares of the n goods. Because the demand equations have been derived from a specific utility function, they automatically satisfy the restrictions imposed by neoclassical demand theory.2 The minimum or subsistence quantities of the goods in equation (B9) are in general unknown and must be estimated. If 7}i is defined as the subsistence ECONOMIC REVIEW, »MARCH/APRIL 1991 level of the ith good consumed by the household, the equation system (B9) can be written in a form amenable to empirical estimation. The estimable system is written as i = ViPi + P, (E-2 ViPj) + uit i= 1,. . •, » , (BIO) J" 1 where the u.s are disturbance terms. e Notes 1. The linear expenditure system can also be derived from a host of other more complicated utility functions, all with their own particular advantages. For examples of these approaches see Klein and Rubin (1947-48), Frisch (1959), Houthakker (I960), Parks (1969), Christensen, Jorgenson, and Lau (1975), and Berndt, Darrough, and Diewert (1977). function. Household demand for motor fuel certainly seems based on factors other than price. In the analysis that follows, household demand for motor fuel is modeled as a function of the price of motor fuel, household income, and the prices of other goods and services. Although simple in form, this demand function is in fact very general and is consistent with the standard consumer utility maximization framework of contemporary microeconomic theory. Household demand functions can be easily transformed into expenditure functions, which represent total household expenditure on a given good as determined by the price of the good in question, household income, and the prices of all other goods that elicit household consumption expenditures. Because these expenditure functions are simple transformations of household demand functions, they share all of the theoretical properties underlying the demand functions derivable from budget-constrained household utility maximization (see the box). 3 The form of the household motor fuel demand or expenditure function used in the empirical analysis reported here can be written as n e m = -nmPm + VP + U m' (1) 2. These conditions include Engel and Cournot aggregation, homogeneity of degree zero in prices and total expenditure, and a symmetric substitution matrix. The restriction that the /3( be bounded by zero and unity means that the substitution matrix will also be negative semidefinite as required by standard demand theory. reflects any errors associated with fitting the function to actual data. Expenditure or demand functions of this type, which have been widely used, have an intuitive interpretation. In terms of total expenditures, this representation assumes that the h o u s e h o l d first purchases a minimum required (that is, nondiscretionary) quantity o f each commodity, then distributes its remaining income over all commodities in fixed proportions. Thus, the household's total income can be decomposed into a subsistence component (the income spent on the minimum required quantities of the commodities) and a supernumerary component (the remaining income spent on all of the commodities). Specialized to the case of motor fuel in equation (1) above, the household's minimum required purchase (subsistence quantity) of motor fuel is represented by the parameter r]m, which is estimated statistically. At the current market price this subsistence quantity of motor fuel costs the household the amount pmVm• The proportion of its supernumerary income (total income minus the total cost of the minimum required amounts of the other goods) spent on motor fuel is given by parameter f3m. This parameter is estimated statistically along with the 17j parameters appearing in the expenditure function. 7-1 (See the box.) As stated earlier, in this formulation the household's total expenditure on motor fuel in a given period, em, is expressed as a function of the price of motor fuel, pm\ the household's total income, E\ and the prices of all other goods consumed by the household, the p j terms. The terms denoted by 17 m, f3m, and -q- are parameters to be estimated statistically. The term denoted by um is a r a n d o m d i s t u r b a n c e term that FEDERAL RESERVE BANK OF ATLANTA Properties of the Demand Function Examining the sensitivity of motor fuel demand to changes in household income and the price of motor fuel gives insight into the behavior of household motor fuel consumption patterns over time. The sensitivity of motor fuel demand to changes in household income is readily obtained by computing 5 the household's income elasticity-of-demand index for motor fuel. This index measures the relationship between a relative change in income and the consequent relative change in motor fuel demand, other things being constant. The income-elasticity index, which will be denoted by r m , can be computed using the formula In this formula, w m is the proportion of the household's total income spent on motor fuel—that is, w m = pmqm/E, where qm is the total quantity of motor fuel purchased, and (3m is the proportionality parameter discussed above. Economic theory requires that the parameter /3m be nonnegative (0 < /3m < 1). Thus, the income-elasticity index for motor fuel is positive, implying that motor fuel is a normal good. Normal goods are those for which demand increases as income increases. Hence, household demand for motor fuel should increase as household income increases, other things being held constant. Own-price elasticity of demand is generally defined as the response of quantity demanded to a given percentage change in the price of a particular good. The own-price elasticity-of-demand index for motor fuel, a m , can be computed using the following formula: am--l+a-pmXv JqJ- C3) The own-price elasticity-of-demand index measures the proportional magnitude that quantity demanded changes as price changes. With (0 < (3 m < 1) and (qm - r] m > 0), it follows that the own-price elasticity index is negative, implying that an increase in the price of motor fuel reduces the quantity demanded. Household Demand for Motor Fuel, 1970-89 The empirical analysis discussed here was carried out by estimating the demand relationship given in equation (1) using quarterly data on U.S. household motor fuel expenditures during the period 1970-89Total quarterly household expenditures on motor fuel, em, were taken from national income and product accounts as a component of personal consumption as estimated by the Bureau of Economic Analysis (BEA) of the U.S. Department of Commerce. The price of motor fuel, pm, is the consumer price index (CPI) component for the series as estimated by the Bureau of Labor Statistics (BLS), U.S. Department of 6 Labor. Total disposable household income, E, is defined as personal income in the United States (less taxes) and is taken from the national income and product accounts as estimated by the BEA. The prices of all other goods consumed by the household are measured using the CPI minus the motor fuel component. This index—based on the overall CPI and the motor fuel component—is derived using the methodology explained in Chester V. McKenzie (1961). The index was calculated for each year separately to constrain changes in relative importance by k n o w n relative share figures released by the BLS. Each series was then linked to the other with 1982-84 having been set as the base period equal to 100. Using this price index, pj, equation (1) can be rewritten in final form as e m= V m P m + i 3 m ( E - V m P m - V I P i ) + um. (4) This household motor fuel demand function is nonlinear in its parameters and thus must be estimated using a nonlinear estimation technique. Equation (4) has been estimated by standard maximum likelihood techniques using the Davidson-Fletcher-Powell algorithm (see A. Ronald Gallant 1990) with an adjustment for the presence of first-order serial correlation of the residuals. Serial correlation occurs when the errors associated with fitting the expenditure function to the data from one period (quarter) carry over into future periods (quarters). First-order serial correlation means that the errors from one quarter are correlated or related to the errors in the ensuing quarter. The presence of this serial correlation detracts from the efficiency of the statistical estimation, making estimated parameter values seem more precise than they truly are. A statistical procedure that corrects for this potential problem has been employed. The parameter values resulting from the estimation are given in Table 1. As shown in Table 1, the key estimated coefficients are statistically significant, being well in excess of their standard errors. This fact indicates that these parameters can be used with confidence in discussing household motor fuel demand relationships during the period of analysis. These parameter estimates also are consistent with underlying economic theory in that they satisfy the qualitative restrictions implied by standard behavioral assumptions and model structure. That is, in line with the discussion in the previous section, the m i n i m u m (subsistence) quantities of motor fuel, measured by r) m , and the index or composite commodity, measured by r\j, are positive, and the supernumerary or marginal-budgetshare parameter, measuring the proportion of supernumerary income spent on motor fuel, is positive ECONOMIC REVIEW, »MARCH/APRIL 1991 e m=7lmPm+ Table 1 Estimated Demand (Expenditure) Function Coefficients for U.S. Household Motor Fuel, 1979-80 Parameter Coefficient Standard Error t- value T] m .560 .032 17.50* ßm .013 .001 13.00* ri I 1.037 .620 1.67 * Statistically significant at the 1 percent level or higher. and less than unity, indicating that the household's marginal utility for motor fuel consumption is decreasing. Nonprice Conservation Theory The nonprice conservation theory of oil consumption's insensitivity to oil prices is based on the notion that national energy conservation efforts following the oil price shock of 1974 should have reduced the economy's dependence on foreign oil. These policies included imposing fuel mileage standards for U.S. automobile manufacturers, excise taxes on "gas guzzlers," and special tax incentives for producing alternatives to fossil fuels. As noted earlier, Brown and Phillips rejected the nonprice conservation theory at the aggregate level. As a way of providing evidence on these conservation measures' effectiveness at the household level, the basic model was tested for the presence of a significant (negative) linear technological or conservationrelated trend in household motor fuel consumption expenditures during the sample period. A statistically significant trend in household motor fuel consumption would suggest that something other than oil prices, income, and prices of other goods was affecting the demand for motor fuel. Testing the model for this trend in consumption expenditures involved estimating equation (4) in the following form: FEDERAL RESERVE BANK OF ATLANTA ßm^E^m PnT P/) (5) where t is a simple time trend variable. Because this time variable captures most systematic influences on demand other than mandated changes, a significant coefficient on the trend variable can be interpreted as consistent with the nonprice conservation theory. If the energy conservation measures were effective, the coefficient 0m on the time trend variable should be negative and statistically significant, supporting the nonprice conservation theory. The estimated value of this time trend coefficient was 0.31, with standard error of 0.23 and a statistically insignificant t-value of 1.35. The values of the other parameters were not significantly different from those given in Table 1. Thus, there was not any statistically significant general d o w n w a r d drift in household motor fuel consumption expenditures during the sample period. 4 O n the basis of this evidence, it seems clear that if conservation measures were at all effective they were effective only in particular years following shocks, not throughout the entire sample period. This finding of very little support for the systematic effectiveness of nonprice conservation measures at the level of the household during the 1970-89 period seems to agree with casual observation and is consistent with the findings reported by Brown and Phillips for aggregate U.S. oil consumption. It should be noted that asymmetries in the responsiveness of household motor fuel expenditures to price changes may also explain why the nonprice conservation theory does not seem to have applied during the whole sample period (1970-89). This period included two episodes of price declines in addition to the two adverse shocks. The mixed pattern of price changes during the period, combined with a possible asymmetric household demand response, contributes to the difficulty in assessing the long-term effectiveness of U.S. energy conservation measures. Asymmetric Response Theory High frequency (monthly and daily) motor fuel price data for the 1970s and 1980s reveal that price increases were sudden while declines tended to be more gradual. Because tests used in the research reported on here are based on quarterly data, they cannot distinguish different price change profiles 7 from more high frequency price series. Thus, the tests of the asymmetric response theory described in this section distinguish price changes only by direction, not speed or persistence. To test for possible asymmetries in the responsiveness of household motor fuel expenditures to changes in the price of motor fuel—that is, that demand responds differently to price increases than to price decreases—a series of statistical tests on the equality of the motor fuel expenditure (demand) equations across subperiods was conducted. Four subperiods were examined. The results of the statistical tests of equality of the expenditure (demand) functions across these subperiods appear in Table 2. The first subperiod—the base period—begins with the first quarter of 1970 (70:1) and runs through Table 2 Tests of Equality of Motor Fuel Expenditure Functions during Periods of Adverse (Rising), Stable, and Favorable (Falling) Motor Fuel Prices* Period Price Trend 1970:1-1973:4 vs. 1974:1-1978:4 Stable 1970:1-1973:4 vs. 1979:1-1985:4 Stable 1970:1-1973:4 vs. 1986:1-1989:4 Stable 1974:1-1978:4 vs. 1986:1-1989:4 Adverse 1979:1-1985:4 vs. 1986:1-1989:4 Adverse Test Result* Reject Adverse Reject Adverse Accept Favorable Reject Favorable Accept Favorable * Likelihood ratio test of equality of expenditure functions at 5 percent level. 8 the end of the fourth quarter of 1973 (73:4). It represents a regime of fairly stable motor fuel and oil prices. The second subperiod covers 74:1-78:4, and the third encompasses 79:1-85:4. Both of these subperiods were dominated by sharp increases in motor fuel prices at the beginning of the period as a result of the 1974 and 1979 adverse oil shocks. The fourth subperiod, 86:1-89:4, was characterized by sharply declining motor fuel and oil prices. Note that prices did not rise steadily throughout either of the periods designated as having adverse oil price trends. Each adverse trend period has been designated to extend from the beginning of one shock to either the next adverse shock or a subsequent dramatic change in the direction of prices. This delineation allows including the response time as part of the period being considered and therefore is consistent with the theories being tested. In this way, the period from 74:1 to 78:4 is deemed an adverse price trend episode even though prices did not rise uniformly over the period. So is the entire 79:1-85:4 period, even though prices declined somewhat at the end of that time. According to this treatment, the economy experienced two consecutive adverse price shocks. Unlike the aggregate demand for oil (see Brown and Phillips 1989a, b), household demand for motor fuel does not appear to behave symmetrically during periods of increasing and decreasing motor fuel prices. This conclusion is evidenced in part by the test results presented in Table 2, where the equality of the 70:1-73:4 subperiod expenditure (demand) function with all other subperiod expenditure functions except the 86:1-89:4 subperiod has been rejected. Further evidence derives from the fact that the expenditure function for the period exhibiting the most adverse, or sharply rising, prices (74:1-78:4) is significantly different from the subperiod during which prices fell sharply (86:1-89:4). No significant difference was found between the 70:1-73:4 base period, characterized by stable motor fuel prices, and the period of sharply falling motor fuel prices. Although the tests recorded in Table 2 indicate that the household motor fuel expenditure (demand) function is not symmetric around motor fuel price increases and decreases, these tests do not shed light on the economic implications of the asymmetric responses. In the next section of this article these economic implications are explored by examining the nature of the income and price elasticities implied by the subperiod expenditure (demand) functions. ECONOMIC REVIEW, »MARCH/APRIL 1991 Income Elasticities of Demand for Motor Fuel The household income-elasticity index can be computed using the formula given in equation (2). This index measures the relationship between a relative change in household income and the consequent relative change in demand for motor fuel, other things constant. As noted earlier, income elasticities are positive for normal goods—that is, dem a n d increases as income increases, other conditions remaining constant. Household income elasticities of demand for motor fuel for each of the subperiods examined in Table 2 are given in Table 3- These income elasticities were computed using equation (2) with the subperiod parameter estimates and the mean values of the other relevant variables for each subperiod. As can be seen in Table 3, all subperiod income elasticities are positive (motor fuel is a normal good), ranging from a high of 1.10 for the 70:1-73:4 base period to a low of 0.34 for the 79:1-85:4 subperiod. The estimated elasticity index of 1.10 in the 70:1-73:4 base subperiod implies that households treated gasoline as a luxury good during this period. A good is Table 3 Household Income Elasticities of Demand for Motor Fuel Income Period Price Trend Elasticity ( r m ) 1970:1-1973:4 Stable 1.10 1974:1-1978:4 Adverse 0.58 1979:1-1985:4 Adverse 0.34 1986:1-1989:4 Favorable 0.74 1970:1-1989:4 FEDERAL RESERVE BANK O F ATLANTA 0.47 considered a luxury good if its income elasticity of demand exceeds unity. That is, for a given percentage increase in the household's income, the demand for the luxury good increases by a more than proportionate amount, other things being equal. Following the 1974 oil shock, all subperiod income elasticities are estimated to be below unity. Thus, it appears that the first oil shock in effect converted gasoline from a luxury good into a necessity. A good is considered a necessity if it is normal with an income elasticity less than unity. The estimated income elasticities dropped dramatically relative to the base-period value during both oil-shock subperiods. These index values suggest that the high price of gasoline, combined with energy conservation measures such as car pooling and improved fuel efficiency of automobiles, may have served to reduce household dependence on motor fuel during these particular subperiods. This idea seems particularly plausible given the time required for these measures to have taken effect and exerted an impact on demand. Because oil and oil products have few close substitutes, it is reasonable to expect income elasticities to diminish as price elasticities increase over the longer run, when alternatives are developed and new habits are solidified. The income-elasticity index increased from a low of 0.34 during the subperiod of the second adverse oil shock to a value of 0.74 during the 86:1-89:4 subperiod, which was characterized by the positive oil shock and falling oil and motor fuel prices. This change is somewhat disturbing because it suggests that reversal of the conservation-related gains following the 1974 and 1979 shocks may have begun during the later subperiod. The rebound in income elasticity implies that the oil price declines of 1986:1 through 1989:4 were seen as permanent. The fixed costs of conservation and conversion were not expected to be incurred again, and the change in behavior was abandoned. The lower relative price of motor fuel may have moved it back toward the luxury-good category. Price Elasticity of Demand for Motor Fuel In the research discussed here price elasticities of demand for motor fuel have been computed using the formula given in equation (3). The priceelasticity-of-demand index measures the proportional magnitude that quantity demanded increases as price falls. Because motor fuel is a normal good, the price-elasticity index should be negative, indicating 9 that demand for motor fuel is inversely related to the price of motor fuel. The price elasticities of demand for each of the four subperiods examined above are given in Table 4. The estimated price elasticities for each of the subperiods and the entire sample period have the anticipated negative sign. The results in Table 4 corroborate those reported in Table 3. For the preshock base subperiod (70:1-73:4) household motor fuel demand was very inelastic or insensitive to the price of motor fuel. However, following the first oil shock in 1974, the sensitivity of the household motor fuel demand function increased noticeably. The index more than doubled in absolute value, increasing from -0.40 during the preshock base subperiod to a value of -0.85 during the subperiod of the first oil shock. On the other hand, the estimated demand elasticity for the 86:1-89:4 subperiod (characterized by dramatic declines in oil and motor fuel prices) was again approaching the value estimated for the base preshock subperiod. This pattern of response in household motor fuel demand to rising and falling gasoline prices lends additional support to the asymmetric price response theory. As can be seen in Table 4, the drop in demand resulting from the motor fuel price increase following the 1974 adverse oil shock exceeds the increase in demand resulting from the lower motor fuel prices that followed the favorable 1986 oil price shock. 5 In ad- Table 4 Price Elasticities of Demand for Motor Fuel Price Period Price Trend Elasticity (a m ) 1970:1-1973:4 Stable -0.40 1974:1-1978:4 Adverse -0.85 1979:1-1985:4 Adverse -0.63 1986:1-1989:4 Favorable -0.47 1970:1-1989:4 10 -0.61 dition, the trend toward a more inelastic household motor fuel demand function since the 1974 subperiod is consistent with the notion that the energy conservation measures introduced following the 1974 adverse oil shock had lost much of their effectiveness by the early 1980s. However, these same results also are consistent with the idea that conservation measures and the development of alternatives have limitations. That is, as prices rise, equal marginal increases in oil prices do not yield equal marginal declines in consumption. The initial price increases elicit conserving behavior and a search for substitutes. Additional marginal conservation efforts yield smaller returns, though, so that, within some limited period of time, additional price increases result in decreasing adjustments to motor fuel demand. Implications and Conclusions The widespread belief among economists that adverse supply shocks lead to changes in behavioral relationships that persist long after a shock's initial impulse has passed is examined in this article. In particular, the question of whether U.S. household motor fuel demand is sensitive to the price of motor fuel is investigated. Casual observation of household gas and oil consumption during the last two decades would suggest that household motor fuel consumption is not very sensitive to motor fuel prices. Two popular theories have been advanced to explain this nonsensitivity—the nonprice conservation theory and the asymmetric response theory. At the household level during the 1970-89 period, the empirical results reported in this article do not support the nonprice conservation theory—which holds that household demand for motor fuel should have become less sensitive to changes in motor fuel prices since the 1974 oil shock as a result of gasoline and energy conservation policies. The differing price and income elasticities associated with various oil price trends support the asymmetric response theory, which hypothesizes that the decrease in demand for motor fuel subsequent to an increase in its price will exceed, in percentage terms, the increase in demand following a comparable decrease in the price of motor fuel. For the time periods covered by the analysis discussed here, the estimates from the econometric model reveal that the change in price elasticity as prices rise is greater than the change as prices fall. Future research efforts will be directed toward investigating this asymmetry and strengthening the basis for elasticity forecasts to allow for esti- E C O N O M I C REVIEW, »MARCH/APRIL 1991 mates of d e m a n d response to price changes not o n l y direction but also different speeds of c o n s u m e r reaction to the price increase will d e p e n d and o n w h e t h e r the price increase is v i e w e d as permanent or temporary. varying persistence. The test results c o n c e r n i n g U.S. h o u s e h o l d m o t o r T h e relatively inelastic d e m a n d for o i l suggests fuel d e m a n d reported in this article suggest that if re- that, at least in the short run, a m o t o r fuel (gasoline) s p o n s e to the 1990 oil s h o c k follows previous pat- tax is a g o o d source o f revenue, considering that the terns, gasoline a n d oil d e m a n d g r o w t h will s l o w as tax base ( m o t o r fuel c o n s u m p t i o n ) is n o t very re- the price elasticity o f d e m a n d increases a n d the in- sponsive to the tax rate. At the s a m e time, a n oil or c o m e elasticity declines. These responses will proba- gasoline excise tax is not likely to elicit c o n t i n u e d or bly b e smaller t h a n those to earlier oil price shocks, p e r m a n e n t conservation efforts unless the tax rate is h o w e v e r , because the payoff for increasing conser- increased o v e r t i m e , b e c a u s e the m a r g i n a l savings vation a n d switching t o substitutes decreases for ev- declines over time. ery m a r g i n a l i n c r e a s e i n o i l p r i c e s . I n addition, Notes 1. This measure is the composite refiners' acquisition cost. 2. For a review of this evidence through 1980, see Bohi (1981). 3. These demand functions are homogeneous of degree zero in prices and income and satisfy the adding-up criterion, and the matrix of substitution terms is symmetric and negative semidefinite. 4. However, it should be noted that, because the coefficients in Table 1 are estimated with a correction for the possibility of serially correlated residuals, it is possible that any technological drift in household motor fuel consumption over the sample as a result of energy or gasoline conservation measures was o f a nontrivial (nonlinear) form captured by the time-series properties of the residuals of the model. The value of the first-order auto correlation coefficient for equation (4) was estimated to be 0.67 and is only marginally statistically significant. 5. Note that there was considerable nonprice rationing of motor fuel coincident with price increases early in the 1974-78 period. This factor may introduce an upward bias to the price elasticity measure for that period. References Berndt, Ernst, M.N. Darrough, and William E. Diewert. "Flexible Functional Forms and Expenditure Distributions: An Application to Canadian Consumer Demand Functions." University of British Columbia, Department of Economics, Discussion Paper 77-10, 1977. Bohi, D.R. Analyzing Demand Behavior: A Study of Energy Elasticities. Baltimore, Md.: Johns Hopkins University Press for Resources for the Future, 1981. Brown, Stephen P.A., and Keith R. Phillips. "The Effects of Oil Prices and Exchange Rates on World Oil Consumption." Federal Reserve Bank of Dallas Economic Review (July 1984): 13-21. . "An Econometric Analysis of U.S. Oil Demand." Federal Reserve Bank of Dallas Working Paper No. 8901, January 1989a. . "Oil Demand and Prices in the 1990s." Federal Reserve Bank of Dallas Economic Review (January 1989b): 1-8. Christensen, Laurits R., Dale N. Jorgenson, and Lawrence J. Lau. "Transcendental Logarithmic Utility Functions." American Economic Review 65 ( J a n u a r Y 1975): 37-54. Frisch, Ragnar A. "A Complete Scheme for Computing All Direct and Cross Demand Elasticities in a Model with Many Sectors." Econometrica 27 (April 1959): 177-96. FEDERAL RESERVE BANK OF ATLANTA Gallant, A. Ronald. Nonlinear Statistical York: John Wiley and Sons, 1990. Models. New Gately, Dermot, and Peter Rappoport. "The Adjustment of U.S. Oil Demand to the Price Increases of the 1970s." The Energy Journal 9 (1988): 93-107. Hogan, W.W. "A Dynamic Putty-Semi-Putty Model of Aggregate Energy Demand." Energy Economics 11 (1989): 53-69. Houthakker, Hendrik S. "Additive Preferences." Econometrica 28 (April I960): 244-57. Klein, Lawrence R., and Herman Rubin. "A Constant Utility Index of the Cost of Living." Review of Economic Studies 15 (1947-48): 84-87. McKenzie, Chester V. "Technical Notes: Relative Importance o f CPI C o m p o n e n t s . " Monthly Labor Review (November 1961): 1233-36. Parks, R.W. "Systems of Demand Equations: An Empirical Comparison of Alternative Functional Forms." Econometrica 37 (October 1969): 629-50. Stone, Richard. "Linear Expenditure Systems and Demand Analysis: An Application to the Pattern of British Demand." Economic Journal 64 (September 1954): 511-27. 11 Beyond Plain Vanilla: A Taxonomy of Swaps Peter A. Abken Since their introduction over a decade ago, swaps have become an important tool for financial risk management. Generally, swaps alter the cash flows from assets or liabilities into preferred forms. Basic swaps have branched into many variants, some more popular and successful than others, each geared toward meeting specific customer needs in various markets. The author describes the features and typical applications of many variants of the four basic swap types—interest rate, currency, commodity, and equity. wap contracts of various kinds have become a mainstay of financial risk management since their introduction in the late 1970s. In the most general terms, a swap is an exchange of cash flows between two parties, referred to as counterparties in the parlance of swap transactions. Swaps, which transform the cash flows of the underlying assets or liabilities to which they are related into a preferred form, have been used in conjunction with positions in debt, currencies, commodities, and equity. Most swap agreements extend from one to ten years, although many have been arranged for much longer periods.1 S The key players responsible for originating and propelling the swaps market are money center banks and investment banks. These institutions benefit from the fee income generated by swaps, which are off-balance-sheet items, and by the spreads that arise in pricing swaps. Innovations in the swaps market, as in other financial services areas, may be characterized as a Darwinian struggle, in which competition heats up and margins narrow as a particular kind of swap becomes accepted and widely used. Such swaps are disparagingly said to be trad- 12 ed "like commodities." That is, little value is added by the dealer in structuring a swap and bringing counterparties together; consequently, little return is realized for the service of intermediation or position taking. Perhaps the most basic, and most popular, swap involves the conversion of interest payments based on a floating rate of interest into payments based on a fixed rate (or vice versa). Because many variants of interest rate and other swaps have emerged over the years, this most basic type has become known as the "plain vanilla" swap. 2 As swap forms take on plain vanilla status, the firms that originated them are compelled to develop new types of swaps to regain their margins, amounting to m o n o p o l y rents, on new products. Some swap variations succeed, while others languish or fail. In this article the plain vanilla swap is a starting point for a detailed taxonomy of the various species and subspecies of swaps. Swap variants are classified along cladistic principles, categorized and compared in terms of their features and applications. Examples illustrate many of the important types of swaps. ECONOMIC REVIEW, »MARCH/APRIL 1991 The Market A Brief History. Before taking a detailed look at swaps, an overview of the market will help put their proliferation into perspective. Although some swaps had been arranged in the late 1970s, the first major transaction was a 1981 currency swap between IBM and the World Bank. This deal received widespread attention and stimulated others. The currency swap actually evolved from a transaction popular in the 1970s, the parallel loan agreement, that produced cash flows identical to a swap's. For example, in one of these agreements a firm in the United States borrows a million dollars by selling a coupon bond and exchanges (swaps) this amount for an equivalent amount of deutsche marks with a German firm, which borrows those deutsche marks in its domestic market. This is the initial exchange of principal. Thereafter, the U.S. firm makes markdenominated coupon payments and the German firm makes dollar-denominated coupon payments. Upon maturity of the underlying debt, the firms swap principal payments. These firms have effectively borrowed in one another's capital markets, although for a variety of reasons (such as foreign exchange controls or lack of credit standing in foreign markets) they could not borrow directly. As Clifford W. Smith, Charles W. Smithson, and Lee Macdonald Wakeman (1990a) point out, the problems with such an agreement were that default by one firm does not relieve the other of its contractual obligation to make payments and that the initial loans remain on-balancesheet items during the life of the agreement for accounting and regulatory purposes. The currency swap, on the other hand, stipulates that a default terminates the agreement for both counterparties and, in general, limits credit-risk exposure to the net cash flows between the counterparties, not the gross amounts. This type of currency swap is essentially a sequence of forward foreign exchange contracts.^ Following the 1981 currency swap, the first interest rate swap, in mid-1982, involved the Student Loan Marketing Association (Sallie Mae). With an investment bank acting as intermediary, Sallie Mae issued intermediate-term fixed rate debt, which was privately placed, and swapped the coupon payments for floating rate payments indexed to the threemonth Treasury bill yield. Through the swap, Sallie Mae achieved a better match of cash flows with its shorter-term floating rate assets.4 At the end of 1982, the combined notional principal outstanding for interest rate and currency swaps stood at $5 billion. Notional principal is the face value of the underlying debt upon which swap cash flows are based. FEDERAL RESERVE BANK OF ATLANTA The commodity swap made its appearance in 1987, when it was approved by a number of U.S. banking regulators (see Schuyler K. Henderson 1990 and Krystyna Krzyzak 1989b, c). Banks had been prohibited from direct transactions in commodities or related futures and forward contracts. In 1987 the Office of the Comptroller of the Currency permitted Chase Manhattan Bank to act as a broker in commodity swaps between an Asian airline and oil producers. Shortly afterward Citicorp also obtained approval for engaging in commodity swaps through its export-trading subsidiary. Regulations were further relaxed in February 1990 to allow national banks to use exchange-traded futures and options to hedge commodity swap positions. However, much commodity swap activity took place offshore because of uncertainties about the Commodity Futures Trading Commission's (CFTC) view of commodity swaps. The CFTC undertook a study of off-exchange transactions in February 1987 to determine whether they came under the CFTC's regulatory jurisdiction. In July 1989 the CFTC established criteria that would exempt commodity swaps from its regulatory oversight.5 Since the CFTC's decision commodity swap activity has been increasing in the United States. As of early 1990, commodity swaps outstanding totaled about $10 billion in terms of the value of the underlying commodities (Julian Lewis 1990, 87). Equity swaps are the newest variety, first introduced in 1989 by Bankers Trust. Based on both domestic and foreign stock indexes, these instruments may take complex forms, such as paying off the greater of two stock indexes against a floating rate of interest. The mechanics of such instruments and their advantages will be discussed below. The Size o f the Market. As of year-end 1989, the size of the worldwide swaps market, as measured by the dollar value of the notional principal, stood at $2.37 trillion. This figure does not include commodity or equity swaps, but these new types of swap have relatively small amounts outstanding compared with interest rate and currency swaps. The International Swap Dealers Association (ISDA), a trade organization, periodically surveys its members, who include most of the major swap dealers. Table 1 displays the survey results for swaps in various categories. The The author is an economist in the financial section of the Atlanta Fed's research department. He is grateful to many people in the swaps market for assistance with his research for this article. He would particularly like to thank Charles W. Smithson and fames M.F. MeVay of Chase Manhattan Bank and Ron Slivka of Salomon Brothers. However, any errors are the author's responsibility. 13 Table 1 U. S. Dollar Interest Rate Swaps 1985-89* Survey Period 1985 1986 1987 1988 1989 Total ISDA User End User Contracts Notional Principal Average Contract Contracts Notional Principal Average Contract Contracts Notional Principal Average Contract 5,918 10,752 16,871 20,381 23,324 $141,834 $235,829 $379,880 $484,272 $622,602 $23.97 $21.93 $22.52 $23.76 $26.69 1,061 3,330 7,472 8,968 13,303 $28,348 $76,921 $161,637 $243,894 $371,144 $26.72 $23.10 $21.63 $27.20 $27.90 6,979 14,082 24,343 29,349 36,627 $170,182 $312,750 $541,517 $728,166 $993,746 $24.38 $22.21 $22.25 $24.81 $27.13 $19.95 $23.49 $23.15 34,127 49,560 73,828 $682,888 $1,010,203 $1,502,600 $20.01 $20.38 $20.35 Total Interest Rate Swaps 1987-89 1987 1988 1989 23,768 35,031 50,193 $476,247 $668,857 $955,492 $20.04 $19.09 $19.04 * All dollar amounts are in millions of dollars in U.S. dollar equivalents. Source: International Swap Dealers Association Market Survey. 10,359 14,529 23,635 $206,641 $341,345 $547,108 interest rate swap market, involving swaps denominated in one currency, composed roughly two-thirds of the market, or $1.5 trillion as of year-end 1989. Of that amount, two-thirds consisted o f U.S. dollar swaps, the most prevalent kind of swap. The average contract size was $20.35 million and $27.13 million for total and total dollar interest rate swaps, respectively. Currency swap market data are given in Table 2. The U.S. dollar is less dominant among currency swaps, for which it represents 41 percent of the total, compared with its 66 percent share of interest rate swaps. For the years during which the survey has been conducted, swaps of every type have grown rapidly. ^ In all categories the position of the end users has been a multiple of those of the swap dealers. Interdealer swaps arise mainly in connection with hedging activities. A certain amount of double counting is therefore involved in the aggregate figures because one swap can set u p a number of others as counterparties hedge their positions. The latest ISDA survey reveals that the most active category for new swaps originated during the period January 1 to June 30, 1990, was non-U.S. dollar interest rate swaps, which grew by 26.4 percent. U.S. dollar swaps increased by 8.2 percent in this period. In contrast, total currency swaps rose by 2.9 percent, with U.S. dollar currency swaps contracted increasing by 4.6 percent. These semiannual growth rates show considerable variability over time and thus do not indicate trend movements. Further discussion of the ISDA survey results appears below in the section on currency swaps. Interest Rate Swaps Interest rate swaps account for the most volume in the swaps market, as seen in the previous section. The explanation to follow covers many of the numerous features that can modify the plain vanilla swap. Though discussed in detail only in relation to interest rate swaps, these alternate forms actually or potentially apply to currency, commodity, and equity swaps as well; they can be combined in innumerable ways to alter any kinds of cash flows. The basic fixed-for-floating interest rate swap involves a net exchange of a fixed rate, usually expressed as a spread over the Treasury bond rate corresponding to the swap maturity, for a floating rate of interest. That floating rate is tied or indexed to any of a number of short-term interest rates. The London Interbank Offered Rate (LIBOR) is the most common. 7 Other rates include the Treasury bill rate, FEDERAL RESERVE BANK O F ATLANTA the prime rate, the Commercial Paper Composite, the Certificate of Deposit Composite, the federal funds rate, the J.J. Kenney index, and the Federal Home Loan Bank System's Eleventh District cost-of-funds index. The Eleventh District index has been used mainly by thrift institutions in California. 8 The J.J. Kenney index is based on short-term tax-exempt municipal bond yields. The fixed rate payer (and floating rate receiver) is said to have bought a swap or to have "gone long" a swap. Similarly, the floating rate payer (and fixed rate receiver) is said to have sold a swap or "gone short" a swap. Swaps are quoted by a dealer (or broker) usually in terms of the spread over the Treasury security of comparable maturity. For example, a swap with seven-year time to maturity, or tenor, might be quoted at 65-72. The dealer is offering to buy a swap (pay fixed) at a rate that is 65 basis points above the seven-year Treasury yield, and offering to sell a swap (receive fixed) at 72 basis points over that yield.9 The dealer is therefore collecting a 7 basis point margin for standing between the counterparties. Like floating rate notes, the floating rate payments on a swap do not necessarily match the timetable of the floating rate index. 10 The payment may be based on the average of the underlying index during some specified interval. The point at which the floating rate is established, based on the floating rate at that time or over some previous period, is termed the reset date. This date is not necessarily the same as the settlement date, when payment on the swap is made to the other counterparty. If reset and settlement dates do not coincide, the swap is said to be paid in arrears, which is also a c o m m o n convention for floating rate notes. The floating rate may be reset daily, weekly, monthly, quarterly, or semiannually, while typically the settlement dates fall monthly, quarterly, semiannually, or annually (Anand K. Bhattacharya and John Breit 1991, 1158). As over-the-counter instruments, interest rate swap terms are open to negotiation. The conventional way to quote a swap rate is relative to the floating rate index "flat." That is, a swap counterparty would pay the fixed rate and receive LIBOR. Swaps can also be arranged to include a spread above or below the floating rate—for example, LIBOR + 10 basis points. In addition, fixed rate payers and floating rate payers can agree to making payments at different periods—quarterly floating rate payments versus semiannual fixed rate payments. 11 However, swap counterparties usually prefer net transactions so that only a difference check passes between them, thereby limiting credit exposure. In the section below the first alteration of the basic plain vanilla structure that is considered encompasses different treatments of a 15 Table 2 U. S. Dollar Currency Swaps 1987-89* Survey Period Contracts Notional Principal 1987 1988 1989 4,665 6,777 9,078 $129,181 $201,374 $257,748 Total ISDA User End User Average Contract $27.69 $29.71 $28.39 Contracts Notional Principal Average Contract Contracts Notional Principal Average Contract 1,366 2,297 3,414 $33,425 $68,103 $96,418 $24.48 $29.66 $28.24 6,031 9,074 12,492 $162,606 $269,477 $354,166 $26.96 $29.70 $28.35 $24.67 $32.30 $27.67 6,612 10,271 15,285 $365,614 $633,642 $869,698 $27.65 $30.85 $28.45 Total Currency Swaps 1987-89 1987 1988 1989 5,173 7,724 11,270 $294,608 $469,092 $647,516 $28.47 $30.37 $28.73 * All dollar amounts are in millions of dollars in U.S. dôllar equivalents. Source: International Swap Dealers Association Market Survey. 1,439 2,547 4,015 $71,006 $164,550 $222,182 swap's notional principal. The second general variation outlined allows for specially tailored coupon structures, and the discussion includes consideration of option-like features. Third, different types of underlying instruments—in particular, asset swaps and their uses in creating synthetic assets—are examined. Finally, option structures are discussed, including options on swaps, known as swaptions. Seasonal Swaps and Roller Coasters. Finally, amortizing and accreting notional principals can be combined to form a seasonal swap, which allows the notional principal to vary according to a counterparty's seasonal borrowing needs such as those retailers typically experience. A swap that allows for periodic or arbitrary but predictable swings in notional principal is called a roller coaster. V a r i a t i o n s o n N o t i o n a l P r i n c i p a l . The plain vanilla swap is nonamortizing. Nonamortizing swaps, known as "bullet" swaps, have a constant underlying notional principal upon which interest payments are made. This structure is easily modified to accommodate any kind of predictable changes in the underlying principal. Uncertainty about the future amount of the principal, which frequently arises with mortgage-backed securities, is usually better handled using option features, which will be discussed shortly. Variations o n C o u p o n Payments. Altering cash flows of underlying securities is one of the primary functions of swaps. In the following section a number of important types of swaps that accomplish this end are discussed, including those with option-like features. Off-Market Swaps. The plain vanilla swap is also characterized as a par value swap. That is, the fixed rate for the swap is established such that no cash payment changes hands when the swap is initiated. The term par value derives from the swap's being viewed as a hypothetical exchange of fixed for floating rate bonds. When arranged at market interest rates, both bonds are equal to their face values (par value). Nonpar, or off-market, swaps involve fixed or floating rates that are different from the par value swap rates. Differences in the fixed rate above or below the par value swap rate entail a cash payment to the fixed rate payer from the floating rate payer if the fixed rate coupon is above the par value swap rate, and vice versa if it is below. The payment's amount is the present value of the difference between the nonpar and par value swap fixed rate payments. Swap counterparties commonly perform this kind of calculation in the process of marking an existing swap to market. An existing swap may be terminated (if permitted in the swap agreement) by such a marking to market of the remaining swap payments. High or low coupon swaps, as off-market swaps are alternatively called, are created simply by doing the calculation at the outset and making or receiving the appropriate payment. One reason for engaging in this type of swap is to change the tax exposure of underlying cash flows. Another is that spreads above or below the floating rate index can be introduced. John Macfarlane, Janet Showers, and Daniel Ross (1991) explain the mechanics of this variation. Amortizing, Annuity, and Mortgage Swaps. Amortizing swaps are typically used in conjunction with mortgage loans, mortgage-backed securities, and automobile- and credit-card-backed securities. All of these tend to involve repayment of principal over time. In general it is difficult to match the amortization schedule of a swap, which usually cannot be changed after its initiation, against the amortization rate on these assets or liabilities; thus, the swapholder runs the risk of being over- or underhedged. A particular example of an amortizing swap is discussed in more detail in the section below on asset swaps. One specific kind is the mortgage swap, which is simply an amortizing swap on mortgages or mortgage-backed securities. The extreme form of an amortizing swap, in which the notional principal diminishes to zero as the principal of a fixed rate mortgage does, is an annuity swap. Accreting Swaps. The flip side of an amortizing swap is an accreting swap, which, as its name suggests, allows the notional principal to accumulate during the life of the swap. Both amortizing and accreting swaps are sometimes also called sawtooth swaps. The accreting swap arises commonly with construction finance, in which a construction company or developer has a floating rate drawdown facility with a bank. That is, a line of credit may be tapped that would lead to increasing amounts of floating rate borrowing. An accreting swap would convert those floating rate payments into fixed rate payments, although again there is a risk of not exactly matching notional principal amounts at each settlement date. It is possible to create amortizing or accreting swaps from bullet swaps of varying tenor instead of arranging a swap specifically with the desired characteristics. FEDERAL RESERVE BANK OF ATLANTA Basis Swaps. A basis swap is an exchange of one floating rate interest payment for another based on a different index. Consider an example in which a bank, First SmartBucks, has invested in two-year floating rate notes that pay the bank one-month LIBOR plus 100 basis points. First SmartBucks has funded this purchase by issuing one-month certificates of deposit. The problem is that LIBOR and the CD rate will not track each other perfectly, exposing 17 First SmartBucks to a so-called basis risk; it may pay more on its CDs than it receives from its floating rate notes. The problem is solved by entering into a basis swap with a swap dealer, who will pay the onemonth CD rate in exchange for LIBOR. Chart 1 illustrates the transaction. Aside from the initial fee for the swap, the cost of this hedging transaction manifests itself as a 10 basis point spread under the CD rate received from the dealer. This hedge may also be less than perfect, however, because the dealer probably would use the Certificate of Deposit Composite, which may not track First SmartBucks's CD rate perfectly, to index his payments. Nevertheless, the swap is likely to mitigate the original basis risk. Yield Curve Swaps. The yield curve swap, a variant of the basis swap, typically is an exchange of interest payments indexed to a short-term rate for ones tied to a long-term rate. For example, a counterparty could contract to make semiannual floating rate payments based on six-month LIBOR and receive floating rate payments indexed to the prevailing thirty-year Treasury bond yield, less a spread to the swap dealer.12 The ten-year Treasury bond yield has also been used for yield curve swaps on the long end, as well as three-month LIBOR on the short end. Yield curve swaps gained popularity in early 1988 when the yield curve began to flatten—that is, when long rates fell relative to short rates (Krzyzak 1988, 29). Savings and loan institutions were major users of this new swap because they found it useful for adjusting the interest rate exposures of their portfolios (see asset swaps below). These swaps were also well suited to speculating on shifts in the yield curve while hedging against changes in its level. Finally, these instruments were combined with a new kind of floating rate debt, called FROGs (floating rate on governments), to transform the FROG's coupon into LIBOR. The coupon was reset semiannually and tied to the yield on newly issued Treasury bonds. 13 This strategy reportedly achieved a lower cost of funding than a standard LIBOR floating rate issue. Caps, Floors, and Collars. A floating rate payer can combine option contracts with a swap to tailor the maximum size of potential swap payments. Interest rate caps, floors, and collars are instruments closely related to swaps that can alter swap cash flows. 14 As an example, consider a plain vanilla swap with a fixed rate of 8 percent (the swap rate). At a reset date, a rise in the floating rate above 8 percent would obligate the floating rate payer to pay the counterparty the net amount of the notional principal outstanding times the difference between the actual floating rate—say, 10 percent—and the swap rate. By buying a 9 percent cap of the same maturity as the swap the user would never pay more than one percentage point above the swap rate. The cap could be obtained from another counterparty, or it could be bundled with the swap in one transaction. However, buying a cap from another counterparty introduces an additional credit risk. Chart 1 A Basis Swap LIBOR Swap Dealer First SmartBucks C D Composite Rate -10 Basis Points l1 LIBOR + 100 Basis Points CD Rate r Assets Liabilities FRNs CDs First SmartBucks transforms LIBOR interest coupons into CD composite rate payments via a basis swap. 18 ECONOMIC REVIEW, »MARCH/APRIL 1991 A counterparty who sells (or writes) a cap is obligated to pay the excess over the cap's strike rate (9 percent in this example). The purchaser in return pays a cap "premium" up front. In fact, caps are sequences of interest rate options with maturities that match the schedule of floating rate payments. Analogous to caps, interest rate floors pay off whenever the floating interest rate falls below the prespecified floor level. To defray some or all of the cost of buying a cap, the floating rate payer could sell a floor with a strike rate less than the swap rate. Such a sale would create an interest rate collar. Thus, rather than paying for the protection of the cap outright, the floating rate payer could give up part of the payments from the swap resulting from large declines in the floating rate below the swap rate. That is, the maximum possible payment from the other swap counterparty would effectively be the difference of the swap rate and the floor strike rate times the notional principal. Synthetic Swaps. The "collaring" of a swap suggests that a floating rate payer could completely offset a swap by buying a cap and selling a floor that both have strike rates equal to the swap rate. Similarly, a fixed rate payer could nullify a swap by selling a cap and buying a floor with strike rates equal to the swap rate. In these cases the floating rate payer would, in effect, be buying a "synthetic" swap and the fixed rate payer would be selling one. However, swaps are not usually unwound in this way because it is generally cheaper simply to buy or sell the corresponding swap; caps and floors may not be sufficiently liqCiid at the desired strike rates to execute these transactions at reasonable prices. (That is, an illiquid—infrequently traded—cap or floor would be quoted with large spreads.) Nevertheless, arbitrage between the swap and cap/floor markets is possible and does occur if rates for these instruments get too far out of line. Participating Swaps. A hybrid version of the fixed rate swap and interest rate cap allows a counterparty to benefit partially from declining rates while not requiring any up-front payment as with a cap. Consider an example using LIBOR. The counterparty would receive LIBOR to pay its floating rate debt. In turn, instead of paying a fixed interest rate as for a plain vanilla swap, a higher fixed rate is established (above the swap rate), which is the maximum rate the counterparty would pay if LIBOR rises above that level. However, if LIBOR falls below this maximum rate, the counterparty's payment would decline less than one-for-one with LIBOR. For example, the swap terms could stipulate that a one percentage point drop in LIBOR would reduce the swap payment by one-half percentage point. The so-called FEDERAL RESERVE BANK O F ATLANTA participation rate in this case is 50 percent. In other words, the counterparty would participate in 50 percent of any decline in LIBOR below the maximum rate. The maximum rate and the participation rate are set to price the swap at zero cost upon initiation. The price of this swap's option feature is paid by giving up part of the gains from falling rates. The participating swap can also be structured to have the counterparty pay LIBOR and receive payments indexed to a fixed schedule. That is, a minimum rate would be specified in the swap, with payments above that minimum determined by the product of the prevailing LIBOR multiplied by the participation rate. A counterparty might want to use such a swap in conjunction with its floating rate assets. Participating swaps can be structured for any interest rates and are also used for currencies and commodities. Reversible Swaps and Roller Coasters. Reversible swaps and roller coasters are a couple of exotic variants on swap structures. A reversible swap allows a counterparty to change status from floating rate payer to fixed rate payer or vice versa at some point during the life of the instrument. The roller coaster takes this concept a step further by having the counterparties reverse roles at each settlement date. Distinct from the earlier type of roller coaster involving variations in notional principal, this one has been used in only a limited number of transactions. Zero Coupon Swaps. As its name implies, all payments on one side of the swap come at the end in one "balloon" payment, while the other side makes periodic fixed or floating rate payments. One use of zero coupon swaps is to transform the cash flows from zero coupon bonds into those of fixed coupon bonds or floating rate bonds, or vice versa. Asset S w a p s . Asset swaps are precisely what their name suggests. They effectively transform an asset into some other type of asset, such as the conversion of a fixed rate bond into a floating rate bond. The conversion results in synthetic securities because of the swap's effects. The analysis of asset swaps actually contains nothing new. The earlier example of First SmartBucks's use of a basis swap, exchanging LIBOR for the CD Composite rate, was a type of asset swap. Asset swaps are usually considered in connection with portfolio management and are low-cost tools for changing the characteristics of individual securities or portfolios. Bhattacharya (1990) discusses an interesting application of asset swaps to a particular kind of mortgage-backed security. The collateral behind mortgage-backed securities is subject to prepayment. For example, homeowners may pay off their mortgage principals early in the event they move 19 or mortgage rates drop sufficiently. Collateralized Mortgage Obligations (CMOs) repackage mortgage cash flows into a variety of securities that carry different prepayment risks. Planned Amortization Class (PAC) bonds are structured to have amortization schedules more predictable than those of other CMO classes. However, the risks are nevertheless sufficient to make PAC bonds trade at fairly wide spreads over corresponding Treasury securities. PAC bonds have been popular candidates for amortizing asset swaps that convert the bonds' fixed coupons into floating rate payments tied to any index. These asset swaps have the potential to make PAC bonds attractive to a broader class of investors and consequently channel more funds to the mortgage market. Such swaps may be a more cost-effective means of altering the characteristics of mortgage-backed securities than having an even broader array of such securities being issued. As a tool for bond portfolio management, asset swaps can change a portfolio's exposure to interest rate risk. The value of a portfolio, and of any bonds within it, fluctuates with shifts in interest rates, tending to fall as market rates rise and vice versa. The sensitivity to interest rate risk is measured by a portfolio's duration, which is based on the future timing and size of its cash flows.15 A portfolio manager can extend a portfolio's duration, increasing its volatility with respect to interest rate movements, by entering into asset swaps to receive fixed rate cash flows and to pay floating rate cash flows. Conversely, a portfolio can be protected or "immunized" against interest rate movements by contracting to make fixed rate cash flows and receive floating rate. The intuition here is that the more a portfolio's (or security's) cash flows move with current market rates, the closer its value will stay to face value. Money market funds, for example, experience little change in asset value because they have very short duration. In contrast, a fund consisting of long-term zero coupon bonds, which have durations equal to their maturities, would have extremely volatile asset values. Asset swaps are particularly useful for adjusting a portfolio when securities sales would result in capital losses. For example, a portfolio manager would be reluctant to change the portfolio's duration by selling off bonds that are "under water" (currently valued below par). As just discussed, an asset swap is ideal for this kind of adjustment.16 As another example, some bonds cannot be traded because they were purchased from an underwriter through a private placement to avoid registration and other costs associated with public issues. Using an asset swap obviates the need to trade the underlying security to alter interest rate exposure. 20 F o r w a r d a n d E x t e n s i o n S w a p s . Forward swaps are analogous to forward or futures contracts as hedging instruments. The difference is that forward or futures contracts hedge cash flows at a single point in the future whereas forward swaps (and swaps generally) hedge streams of cash flows. Extension swaps are an application of forward swaps. Forward Swaps. Financial managers, such as corporate treasurers, often want to hedge themselves against rising interest rates when considering a future debt issue. For example, selling a new issue of bonds may be necessary to refund outstanding corporate bonds that mature in one year. The yield on that issue is unknown today but could be locked in using a forward or deferred swap. If rates have risen when the outstanding bonds mature, the firm sells the swap, realizing a gain equal to the present value "As a tool for bond portfolio management, asset swaps can change a portfolio's exposure to interest rate risk." o f the difference b e t w e e n the cash flows b a s e d o n the current s w a p rate a n d those b a s e d o n the l o w e r fixed rate o f the forward s w a p . This gain w o u l d offset the higher c o u p o n p a y m e n t s o n the n e w l y issued fixed rate b o n d ; the effective rate p a i d w o u l d b e the s a m e as the forward s w a p rate. H o w e v e r , a fall in rates w o u l d translate into a loss o n the forward s w a p u p o n sale, a l t h o u g h the n e w l y i s s u e d fixed rate b o n d w o u l d itself carry a l o w e r rate. T h e effective rate o n the fixed rate issue w o u l d a g a i n b e the f o r w a r d s w a p rate, neglecting differences in transactions costs. T h e forward s w a p in this e x a m p l e is u s e d as a h e d g i n g tool, establishing a certain fixed rate today instead o f a n u n k n o w n fixed rate at the future date for d e b t issuance. Extension Swaps. A n extension s w a p is merely a f o r w a r d s w a p a p p e n d e d to a n existing s w a p before its term e n d s t o e x t e n d it b y s o m e a d d i t i o n a l p e r i o d (Jeffry B r o w n 1991, 127). If the forward s w a p is arr a n g e d b a s e d o n current f o r w a r d interest rates, the ECONOMIC REVIEW, »MARCH/APRIL 1991 extension swap w o u l d be obtained at n o cost. However, if a counterparty wants the forward swap rate to match an outstanding swap's rate, an upfront cash payment (or receipt) might be necessary to compensate for the change in market rates since the outstanding swap's origination. The extension swap in this case would be a type of off-market swap. Swaptions. The earlier discussion of amortizing swaps and the example of an asset swap involving a PAC bond emphasized the risk inherent in mismatches of principal with notional principal. The amount of principal is not always perfectly predictable, especially for many new types of assetbacked securities. Option contracts are designed to handle contingencies of this kind, and, not surprisingly, a market has developed for options on swaps, known as swaptions. (There is also a market for op- "For hedging applications, perhaps the swaption's most basic use is to give a swap counterparty the option to cancel a swap, at no further cost beyond the initial swaption premium." tions on caps and floors, which, as one might guess, are called captions and floortions.) Like any option, swaptions entail a right and not an obligation on the part of the buyer. Unfortunately, the nomenclature for swaptions is confusing, so the details are often simply spelled out in talking about them. A call swaption (a call option on a swap or payer swaption) is the right to buy a swap—pay a fixed rate of interest and receive floating. A put swaption (put option on a swap or receiver swaption) is the right to sell a swap—pay floating and receive fixed. The swaption on the plain vanilla swap is the most common, although swaptions can be written on more complicated swaps. Both the maturity of the swaption and the tenor of the underlying swap, which commences at a stipulated future date, must be specified. Also like options, swaptions come in both American and European varieties. The European swaption, which accounts for about 90 percent of the market, may be exercised only upon its matu- FEDERAL RESERVE BANK OF ATLANTA rity date, whereas the American swaption may be exercised at any time before maturity (Robert Tompkins 1989, 19). Only European swaptions will be considered in this discussion, unless otherwise noted. A call swaption would be exercised at maturity if the swaption strike rate—the fixed rate specified in the contract—is lower than the prevailing market fixed rate for swaps of the same tenor. The swaption could be closed out by selling the low fixed rate swap obtained through the swaption for a gain, rather than entering into that swap. Similar reasoning applies to the decision to exercise a put swaption. Swaptions are quite different from caps and floors, although these instruments are frequently used in similar situations. A swaption involves one option on a swap, while a cap (or floor) represents a series of options expiring at different dates on a floating interest rate. In addition, cap prices depend partly on the volatility of near-term forward rates, whereas swaption prices reflect the volatility of future swap rates, which in turn are averages of more distant, less volatile forward rates. Consequently, swaptions are much cheaper than caps or floors. Like options, swaptions require up-front payments, but these have recently fallen in the range of 20-40 basis points as compared with 200-300 basis points for caps or floors (Krzyzak 1989a, 13). American swaptions would be slightly more costly than European swaptions because of the additional right to exercise the instrument before maturity. Callable, Puttable, and Reversible Swaps. For hedging applications, perhaps the swaption's most basic use is to give a swap counterparty the option to cancel a swap, at no further cost beyond the initial swaption premium. A fixed-for-floating swap bundled together with a put swaption is known as a callable swap. The swap can be canceled upon the maturity of the embedded swaption if, for example, interest rates have fallen. Exercising the swaption creates an offsetting floating-for-fixed swap. A floating-for-fixed swap combined with a call swaption is called a puttable swap. The swap can be terminated if interest rates have risen—that is, if a higher fixed rate could be received from a new swap. Another example of a swaption application involves the PAC bond considered earlier. The amortizing swap to pay fixed and receive floating could be hedged against the possibility that the rate of amortization is faster than that structured in the swap. A put swaption purchased along with the original fixed-for-floating swap would (partially) hedge this risk. The purchaser would buy a swaption(s) in 21 the amount necessary to partially offset the underlying swap in order to cover the potential additional amortization of principal. An American swaption would be appropriate for this application. The reversible swap described earlier can be synthesized by a fixed-for-floating plain vanilla swap combined with put swaptions for twice the notional principal of the underlying swap. Assuming a swaption has the same notional principal amount as the swap, the first swaption cancels the existing swap and the second creates a floating-for-fixed swap upon maturity, running for the remaining term of the original swap. Extendable Swaps. As the name suggests, an extendable swap contains the option to lengthen its term at the original swap rate. Such a swap simply amounts to an ordinary swap with a swaption expiring at the end of the swap's tenor. Note the difference between an extendable swap and an extension swap. The former gives the holder the option to extend a swap; the latter is a commitment. The same distinction applies to swaptions and forward swaps. Leveraged Buyout Hedging. Another application of swaptions has been in leveraged buyouts, in which a firm's management takes on large amounts of debt to "take a firm private." Lenders, such as commercial banks, often require the firm to hedge its debt, which typically is floating rate. A call swaption with a strike rate at a level the firm could safely meet would accomplish this end. Should the floating rate rise sharply, the swaption would be exercised, converting the remaining floating rate payments to manageable fixed rate payments. However, lenders involved in leveraged buyout financing often prefer to sell caps because a swaption, if exercised, makes its writer a counterparty to a highly leveraged (and often low-rated) firm. A cap writer faces no credit risk from the cap buyer. Synthetic Straight Debt. A final example of swaption usage is in stripping callable debt. This strategy has been popular in the swaption market's brief history. Corporate bonds are frequently issued with options allowing the issuer to refinance the debt issue at a lower coupon if interest rates fall before the bonds mature. The issuer usually cannot exercise the embedded call until after some prespecified date. The callable debt's buyer has effectively written a call option on the price of the bond to the issuer, the firm. If bond prices rise above the strike price of the calls (implying that interest rates have fallen sufficiently), the issuer has the right to call the bonds away after paying the strike price. Because many participants in these markets have believed that the calls attached to these bonds are 22 undervalued, the following arbitrage strategy developed. Firms wanting fixed rate debt issued callable bonds and "stripped" the embedded call options by selling call swaptions, with the net result of creating synthetic noncallable or "straight" bonds at a lower yield than that prevailing on comparable fixed rate bonds. The yield reduction stemmed from selling the undervalued bond market calls at a profit in the swap market.17 As an illustration of the basic strategy, assume the bond is callable at par. That is, if at the call date the relevant interest rate is at or below the original coupon rate, the bond will be called. To strip the call option, the issuer writes a put swaption, which, if exercised, obligates the firm to pay fixed and receive floating on a swap commencing on the bond's first call date and ending at the bond's maturity date. In this example the swaption strike would be set to the bond's coupon rate. If interest rates fall, the put swaption is exercised. In turn, the firm would call its debt and simultaneously issue floating rate debt, whose coupon payments would be met by the floating rate payments coming from the swap counterparty. On balance, the firm would continue to make fixed rate payments, though to the swap counterparty instead of to the bondholders. There are many variations on this strategy. Also, embedded put options can be stripped from bonds in a similar way. 18 The Size of the Swaption Market. As of year-end 1989, $79.7 billion in U.S. dollar and non-U.S. dollar swaptions was outstanding, as measured by the value of the underlying notional principal. 19 The market grew 118 percent compared with the figure for yearend 1988, the first year the survey included swaptions. The size of the caps, collars, and floors market was considerably larger. For year-end 1989, the total U.S. dollar and non-U.S. dollar value of the notional principal for caps, collars, and floors was $457.6 billion, representing a 57 percent increase over the previous year's figure. Non-U.S. D o l l a r D e n o m i n a t e d Interest Rate S w a p s . The interest rate swap market is active worldwide. About one-third of interest rate swaps outstanding involved currencies other than the U.S. dollar. Table 3 reports the latest International Swap Dealers Association survey results for yearend 1989 reflecting swaps involving a single currency. The dollar e q u i v a l e n t of the n o t i o n a l principal outstanding is shown, ranked by currency. The Japanese yen is a distant number two to the U.S. dollar, accounting for 8.5 percent of the market. The British pound and deutsche mark are next in order, with the New Zealand dollar ranking last. ECONOMIC REVIEW, »MARCH/APRIL 1991 Table 3 Interest Rate Swaps as of December 31,1989* Currency U.S. Dollar Equivalent End-User Counterparty (percent) U.S. Dollar Yen Sterling Deutsche Mark Australian Dollar French Franc Canadian Dollar Swiss Franc European Currency Unit Dutch Guilder Hong Kong Dollar Belgian Franc New Zealand Dollar $993,746 $128,022 $100,417 $84,620 $67,599 $42,016 $29,169 $28,605 $18,988 $5,979 $2,149 $835 $444 62.65 52.25 60.13 61.46 84.35 89.92 87.66 55.65 58.51 65.14 60.12 79.16 82.66 ISDA Counterparty (percent) Currency as Percentage of Total ($1,502.6 billion) 37.35 47.75 39.87 38.54 15.65 10.08 12.34 44.35 41.49 34.86 39.88 20.84 17.57 66.14 8.52 6.68 5.63 4.50 2.80 1.94 1.90 1.26 .40 .14 .06 .03 * All dollar amounts are in millions of dollars in U.S. dollar equivalents. Source: International Swap Dealers Association Market Survey. Currency Swaps Basic currency swaps were described earlier in connection with their evolution from parallel loan agreements. The fixed-for-fixed currency swap is the most rudimentary type of swap and is roughly equivalent to a series of forward foreign exchange contracts. For example, a firm could borrow yen at a fixed interest rate and swap its yen-dominated debt for fixed rate dollar-denominated debt. The exchange rate for converting cash flows throughout the life of the swap would be established at the outset. Forward foreign exchange contracts, if they were available in long-dated maturities, could also lock in the exchange rate for future cash flows. All of the features enumerated for interest rate swaps can be applied singly or in combination to swaps involving different currencies. A number of applications of currency swaps are discussed below. Currency C o u p o n Swaps. One of the currency swap's early variants is the currency coupon swap, otherwise known as the cross-coupon swap. This swap is like a plain vanilla swap in which the fixed interest rate is paid in one currency while the floating FEDERAL RESERVE BANK OF ATLANTA rate is paid in another. However, the principal involved in the transaction is usually exchanged as well. An Example. Consider a hypothetical transaction between a U.S. firm, USTech, and a British bank, BritBank. A U.S. swap dealer intermediates the transaction, in part because this institution has the relevant credit information about the swap counterparties that they lack individually. USTech is setting up a British subsidiary and issuing dollar-denominated floating rate bonds tied to LIBOR to finance this operation. USTech wants to hedge itself on two counts, though: first, it wants protection against foreign exchange rate fluctuations because the subsidiary's sales revenue will be in sterling but will be needed to service the dollar-denominated floating rate debt; second, USTech prefers to make fixed rate payments. A currency coupon swap would enable the firm to make sterling-denominated fixed rate payments while receiving dollar-denominated LIBOR, which it would pass to its floating rate bondholders. O n the other hand, BritBank would like sterling-denominated fixed rate cash flows instead of dollar-denominated LIBOR payments from floating rate notes that it holds in a portfolio within its trust department. The bank wants the fixed rate sterling cash flows to extend the duration of its portfolio. 23 As is typical of currency swaps, this one involves exchanges of principal at the beginning and end of the swap. The dealer collects his margin on the fixed rate side of the swap. Like the fixed rate currency swap, the exchange rate for the currency coupon swap is established at the outset and prevails at each of the subsequent settlement dates. Payments at those dates are for the gross amounts of the cash flows, not the net amount as with interest rate swaps, although some swaps stipulate that net amounts be exchanged. European Currency Unit Swaps. The European Currency Unit (ECU) has become an increasingly important "currency" in the Eurobond market. If progress is made toward monetary union of the European Community (EC), the ECU may become European markets' official unit of account. It currently is valued as a weighted average of twelve EC currencies. Although growing rapidly, the number of outstanding ECU-denominated bonds constitutes only about 4 percent of the outstanding amount of publicly issued Eurobonds (Graham Bishop 1991, 72.) Cross-coupon ECU swaps have been used to transform both principal and coupon payments denominated in the ECU into other currencies and vice versa. Terry Shanahan and Jim Durrant (1990) discuss an example in which a U.S. multinational firm needed to finance subsidiaries in France, Belgium, and the Netherlands. The firm borrowed in the Eurobond market by floating ECU-denominated fixed rate debt and converted the issue via a cross-coupon swap into floating rate debt with payments in French francs, Belgian francs, and Dutch guilders. The firm exchanged the principal, consisting of a basket of currencies in proportion to each currency's share in the ECU, raised from the bond buyers. In return, the firm received an equivalent value of the three currencies from the swap counterparty. During the life of this five-year swap, the firm received annual ECU coupon payments from the counterparty, which the firm passed on to the bondholders, and it made annual floating rate payments in guilders and Belgian francs and semiannual floating rate payments in French francs to the counterparty. Upon maturity of the swap, the initial transfer of principal was reversed. The counterparty exchanged ECU principal for repayment in the three currencies from the U.S. firm. In turn, the firm redeemed its bonds with the ECU payment from the counterparty. S w a p p i n g I l l i q u i d B o n d s a n d Private Placements. A major impetus for the growth of currency swaps has been and continues to be the portfolio management of illiquid securities. The earlier discussion of portfolio duration adjustment showed a 24 basic rationale for u s i n g s w a p s , w h i c h h o l d s particularly true i n t h e E u r o b o n d m a r k e t , w h e r e many b o n d s lack the liquidity to b e traded readily. I n add i t i o n , for i n t e r n a t i o n a l l y diversified portfolios, b o n d trading m a y b e desired t o c h a n g e portfolios' e x p o s u r e s to e x c h a n g e rate fluctuations. Currency s w a p s fulfill portfolio m a n a g e r s ' n e e d s for s u c h risk management. Currency (and interest rate) swaps have been especially useful in managing portfolios of privately placed bonds. In terms of a number of costs to the issuer, these bonds are significantly cheaper than publicly placed bonds. Use of privately placed bonds avoids the public disclosure and registration requirements as well as compliance with U.S. accounting regulations; it also minimizes legal costs, reduces underwriting costs, and speeds placement. Yet such securities appeal to a much narrower class of investors because of their illiquidity. In April 1990 the Security and Exchange Commission approved Rule 144A, which greatly simplifies disclosure requirements for private placement issuers (Franklin Chu 1991, 55). Non-U.S. corporations that need to fund their U.S. subsidiaries will find it much easier to raise capital through private placements. The disadvantages of holding these relatively illiquid securities is expected to be lessened both by the use of swaps in portfolio management and by the growth of a secondary market for private placements (Brady 1990, 86). The Size o f the Market. The U.S. dollar is the preeminent currency in the currency swaps market. Table 4 shows that the dollar has a 41 percent share in the currency swaps market, followed by the Japanese yen with a 23 percent share. The Swiss franc, Australian dollar, and German mark occupy the next ranks, with the Hong Kong dollar taking the smallest share of the market for the surveyed currencies. Commodity Swaps Commodity swaps are straightforward extensions of financial swaps, though a number of institutional factors make commodity swapping much riskier than the financial variety. As mentioned earlier, only about $10 billion in notional value has been transacted in this relatively new market. However, commodity prices historically have been much more volatile than financial asset prices, and volatility tends to promote the development and use of hedging instruments. Commodity swaps' volume has reportedly doubled in the past year and is expected to do so E C O N O M I C REVIEW, »MARCH/APRIL 1991 Table 4 Currency Swaps as of December 31, 1989* Currency U.S. Dollar Yen Swiss Franc Australian Dollar Deutsche Mark European Currency Unit Sterling Canadian Dollar Dutch Guilder French Franc New Zealand Dollar Belgian Franc Hong Kong Dollar U.S. Dollar Equivalent End-User Counterparty (percent) ISDA Counterparty (percent) Currency as Percentage of Total ($869.7 billion) $354,166 $201,145 $64,823 $61,768 $53,839 $39,948 $33,466 $32,580 $10,132 $8,435 $5,818 $2,997 $583 72.78 71.83 77.42 70.77 79.93 83.06 74.11 81.72 82.53 88.74 81.90 86.89 90.39 27.22 28.17 22.58 29.23 20.07 16.94 25.89 18.28 17.47 11.26 18.10 13.11 9.61 40.72 23.13 7.45 7.10 6.19 4.59 3.85 3.75 1.17 .97 .67 .34 .07 * All dollar amounts are in millions of dollars in U.S. dollar equivalents. Source: International Swap Dealers Association Market Survey. again in 1991 ( J a n e t Lewis 1990, 207). Another impetus is likely to be the resolution of some regulatory uncertainties, as discussed above. Energy-related commodities hedged via swaps to date include crude oil, heating oil, gasoline, naphtha, natural gas, jet fuel, maritime diesel fuel, and coal. Swap maturities have ranged from one month to five years. A relatively smaller number of swaps have been arranged for gold and for base metals, mainly copper and aluminum, as well as a few in nickel and zinc (Brady 1990, 87). cost of the hedge. In contrast, over-the-counter oil swaps are well suited to hedging intermediate-term risks that cannot be handled by simple positions in futures having relatively short maturity. At the same time, the implication is that swap intermediaries face greater risks because of difficulties they encounter in hedging their swap positions (see Janet Lewis 1990). Oil trading firms have an advantage in acting as dealers because they also carry out transactions in the underlying commodities, giving them additional flexibility in hedging. The most popular commodity swap has been the plain vanilla fixed-for-floating swap, very much akin to the plain vanilla interest rate swap. End users turn to swaps for hedging for essentially the same reasons that they take positions in commodity futures contracts. Their pricing decisions can be based on a known future cost of inputs or revenue from outputs, allowing the appropriate margins to be built in. The end users avail themselves of hedging instruments to transfer the risk to others who specialize in managing that risk.20 Exchange-traded futures and options contracts tend to be liquid for contracts with time to maturity of only a few months. Hedging large positions farther out in time would cause the futures prices to move against the hedger, raising the A commodity swap may be important as a hedge for a firm that is considering financing a project using debt.21 The same is true for interest rate and currency swaps as well, but commodity prices are notoriously volatile, giving lenders ample reason to require a commodity swap hedge.22 In other words, swaps can increase a firm's ability to borrow. A n E x a m p l e . A U.S. producer of oil, TexOil, Inc., sells oil at the spot price but wants to hedge against any large drops in the price of oil that would make production uneconomical. Another counterparty, a charter luxury liner company, LuvBoats Ltd., wants to hedge the proceeds from advanced ticket sales for the coming year. Maritime diesel fuel, purchased at the spot price, is a major FEDERAL RESERVE BANK OF ATLANTA 25 operating cost for LuvBoats's ships. Chart 2 depicts a pair of plain vanilla swaps with a swap dealer intermediating the transaction. As with any kind of swap transaction, a dealer does not necessarily need an offsetting counterparty to enter into a swap with another counterparty. The swap involving LuvBoats Ltd. is actually tied to the price of No. 2 heating oil, which is a more actively traded commodity than maritime diesel. The spread to the counterparty is lower because the swap dealer can better hedge its position, for example by using No. 2 heating oil futures contracts. LuvBoats is willing to bear some basis risk—the risk that maritime diesel and heating oil price movements will be less than perfectly correlated—to avoid paying the dealer a larger spread to index a swap to the price of maritime diesel. TexOil receives a fixed price of $25 per barrel of crude from the swap dealer, while LuvBoats pays a fixed amount of 74 cents per gallon of heating oil. Since the swap's origination, oil and refined product prices have declined, resulting in a $4.52 per barrel net payment to TexOil and a 9 cent per gallon net payment from LuvBoats at the current payment date. Oil swaps can assume more complex forms. For example, they can be combined with currency and interest rate swaps to convert uncertain, dollardenominated spot market purchases of oil into fixed deutsche mark payments. To meet regulatory guidelines, commodity swaps require the inclusion of caps and floors, although these are usually set at prices far from the prevailing commodity price and thus are unlikely to be reached. Caps, collars, floors, participating swaps, swaptions, and many other instruments have been adapted to the commodity markets. Also, oil and other commodity swaps typically reset based on daily averages of spot market prices for the underlying commodity. Averaging tends to make the floating side of a swap have a better correspondence with actual spot market purchases by the counterparties. A swap reset based on a single day's price would be less likely to be representative of such purchases. Equity Swaps Equity swaps are the newest type of swap and are a subset of a new class of instruments known as synthetic equity. 23 Equity swaps generally function as an asset swap that converts the interest flows on a bond portfolio into cash flows linked to a stock index. The stock indexes that have been used include the Standard and Poor's (S&P) 500, the Tokyo Stock Price Index (TOPIX) and Nikkei 225 (Japan), the Chambre des Agents de Change (CAC) 240 (France), the Financial Times Stock Exchange (FTSE) 100 (United Kingdom), the Toronto Stock Exchange (TSE) 300 (Canada), as well as others (see Chart 2 Commodity Swaps Crude Oil Swap No. 2 Heating Oil Swap TexOil receives $25 per barrel and pays the spot price for crude oil. LuvBoats receives the spot price for No. 2 heating oil and pays a fixed price for heating oil. LuvBoats in turn buys maritime diesel at the spot price for diesel. There are 42 gallons in a barrel. 26 E C O N O M I C REVIEW, »MARCH/APRIL 1991 Salomon Brothers, Inc. 1990; Saul Hansell 1990; and Richard Metcalfe 1990, 40). Linking portfolio performance to an index means that dividends are not received as with actual equity ownership; the portfolio tracks only the capital gain component of the underlying stocks. One of the advantages of using a synthetic swap is that transactions costs are mitigated, especially in dealing w i t h less l i q u i d foreign stock markets (Hansell 1990, 56). O n the other hand, such swaps are also illiquid, which implies that their use be predicated on a buy-and-hold strategy for an investment portfolio. Equity swaps have been structured to have one- to five-year tenors and usually have quarterly or semiannual reset dates. The mechanics of an equity swap are similar to the workings of other kinds of swaps. Typically, an investor will swap either fixed or floating rate interest payments for payments indexed to the performance of a stock index such as the S&P 500. If the index appreciates during the interval between settlement dates, the investor receives a payment from the counterparty equal to the rate of appreciation times the swap's notional principal. At the same time, the investor pays, for example, LIBOR less a spread representing the margin to the dealer. Actual settlement would involve only the difference between these bases. In the event the S&P 500 falls, the investor w o u l d pay the rate of depreciation times notional principal and LIBOR less a spread. O f course, the investor is receiving LIBOR or another floating rate from his or her investment portfolio. The net result of the swap is that the portfolio's income behaves like that of an index equity portfolio. A variation of the basic equity swap—the asset allocation swap—links the equity side of the swap to the maximum of two indexes. For example, the swap agreement could stipulate that the counterparty receive the maximum of the rate of appreciation (or pay the maximum rate of depreciation) on the S&P 500 or Nikkei 225 at each settlement date. This kind of swap effectively swaps a portfolio into a foreign stock portfolio or domestic stock portfolio instantly, without transactions costs (apart from those associated with the swap). There are many other possibilities for asset allocation swaps. As another example, the swap could be indexed to the maximum of the S&P 500 or a bond index. Index options could be embedded in the swaps to trade away upside exposure in exchange for downside protection from index moves. FEDERAL RESERVE BANK O F ATLANTA Conclusion Swaps are but one kind of instrument that has been spawned in the profusion of financial innovation during the last two decades.24 In the most general terms, swaps are contracts that transform cash flows from underlying assets or liabilities. They have been designed to incorporate great flexibility in that task and hence are frequently described as instruments that tailor cash flows. This article encompasses the four basic types of swap: interest rate, currency, commodity, and equity. Each group in turn branches into a variety of forms that can accommodate virtually any application. However, novelty does not guarantee success. The most successful swaps have frequently been the simplest, plain vanilla variety. Swaps integrate credit markets. By the nature of their function, swaps can link money markets (shortterm financing) and capital markets (long-term financing). Swaps also play a significant role in the so-called globalization of financial markets because they obviate the need for many investors to carry out transactions in underlying foreign securities, thereby contributing to the international diversification of portfolios. International arbitrage of securities and swaps markets is left to those participants w h o have the lowest transactions costs, increasing global market efficiency. Swaps are an important tool for simplifying financial transactions that cross national borders. At the same time, they pose potential risks to the stability of financial markets. Recent concern about the strength of both banks and investment banks has focused the attention of swap market participants on counterparties' creditworthiness, upon which the financial obligations contracted through a swap agreement depend.2"5 However, part of the reason that swaps evolved was to reduce the credit exposure of counterparties involved in similar financial arrangements. Swaps generally confine credit risk to exposure to the net difference in cash flows, not the gross amounts or exposure of underlying principal, and defaults have been rare occurrences.26 The implementation of the Basle Agreement in 1992 will establish more uniform capital standards for the world's commercial banks and should help to further reduce credit risks in the swap market. 27 27 Notes 1. Shirreff (1989) reports that swaps with thirty-year maturities or "tenors" have been arranged. Such long-lived swaps typically involved counterparties with top credit ratings or relied on third-party credit enhancements. 2. Wall and Pringle (1988) discuss the plain vanilla swap in detail and consider the reasons for using swaps. 3. A forward contract commits the buyer to purchase the underlying asset at a prespecified price (the forward price) u p o n maturity of the contract. A call option gives the buyer the right, but not the obligation, to purchase an underlying asset at a prespecified price on or sometime before the maturity date of the option. The put gives the corresponding right to sell at a prespecified price. These instruments will be described further at appropriate places in the exposition. may also be used for this purpose. Brown and Smith discuss many subtleties of these strategies. 18. Krzyzak (1988, 29; 1989a, 9) reports that the embedded calls were overvalued and that call monetization was used to undo the expensive call. In this case, call monetization would not be an arbitrage. 19. Chew (1991) discusses recent activity in the non-U.S. dollar swaptions markets, particularly deutsche mark instruments. 20. This point of view is not universal or uncontroversial. Williams (1986) argues that risk aversion has nothing to d o with the use of futures. Rather, futures contracts reduce transactions costs in dealing with underlying commodities. His model assumes that all futures market participants are risk neutral. 4. See McNulty and Stieber (1991) for a more detailed account. 5. See Henderson (1990) for details about the CFTC's criteria. 21. Also, Smith, Smithson, and Wilford (1990) discuss a conflict between stockholders and bondholders of a corporation, known as the underinvestment problem, that swaps can mitigate. 6. The growth may be exaggerated by these figures because the number of survey respondents, not reported in the tables, has also been increasing. However, the ISDA points out that the major swap dealers have consistently participated in their surveys. 22. See Spraos (1990) for a case study of a complex copper swap required in part for this reason. 7. See Kuprianov (1986) for a background discussion of Eurodollar futures and LIBOR. 8. See McNulty and Stieber (1991, 100-101) for information about the Eleventh District cost-of-funds rate. 9. A basis point is a hundredth of a percentage point. 10. Ramaswamy and Sundaresan (1986) analyze floating rate securities and discuss the characteristics of such securities. 11. See Macfarlane, Showers, and Ross (1991) for a discussion of nonstandard swap terms. This article gives a detailed account of swap terminology and conventions. 12. Ordinarily, comparisons of yields along the yield curve are made using instruments of comparable default risk. Yield curve swaps exchange floating payments on debt bearing different default risks. Because the underlying three-month Eurodollar time deposit is default risky, LIBOR is greater than the riskless three-month Treasury bill yield. The swap therefore exchanges credit spreads as well as yield curve spreads. 13. See G o o d m a n (1991, 160-61) for details about this strategy. 14. See Abken (1989) for an introduction to these instruments. 15. See Bodie, Kane, and Marcus (1989) for an introduction to duration analysis. 16. This example is cited by Bhattacharya (1990, 56). 17. Goodman (1991) and Brown and Smith (1990) discuss call monetization using several strategies. Forward swaps 28 23. Other examples of synthetic equity include over-thecounter equity options, public warrant issues, and bonds containing equity options. See Hansell (1990). Index-linked certificates of deposit were a retail form of synthetic equity offered by a number of commercial banks and savings and loans in 1987. 24. See Finnerty (1990) for a comprehensive survey of financial innovations since the 1970s. 25. Krzyzak (1990) and Brady (1991) describe the concerns and difficulties experienced by low-rated swap dealers in dealing with higher-rated counterparties. See Abken (1991) for a model of swap valuation in which swaps are subject to default by the participating counterparties. 26. Aggarwal (1991) reports several sources giving a figure of $35 million in write-offs resulting from swap defaults as of year-end 1988. The collapse of Drexel, Burnham, Lambert in 1989 brought with it potential defaults on its swap book. Most o f these swaps were closed out or rearranged with other swap dealers, avoiding defaults that would have shaken the swaps market. See Perry (1990) for an account of the Drexel collapse and its aftermath on the swaps market. Evans (1991) reports that U.S. and foreign banks face potential defaults of u p to SI billion because of to a British court ruling that nullifies swap contracts with about 80 British municipalities. 27. See Wall, Pringle, and McNulty (1990) for a discussion of the Basle Agreement and its treatment of swaps under the new capital standards. Levis and Suchar (1990) give further discussion and detailed examples. E C O N O M I C REVIEW, MARCH/APRIL 1991 References Abken, Peter A. "Interest-Rate Caps, Collars, and Floors." Federal Reserve Bank of Atlanta Economic Review 74 (November/December 1989): 2-24. . "Valuation of Default-Risky Interest-Rate Swaps." Federal Reserve Bank of Atlanta working paper, forthcoming, 1991. Aggarwal, Raj. "Assessing Default Risk in Interest Rate Swaps." In Interest Rate Swaps, edited by Carl R. Beidleman, 430-48. H o m e w o o d , 111.: Business O n e Irwin, 1991. Bhattacharya, Anand K. "Synthetic Asset Swaps." Journal of Portfolio Management 17 (Fall 1990): 56-64. , and J o h n Breit. "Customized Interest-Rate Risk Agreements and Their Applications." In The Handbook of Fixed Income Securities, 3d ed., edited by Frank J. Fabozzi, 1157-89- Homewood, 111.: Business O n e Irwin, 1991. B i s h o p , G r a h a m . "ECU Bonds: Pioneer o f Currency Union." Euromoney (January 1991): 7Iff. Bodie, Zvi, Alex Kane, and Alan J. Marcus. Investments. Homewood, 111.: Richard D. Irwin, Inc., 1989. Brady, Simon. "How to Tailor Your Assets." Euromoney (April 1990): 83-89. . "Time Runs O u t for Low-Rated Swappers." Euromoney (February 1991): 9-10. Brown, Jeffry P. "Variations to Basic Swaps." In Interest Rate Swaps, edited by Carl R. Beidleman, 114-29. Homewood, 111.: Business O n e Irwin, 1991Brown, Keith, and D o n a l d J. Smith. "Forward Swaps, Swap Options, and the Management of Callable Debt." Journal of Applied Corporate Finance 2 (Winter 1990): 59-71. Chew, Lillian. "Strip Mining." Risk 4 (February 1991): 20ff. Chu, Franklin J. "The U.S. Private Market for Foreign Securities." Bankers Magazine (January/February 1991): 55-60. Evans, John. "British Court Rules Swaps by Municipalities Illegal." American Banker, January 25, 1991, 13. Finnerty, John D. "Financial Engineering in Corporate Finance: An Overview." In The Handbook of Financial Engineering, edited by Clifford W. Smith and Charles W. Smithson, 69-108. Grand Rapids, Mich.: Harper Business, 1990. Goodman, Laurie S. "Capital Market Applications of Interest Rate Swaps." In Interest Rate Swaps, edited by Carl R. Beidleman, 147-74. Homewood, 111.: Business O n e Irwin, 1991. Hansell, Saul. "Is the World Ready for Synthetic Equity?" Institutional Investor (August 1990): 54-61. H e n d e r s o n , Schuyler K. "A Legal Eye o n H e d g i n g ' s Newest Club." Euromoney (May 1990): 95-96. Krzyzak, Krystyna. "Don't Take Swaps at Face Value." Risk 1 (November 1988): 26-31. . "Swaptions Deciphered." Risk 2 (February 1989a): 9-17. . "From Basis Points to Barrels." Risk 2 (May 1989b): . " S w a p s Survey: A r o u n d the H o u s e s . " Risk 3 (September 1990): 51-57. Kuprianov, Anatoli. "Short-Term Interest Rate Futures." Federal Reserve Bank of Richmond Economic Review (September/October 1986): 12-26. Levis, Mario, and Victor Suchar. "Basle Basics." Risk 3 (April 1990): 38-39. Lewis, Janet. "Oil Price Jitters? Try Energy Swaps." Institutional Investor (December 1990): 206-8. Lewis, Julian. "The Bandwagon Starts to Roll." Euromoney (May 1990): 87-94. Macfarlane, John, Janet Showers, and Daniel Ross. "The Interest-Rate Swap Market: Yield Mathematics, Terminology, and Conventions." In Interest Rate Swaps, edited by Carl R. Beidleman, 233-65. H o m e w o o d , 111.: Business O n e Irwin, 1991. McNulty, James E., and Sharon L. Stieber. "The Development and Standardization of the Swap Market." In Interest Rate Swaps, edited by Carl R. Beidleman, 97-113Homewood, 111.: Business O n e Irwin, 1991. Metcalfe, Richard. "Out of the Shadows." Risk 3 (October 1990): 40-42. Perry, Phillip M. "Drexel Redux? Credit Quality Is a Hot Topic." Corporate Risk Management (May/June 1990): 27-29. Ramaswamy, Krishna, and Suresh M. Sundaresan. "The Valuation of Floating-Rate Instruments: Theory and Evidence." Journal of Financial Economics 17 (December 1986): 251-72. Salomon Brothers, Inc. "Equity-Linked Index Swaps." Sales brochure, 1990. Shanahan, Terry, and Jim Durrant. "Driving Factors." Risk 10 (November 1990): I4ff. Shirreff, David. "Where Others Fear to Tread." Risk 8 (September 1989): 11-16. Smith, Clifford W., Charles W. Smithson, and Lee Macdonald Wakeman. "The Evolving Market for Swaps." In The Handbook of Financial Engineering, edited by Clifford W. Smith and Charles W. Smithson, 191-211. Grand Rapids, Mich.: Harper Business, 1990. Smith, Clifford W., Charles W. Smithson, and D. Sykes Wilford. "Financial Engineering: Why Hedge?" In The Handbook of Financial Engineering, edited by Clifford W. Smith and Charles W. Smithson, 126-37. Grand Rapids, Mich.: Harper Business, 1990. Spraos, Paul B. "The Anatomy of a Copper Swap." Corporate Risk Management 2 (January/February 1990): 8, 10. Tompkins, Robert. "Behind the Mirror." Risk 2 (February 1989): 17-23Wall, Larry D., and John J. Pringle. "Interest Rate Swaps: A Review of the Issues." Federal Reserve Bank of Atlanta Economic Review 73 (November/December 1988): 22-37. 8-12. Wall, Larry D., J o h n J. Pringle, and James E. McNulty. "Capital Requirements for Interest-Rate and Foreign-Exchange Hedges." Federal Reserve Bank of Atlanta Economic Review 75 (May/June 1990): 14-27. . "Copper-Bottomed Hedge." Risk 2 (September 1989c): 35-39. Williams, Jeffrey. The Economic Function of Futures Markets. New York: Cambridge University Press, 1986. FEDERAL RESERVE BANK OF ATLANTA 29 FYI: The Impact of Private-Sector Defense Cuts on Regions of the United States David D. Whitehead n light of the reduced threat of a major conflict with the Soviet Union, both Congress and the administration last year acknowledged the opportunity to cut national defense expenditures. Because defense s p e n d i n g constituted 24 percent of all government outlays and slightly more than 5 percent of the gross national product in 1990, even small cuts could have significant negative effects on employment of workers producing military goods and support services as well as the locales in which they work. Assessing the potential effects of the defense cuts proposed in the 1992 Bush administration budget is the purpose of this article. The focus is on the short-term, initial consequences for employment in various regions of the country. Longer-term employment results of eliminating defense-related jobs and redistributing government defense savings throughout the economy are beyond the scope of this study.1 I Background on Proposed Defense Cuts In the early 1970s, following the Vietnam War, real (inflation-adjusted) defense outlays declined rather rapidly and then stabilized at middecade, remaining fairly constant from 1975 through 1980. The end of The author is the research officer in charge of the regional section of the Atlanta Fed's research department. 30 the Carter administration and the beginning of Reagan's presidency saw double-digit nominal percent increases in defense outlays, resulting in an increase of real outlays. That increase continued throughout most of the Reagan years, reaching a plateau at the $240 billion to $250 billion range (in 1982 dollars) for fiscal year (FY) 1986 through FY 1990. It is from this level that the administration and Congress are currently proposing cuts.2 Table 1 (page 35) shows the budgeted cuts in nominal defense outlays for 1991 through 1995. To demonstrate the relative magnitude of the cuts, the 1992 budget presents them as a reduction from a nominal spending level that would have maintained 1990's real defense spending level, providing a baseline. For instance, actual defense expenditures in 1990 amounted to $299 billion. Those same purchases in 1991 would have cost $307 billion—an $8 billion difference representing an increase of 2.6 percent. The percentage implied by the budget document is the 1991 rate of inflation in defense item prices. Expenditures totaling less than $307 billion would reflect a reduction in the real level of expenditures. The budget calls for defense outlays of $299 billion in 1991—the same dollar figure as in 1990. However, because of inflation this amount represents a reduction of 2.6 percent in constant dollars. For 1992 the baseline for constant real defense expenditures converts to $318 billion in 1992 dollars, some $19 billion, or 6.4 percent, more than the 1990 figure. The 1992 budget calls for defense outlays of $295 billion, a cumulative decline of 7.2 percent ECONOMIC REVIEW, »MARCH/APRIL 1991 from the $318 billion that would equal the 1990 level of defense purchases. This amount represents a 4.6 percent reduction, in real terms, from 1991. By 1995 the cumulative cuts from the 1990 baseline will amount to a 17.4 percent reduction in real defense outlays—$182 billion, with almost one-third called for in 1995. A reduction in real outlays implies a likely nearterm loss of employment for those producing the eliminated goods or services. Cutting defense spending, no matter how it is accomplished, will affect economic activity throughout the nation, but the impact will be felt most acutely by regions and local communities that have been highly dependent on Defense Department expenditures. Regional Impact of Defense Cuts As acknowledged earlier, estimating the overall consequences of defense spending reductions would require a complex analysis outside the scope of this article. The initial effect on employment in each state or region targeted for Defense Department spending cuts can more reasonably be approximated. However, the extent to which the current employment levels will be affected depends on decisions about specific base closings, realignments, and overall reductions in troop levels. Direct employment generated by defense spending can be divided into three general categories: military personnel, Department of Defense civilian personnel, and civilian workers employed as a direct result of defense purchases from private-sector firms. The number of workers in the first two categories is easily measured and their various locations identified. The first two columns of Table 2 (pages 36-37) indicate active duty military and Department of Defense civilian personnel in 1988 for each state and census region.3 As discussed in the box on page 32, anticipated base closings and realignments are expected to have little effect on employment. Because overall reductions in troop levels depend on a number of factors, any projection about the effects of such reductions—accomplished primarily through attrition, curtailed recruitment, and distribution of these cuts according to a set of unknown military priorities—would be highly speculative at this point. No direct measure of the third category—privatesector defense-related employment—is available, but estimates can be based on Department of Defense "prime contract awards." These contracts are a measure of certain expenditures that, like others, generate income and employment. If one accepts the FEDERAL RESERVE BANK OF ATLANTA assumption that related employment is distributed in the same way as prime contract dollars, the relationship between these expenditures, which are reported on a state-by-state basis, and resulting employment can be roughly approximated. It is true that this assumption injects some error into the estimates because various types of contracts are not distributed uniformly across states. In addition, some of the work associated with prime contract awards may be subcontracted to out-of-state firms. Although these points are troublesome for estimating local effects, numerous studies confirm that contract awards are sufficiently detailed to identify the location of most production and that subcontract work flows in both directions—into and out of a state—largely offsetting any impact. Thus, estimating the geographic distribution o f e m p l o y m e n t o n the basis o f prime contract awards appears to be a reasonable approach. 4 The third column of Table 2 shows estimates of the level of private-sector defense-related employment based on the average value of contracts awarded to private firms in each state during the period 1986 to 1988. The three-year average is used because contracts awarded in one year are usually not completed in a single twelve-month interval and therefore do not represent that year's actual outlays. Each state's percentage of total prime contract awards in the nation is calculated and then used to allocate to each state its share of the estimated 3,400,000 jobs supported annually by prime contracts awarded nationwide. 5 For example, Connecticut received an average of 3.9 percent of all prime contract awards in the nation during FY 1986-88—134,018 of the nation's private-sector defense-related jobs. In terms of major census regions, 1988 estimates place the highest concentrations of privatesector defense workers in the South (32 percent) and the West (30 percent). Far exceeding all other states, California contains more than 19 percent of the nation's total. More important than these workers' location is each area's dependence on privatesector defense employment. The sixth column of Table 2 shows private-sector defense-related employment as a percentage of total employment in census regions, subregions, and states. O f major census regions, the West and Northeast are most economically dependent on these jobs. In terms of subregions, New England has the highest proportion of private-sector defense jobs—5.9 percent of the area's total e m p l o y m e n t . The Pacific area ranks second highest in terms of shares of these workers, and the South Atlantic and West North Central subregions third. Least d e p e n d e n t o n private-sector defense-related employment is the 31 Direct Employment Impact of Base Closings nation. While the Northeast, North Central, and West census regions will all lose employment under the 1988 legislation, the South will gain, but only marginally. In January 1990, just as the first round of base closings began, the Department of Defense proposed a second round of base closings involving thirty-five bases and the realignment of forces at several dozen more. These recommendations have not been acted upon. Congress instead adopted legislation establishing a commission that may accept or amend the Department of Defense's base-closing recommendations before passing them on to the president. The president then has fifteen days to pass the recommendations to Congress for approval or rejection. He may recommend either approval or rejection of the list. Legislation passed in 1988 cleared the way to close or realign military bases in various areas of the nation, thereby directly affecting military and civilian base workers. The legislation targeted eighty-six bases for closing, five for partial closing, and fifty-four for realignment involving either expanding or downgrading military and civilian employment. In terms of net employment, 12,900 military and 7,742 civilian workers are expected to be affected nationwide. Total numbers of active duty military or Department of Defense civilian workers are less than 1 percent. The table below lists the sixty-six bases slated to lose or gain 100 or more employees. With a net loss of more than 11,000 and 17,000 jobs, respectively, the West and, more specifically, California will be hardest hit. In 1988 California housed more than 204,000 active duty military personnel and more than 130,000 Department of Defense civilian workers, significantly more than any other state. The state's loss equals 5 percent of those employees in its borders, compared with a loss of less than a 1 percent nationwide. However, these figures represent a very small part of the total employment picture—one-tenth of one percent in California and less than two one-hundredths of one percent in the This new process was set in motion on April 12, 1991, when Defense Secretary Cheney recommended closing thirty-one major military bases and twelve minor installations as well as realignment of twenty-eight others. Because these most recently proposed closings and realignments are still speculative at this point, even a ball-park estimate of potential losses or gains they may create in direct military or Department of Defense civilian employment is virtually impossible. Bases Gaining or Losing 100 or More Employees under the 1988 Legislation, Aggregated by State and Census Region Region Number of Bases by Service Net Employment Change Military Net Employment Change Civilian, P O D 1 Total Net Employment Change -761 1,633 872 NORTHEAST Massachusetts 2 Army New Hampshire 1 Air Force -2,250 -400 -2,650 N e w Jersey 3 Army -3,199 -1,525 -4,724 N e w York 1 Air Force 2 Navy 101 3 104 Pennsylvania 2 Army 1 Navy -442 812 370 -6,551 523 -6,028 -3,116 -2,637 -5,753 600 319 919 1 100 101 300 52 352 -2,215 -2,166 -4,381 TOTAL NORTHEAST NORTH CENTRAL Illinois 1 Air Force 2 Army Indiana Michigan Missouri TOTAL N O R T H CENTRAL 32 2 Army 1 Army 1 Army ECONOMIC REVIEW, MARCH/APRIL 1991 Number of Bases by Service Net Employment Change Military Net Employment Change Civilian, DOD 1 Total Net Employment Change SOUTH Kentucky 2 Army 265 -1,056 -791 Louisiana 1 Navy -338 -24 -362 Maryland 2 Army -88 -181 -269 Mississippi 1 Air Force 114 77 191 South Carolina 1 Army 661 126 787 Texas 3 Air Force 1 Army 2 Navy 1,683 380 2,063 Virginia 3 Army 146 -986 -840 2,443 1,664 779 TOTAL S O U T H WEST Arizona 1 Air Force 2 Army -783 -62 -845 California 6 Air Force 3 Navy 2 Army -12,817 -4,507 -17,324 1 Air Force 545 -5 540 2,267 30 2,297 1,859 90 1,949 1,448 195 1,643 -3 -161 -164 -26 -159 -185 894 144 1,038 -6,616 -4,435 -11,051 -12,939 -7,742 -20,681 Colorado 2 Army Hawaii Idaho New Mexico Oregon Utah Washington 1 Navy 1 Air Force 2 Air Force 1 Army 2 Army 2 Air Force 2 Navy 1 Army TOTAL WEST TOTAL U N I T E D STATES 1 U.S. Department of Defense. Source: Calculated by the Federal Reserve Bank of Atlanta from data in Congressional Quarterly Almanac 44 (1988): 442-43. East North Central subregion, where these jobs account for only 1.5 percent of total employment. At the state level, those with a high dependence on private-sector defense employment will, of course, be affected by defense spending cuts. While private-sector defense-related work accounts for 3-0 percent of employment nationally, such jobs make FEDERAL RESERVE BANK O F ATLANTA up more than 4 percent of total employment in ten states and the District of Columbia. The latter leads with 10.6 percent of its workers in such jobs. Connecticut and Massachusetts show the second and third greatest dependencies on defense expenditures to support private-sector jobs, contributing to New England's position as the most dependent subregion. 33 Statistics on the geographic concentration of workers and reliance on defense spending for private-sector employment reveal little about an area's ability to absorb employment dislocations that may result from defense spending cuts. The effects of these reductions could be mitigated by the vitality and diversity of local economic activity. Therefore, to identify areas most likely to suffer relatively larger and longer employment dislocations, one must determine the time it will take to integrate those affected back into the economy as productively employed workers. Two major factors are the rate at which new jobs are being created in an area and the transferability of displaced workers' skills to nondefense business activity. The absorption of these workers will also depend on demand in occupations producing civilian goods and services. The number of dislocated workers relative to employment growth in an area provides a standardized measure of how quickly a state or region may be able to reabsorb these individuals into the ranks of the employed. Additionally, assessing the types of activities involved in defense-related jobs may help evaluate, albeit roughly, the versatility of a worker's abilities. For instance, a steel worker and an office worker will have developed different skills, some perhaps more in demand by other employment sectors. An absorption rate is simply the number of potentially dislocated workers divided into the net amount of new employment in an area during a given time period (hereafter used to refer to one year). In direct relationship, slower total employment growth results in a lower absorption rate for the same absolute number of dislocated workers. The higher the annual absorption rate, the lower the number of displaced workers relative to employment growth and the easier it should be for these workers to find alternative jobs in a given area. The first and second columns of Table 3 (pages 38-39) show the number of private-sector defense-related workers that could be dislocated annually in each state and region if these cuts are spread proportionately over all regions. The third c o l u m n shows the state's annual change in total employment for 1988-89, a recent and not unusual year for economic growth. The fourth and fifth columns show the number of times in a given year that the assumed dislocated workers could be absorbed into the work force based on the area's employment growth from 1988 to 1989- For example, a 3 percent annual cut in real defense spending would displace an estimated 102,000 workers nationwide in one year. Total annual average employment in the nation grew by 2.3 million from 1988 to 1989- Therefore, the estimated number of workers displaced in 1989 34 could have been absorbed almost twenty-three times. With a 6 percent cut, the rate of absorption w o u l d fall by 50 percent to slightly higher than eleven. Table 3 suggests that the Northeast will have more difficulty than the rest of the nation dealing with the employment impact of defense budget cuts. New England's absorption rate of 2.9 percent is significantly smaller than all other subregions', as expected in light of the area's dependence on defense spending. The West North Central and Middle Atlantic subregions also show absorption rates significantly lower than the nation's. If relative absorption rates hold, the two areas least likely to be negatively affected by defense cuts are the East North Central and East South Central, each showing absorption rates two to three times the national average. The five states most likely to suffer severely because of defense outlay cuts are Connecticut, Massachusetts, Virginia, Missouri, and Colorado. For the times measured, these states typically have had a larger-than-average share of employees tied to defense and a comparatively slow rate of overall employment growth. In Connecticut 8 percent of total employment is estimated to be civilian defense-related. In Massachusetts the figure is 7 percent, and in Virginia, 6.7 percent. Missouri, with 6 percent of its total e m p l o y m e n t private-sector defense-related, ranks fourth, and Colorado ranks tenth with 4 percent. Texas, New York, and Florida also have relatively large shares of total defense purchases—4 to 7 percent—but their private-sector defense employment as a percent of total employment is smaller than in the states identified above. Across the board, defense cuts even at the 6 percent level will have only marginal impact on most local economies, while a few states will be rather hard hit. The other significant indicator of the difficulty an area will have in assimilating dislocated workers back into the labor force lies in the types of jobs likely to be affected. Defense-related civilian employment nationwide tends to be concentrated in durable manufacturing activities. Based on prime contract awards, Table 4 (page 40) shows the percentage of total Department of Defense nonwage and salary expenditures flowing to the six industries receiving the greatest shares of these monies. Aircraft and parts, ordnance and accessories, and other transportation equipment (basically shipbuilding and repair)—heavy manufacturing activities—account for approximately 35 percent of all Defense Department direct purchases from the private sector. Radio, television, and communications equipment, classified as ECONOMIC REVIEW, MARCH/APRIL 1991 light manufacturing, represents almost 20 percent of total private-sector direct defense expenditures, while business services and new construction represent 13 percent and 5 percent, respectively. The portions of these various industries' expenditures differ significantly among the four main census regions. For instance, business services and radio, television, and communications equipment constitute the largest shares of private-sector direct defense spending in the South. This configuration is in sharp contrast to the North Central region, which receives the major share of spending for aircraft and parts and ordnance and accessories. The general level of local economic activity certainly contributes to how long workers remain unemployed if defense-related jobs are discontinued. As observed earlier, each worker's versatility of skills is also a factor. Those who have experience easily transferable to private-sector production are more likely to shift to producing private-sector goods at the same firm or at another. For example, nonmanufacturing constitutes a larger proportion of private-sector jobs, so nonmanufacturing skills tend to be easier to transfer to private-sector employment.6 Workers with skills unique to the defense sector or little used in privatesector operations will find it harder to find alternative employment. Examining the composition of civilian defenserelated employment in each of the five states that defense cuts are likely to affect the most can provide clues about the degree of difficulty workers may face in seeking new positions. For these states, which receive the highest portion of prime contract awards, Table 5 (page 40) indicates the percentages of total expenditures in the six industries. The figures suggest the share of defense-related civilian employment associated with each industry in these states, even though the relationship between expenditures and number of workers differs by industry. For instance, in Virginia defense dollars tend to flow more toward—and, it is assumed, generate more employment in—business services than the other categories. This concentration of defense dollars suggests, of course, that if defense cuts are proportional across industry lines Virginia will feel the strongest employment influences in this sector. Along with Virginia, Colorado derives more service-sector employment from defense expenditures than do the three other states included in Table 5. Because the skills associated with service jobs can usually be transferred, the dislocation shocks in these two states will probably be less severe than in the other three states. Connecticut and Missouri are the states most dependent on manufacturing-related Table 1 National Defense Outlays, FY 1991-95 (billions of dollars) 1991 1992 1993 1994 1995 July 1990 Gramm-Rudman-Hollings Baseline (no real growth) 307 318 327 342 350 February 1991 Budget Document 1 299 295 292 287 289 Cumulative Real Cut from Baseline (percent) 2.6 7.2 10.7 16.1 17.4 Annual Real Cuts (percent) 2.6 4.6 3.5 5.4 1.3 1 Excludes effects of Desert Shield and Desert Storm. Source: Budget of the United States Government, Fiscal Year 1992, Part Two, 184. FEDERAL RESERVE BANK OF ATLANTA 35 Table 2 Defense-Related Employment by Census Region and State (thousands) Region NORTHEAST New England Connecticut Maine Massachusetts New Hampshire Rhode Island Vermont Total New England Middle Atlantic New Jersey New York Pennsylvania Total Middle Atlantic 1988 Active Duty1 1988 D O D Civilian Private Sector2 455.0 6,697 84.6 236.4 97.3 128.3 284.1 156.0 3,824 8,168 5,471 97.1 418.3 568.4 17,463 1.8 2.4 810.2 1,023.4 24,160 3.4 42.1 54.3 47.2 127.9 24.5 101.0 5,414 2,658 4,198 5,002 2,460 0.8 15.0 11.9 33.3 3.3 84.5 296.0 1.5 14.9 36.0 59.1 147.4 0.6 33.8 25.1 19.4 52.6 TOTAL NORTHEAST 82.3 130.9 NORTH CENTRAL East North Central Illinois Indiana Michigan Ohio Wisconsin 37.9 5.8 8.9 11.5 0.9 21.0 16.8 75.1 68.0 172.7 28.7 445.5 19,732 16.8 1,433 Total East North Central 65.0 West North Central Iowa Kansas Minnesota Missouri Nebraska North Dakota South Dakota 0.4 23.2 0.9 15.2 13.1 10.9 6.7 20.1 70.4 38.8 269.8 10.1 379.0 135.4 123.3 565.8 1.8 5.5 33.6 145.2 76.3 Total West North Central TOTAL NORTH CENTRAL SOUTH South Atlantic Delaware District of Columbia Florida Georgia 4.6 12.6 77.0 62.1 7.9 2.9 7.0 391.9 0.1 29.3 214.3 12.4 11.3 3.1 6.1 53.0 Private-Sector DefenseRelated as a Percentage of Total Employment3 1,687 575 3,052 583 509 291 5.0 10.7 11.7 1.5 4.3 28.3 Total 1988 Employment 146.4 33.0 234.7 17.8 19.3 3.8 134.0 7.4 5.5 8.7 3.9 3.7 18.6 Total Defense-Related Employment 6.8 3.0 4.0 1.9 1.5 17.0 32.0 37.8 6.2 4.3 1.9 66.0 63.0 182.7 23.3 17.1 1,221 2,233 2,446 790 316 343 2.1 2.2 1.1 5.9 2.2 2.9 2.0 1.1 2.6 1.0 1.5 1.0 2.9 2.6 6.0 0.8 1.4 8,782 0.6 3.1 824.5 28,514 2.0 11.9 63.2 254.2 176.2 338 316 5,777 2,970 1.6 10.6 2.5 2.6 Maryland North Carolina South Carolina Virginia West Virginia Total South Atlantic East South Central Alabama Kentucky Mississippi Tennessee Total East South Central West South Central Arkansas Louisiana Oklahoma Texas 34.3 89.9 41.8 42.0 15.6 20.3 100.8 106.2 1.7 423.5 117.2 30.0 13.9 203.3 2.7 193.5 135.5 76.0 410.3 4.8 274.4 2,342 3,199 1,586 2,980 665 0.4 627.7 1,325.6 20,173 3.1 20.6 38.4 14.4 25.6 13.5 42.4 1,738 1,578 1,045 2,203 2.4 0.7 4.4 1.3 0.4 10.0 83.4 9.1 22.8 34.0 126.4 5.0 0.9 0.9 6.8 11.1 7.6 45.8 28.3 88.6 63.1 71.3 45.9 57.8 127.7 268.9 6,564 1.9 19.6 34.3 1,024 1,707 1,421 7,744 1.9 2.0 246.5 33.7 66.2 74.0 435.2 11.2 192.3 5.0 9.1 23.8 62.3 100.2 316.6 609.1 11,896 2.7 699.2 432.4 1,072.0 2,203.6 38,633 WEST Mountain Arizona Colorado Idaho Montana Nevada New Mexico Utah Wyoming 2.8 24.8 38.3 5.6 4.2 9.2 15.2 5.8 3.9 10.4 13.7 1.3 1.3 2.0 108.9 117.3 8.4 7.5 16.3 40.9 51.6 Total Mountain 107.0 61.6 73.7 65.3 1.5 2.0 5.1 15.7 24.1 1.7 189.1 6.8 357.7 1,556 1,598 451 375 554 634 722 223 6,113 Pacific Alaska California Hawaii Oregon Washington 22.5 204.8 44.8 0.7 42.0 Total Pacific 5.0 131.2 20.6 3.2 28.7 13.0 657.9 13.5 7.6 70.7 314.0 40.5 993.9 78.1 11.5 141.4 227 13,385 502 1,343 2,172 188.7 762.7 1,265.4 17,629 3.3 4.3 421.0 250.3 951.8 1,623.1 23,742 4.0 1,337.9 936.9 3,399.8 5,674.6 115,049 3.0 Total West South Central TOTAL SOUTH TOTAL WEST TOTAL UNITED STATES 1 10.0 21.7 1.2 16.2 1.1 3.2 4.7 4.1 0.3 0.5 0.9 2.5 3.3 0.8 3.1 5.7 4.9 2.7 0.6 Data for 1988 are the latest for which all series necessary for estimating private-sector defense-related employment are available. Based on the average value of prime contract awards during 1986-88. Percentages are based on the actual numbers, not the rounded numbers shown. Sources: Data for columns 1-4 are from U.S. Department of Defense, Directorate for Information Operations and Reports, Atlas/Data Abstract for United States and Selected Areas, Fiscal Year 1988, Office of the Secretary of Defense Washington Headquarters Services. For column 5, data are from U.S. Department of Labor, Bureau of Labor Statistics, Employment and Earnings, May 1990. Column 6 percentages were calculated by the Federal Reserve Bank of Atlanta. 2 3 Table 3 Potential Absorption of Displaced Private-Sector Defense-Related Workers by Census Region and State Region NORTHEAST New England Connecticut Maine Massachusetts New Hampshire Rhode Island Vermont Total New England Middle Atlantic New Jersey New York Pennsylvania Total Middle Atlantic TOTAL NORTHEAST N O R T H CENTRAL East North Central Illinois Indiana Michigan Ohio Wisconsin Total East North Central West North Central Iowa Kansas Minnesota Missouri Nebraska North Dakota South Dakota Total West North Central TOTAL N O R T H CENTRAL SOUTH South Atlantic Delaware District of Columbia Number of Employees Displaced by 3 Percent Cut (thousands) Number of Employees Displaced by 6 Percent Cut (thousands) Change in Annual Average Total Employment 1988-89' (thousands) 4.0 8.0 10 0.1 6.4 0.4 0.3 1.0 12.9 0.7 0.7 16 Absorption Rate 1989 2 3 Percent Cut 2.5 31.8 6 Percent Cut 1.2 15.9 0.2 0.1 7 -5 5 18.9 0.0 53.6 34 2.9 26.8 1.4 2 71 122 0.8 10.0 1 9.4 0.0 0.1 0.2 23.5 2.5 7.1 2.9 12.6 5.1 14.2 5.8 41.8 0.4 5.0 20.9 25.1 195 15.5 7.8 24.3 48.6 229 9.4 4.7 1.3 2.5 3.3 242 87 68 117 37 95.9 26.7 24.0 15.2 25.2 16.8 11.8 8.9 17.8 551 191.8 53.5 48.0 30.5 50.3 62.1 0.4 0.1 15 33.6 1.1 1.8 2.2 1.6 1.4 3.8 0.7 2.8 7.7 1.5 11 10.2 4.5 5.7 31.0 5.1 2.3 2.8 0.0 3.9 25.7 0.2 0.1 0.4 0.3 8.1 16.2 8 25 -3 1 3 60 17.0 34.0 611 36.0 18.0 0.2 0.3 2.0 11 66.4 33.2 -17 0.0 0.0 4.4 0.1 1.0 3.5 8.8 0.1 0.0 7.8 51.5 7.4 3.7 Florida Georgia Maryland North Carolina South Carolina Virginia West Virginia Total South Atlantic East South Central Alabama Kentucky Mississippi Tennessee Total East South Central West South Central Arkansas Louisiana Oklahoma Texas Total West South Central TOTAL SOUTH WEST Mountain Arizona Colorado Idaho Montana Nevada New Mexico Utah Wyoming Total Mountain Pacific Alaska California Hawaii Oregon Washington Total Pacific TOTAL WEST TOTAL UNITED STATES 1 4.4 2.3 3.5 0.9 0.4 6.1 0.1 8.7 4.6 7.0 1.8 0.8 12.2 0.2 18.8 68 48 88 73 29 42 36 15.6 21.0 25.0 81.0 69.3 6.9 442.5 7.8 10.5 12.5 40.5 34.7 3.4 221.2 37.7 378 20.1 10.0 1.3 0.3 1.4 0.8 2.5 0.7 2.7 1.7 3.8 35 56 30 44 27.5 166.0 21.8 51.8 13.7 83.0 10.9 25.9 7.7 165 43.0 21.5 0.6 1.0 0.5 7.4 1.2 2.1 1.0 14.9 9.5 29 42 17 116 49.2 40.8 34.9 15.7 24.6 20.4 17.5 7.8 19.0 204 21.5 10.7 32.2 64.3 747 23.2 11.6 2.2 2.0 0.0 0.1 0.2 0.5 0.7 0.1 4.4 3.9 0.1 0.1 0.3 0.9 1.4 0.1 5.7 11.3 62 -1 13 6 18 18 30 1 28.1 0.0 293.2 99.1 118.8 38.2 41.5 19.9 14.0 0.0 146.6 49.5 59.4 19.1 20.7 9.9 147 25.9 13.0 0.4 19.7 0.4 0.2 2.1 0.8 39.5 0.8 0.5 4.2 22.9 9 396 10 48 128 23.1 20.1 24.7 209.8 60.4 11.5 10.0 12.3 104.9 30.2 45.8 591 25.8 12.9 28.5 57.1 738 25.8 12.9 102.0 204.0 2,325 22.8 11.4 Changes are based on the actual numbers. The number of times displaced defense workers could have been absorbed, given 1988-89 employment growth. Source: Calculated by the Federal Reserve Bank of Atlanta from data in previous tables. 2 Table 4 Direct Defense Expenditures in 1989, by Region and Major Expenditure Category1 (percent of total direct expenditures) Radio, TV, and Communications Equipment Aircraft and Parts Ordnance and Accessories Business Services Other Transportation Equipment Northeast 28.90 17.08 7.06 13.10 5.78 .60 North Central 13.62 22.24 14.21 7.14 .94 3.56 South 17.65 12.32 5.92 18.23 9.66 9.48 West 19.01 14.86 26.88 11.72 3.17 5.45 United States 19.90 15.87 14.32 13.12 5.23 5.21 Region New Construction 1 Based on prime contract awards. Source: Department of Defense Projected Defense Purchases: Detail by Industry and State, Calendar Years 1989-1994, Directorate for Information Operations and Reports, November 1989. Table 5 Direct Defense Expenditures in 1989, by Selected States and Major Expenditure Category1 (percent of total direct expenditures) State Radio, TV, and Communications Equipment Aircraft and Parts Ordnance and Accessories Business Services 36.31 Other Transportation Equipment New Construction 23.35 4.26 Virginia 17.55 Colorado 20.80 5.65 18.13 20.69 Massachusetts 33.75 11.33 9.19 20.24 Missouri 11.91 45.12 19.83 5.42 6.94 44.79 3.69 8.27 18.49 19.90 15.87 14.32 13.12 5.23 Connecticut United States 1 11.79 3.47 5.21 Based on prime contract awards. Source: Department of Defense Projected Defense Purchases: Detail by Industry and State, Calendar Years 1989-1994, Directorate for Information Operations and Reports, November 1989. 40 E C O N O M I C REVIEW, »MARCH/APRIL 1991 defense dollars and, therefore, most likely to have a difficult time absorbing workers displaced by defense budget cuts. effect nationally, and additional base closings remain too speculative at this point to assess their impact. In addition, the Defense Department asserts that normal attrition and curtailed recruitment could be used to accomplish as much as a 10 percent reduction in active duty military and direct civilian employment during the next three to five years. Conclusion The recently released 1992 budget for the United States proposes defense outlay cuts ranging from 3 to 6 percent in real terms over the next five years. Congress and the Bush administration generally agree that cuts in defense spending are appropriate and desirable, given the size of the country's budget deficits and the reduced threat of a conflict with the Soviet Union. Desert Shield and Desert Storm are not accounted for in the 1992 budget, and the impact the Gulf crisis may have on future defense cuts is uncertain at this point. However, given the quick resolution in the Gulf crisis, plans to cut defense expenditures are likely to move forward. In terms of employment, defense outlay cuts in the range expected generally will not result in significant initial adverse effects for the nation as a whole or for most regions and states. Base closings as well as the overall cuts in defense spending will affect active duty military and Department of Defense civilian workers. The closings announced in 1988 will have only a marginal Prime contract awards by the Department of Defense are good indicators of private-sector employment directly related to defense expenditures. On the basis of these awards, employment likely to be affected by the budget cuts can be identified by industry within a state or census region. Absorption rates indicate that most areas have enjoyed sufficient economic growth to absorb potentially displaced private-sector defense workers back into the work force in a reasonable fashion. The Northeast census region proved to be the exception. Five states—Virginia, Colorado, Massachusetts, Connecticut, and Missouri—will potentially be hardest hit by defense cuts. These states have been generally more dependent on defense spending, and they also showed less employment growth than other states in the fairly representative year of 1989. Of course, movement of these workers to places offering more new jobs may mitigate their difficulties in getting other work, but out-migration in these hard-hit areas may worsen the consequences for local communities. Notes 1. For a thorough discussion and a model of these relationships, see Brian Coshell, "Defense Spending Cuts: Implications for Deficit Reduction and the Economy," Congressional Research Service, Library of Congress, January 5, 1990; and Mark A. Wynne, "The Long-Run Effects of a Permanent Change in Defense Purchases," Federal Reserve Bank of Dallas Economic Review (January 1991): 1-16. 4. Much of the data and methodology for approaching this issue was obtained from a report by Joseph V. Cartwright titled "Potential Defense Work Force Dislocations and U.S. Defense Budget Cuts: An Illustration," prepared as a staff paper for the Department of Defense in March 1990. This study also discusses the merits of using prime contract awards to allocate direct privatesector defense-related employment. 2. Expenditures related to operations Desert Shield and Desert Storm are excluded from both the 1990 actual and 1991 estimated defense outlays in the 1992 budget. An assessment of these expenditures' size or potential impact is beyond the scope of this article. 5. This estimate is from the U.S. Department of Defense Directorate for Information Operations and Reports, Selected Manpower Statistics, Fiscal Year 1988. 6. For a more detailed discussion of the impact of defense cuts by industry and occupation see Norman C. Saunders, "Defense Spending in the 1990s: The Effect o f D e e p e r Cuts," Monthly Lahor Review ( O c t o b e r 1990): 3-15. 3. Data for 1988 are the latest for which all series necessary for estimating the third category of defense-related employment—private-sector workers—are available. FEDERAL RESERVE BANK O F ATLANTA 41 Book Review The Interstate Banking Revolution: Benefits, Risks, and Tradeoffs for Bankers and Consumers by Peter S. Rose. Westport, Conn.: Greenwood Press Inc., 1989. 230 pages. $45.00. he character of U.S. banking has changed dramatically since the early 1980s as interstate banking has mushroomed. During the 1981-90 period more mergers and acquisitions have occurred in banking and finance—on average, more than 300 a year—than in any other industry. Until the early 1980s the McFadden Act and the Douglas Amendment had effectively closed the door to interstate banking, but banks had exploited a number of regulatory loopholes to cross state lines. Several bank holding companies moved to acquire nonbank businesses—including finance companies, mortgage banks, leasing firms, and the like—in the 1960s and 1970s. Before the Competitive Equality in Banking Act of 1987 effectively prohibited the establishment of new nonbank banks, some banks sought to avoid geographic restrictions by acquiring such firms, which could perform only some of the functions of a full-service bank. Finally, many banks crossed state lines by acquiring troubled thrift institutions.1 T During the past decade, however, full-service offices have become the dominant mode of interstate banking as more and more states have enacted laws allowing entry from other states (B. Frank King, Sheila L. Tschinkel, and David D. Whitehead 1989). By the end of 1988 more than half of the 14,600 interstate bank offices were permitted to offer all banking services. The ready availability of data on bank mergers has encouraged research on diverse issues. In The Interstate Banking Revolution: Benefits, Risks and Tradeoffs for Bankers and Consumers, Peter S. Rose examines the evidence, chiefly through a literature survey, in three major research areas: the motives underlying interstate banking; the potential benefits and costs of interstate banking to banks, shareholders, and the public; and the track record of and public policy issues arising from the interstate banking movement. The author, w h o holds the Jeanne and 42 John Blocker Chair of Business Administration in the Department of Finance at Texas A&M University, supplements his discussion of the literature with anecdotes drawn from conversations with bankers. The book's main contribution is the valuable insights it provides into the strengths and weaknesses characterizing studies of the motives behind and the public policy issues raised by interstate bank mergers. Its largest flaw is its lack of any discussion of capital adequacy and deposit insurance reform and how these issues may be expected to affect interstate banking. Motives for Bank Mergers One way to gain insight into the different implications of bank mergers for the acquirer's and target's shareholders and for social welfare is to analyze the motives for mergers. Rose devotes an entire chapter to discussing some of the economic, technological, and legislative incentives for interstate banking. Economic considerations commonly cited include banks' efforts to attract capital to struggling local economies, to broaden their pool of management talent, to profit from improved stock prices for target banks, and to restore a customer base suffering from the departure of large business clients. Technological advances, too, which have reportedly led to economies of scale and scope at financial institutions, may have motivated bankers to seek out new markets and services. Legislative developments that have hastened interstate banking include changes to help stem the rising number of savings and loan failures and to promote regulatory equity between bank and nonbank firms. The failure to liberalize banks' service powers may also have been an important underlying factor moti- E C O N O M I C REVIEW, MARCH/APRIL 1991 vating interstate banking. Rose mentions that a more liberal antitrust environment may have facilitated interstate banking. However, because antitrust laws have a direct effect only on concentration in local banking markets, it is unlikely that they have made much difference to interstate banking, which generally has more influence on national or regional banking concentration.2 To determine whether empirical support exists for these explanations, the author examines the literature on bank merger motivation. Most of these studies are based on the premise that profitability, gains in efficiency, diversification, and possible reductions in risk should play important roles in shaping bank mergers and acquisitions. Achieving economies of scale is a powerful economic motive for bank mergers. In addition, banks may hope to establish scope economies—the reduction in costs from providing an array of services— through mergers. The author concludes that the optimal bank size for achieving scale economies is around $100 million in total assets and that cost savings are not a key consideration. Rose bases his conclusion, however, on cost studies that used Functional Cost Analysis (FCA) data, which provide little information about large banks' cost structures.3 A large proportion of the dollar v o l u m e of interstate transactions involves banks with total assets greater than $1 billion, for which FCA data are nonexistent. 4 Researchers have more recently examined economies of scale at large banks using the intermediation approach, defining output in terms of dollars (for example, William C. Hunter and Stephen G. Timme 1991 and Sherrill Shaffer and Edmond David 1986). These studies point to economies of scale for banks with u p to $5 billion in assets. Shaffer and David suggest that, although these scale economies may be statistically small, they are quite important economically. Thus, contrary to Rose's findings, scale economies appear to be a valid motive for bank mergers. Further, Rose fails to mention that existing cost studies do not control for several factors typically associated with large bank mergers—such as greater convenience, a wider array of services, lower service fees, and a wider pool of skilled management. These scale economies can benefit both the public and shareholders.5 The search for greater market power is a second important factor driving banks to cross state lines. Banks expand geographically into less-than-competitive markets to generate higher profits. Interest in interstate banking's effect on local banking market concentration arises because of the apparent relationship between market concentration and competition and this FEDERAL RESERVE BANK OF ATLANTA relationship's impact on shareholders and the public. 6 Although banks' shareholders can benefit if a less competitive environment translates into higher profits or reduced risk, consumers and small businesses may ultimately pay higher prices for bank services. The market power explanation for bank mergers has several problems that Rose neglects to mention. First, if one assumes that expanding market power raises profits for some banks, this factor should be reflected in the purchase price for target banks. Second, though the evidence appears to indicate a link between structure and performance, there is also some support for the alternative "efficient market structure" hypothesis proposed by Michael Smirlock (1985) and Douglas D. Evanoff and Diana L. Fortier (1988). They argue that a positive correlation between market concentration and profits is merely the result of a large efficiency gap among different banks in the same market. Finally, the rather narrow view of banking Rose adopts in his discussion of antitrust issues does not take into account the fact that nonbank firms dilute market concentration by providing many of the same products banks offer. In other words, banking concentration may overstate actual product market concentration. Overall, the evidence regarding bank market structure issues is inconclusive. Rose is quick to point out that the mixed evidence in this area may be partly the result of methodological and data limitations of the studies. Aside from motivations associated with economies of scale and market power, a third reason banks may seek mergers is to diversify earnings, both geographically and by customer type. Such diversification can reduce the overall credit risk of a bank's asset portfolio. The acquirer may also broaden its base of core deposits, which are more stable than purchased funds. Shareholders can benefit from both types of diversification. Hunter and Larry D. Wall (1989) argue that whether the public benefits is debatable. Although banks with larger core deposits may gain more stability, they may do so at the expense of market discipline resulting from depositors' tendency to withdraw purchased funds when a bank encounters financial problems. Rose points out that regional banking compacts limit gains from diversification because the economic characteristics of neighboring states are similar to an interstate bank's home state. However, banks increasingly have participated in loan syndications originated by institutions based outside their regions. In addition, banks are more and more selling off many of their loans into mortgage- and other asset-backed securities. These trends undermine the argument that asset diversification is a driving force behind mergers. 43 To gain additional insights into the motives underlying bank mergers, Rose surveys studies of a c q u i r i n g a n d a c q u i r e d b a n k s ' p r e m e r g e r characteristics. He concludes that interstate bank mergers have been motivated by an effort to produce and deliver retail banking services, implying that mergers are undertaken partly to protect and increase the deposit base. Evidence regarding profitability of mergers is mixed, however. For example, Rose cites two of his own earlier studies that suggest earnings and efficiency problems and somewhat greater risk exposure among both interstate acquirers and the banks they purchased. In contrast, Hunter and Wall (1989), analyzing the characteristics of a sample of 559 target banks during the 1981-86 period, report that the most valued merger targets exhibited higher profitability and faster premerger growth in core deposits, had higher ratios of loans to earning assets, and used more financial leverage. Which Markets to Enter and Which Banks to Target? Rose also addresses the questions of how banks assess the desirability of states for future entry and how they select targets for acquisition. He studies bank entry at the state level and concludes that interstate banking acquisitions have not spread randomly or uniformly across the country but have been concentrated in selected states and regions. The principal factors driving bank expansion at the state level are market size; growth rates of personal income, population, and deposits; total deposits; and number of banking offices per capita. Analyzing the attractiveness of individual states is of limited use, however, because it obscures differences across markets within a state. Recent research supports this conclusion. Using the local market as the unit of observation, Dean F. Amel (1989) examines the characteristics of banking markets that have been attractive to entry by merger. Amel finds that urban markets are five times more likely to experience entry than are rural areas. Furthermore, high market profitability is very important in rural areas, whereas low market concentration is more critical in urban areas. In addition to assessing the desirability of state and local markets for entry, acquiring banks also scrutinize target banks according to various criteria. Rose suggests that the key factors underlying the choice of target banks are earnings per share, the stock-price-to-earnings ratio, and the merger premi- 44 um. However, he does not discuss the economic determinants of the purchase price. Hunter and Wall (1989) found that target banks likely to attract the highest purchase prices have high core deposit growth and offer greater potential for diversifying the acquirer's risk. This finding appears to support Rose's conclusion that interstate banks have tended to pursue acquisitions with a strong retail banking orientation. Effects of Mergers on Shareholders and the Public Shareholders of both acquiring and acquired interstate banks may benefit from mergers, Rose finds, especially in large merger transactions. His conclusion is based mainly on studies by Sankar De and Marcia Millon Cornett (1988) and Jack W. Trifts and Kevin P. Scanlon (1987). More recent studies, though, suggest that bank acquisitions are undertaken for reasons other than to maximize shareholder value. David M. Dubofsky and Donald R. Fraser (1989) found that, on average, stockholders of acquiring banks did not gain. Moreover, with the increase in the number of potential bidders after 1981, Dubofsky and Fraser note that managers placed greater value on the asset size of the firm than on stockholders' wealth. Interstate banking's impact on the public is more difficult to assess. In the studies he surveys Rose finds no evidence that interstate banking diminishes competition in local markets or raises the prices of financial services. In fact, he suggests that the retail orientation of most interstate acquisitions may have increased competition, especially in those communities with large numbers of households and small businesses. Moreover, he discovers no support for the hypothesis that interstate acquirers drain funds away from local communities. O n the other hand, Rose finds no undisputed evidence that interstate banking enhances economic growth, increases bank profitability, or reduces the risk of bank failure. Overall, he concludes, the "expected benefits from full-service interstate banking will probably remain small or nonexistent for most American communities for the foreseeable future." Implications for Public Policy Interstate banking raises some major public policy issues. One that Rose addresses is the probable increase in regional and national banking concentra- ECONOMIC REVIEW, »MARCH/APRIL 1991 tion. Rose emphasizes that, even if the number of banks declined sharply, the system would remain competitive as long as enough institutions compete in local banking markets. However, regional banking concentration could diminish competition in the financial services market for middle-market companies—those with annual sales between $50 million and $250 million. Large commercial banks supply services to middle-market businesses over geographic areas that are typically bigger than local banking markets but smaller than the nation. Constance R. Dunham (1986) terms these regional banking markets. In assessing the need to redefine geographic banking markets in response to this concern, Rose cites evidence suggesting that expanding the focus of competitive analysis to include regional banking markets would restrict mergers between large banks. Another key interstate banking issue, which Rose fails to emphasize, is the need for reform in the deposit insurance system. Interstate bank mergers have increased the number of institutions that may be deemed too big to fail, thereby increasing the burden on the Federal Deposit Insurance Corporation (FDIC). In the long run, either the fees banks pay for federal deposit insurance need to be scaled 1. The need to provide avenues of rescue for failing banks and savings and loans prompted states such as Texas to enact interstate banking legislation during the 1980s. 2. Even in those interstate bank mergers in which local market concentration has been affected, the regulatory authorities have not invoked the potential competition doctrine. 3. The Federal Reserve's Functional Cost Analysis data set provides the most comprehensive sample of banks for which the number of loan and deposit accounts are available. Bank involvement in the FCA program is voluntary, and many banks d o not participate. In particular, the FCA sample has too few banks with assets in excess of SI billion to estimate their cost function reliably. 4. See Goudreau and Wall (1990) for evidence on southeastern bank mergers. to reflect institutions' riskiness or other measures that introduce market discipline need to be implemented. Conclusions On balance, Rose's work is an important contribution to the literature on interstate banking. The book provides a timely summary of current debate and should be of interest to academicians, bankers, and regulators, although it offers few original insights and fails to consider several relevant studies and concerns. The issues surrounding interstate banking take on added importance with the recently released Treasury proposal, which seeks to amend the McFadden Act and the Bank Holding Company Act to allow interstate branch banking as well as interstate bank acquisitions on a nationwide basis. A r u n a Srinivasan The reviewer is an economist in the financial Atlanta Fed's research department. section of the 5. Hunter and Wall (1989) suggest that mergers may be motivated by a desire to become "too big to fail" or "too large to be acquired." Shareholders can benefit from a bank's becoming too big to fail; however, the publicmay suffer because of increased risk-bearing by the FDIC. Both shareholders and the public may lose in mergers motivated by a desire to become too large to be acquired because these transactions may reduce the banking system's overall efficiency. 6. Most market concentration studies use intrastate merger data. However, the findings are relevant for interstate mergers because even though the initial entry into a state is a market extension merger, subsequent acquisitions may involve horizontal mergers and raise antitrust questions. References Amel, Dean F. "An Empirical Investigation of Potential Competition: Evidence from the Banking Industry." In Bank Mergers: Current Issues and Perspectives, edited by Benton E. G u p , 29-68. Boston: Kluwer Academic Publishers, 1989. Dubofsky, David M., and Donald R. Fraser. "Regulatory Change and the Market for Bank Control." In Bank Mergers: Current Issues and Perspectives, edited by Benton E. Gup, 29-68. Boston: Kluwer Academic Publishers, 1989. De, Sankar, and Marcia Millon Cornett. "An Examination of Stock Market Reactions to Interstate Bank Mergers." Unpublished paper, Southern Methodist University, 1988. Dunham, Constance R. "Regional Banking Competition." Federal Reserve Bank of Boston New England Economic Review (July/August 1986): 3-19- FEDERAL RESERVE BANK O F ATLANTA 45 Evanoff, Douglas D., and Diana L. Fortier. "Re-Evaluation of the Structure-Conduct-Performance Paradigm in Banking." Journal of Financial Services Research 1 (1988): 277-94. Goudreau, Robert G., and Larry D. Wall. "Southeastern Interstate Banking and Consolidation: 1984-89." Federal Reserve Bank of Atlanta Economic Review 75 (November/December 1990): 32-54. Hunter, William C., and Stephen G. Timme. "Technological Change in Large U.S. Commercial Banks." Journal of Business 64 (July 1991): 206-45. Hunter, William C., and Larry D. Wall. "Bank Merger Motivations: A Review of the Evidence and an Examination of Key Target Bank Characteristics." Federal Reserve Bank of Atlanta Economic Review 74 (September/October 1989): 2-19- Federal Reserve Bank of Atlanta Economic Review 74 (May/June 1989): 32-51. Shaffer, Sherrill, and Edmond David. "Economies of Superscale and Interstate Expansion." Federal Reserve Bank of New York Research Paper No. 8612, November 1986. Smirlock, Michael. "Evidence on the (Non)Relationship between Concentration a n d Profitability in Banking." Journal of Money, Credit, and Banking Y7 (February 1985): 69-83Trifts, Jack W., and Kevin P. Scanlon. "Interstate Bank Mergers: The Early Evidence." Journal of Financial Research 10 (1987): 305-11. U.S. Department of the Treasury. "Modernizing the Financial System: Recommendations for Safer, More Competitive Banks." February 1991. King, B. Frank, Sheila L. Tschinkel, and David D. Whitehead. "Interstate Banking Developments in the 1980s." 46 ECONOMIC REVIEW, MARCH/APRIL 1991 ? S® ^ 1 SVí S s y - '••"' -í? Sf -•J-;.»;' s s ¥< II In) -' * i • • • --.:•-.:-•-•• • - - * '••••• • ••••:.:• ' ' mmm i».--,>y.-: ' < 1 Wm ' & • •. :, -•••• ' t - ••• - • éWWf • n Economic Review Federal Reserve Bank of Atlanta 104 Marietta St, N.W. Atlanta, Georgia 30303-2713 Address Correction Requested B u l k Rate U.S. Postage PAID Atlanta, GA P e r m i t 292