The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Is Low Unemployment Inflationary? R O B E R T O C H A N G The author is a research officer in the macropolicy section of the Atlanta Fed’s research department. He thanks Robert Eisenbeis, Frank King, Mary Rosenbaum, and Eric Leeper for useful comments and suggestions. F ROM A VARIETY OF PERSPECTIVES, THE MACROECONOMIC PERFORMANCE OF THE U.S. ECONOMY HAS BEEN VERY SATISFACTORY IN RECENT YEARS. IN PARTICULAR, BROAD-BASED MEASURES OF INFLATION HAVE STABILIZED BETWEEN BELOW 51⁄2 PERCENT. HOWEVER, 21⁄ 2 AND 3 PERCENT WHILE UNEMPLOYMENT HAS FALLEN MANY OBSERVERS REMAIN UNEASY, BELIEVING THAT THE CUR- RENT SITUATION IS FRAGILE AND TEMPORARY. THIS THAT THE CURRENT RATE OF UNEMPLOYMENT IS BELIEF IS, IN TURN, ROOTED IN A LESS OBVIOUS VIEW “TOO LOW” TO BE CONSISTENT WITH LOW AND STABLE INFLATION. This state of affairs becomes most visible on the first Friday of every month, when the Bureau of Labor Statistics releases the latest data on employment and unemployment in the United States. Recently, these data releases have often been followed by sharp changes in financial markets. In particular, markets have taken lower-than-expected unemployment rates to mean that inflation is about to accelerate, resulting in falling stock prices and increasing interest rates. The average citizen would find this to be a rather strange ritual. Isn’t low unemployment good for the country? And why is low unemployment supposed to lead to higher inflation anyway? These are important and difficult questions. An influential economic theory, however, argues that the answers are easy and widely found in macroeconomics textbooks. Low unemployment, this theory implies, is unambiguously “good” only up to a point. If unemployment falls below this point, known as the nonaccelerating inflation rate of unemployment (NAIRU), inflation tends to accelerate.1 Opinions about the current location of the NAIRU vary, but many published estimates place it close to 6 percent.2 Since recent unemployment figures have been consistently below that 4 range, adherents to this theory predict that inflation will accelerate. So far these predictions have turned out to be wrong. That their failure should not be a surprise is one of the themes of this article. More precisely, this article argues that the concept of the NAIRU is of very limited use for predicting inflation, understanding its causes, or forming policy. There is both empirical and theoretical support for this view. On the empirical side, the article discusses evidence showing that the NAIRU is highly variable and that there is a great deal of uncertainty about where it is at any particular point in time. These findings imply that, in practice, one cannot know if unemployment is above or below that value supposedly consistent with stable inflation. On the theoretical side, this article argues that even if such a value became known, it would be irrelevant. Contemporary economic theory implies that movements of the unemployment rate may be positively or negatively related to inflation, depending on the nature of the fundamental shocks causing the unemployment changes. Identifying such shocks is possible and helpful for predicting the inflation implications of unemployment changes; but given that these shocks can be identified, whether current unemployment Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 is above or below the NAIRU provides no additional useful information about prospective changes in inflation. The Phillips Curve, the Natural Rate, and the NAIRU he idea that unemployment may be too low to be consistent with stable inflation is of relatively recent vintage. Its origins can be traced to the 1960s and 1970s discussion about how to interpret the then recently discovered “Phillips curve,” an empirical association between inflation and unemployment. Some aspects of this debate are useful for the discussion that comes later. As some readers may recall, Phillips’s (1958) analysis of almost a century of U.K. data shocked the economics profession. Phillips focused on the relationship between wages and unemployment and discovered the striking fact that the rate of change in nominal wages had a negative correlation with unemployment. Soon afterward, Samuelson and Solow (1960) showed that a similar relation held in U.S. data. Moreover, Samuelson and Solow argued that changes in nominal wages were positively related to overall inflation, thus recasting the wageunemployment relation discovered by Phillips into the inverse relation between price inflation and unemployment commonly known as the Phillips curve. The discovery of the Phillips curve generated a heated debate about its implications for economic policy. In particular, research focused on whether a monetary authority such as the Federal Reserve could “buy” less unemployment at the cost of faster inflation. Some argued that the existence of a Phillips curve implied that unemployment could be permanently lowered if inflation were kept at a permanently higher level. Others, in particular Friedman (1968) and Phelps (1968), argued that there was an inflation-unemployment trade-off in the short run but not in the long run. In justifying their thesis, Friedman and Phelps coined the term “natural rate of unemployment.” To understand Friedman and Phelps’s argument, consider first an economy without price surprises, so actual inflation is always equal to previously expected inflation. In such an economy, some workers would always be observed to be unemployed and looking for a job. This phenomenon may simply reflect the fact that, since workers and jobs are heterogeneous, unemployed workers and T firms may take time to search for adequate matches. Hence, even if inflation were always perfectly foreseen, the economy would experience a positive rate of unemployment that Friedman and Phelps called “natural.” What if inflation were not perfectly foreseen? Friedman and Phelps argued that unexpectedly high inflation would make actual unemployment fall below its natural rate, but only in the short run. This decline would happen, in particular, if wage contracts had been negotiated on the basis of previously expected inflation, in which case an inflation surprise would reduce real (inflationadjusted) wages and stimulate employment. One implication is that a monetary authority Contemporary economic could indeed “buy” lower unemployment theory implies that by inducing inflation movements of the unto rise above previemployment rate may be ously expected inflation. But this effect positively or negatively would be only temporelated to inflation, rary because ecodepending on the nature nomic agents would eventually learn to of the fundamental shocks forecast inflation causing the unemployment correctly, and the changes. difference between expected inflation and actual inflation would tend to disappear.3 Although unexpected accelerations in inflation, engendered by monetary policy, may “cause” unemployment to fall below the natural rate, the converse need not hold. Given monetary policy, the Friedman-Phelps theory had no implications for whether movements in the unemployment rate have an independent effect on inflation. In subsequent research, a subtly but clearly different view on the relation between inflation and unemployment emerged. According to this view, inflation tends to accelerate whenever unemployment falls below a particular number, which has come to be known as the “nonaccelerating inflation rate of unemployment,” or NAIRU. The NAIRU concept was first proposed by Modigliani and Papademos, who posited the existence of a rate of unemployment “such that, as long as unemployment is above it, inflation can be expected to decline” (1975, 142). 1. It is worth noting that several authors have also used the term natural rate to refer to the NAIRU. To avoid confusion, this article will make a distinction between the two concepts. The distinction is important because, as discussed below, the natural rate concept developed by Friedman (1968) and Phelps (1968) is not the same as the NAIRU concept, although their values may coincide. 2. For instance, the Congressional Budget Office (1996, 5) recently estimated that the NAIRU was 5.8 percent in 1995. 3. The Friedman-Phelps view further implies that the monetary authority can keep the unemployment rate permanently below its natural rate only by perpetually engineering unexpected increases in inflation. If people’s expectations of inflation are relatively slow to adjust, ever-accelerating inflation is required. This implication does not hold true if expectations of inflation are fully “rational,” as first demonstrated by the work of Lucas (1972) and Sargent and Wallace (1976). Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 5 Ye a r - o v e r - Ye a r P e r c e n t C h a n g e C H A R T 1 Inflation and Unemployment, 1955-96 (Monthly) Unemployment Consumer Price Index Inflation 12 8 4 0 1955 1965 The intuition is that low unemployment is likely to intensify wage pressures and consequently to result in a generalized wage increase. Assuming that firms manage to pass this cost increase to consumers in the form of higher prices, a fall in unemployment is likely to be associated with an increase in inflation. Similarly, an increase in unemployment must result in a fall in inflation. There must therefore be a level of unemployment such that inflation can be expected to remain constant; this level is the NAIRU. Readers may note that the Friedman-Phelps natural rate and the NAIRU are different concepts. Friedman and Phelps defined the natural rate as an equilibrium whose value was determined by the characteristics of the labor market. In contrast, the NAIRU is posited as an empirical value rather than an equilibrium value. More importantly, the theory of the NAIRU implies that low unemployment may cause inflation to increase independently of the causes of the low unemployment and, in particular, of monetary policy; this is not an implication of the Friedman-Phelps natural rate theory. The NAIRU concept pervades current policy discussions and is the main concern of this article. Since Modigliani and Papademos’s article numerous studies have, not surprisingly, focused on the estimation of the NAIRU. If there were in fact a strong, stable relation between unemployment, a known NAIRU, and inflation, then one could compare current unemployment with the NAIRU to accurately predict future inflation. But this is not the world we live in, as the discussion below will show. A First Look at the NAIRU o understand the details and some of the problems associated with estimating the NAIRU, it is helpful to have a broad idea of the historical behavior of T 6 1975 1985 1995 unemployment and inflation. Chart 1 depicts the behavior of civilian unemployment and the inflation rate since 1955. The most notable aspect of this chart is the dramatic increase in inflation that started in the mid-sixties and ended in the early eighties, and the subsequent disinflation. In 1965 inflation was less than 2 percent a year; in 1980 it surpassed 13 percent. Much of the increase is widely believed to have been caused by the oil shocks of 1973 and 1979, and in fact the chart shows a rapid increase in inflation following each of these dates. It has to be noted, though, that the increase in inflation actually started much earlier. The trend toward higher inflation was broken around 1980, and the first half of the 1980s witnessed a rapid decrease in inflation toward the 4 to 6 percent range. This change in inflation behavior has been widely attributed to strongly contractionary monetary policy starting in October 1979 and called, accordingly, the “Volcker deflation.” Finally, inflation since 1991 has been surprisingly stable at around 3 percent. In Chart 1, unemployment shows a slightly upward trend over the 1955-96 period but considerable variation around its trend. Until the early seventies, unemployment was about 5 percent on average. Its fluctuations were much larger during the decade following the first oil shock; average unemployment during that period was 7 1⁄ 2 percent, and it surpassed 10 percent in 1982, reflecting the contractionary effects of the Volcker deflation. Since 1984 unemployment has been declining, averaging a little less than 6 percent. In Chart 1 it is not obvious that there may be a special level of the unemployment rate above which inflation would be expected to decline. However, the ModiglianiPapademos definition of the NAIRU suggests a different and more informative way to look at the same data. Their definition implies that inflation is expected to decline Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 C H A R T 2 Unemployment versus Inflation Changes, 1955-96 1.6 1.2 Inflation 0.8 1955–73 0.4 1974–83 1984–96 0 –0.4 –0.8 –1.2 3.2 4.8 6.4 8.0 Unemployment whenever unemployment is above the NAIRU. Hence there should be a negative relation between expected changes in inflation and the difference between unemployment and the NAIRU. If, in addition, this relation is assumed to be linear, a plausible hypothesis for inflation, unemployment, and the NAIRU is given by Et[π(t + 1) – π(t)] = b[u(t) – u*], (1) where π(t) denotes the inflation rate in period t, u(t) the unemployment rate in period t, u* the NAIRU, b a coefficient whose expected sign is negative, and Et[.] is short notation for “expectation given information available up to period t.” That is, provided that b is indeed negative, an equation such as (1) implies that, if this month’s unemployment is above the NAIRU u*, inflation should be expected to decline next month. Neither b nor u* are directly observable; instead, they must be inferred from the data. Versions of equation (1) are the basis of most attempts to estimate the NAIRU. The intuition for work of this kind, and some of the problems associated with it, can be grasped from a scatter diagram of unemployment against subsequent changes in inflation, such as Chart 2. Each point in the chart represents a particular month’s unemployment rate, measured against the horizontal axis, and the subsequent month’s change in inflation, measured against the vertical axis. In particular, observations above the horizontal axis represent months in which inflation increased. If equation (1) were to hold in the data, the observations depicted in Chart 2 would be distributed around a line of negative slope that intersects the horizontal axis at precisely the NAIRU, u*. At first glance the chart suggests that there is a negative relation between the unemployment rate and sub- 9.6 11.2 sequent changes in inflation, in particular for extreme values of the unemployment rate. When unemployment has been below 5 percent, inflation has historically increased more often than decreased. Conversely, unemployment rates above 8 percent have been mostly associated with falling inflation rates. However, the relationship in the middle range of unemployment rates between 5 and 8 percent is much less clear: for each value of unemployment in that range, the number of observations above the horizontal axis is about the same as the number of observations below it. Hence, if the NAIRU is defined as “a rate of unemployment such that inflation is as likely to increase as to decrease,” Chart 2 suggests that the data are not likely to yield a very precise estimate of the NAIRU. A closer look at Chart 2 should reveal a second aspect of the data that is relevant for this discussion: the relationship between changes in inflation and unemployment, and by inference the NAIRU, has moved significantly over time. To illustrate this behavior, observations in Chart 2 are distinguished by different symbols corresponding to three different subperiods, 1955-73, 1974-83, and 1984-96. Splitting the sample in this way is loosely motivated by the fact that the oil shocks of 1973 and 1979 and the Volcker disinflation were large, unusual events that affected both unemployment and inflation. It is clear that the observations in Chart 2 are not just randomly distributed across the three periods under consideration. Observations before 1974, the squares in the chart, appear mostly to the left, showing that unemployment was relatively low. Observations between 1974 and 1983, depicted as diamonds, appear to the right, because unemployment in that period was relatively high. Finally, observations from 1984 on, marked by triangles, are in the middle part of the chart. Using these data to Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 7 look for a value of the unemployment rate at which inflation is “as likely to increase as to decrease,” it is hard to deny that the value is different for each of the three periods considered. In other words, Chart 2 suggests that the NAIRU has not been constant. It seems to have been lower in the first period than in the second and to have fallen after 1983 (although not to the pre-1974 level). Chart 2 thus suggests that the NAIRU is an elusive number. Its precise location is difficult to infer from the data, and it moves around over time. These conclusions, obtained from the visual inspection of Chart 2, are confirmed by more formal statistical evidence, to which the discussion now turns. There Is No Reliable NAIRU Estimate o understand the statistical estimation of the NAIRU, start by supposing that equation (1) is true. Estimating the NAIRU then amounts to estimating the parameters b and u* in (1), which requires some If there were a strong, quick manipulations. stable relation between To simplify notation, define the change unemployment, a known of the monthly inflaNAIRU, and inflation, then tion rate, ∆π(t + 1) = one could compare current π(t + 1) – π(t). A minor difficulty is unemployment with the that the left-hand NAIRU to predict future side of (1) is not the inflation. But this is not observed value of ∆π(t + 1) but its the world we live in. expected value conditional on information available up to period t, which is unobservable. To handle this problem, let e(t + 1) = ∆π(t + 1) – Et ∆π (t + 1) denote the error in predicting ∆π(t + 1). One can now replace the lefthand side of (1) by the difference between the observed change in inflation and the error in predicting it, ∆π (t + 1) – e(t + 1). Finally, define c = bu*. Inserting these definitions in (1), one obtains T ∆π(t + 1) = –c + bu(t) + e(t + 1). (2) Equation (2) is very useful. It is a linear equation, and its parameters b and c can be estimated using ordinary least squares.4 Since c is equal to bu*, an estimate of u* is simply given by the estimate of c divided by the estimate of b. This way of estimating the NAIRU is fairly common in the literature.5 Once an equation such as (2) has been estimated, standard statistical techniques allow assessment of the precision of the estimated coefficients c and b and consequently the precision of the estimated NAIRU u*. The degree of precision is usually summarized using confi8 dence intervals. A 95 percent confidence interval for u*, in particular, provides upper and lower bounds for the unemployment rate that should contain u* with 95 percent probability. Standard techniques are also available for testing whether assuming that b, c, and the NAIRU are constant, as is implicit in equations (1) and (2). Recalling the discussion of Chart 2, it is clear that testing for the stability of the NAIRU is particularly important given that the data suggest that the NAIRU has changed over time. Equation (1) is unduly restrictive in that it restricts the expected change in inflation to respond only to the current value of unemployment relative to the NAIRU. It may be more realistic to assume that the expected change in inflation also responds to past unemployment, as would be the case if increases in wages were translated only gradually to consumer prices. Similarly, people’s expectations concerning the change in inflation may depend on current and previous changes in inflation. These considerations imply that lags of unemployment, relative to NAIRU, and lags of inflation changes should be included as explanatory variables in (1). Accordingly, the following generalization of (1) was analyzed: (3) Et∆π (t + 1) = b0[u(t) – u*] + b1[u(t – 1) – u*] + … + b11[u(t – 11) – u* + a0∆π(t) + … + a11 ∆π(t – 11). The alphas and betas are coefficients to be estimated, in addition to the NAIRU, u*. In contrast to (1), equation (3) allows the expected change in inflation between periods t and (t + 1) to be influenced by the deviations of unemployment from the NAIRU in the previous twelve months. Also, the expected change of inflation can be influenced by the twelve previous changes in inflation. The coefficients of (3) were estimated by ordinary least squares. For brevity, details are omitted and only the main findings are reported here. For the full 1955-96 sample displayed in Chart 1, NAIRU was estimated to be 6.14. This estimate is close to others found previously.6 A second finding was that this estimate of NAIRU is rather imprecise: a 95 percent confidence interval is given by the range of unemployment rates between 5.38 and 6.90. The wide range for the location of the NAIRU is consistent with the lack of precision found in previous work. Indeed, the results presented here are on the optimistic side. In a recent, convincing paper Staiger, Stock, and Watson examined a large number of alternative procedures and concluded, “Our main finding is that the natural rate is measured quite imprecisely. For example, we find that a typical value of the NAIRU in 1990 is 6.2 percent, with a 95 percent confidence interval for the NAIRU in 1990 being 5.1 to 7.7 percent” (1996, 2). Readers may note that Staiger, Stock, and Watson’s “typical” confidence interval is much larger than the one estimated here. In fact, for a specification very close to Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 (3), Staiger, Stock, and Watson argue that perhaps a better way to estimate a 95 percent confidence interval yields an interval from 4.74 to 8.31.7 From the perspective of policy analysis, the main implication of these findings is that the range of uncertainty about the location of the NAIRU is often too large to be useful. Consider, in particular, the recent debate about whether the unemployment rate signals increasing inflation. Since unemployment has been in the 5.5 to 6.5 percent range for the last three years, there is no way to tell, on the basis of observed data, whether current unemployment is above or below the NAIRU; it is well within the confidence bands. Returning to equation (3), further analysis shows that the coefficients and hence the implied value of the NAIRU have changed over time. One can test and easily reject the hypothesis that the coefficients of (3) are constant over the three subperiods underlying the discussion of Chart 2 above, namely 1955-73, 1974-83, and 1984present.8 Estimates of the NAIRU were at 5.91 for the first subperiod, 7.44 for the second, and 6.04 for the most recent subperiod. Hence the statistical analysis is consistent with the hypotheses obtained from the visual inspection of Chart 2. It also agrees with most of the literature. The instability of the NAIRU raises further difficulties. Estimating a moving NAIRU requires specifying how the NAIRU moves over time but, unfortunately, current economic theory provides little guidance on the economic determinants of changes in the NAIRU. As a consequence, recent studies have focused on models in which the u* is viewed as slowly varying over time but in a way that has no relation with other economic events. This strategy does imply that the NAIRU’s location at different times can be estimated and its changes predicted. However, there are at least two important problems with it. The first is that the work of Staiger, Stock, and Watson has shown that forecasts of the NAIRU obtained in this manner are also subject to very wide confidence intervals. For example, when they estimated a version of (3) allowing u* to be a smooth function of time, they found the 95 percent confidence interval for u* in January 1990 to be 4.17 to 8.91. The second problem is that the instability of the estimated NAIRU may be a symptom of a deeper problem, namely, that equations such as (1) and (3) may be incorrectly specified. To illustrate this problem, assume in equation (1) that the NAIRU is in fact constant. The hypothesis expressed by that equation is that inflation next month is expected to be the same as it is this month, unless this month’s unemployment deviates from the NAIRU. Intuition suggests that such a hypothesis is too extreme: even if there is no such deviation, it may be the case that inflation is not expected to stay the same. For example, it is plausible that people’s expectations of inflation would have increased following the oil shocks of 1973 and 1979 even if the unemployment rate had not deviated from the NAIRU. Conversely, it is likely that expectations of inflation decreased following the October 1979 Federal Reserve announcement of a change toward money targets to fight accelerating inflation. These considerations suggest that (1) should be changed to something like Etπ(t + 1) – π(t) = s(t) + b[u(t) – u*], (4) where s(t) represents the expected change in inflation when unemployment is at the NAIRU. As shown by Chart 1, s(t) was probably a positive number during the subperiod from 1974 to 1983 in which inflation was accelerating. Suppose that s(t) was zero until 1973, a positive number between 1974 and 1983, and a negative number since 1984. Then it is not hard to show that the estimated NAIRU for the second subperiod would be higher than that for the first or third subperiods, just as indicated before.9 But this would just be the result of the incorrect omission of s(t), for the earlier assumption was that the NAIRU had stayed constant at u* for the whole 1955-96 period. Some researchers have tried to deal with this specification problem by adding measures of supply shocks, or 4. A property of expectations conditional on an information set is that prediction errors are independent of any variable in the information set (see Goldberger 1991, chap. 5). Since e (t + 1) is a prediction error and u(t) is assumed to be known by the public at t, e (t + 1) and u(t) must be independent. This condition guarantees that OLS estimates of b and c are at least consistent (see Goldberger 1991, chap. 13.) 5. Recent examples include Weiner (1993), Tootell (1994), Fuhrer (1995), and Staiger, Stock, and Watson (1996). 6. See footnote 5 for published estimates. 7. Two reasons underlie the differences between the estimates given here and those of Staiger, Stock, and Watson. The first is that they include in their regressions additional explanatory variables intended to control for supply shocks and Nixon-era price controls. The second reason is that Staiger, Stock, and Watson assumed that the error term in the estimated equation is normally distributed and derived exact confidence intervals based on that assumption. This study did not assume normality and derived approximate intervals based on the so-called Delta method, which is valid in large samples. For a description of the Delta method, see Goldberger (1991, 102). 8. The test is a standard one for stability of coefficients over time, as described by Harvey (1989, chap. 2). 9. For example, the assumption in the text is that s(t) is equal to some positive number, say n, for the 1974-83 subperiod. If one estimates (2) the regression constant c is equal to bu* – n, and the true value of the NAIRU should be calculated to be (c – n)/b and not c/b. Hence the standard procedure would overestimate the NAIRU for the 1974-83 subperiod. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 9 CHART 3 Effect of an Aggregate Demand Shock P1 E0 P* AD1 P r i c e L e v e l ( P ) AS E1 AD0 Y* O u t p u t Y1 ( Y ) policy shocks that may shift inflationary expectations, to (1) or (3). Also, some have examined specifications that do not constrain expected inflation changes to be zero when unemployment is at the NAIRU. But once these changes are made, the NAIRU idea loses its simplicity and, hence, much of its appeal. In addition, such procedures neglect an essential point: the bulk of the variation of inflation may be due to changes in expected inflation that are unrelated to the deviation of unemployment from the NAIRU. Consequently, modeling efforts should be directed toward understanding the other determinants of expected inflation. To summarize, statistical analysis confirms that the value of the NAIRU is too elusive to be a reliable policy concept. In addition, the estimated NAIRU seems to move around, increasing uncertainty about its location and raising questions about the statistical assumptions that have to hold to estimate it. Why Low Unemployment Is Not Always Followed by Higher Inflation erhaps the most compelling argument against using estimates of the NAIRU to predict inflation is a theoretical one. Most contemporary models of actual economies treat both inflation and unemployment as endogenous outcomes. These models treat movements in inflation and unemployment as simultaneously determined responses to more basic forces or “shocks,” which are typically modeled as random disturbances. In practice, shocks are of different kinds and hit actual economies all the time. Inflation and unemployment will move in opposite directions in response to some shocks and in the same direction in response to others. Hence, declines in unemployment will be associated with increasing inflation in some cases but with decreasing inflation in others. This relationship holds even if there is a well-defined Friedman-Phelps natural rate to which P 10 unemployment must return in the long run and even if a NAIRU can be estimated. The relation between inflation, unemployment, the natural rate, and the NAIRU is not unambiguous but depends on the nature of the shocks hitting the system. It also follows that predictions of inflation should be based, if possible, on an assessment of these shocks rather than on unemployment movements alone. As a consequence, in contemporary economic models it is incorrect in general to assume that the probability of an increase in inflation has gone up just because the unemployment rate has fallen below the estimated NAIRU. The correct inference depends on the kind of shock that has made the unemployment rate fall. It is clearly possible for some shocks to cause a fall in unemployment and to raise the likelihood of a decrease in inflation. Is it possible to empirically identify the different shocks that affect the economy? The answer is yes, as shown recently by Sims (1986), Bernanke (1986), Gordon and Leeper (1994), Leeper (1995), and Sims and Zha (1996). These studies have shown how to identify the underlying causes of low unemployment and thereby to infer their expected effect on inflation. They present a promising alternative concept for inflation forecasting and policy formulation. If shocks can be identified, as this line of research demonstrates, knowledge of the NAIRU relative to the actual unemployment rate will add no new information for predicting inflation. It may be helpful to illustrate these ideas in a particular and perhaps familiar context. As described by intermediate macroeconomic textbooks, the outcome of many macroeconomic models can be summarized by the intersection of two curves called aggregate demand (AD) and aggregate supply (AS). The AD and AS curves are not ordinary supply and demand curves. Rather, each summarizes the equilibriums of several markets, which can be kept in the background for this discussion. The aggregate supply (AS) curve summarizes the combinations of output (Y) and price levels (P) such that the market for labor is in equilibrium. In Chart 3, an AS curve is shown with a positive slope, which is consistent with the traditional view that nominal wages adjust sluggishly. Under that view, an increase in the price level (P) reduces real wages and induces firms to increase hiring, thus resulting in a reduction in unemployment and a consequent increase in output (Y); hence P and Y must be positively related for labor market equilibrium. The Friedman-Phelps natural rate hypothesis implies, in this context, that the AS curve can only be upward-sloping in the short run because eventually nominal wages will adjust to fully incorporate the effect of unexpected inflation; hence the real wage, unemployment, and output will eventually return to their original, so-called natural levels. Given that our emphasis is on the short run, the distinction between the short and the long run is not Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 CHART 4 Short-run Effect of an Aggregate Supply Shock L e v e l ( P ) AS0 E0 P* E1 AS1 P1 AD P r i c e important for the purposes of this discussion. Accordingly, in Chart 3, the output level associated with the natural rate is denoted by Y*, and the AS curve can be taken to be a short-run one. The aggregate demand (AD) curve depicts combinations of output and price levels consistent with equilibriums in the markets for output and money, which, again, need not be explicitly shown for the present purposes. It is typically assumed that the AD curve is downward-sloping. Given the nominal supply of money, an increase in the price level (P) causes a fall in the real quantity of money. The resulting excess demand for money is usually assumed to be eliminated by an increase in the interest rate, which reduces the demand for output (Y). Hence goods and money markets equilibriums require a higher P to be associated with a larger Y or, in other words, the AD curve to slope down. The analysis of the effects of a particular shock to the economy proceeds by comparing the outcomes before and after the shock. In Chart 3, the aggregate supply curve and the aggregate demand curve in the absence of economic shocks are displayed as AS and AD0, respectively. The intersection of AS and AD0,, the point E0, gives the outcome of this model in the absence of shocks: if output is Y* and the price level is P*, all markets are in equilibrium. Now, consider a shock in either the output or the money market that displaces the aggregate demand curve to the right, say, to AD1. An example of such an “AD shock” is an unexpected decrease in the demand for money. For any given price level P, and assuming that the nominal supply of money has not changed, the real supply of money must increase. Equilibrium in the money market then requires a matching increase in the real demand for money. This increase is brought about, in turn, by a fall in the interest rate and an increase in income and output (Y). Hence Y must be larger for each P, or the AD curve must shift to the right, if the markets of money and goods are to be in equilibrium. For all markets to be in equilibrium after the AD shock, the new price level must be P1 and output Y1, as given by the intersection of AD1 and AS. The upshot is that prices and output both increase. The increase in output must be, in turn, associated with lower unemployment. Hence as a result of an AD shock unemployment falls below the natural rate and inflation increases. Chart 4 illustrates the effects of a different kind of shock, one that affects supply. Initially, aggregate demand and supply are given by AD and AS0, respectively. Suppose there is a shock that shifts the aggregate supply curve to the right, say, to AS1. An example is an unexpected increase in labor productivity that, given prices, induces firms to increase employment and production. Given all prices, and hence P, labor market equilibrium thus requires increased output (Y), and the AS must move to the right. Y* O u t p u t Y1 ( Y ) Chart 4 shows that after the AS shock, the model’s equilibrium moves from E0 to E1. Since Y is above Y*, unemployment must fall below the natural rate. On the other hand, inflation must fall. Hence unemployment and inflation respond in the same direction to an AS shock. In this simple context it is possible that the natural rate is such that “when unemployment is above it, inflation is expected to decline,” at least on average; in other words, the natural rate and the NAIRU may coincide. This would be the case if the economy were mostly affected by AD shocks, which move unemployment and inflation in the opposite direction. Note that, from this perspective, the imprecision in the econometric estimation of equations such as (2) or (3) may be attributable to the existence of significant AS shocks. Whether the estimation of the NAIRU yields precise estimates is, however, beside the point. The reason to estimate the NAIRU is, clearly, to forecast the direction of prospective changes in inflation. But if the economy is subject to AD and AS shocks, one should be able to make better predictions by relying on all available information, not only on the comparison between the NAIRU and observed unemployment. To see why, suppose that one observes prices only with a delay and observes that unemployment falls below the natural rate. Is it a good idea to bet that inflation will increase? Clearly not, if one knows that the fall in unemployment has been caused by an AS shock. In such a case blind adherence to the concept that “unemployment below the NAIRU is likely to be associated with increased inflation” would lead to a mistaken inference. These considerations are, of course, consequential for economic policy. The effect of contractionary monetary policy changes, such as increases in the federal funds rate, in the AD-AS model is to shift the AD curve toward the origin. Suppose, as in the previous paragraph, that Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 11 unemployment falls below the natural rate. If the Federal Reserve’s objective is to keep a stable price level, should it adopt a contractionary monetary policy? The answer is positive only if the cause of low unemployment is an AD shock, as in Chart 3. If unemployment is instead low because of a favorable AS shock, as in Chart 4, contractionary monetary policy will have the undesired effect of exacerbating the fall in the price level. Note that this simple example illustrates that it is not necessary to deny the existence of a stable negative correlation between unemployment and inflation to assert that the NAIRU concept is of little use in predicting inflation. In the example, such a Phillips curve-type correlation and perhaps even a stable NAIRU could be observable if most shocks to the economy came from the demand side. It is noteworthy that the NAIRU concept was developed during an era in which it was widely believed that demand shocks were responsible for the bulk of economic fluctuations. Beliefs have since changed due to two developments. The first is that since 1973 oil prices have become a main source of concern for the U.S. economy. The second is the emergence of a school of thought, called real business cycle theory, that asserts that macroeconomic fluctuations are mostly caused by shocks to the aggregate production function, which affect the supply side. While many of the implications of real business cycle theory are controversial, it is safe to say that the evidence presented by its proponents has changed the beliefs of most macroeconomists toward assigning supply shocks a more important role in explaining fluctuations. It is not difficult to find examples of favorable aggregate supply shocks that may help explain why recent unemployment has remained low without accelerating inflation. Oil prices are one example: except for shortlived episodes mostly related to the U.S. conflict with Iraq, the real price of oil has been surprisingly low for many years. Another example is the furious pace of technological innovation in computers and communications. While the quantitative effect of these shocks remains to be determined, the point is that the recent performance of the U.S. economy may be explained to a large extent by good luck on the supply side. This discussion has assumed that it is possible to observe shocks to aggregate demand and aggregate supply or, more generally, the fundamental shocks that affect actual economies. This is in fact a valid assumption in view of the existence of a large literature devoted to identifying the sources of macroeconomic fluctuations. Recent studies, in particular, have followed the lead of Bernanke (1986) and Sims (1986) in trying to decompose the sources of fluctuations by imposing mild theoretical conditions on estimated vector autoregressions (VARs).10 Using the Bernanke-Sims methodology, subsequent papers have analyzed the nature of the shocks that 12 ultimately cause fluctuations in the U.S. economy and elsewhere. For example, Gali (1992) showed that an ADAS model similar to that described above can successfully explain the U.S. data. More recent studies have refined estimates and proposed alternative models; a good exposition of technical details and recent developments is the article by Leeper (1995). For present purposes, the point to be noted is that isolating the fundamental shocks hitting actual economies is feasible and has been done in practice. The arguments just advanced can be summarized and rephrased in a perhaps more appropriate way: concluding that inflation is likely to increase just because the unemployment rate has fallen below the NAIRU ignores useful and available information. Techniques are available for identifying the different shocks that impinge on the economy and cause unemployment to fall. Once these shocks are identified, the deviation of the unemployment rate from the NAIRU provides no additional information for the prediction of changes in inflation. Conclusion: There Are Better Ways his article has advanced theoretical and empirical reasons to show that, in practice, the concept of a noninflation accelerating rate of unemployment is not useful for policy purposes. To make this point, the discussion has focused on three facts. First, the NAIRU moves around. Second, uncertainty about where the NAIRU is at any point of time is considerable. Third, even if we knew where the NAIRU were, it would be suboptimal to predict inflation solely on the basis of the comparison of unemployment against the NAIRU. It is not hard to find additional justification for the views expressed here. A policy of raising the fed funds rate when unemployment falls below the NAIRU may be ineffective, for example, even if the NAIRU were constant, its location were known, and all shocks to the economy were to come from the demand side. Implementing such policy would likely induce changes in the expectations and behavior of the private sector, a point made forcefully many years ago by Lucas (1976). Because the “Lucas critique” has been emphasized in past discussions of the NAIRU and the Phillips curve, this article has not elaborated on it. However, the Lucas critique is an important additional reason to be skeptical about using the NAIRU for policy. This article’s position has been critical of the NAIRU concept, but economists remain divided about the policy relevance of the NAIRU.11 The arguments discussed above highlight the need to further develop models in which the fundamental forces behind macroeconomic fluctuations can be identified and analyzed. Fortunately, a body of outstanding research in that direction is already in progress. T Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 R E F E R E N C E S BERNANKE, BEN S. 1986. “Alternative Explanations of the Money-Income Correlation.” Carnegie-Rochester Conference Series on Public Policy 25:49-99. CONGRESSIONAL BUDGET OFFICE. 1996. The Economic and Budget Outlook: Fiscal Years 1997-2006. May. FRIEDMAN, MILTON. 1968. “The Role of Monetary Policy.” American Economic Review 58 (March): 1-17. FUHRER, JEFFREY C. 1995. “The Phillips Curve Is Alive and Well.” New England Economic Review (March/April): 41-46. GALI, JORDI. 1992. “How Well Does the IS-LM Model Fit Postwar U.S. Data?” Quarterly Journal of Economics 107 (May): 709-38. GOLDBERGER, ARTHUR S. 1991. A Course in Econometrics. Cambridge, Mass.: Harvard University Press. GORDON, DAVID B., AND ERIC M. LEEPER. 1994. “The Dynamic Impact of Monetary Policy: An Exercise in Tentative Identification.” Journal of Political Economy 102 (December): 1228-47. GORDON, ROBERT J. 1997. “The Time-Varying NAIRU and Its Implications for Monetary Policy.” Journal of Economic Perspectives 11 (Winter): 11-32. HARVEY, A.C. 1989. The Econometric Analysis of Time Series. Cambridge, Mass.: MIT Press. LEEPER, ERIC M. 1995. “Reducing Our Ignorance about Monetary Policy Effects.” Federal Reserve Bank of Atlanta Economic Review 80 (July/August): 1-38. LUCAS, ROBERT E. 1972. “Expectations and the Neutrality of Money.” Journal of Economic Theory 4 (April): 103-24. ———. 1976. “Econometric Policy Evaluation: A Critique.” Carnegie-Rochester Conference Series on Public Policy 1:19-46. MODIGLIANI, FRANCO, AND LUCAS PAPADEMOS. 1975. “Targets for Monetary Policy in the Coming Year.” Brookings Papers on Economic Activity 1:141-63. PHILLIPS, A. WILLIAM. 1958. “The Relation between Unemployment and the Rate of Change of Money Wage Rates in the United Kingdom, 1861-1957.” Economica 25 (November): 283-99. ROGERSON, RICHARD. 1997. “Theory ahead of Language in the Economics of Unemployment.” Journal of Economic Perspectives 11 (Winter): 73-92. SAMUELSON, PAUL A., AND ROBERT M. SOLOW. 1960. “Analytical Aspects of Anti-Inflation Policy.” American Economic Review 40 (May): 177-94. SARGENT, THOMAS J., AND NEIL WALLACE. 1976. “Rational Expectations and the Theory of Economic Policy.” Journal of Monetary Economics 2:169-83. SIMS, CHRISTOPHER A. 1986. “Are Forecasting Models Usable for Policy Analysis?” Federal Reserve Bank of Minneapolis Quarterly Review 10 (Winter): 2-16. SIMS, CHRISTOPHER A., AND TAO ZHA. 1996. “Does Monetary Policy Generate Recessions?” Yale University. Photocopy. STAIGER, DOUGLAS, JAMES H. STOCK, AND MARK W. WATSON. 1996. “How Precise Are Estimates of the Natural Rate of Unemployment?” National Bureau of Economic Research, Working Paper 5477, March. ———. 1997. “The NAIRU, Unemployment, and Monetary Policy.” Journal of Economic Perspectives 11 (Winter): 33-50. STIGLITZ, JOSEPH. 1997. “Reflections on the Natural Rate Hypothesis.” Journal of Economic Perspectives 11 (Winter): 3-10. TOOTELL, GEOFFREY M.B. 1994. “Restructuring, the NAIRU, and the Phillips Curve.” New England Economic Review (September/October): 31-44. WEINER, STUART E. 1993. “New Estimates of the Natural Rate of Unemployment.” Federal Reserve Bank of Kansas City Economic Review (Fourth Quarter): 53-69. PHELPS, EDMUND S. 1968. “Money Wage Dynamics and Labor Market Equilibrium.” Journal of Political Economy 76 (July/August): 678-711. 10. As the name suggests, a VAR model is one in which a vector of variables is “regressed” against its own lags. Thus, for example, one may define the vector x(t) to be the triple of output, inflation, and a short interest rate and estimate the system x(t) = A(1) x(t – 1) + A(2) x(t – 2) + … + A(n)x(t – n) + e(t), where e(t) is a (three-dimensional) disturbance, n is the number of lags, and A(1), … , A(n) are 3 x 3 matrices of coefficients to be estimated from the data. Typically, the estimates of the e(t)’s will not directly yield the fundamental shocks hitting the economy but, rather, linear functions of such shocks. However, Bernanke and Sims showed how one can recover the fundamental shocks by imposing a priori restrictions drawn from economic theory. 11. The curious reader is encouraged to read the Winter 1997 issue of the Journal of Economic Perspectives, which includes several recent papers on the subject. Of these, the studies by Staiger, Stock, and Watson (1997) and Rogerson (1997) are particularly critical of the NAIRU. More sympathetic views are developed by Stiglitz (1997) and Gordon (1997). Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 13 Taking Note of the Deposit Insurance Fund: A Plan for the FDIC to Issue Capital Notes L A R R Y D . W A L L The author is a research officer in the financial section of the Atlanta Fed’s research department. He thanks George Benston, Rob Bliss, Mark Carey, Jerry Dwyer, Bob Eisenbeis, Mark Flannery, Ed Kane, Michael Gordy, Paul Kupiec, Saikat Nandi, Will Roberds, Steve Smith, Ellis Tallman, and workshop participants at the Board of Governors of the Federal Reserve System and the Office of the Comptroller of the Currency for helpful comments. The views expressed are the author’s. A RE EXISTING REGULATORY POLICIES ADEQUATELY LIMITING TAXPAYERS’ EXPOSURE TO DEPOSIT INSURANCE LOSSES? LOOKING TO THE PAST, IF THERE HAD BEEN A CREDIBLE, INDEPENDENT ANSWER TO THIS QUESTION, THERE MIGHT HAVE BEEN FEWER LOSSES FROM THE THRIFT DEBACLE OF THE 1980S. BOTH REGULATORS AND CONGRESS MIGHT HAVE RESPONDED MORE RIG- OROUSLY TO THE THRIFT PROBLEM IF A MORE CREDIBLE SIGNAL HAD BEEN GIVEN ABOUT ITS SERIOUSNESS. Looking to the future, although some subsequent policy changes should help forestall such scenarios, breakdowns that would expose the taxpayer to losses remain possible. The likelihood that in the event of substantial bank failures taxpayer funds would have to bail out the deposit insurance fund has been used as an argument for continuing regulatory controls on what activities may be affiliated with banks. This article argues that the interests of both taxpayers and banks may be best served by developing an independent monitor of the insurance fund and outlines a proposal that would provide such a monitor. The proposal calls for the Federal Deposit Insurance Corporation (FDIC) to issue securities for which the promise of payment is contingent on the state of the insurance fund.1 14 For example, if the fund required taxpayer contributions to satisfy its deposit insurance obligations, then the securities would receive no payment. The notes would be called capital notes because their promised payments would depend on the level (or capital) in the deposit insurance fund. Although capital noteholders would necessarily be subject to risk, the primary purpose of the notes is not to substitute for bank-supplied funds in absorbing the risk of loss but to produce information about the risks facing the fund. This proposal provides a way of tapping private investors’ information.2 Private investors already gather a substantial amount of information about the state of the banking industry, and thus they are in a position to make informed judgments about the state of the deposit insur- Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 ance fund. However, investors currently have no incentive to focus on the implications of their knowledge for the insurance fund, nor do they have any organized vehicle for aggregating and expressing their views of the fund’s health. The capital note proposal gives investors both an incentive to examine the fund and a mechanism for expressing their views. Capital notes would also help regulators by providing an independent assessment of the risks facing the fund and by enhancing the incentives of senior regulators to protect the fund.3 Such supplemental information might also be useful to Congress in evaluating both the condition of insured intermediaries and the performance of the system for disciplining banks’ risk taking. The plan presented here is designed to produce useful information with little or no net cost to taxpayers and banks. It imposes virtually no direct costs on the taxpayers because the sole source of funds for paying the notes is the proceeds of deposit insurance premiums paid by insured banks. Although the plan imposes costs on banks, the net present value of this cost is likely to be near zero because receipts from issuing notes would be used to offset banks’ current insurance payment obligations. In the United States these receipts could be applied to payments on Financing Corporation (FICO) bonds.4 Indeed, the plan would have a positive net present value to banks to the extent that reducing policymakers’ concerns about the insurance fund could result in more extensive deregulation. A secondary rationale for this proposal is to encourage greater consideration of the use of carefully crafted financial market contracts to reduce government exposure to deposit insurance losses. The government will continue to bear some residual risk from the failure of depositories as long as it provides deposit insurance.5 However, through thoughtfully designed financial contracts, the government could enlist private-sector help in monitoring and perhaps even reducing its deposit insurance exposure.6 This recommendation is not presented as a substitute for mechanisms that monitor the riskiness of individual banks and limit losses at failed banks. It is designed instead to provide information about the overall state of the insurance fund, not about any individual bank. The discussion below focuses on the United States and is based on the current system for monitorAs long as the governing and disciplining ment is at risk of loss banks as well as the when depositories fail, system for distributing losses at failed some mechanism for banks. Thus, the prosignaling the magnivisions of the Federal tude of that exposure Deposit Insurance Corporation Imis desirable. provement Act of 1991 (FDICIA), such as those requiring prompt corrective action and least costly resolution, are assumed to remain in force. However, the capital note proposal has the potential for broader applications, as shown in Box 1, which considers the benefits of incorporating capital notes into alternative regulatory regimes such as crossguarantees and narrow banking. The remainder of this article reviews the need for a way to monitor the health of the fund, and lays out the capital note plan in greater detail, explaining how capital 1. For simplicity, the proposal assumes the existence of only one fund, but the plan could easily be extended to each of the two existing funds in the United States, the Bank Insurance Fund and the Savings Associations Insurance Fund. 2. The Federal Deposit Insurance Corporation Improvement Act of 1991 (FDICIA) provides for ex post monitoring of the fund’s condition via studies of excessively expensive bank resolutions and through its requirement for higher insurance premiums if the systemic risk exception is invoked. In contrast, the capital note proposal allows for ex ante information production by private market participants. 3. The federal bank regulatory agencies in the United States are the FDIC, the Federal Reserve System, the Office of the Comptroller of the Currency (OCC), and the Office of Thrift Supervision (OTS). 4. FICO was created by Congress to help finance the resolution of failed thrifts. Its obligations are to be met by the collection of insurance premiums from thrifts and banks. 5. Box 1 notes that the removal of de jure deposit insurance does not imply the end of de facto insurance. 6. Two casual suggestions proposed in a seminar as alternatives to the capital notes plan call for contracts that could produce the same information as capital notes. One suggestion was that the United States issue option contracts that are payable only if the deposit insurance fund requires a taxpayer bailout. The capital note proposal seems preferable to issuing options contracts because Congress could have a very difficult time understanding why it should make payments to optionholders when it also has to appropriate taxpayer money to bail out the insurance fund. Paul Kupiec suggested that the FDIC issue contracts that would require payments to the agency in the event the insurance fund needed additional funding. The capital note proposal also seems to have an advantage over Kupiec’s suggestion for issuing reinsurance contracts because the capital note plan would obtain the investors’ funds before a crisis rather than try to obtain the money after a crisis. However, while the capital note arguably offers distinct advantages, both of these alternative proposals seem capable of providing independent information about the state of the insurance fund and hence could result in a significant improvement over the current system of monitoring the fund. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 15 B O X 1 The Usefulness of Capital Notes under Alternative Regulatory Systems he capital note proposal for monitoring the health of the FDIC fund is set in the context of existing U.S. banking supervision and closure policies. A number of alternative approaches to bank supervision and closure have been proposed over the years. However, the attractiveness of the capital note proposal hinges on whether the government is at risk from bank failures. As long as the government is at risk of loss when depositories fail, some mechanism for signaling the magnitude of that exposure is desirable. The analysis that follows argues that none of the existing proposals for bank supervision and closure would eliminate the government’s exposure to loss, and some of the proposals may actually increase the government’s risk by shifting its exposure to unregulated entities. Thus, the benefits of implementing the capital note proposal are independent of the system of bank supervision and closure. However, the capital note proposal would be feasible only in systems that explicitly recognize a contingent government liability. T Proposals That Recognize Contingent Government Liability The capital note proposal would be both desirable and feasible with reform proposals that explicitly permit government deposit insurance as a backup to private-sector systems for monitoring banks and absorbing losses from failed banks. Cross-Guarantees. A deposit insurance reform proposal from Ely (1994) is based on cross-guarantees among banks. The idea behind this proposal is that the parties in the best position to evaluate a depository’s risk exposure are other depositories managing similar risks. The proposal has some historical support in the work of Calomiris (1990) on state-run deposit insurance systems in the 1800s. If Ely’s proposal worked as intended, then government guarantees would be unnecessary because no bank would fail with large losses to depositors. However, banks in such a system could collectively choose to take on risk, knowing that if enough of them failed the government would probably step in to protect depositors. Indeed, by tying the fortunes of many banks together, a cross-guarantee system could force systemic concerns and a government bailout if a subset of the depositories suffered sufficiently large losses. While the cross-guarantee system would not eliminate the government’s exposure to loss, this approach is not 16 inherently inconsistent with a government backup insurance fund that issues capital notes. Under such a system, the losses at failed banks would first be the responsibility of the other banks in the cross-guarantee system and would become a government problem only if the losses exceeded the capacity of the cross-guarantee system to absorb them. The backup government insurance scheme, including the capital note issue, could be funded by the individual crossguarantee funds. The note price and interest rate would then serve as a signal about the financial condition of the cross-guarantee systems. Specialized Guarantors and Puttable Subordinated Debt. The idea of shifting monitoring responsibility and the risk of loss to private parties unaffiliated with banks is shared by proposals by Kane, Hickman, and Burger (1993) and Wall (1989). The Kane-Hickman-Burger proposal shifts the risk of loss and responsibility for monitoring to a private surety. Wall’s proposal requires banks to either maintain a minimum amount of subordinated debt in a form that could be withdrawn or face automatic closure. Both of these proposals avoid some of the conflict of incentives facing crossguarantees by placing the responsibility and risk outside the banking system. Both systems would provide for marketdetermined early closure of depositories should their equity values erode gradually over time. However, both are vulnerable to the possibility that a sudden, very large drop in asset prices could change the incentive of the private monitors.1 Should the sudden loss exceed the value of a depository’s equity, the risk arises that private monitors’ claims will become more like equity than debt. That is, the value of the private monitor’s claim may be greater if it forbears from closing an institution in the hope that the depository will recover its financial strength. Both the Kane, Hickman, and Burger proposal and the Wall proposal give market participants an incentive to monitor banks and signal when individual banks are financially distressed; thus, both are similar in spirit to the capital note proposal. Neither plan would necessarily be operationally inconsistent with the capital note plan. The guarantors under the Kane-Hickman-Burger proposal could contribute to a government backup insurance fund in a manner similar to that suggested for the cross-guarantee system. Under the puttable debt proposal, the banks themselves would contribute to the government insurance fund. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 Proposals That Claim to Eliminate Contingent Government Liability Proposals designed to eliminate the government’s contingent liability are unlikely to accomplish their intended goal, as Benston (1995) notes. In any event, they would complicate the use of capital notes because of questions about which institutions are likely to receive de facto insurance coverage. Eliminate Deposit Insurance. Proposals to eliminate deposit insurance seem to obviate the use of capital notes. If the government is not at risk, then monitoring systems seem unnecessary. The problem with this approach is that eliminating statutory provisions for deposit insurance is not the same as eliminating either the expectation or reality that deposit insurance will be provided at failed depositories. Neither the states of Ohio or Maryland explicitly backed their deposit insurance systems, but when the funds went bankrupt the states bailed out the depositors. Similarly, banks have failed in numerous countries around the world and depositors were ultimately bailed out, despite the absence of deposit insurance or tight limits on the extent of de jure insurance coverage. The bottom line is that democratic governments in the later part of this century have often been unwilling to let depositors suffer. While eliminating deposit insurance statutes is not necessarily a feasible way of eliminating deposit insurance, it would nevertheless render the capital note plan unworkable. First, the existence of capital notes based on government deposit insurance liabilities would further undercut the market’s perception that deposit insurance had truly been eliminated. Second, it is not clear who would pay the interest on the notes. Without an insurance fund, nobody would pay insurance premiums that could go toward paying off the noteholders. Third, it would be less clear what set of government bailouts might place the noteholders at risk. Safe Bank Proposals. Another proposal that seems to eliminate risk to the FDIC is some version of the safe bank or narrow bank proposal (see, for example, Litan 1987 and Pierce 1991). Such a proposal would limit deposit insurance to accounts backed by short-term, highly liquid securities with low credit and market risk. These securities could be marked to market on a daily basis so that the exposure of the insurance fund to losses would be minimal. If such a proposal worked as intended, it too could eliminate the advantages of preferred stock in the insurance fund. The problem with safe bank proposals is their implicit assumption that the combination of making short-term, information-intensive loans and issuing short-term, highly liquid deposits in banks is a historical accident with no economic basis for its continued existence. However, the widespread combination of these functions in banklike institutions around the world suggests that there is some economic benefit in combining the two functions. Flannery (1994) argues that this combination is an efficient way of dealing with the agency costs associated with making shortterm, information-intensive loans. Rajan (1996) argues that businesses’ demand for large loans at short notice makes it desirable to have lending concentrated in institutions that specialize in managing liquidity, which is what deposittaking banks must do. Whether or not Flannery or Rajan are correct, or whether the loan-deposit combination exists for some other reason, the weight of banking practice worldwide suggests that banks cannot be neatly divided into a deposittaking function and a lending function. Yet, if this division is not possible, then limiting deposit insurance to narrow banks only means that banking problems currently troubling policymakers will reappear outside the narrow banks. Thus, while adopting the narrow bank proposal seems unlikely to make deposit insurance irrelevant, it nevertheless has the effect of eliminating regulation of those institutions for which it is relevant. The narrow bank proposal creates the same sort of operational problems for the capital note proposal as the elimination of deposit insurance—namely, that those institutions most likely to require government support are not formally covered by the system. 1. This vulnerability is less than it may appear based on a review of banks’ financial statements. Often, what appears, from the perspective of the financial statements, to be a sudden loss is actually a series of losses accumulated over a longer period. The losses only seem to have occurred suddenly because the bank deferred recognition of the losses until delay either no longer benefited the bank or release became unavoidable. For example, many banks deferred recognizing losses on loans to Latin American borrowers in the 1980s for a long period of time until the banks decided that recognition would be desirable. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 17 notes would reduce taxpayers’ risk in the event of deposit insurance losses. The discussion includes a consideration of some possible disadvantages. The Need for Monitoring rior to the creation of the Federal Deposit Insurance Corporation in 1933, both depositors and the government monitored the condition of individual banks. Depositors monitored banks because they were exposed to losses should the bank fail.7 The government played an important role by examining banks’ confidential records and certifying that the records did not contain adverse information. However, the creation of the FDIC, combined with the agency’s historic policy of seeking to protect all depositors, resulted in less depositor discipline, especially in weaker financial institutions. This relaxing of depositor monitoring shifted the An alternative to burden of disciplinenhanced government ing banks’ risk taking discipline of banks is to federal bank regulators. to shift risk back to Federal bank regthe private sector in ulators provided an an attempt to enlist adequate level of discipline over most of market discipline. the period from the mid-1930s through the mid-1970s, helped in no small part by restrictive regulations that boosted banks’ charter values and simplified banks’ operations by limiting their activities largely to gathering deposits and making short-term loans.8 However, regulators’ ability to maintain restrictive regulations has been eroding continually since the 1970s. Bank charter values have declined as the combination of improving technology (communication, data processing, and financial), increases in the level and volatility of interest rates, and changes in regulation has increased competition among banks and between banks and nonbank firms. The complexity of banks, especially the largest banks, has increased dramatically with the use of technology to provide a wide variety of financial services. As the task of supervising banks became more difficult, two supervisory problems surfaced. The first was that of conflicts in supervisory incentives, revealed during the thrift debacle of the 1980s.9 Kane (1989a, 1989b) points out that although bank supervisors are in effect agents for the taxpayers, their personal objectives often include other goals that are inconsistent with reducing taxpayer exposure to failed depositories.10 Congress sought to mitigate this conflict in incentives through passage of the FDIC Improvement Act of 1991. In particular, the act P 18 provides for prompt corrective action that prescribes a series of mandatory and optional regulatory responses to falling capital ratios.11 However, prompt corrective action depends in large part on the accurate measurement of banks’ financial condition, and such accurate measurement is not always feasible. A large fraction of a depository’s assets are either not traded at all or are traded in very illiquid markets where prices may vary from full-information value (Berger, King, and O’Brien 1991). The determination of the economic value of these assets is necessarily subject to potentially large measurement error. Thus, regulators still have room to exercise discretion so that, as Kane (1995) points out, incentive problems have not been eliminated by FDICIA.12 The second problem is how to supervise banking organizations effectively, given that changing technology has made these organizations far more complex. The banking, securities, and insurance industries have each used advances in technology and innovative legal approaches to offer products that are functionally similar to products offered by the other two industries but that have a very different regulatory status; for example, money market mutual funds are substituted for bank deposits. As a consequence, current systems of regulation that designate separate regulatory bodies for each of the different types of financial services are being made obsolete by developments in the financial marketplace. Whether regulators could adequately limit the risk to the FDIC fund in the emerging complex, highly competitive financial marketplace is an open question.13 Some analysts, such as Pierce (1991, 98-100), advocate severely restricting insured banks’ choice of assets, in part because they are pessimistic about regulators’ ability to monitor sophisticated financial services firms.14 An alternative to enhanced government discipline of banks is to shift risk back to the private sector in an attempt to enlist market discipline. FDICIA has sought to enlist depositor discipline through adoption of least costly resolution.15 This provision retains deposit insurance coverage up to the de jure coverage level of $100,000; however, the FDIC absorbs losses from larger deposits only if doing so reduces the overall cost of resolution to the agency. If this stipulation is enforced for all bank failures, it should substantially increase depositor monitoring, especially at the largest banks, which tend to be the most complex. However, FDICIA also provides for the suspension of least costly resolution in systemic risk situations, thus tending to reduce depositor monitoring, particularly at large banks. The ultimate effectiveness of least costly resolution in providing depositor discipline at large banks therefore remains an open question, in part because no very large bank has failed since the adoption of FDICIA.16 Even if systemic risk considerations (real or imagined) limit the potential effectiveness of depositor discipline, other opportunities for increasing market Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 discipline remain. One set of options would be to shift more of the losses to the private sector without giving market participants direct authority to force troubled banks to close. Examples of such options include increased issuance of subordinated debt, as proposed by Benston and others (1986, 192-96), and the use of private coinsurance to help set insurance premiums for individual banks. Market discipline in these instances arises from the risk premiums that financial markets would charge individual banks. The limitation of these proposals is that, while they give government regulators more time to close failing banks before the FDIC absorbs any losses, they do not directly address the problems of conflicting regulatory incentives or the problems of supervising very complex financial firms. Another set of options for increasing market discipline would give private parties both a larger fraction of the risk of loss in bank failure and the ability to close a failing bank. Examples of such proposals (which are discussed in Box 1) include those of Kane, Hickman, and Burger (1993) for private sureties and Wall (1989) for puttable subordinated debt. These strategies, which give private participants the authority to close any institution that cannot demonstrate its solvency or reasonable risktaking policies, go further to address the regulatory problems of conflicting incentives and the difficulty of supervising complex organizations. However, these proposals are limited in that sudden, very large losses can effectively force the conversion of the private monitors into de facto equityholders in a failed depository. Thus, in some circumstances the health of the FDIC would ultimately depend on government supervisors, even under the provisions of these proposals. Ultimately, then, although the primary system for protecting taxpayers from losses at financial intermediaries has been improved, and perhaps could be further 7. This monitoring occurred for all national banks and those state-chartered banks that did not have state-sponsored deposit insurance. 8. Charter or franchise value is the net present value of the economic profits resulting from owning a bank charter. Keeley (1990) provides evidence of a decline in franchise value. Demsetz, Saidenberg, and Strahan (1996) provide evidence that banks with higher franchise values take less risk. 9. See Kane and Yu (1995) for evidence that the U.S. Federal Home Loan Bank Board (FHLBB) dramatically understated the true reserves of the Federal Savings and Loan Insurance Fund for the period 1985-88. They further argue that the misinformation supplied by the FHLBB played an important role in actions taken by Congress in 1987. Cole and Eisenbeis (1996) provide evidence that delays in closure increased the cost to taxpayers. Other discussions that suggest that regulatory problems have contributed to actual or potential taxpayer losses include Kane and Kaufman (1993) on problems in Australia; De Krivoy (1995) on Venezuela; Cargill, Hutchison, and Ito (1995) on Japan; and Gall (1996) on Brazil. 10. Several theoretical models reach varying conclusions about optimal closure policies. Examples of these studies include Acharya and Dreyfus (1989), Davies and McManus (1991), Kumar and Morgan (1994), Mailath and Mester (1994), and Noe, Rebello, and Wall (1996). However, the previously cited empirical work appears to suggest that Kane’s model better explains most past regulatory forbearance than do these models. 11. For a presentation of the structured early intervention and resolution proposal that formed the basis for prompt corrective action provisions of FDICIA, see Benston and Kaufman (1988). For a discussion of FDICIA’s key safety and soundness provisions, see Wall (1993). 12. Jones and King (1995) and Peek and Rosengren (1996) show that a large fraction of the banks with high insolvency risk in their respective samples would not have come under prompt corrective action requirements given existing capital standards. 13. Chan, Greenbaum, and Thakor (1992) question whether fairly priced deposit insurance is feasible in a competitive marketplace. Their results may be put in more general terms by recognizing that in order to own a bank charter in the United States, a firm must generally participate in the deposit insurance system and accept other regulatory requirements. When restated in this manner, their results imply that an intermediary will prefer to be chartered as a bank only if the rents from owning a charter (including any government subsidy through deposit insurance) exceed the regulatory costs (both safety and soundness costs and social regulation costs) of owning a charter. As the government reduces or eliminates the deposit insurance subsidy, the ratio of rents to taxes from owning a bank charter will become less favorable and less intermediation will take place through the banking system. However, even if one believes that the ratio of rents to taxes from owning a bank charter should be increased to encourage charter ownership, it is not obvious that the most efficient way to do so is by subsidizing bank risk taking. 14. Box 1 discusses the narrow bank approach in greater detail but reaches the conclusion that the approach is unlikely to reduce FDIC’s exposure to the degree suggested by its proponent. 15. Another legislative change is a depositor preference provision, which states that domestic depositors (including the FDIC, which stands in the depositor’s place when it makes an insurance payment) in a failed depository are entitled to full repayment before nondeposit creditors receive any payment. In theory this provision should lead to an increase in monitoring by nondeposit creditors, at least in those states that did not previously follow depositor preference. In practice, the increased risk to nondeposit creditors can be substantially reduced by having the bank put up collateral to back the nondeposit liability. See Osterberg (1996) for a review and empirical analysis of depositor preference laws. 16. FDICIA also provides regulators with various powers, such as final net settlement, designed to enhance their ability to implement least costly resolution. The credibility of least costly resolution could be enhanced if regulators developed and advertised plans to implement least costly resolution in a way that was unlikely to generate systemic risk. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 19 improved, it remains vulnerable to failure. Drawing on an engineering analogy, when the failure of a primary system could have catastrophic consequences, backups must be incorporated into the design to reduce the risk of damage. Given the potentially high stakes from a failure of the system to limit banks’ risk to the fund, Kane (1995) argues that it would be desirable to monitor the condition of the FDIC. He contends that information about the FDIC’s condition would help taxpayers assess their exposure as well as assist labor markets in evaluating the performance of top government officials. In particular, Kane recommends implementing some mechanism that would encourage production and dissemination of private information about the performance of the deposit insurer. He argues that “the trick would be to make sure that projected cash flows respond to loss exposures occasioned by inadequacies in federal loss-control and pricing policies” (1995, 454). However, while his specific proposals would produce direct information on individual banks or subsets of banks, they would produce only indirect information on the fund per se. These proposals are essentially substitutes for FDICIA’s approach to dealing with individual bank risk and do not constitute backups.17 The proposal that follows is a low-cost backup system. A Plan for a Low-Cost Backup System he plan outlined here calls upon the deposit insurer to issue capital notes as a mechanism for signaling taxpayers about the riskiness of the deposit insurance fund. Subsequent discussion explains the reason for the plan’s structure, shows how the plan will meet its goals, and analyzes some potential problems. 1. The FDIC’s insurance fund would issue coupon notes that pay interest and principal except in circumstances stipulated below. These notes would vary in maturity from one to five years. New note issues would occur at least semiannually. 2. Interest payments would be suspended if the fund received a loan from the government to finance the resolution of failing banks. If such a suspension occurred then insurance premiums would automatically increase to a historically high level (such as a minimum premium of $0.25 per $100 of deposits). 3. The capital notes’ right to interest (future and cumulative unpaid interest from prior periods) would be terminated when the fund reaches zero, and any source of funds other than depository insurance premiums would be appropriated by Congress to absorb deposit insurance losses. 4. The quantity of capital notes would be determined by the FDIC, based on the goal of maximizing the information content of the notes about the health of the insurance fund. Depositories, their affiliates, and the accounts they manage for other parties (such as trust accounts and mutual funds) would not be permitted to hold the notes as an investment. T 20 Small inventories of the notes for securities affiliates that act as market makers would be permitted. 5. Every time the FDIC issued new notes it would report on the status of the fund and the risks facing the banking system. The report would also contain an estimate of the distribution of potential losses to the insurance fund (under at least three sets of economic assumptions) and a description of the methods used to estimate losses. Further, if the yield to maturity at the time of issue of a note exceeded that of comparable maturity investment-grade securities, then three reports would be required.18 First, the FDIC would be obligated to report to Congress on the risk that the insurance fund would become inadequate, on any steps the agency planned on taking to reduce the risk, and on any additional legislation that would be helpful in reducing the risk to the insurance fund. Second, the Secretary of the Treasury would be required to report on the budget implications of the risk to the fund. Finally, the General Accounting Office (GAO) would be required to review the problems, regulatory responses, and any appropriate congressional responses. 6. Part of the pension plan provided for the FDIC directors would be invested in capital notes while the directors were in office; for example, one-quarter of a director’s contribution to the pension fund during his or her term would be automatically invested in the notes. A director would be able to sell the notes over some time period after retiring from office.19 In order to reduce the exposure to banks not directly under its regulation, the FDIC would have the authority to examine any insured bank at its sole discretion. 7. The proceeds from the capital note issue would go to repaying FICO bonds.20 The premiums levied on banks to pay the interest and principal on FICO bonds would be proportionately reduced. The plan outlined here relies on signals from market pricing to indicate when the insurance fund is likely to be under stress. The reason for requiring the FDIC to return to the financial markets at least every six months to issue new securities is that an issue of long-term capital notes could have an illiquid secondary market so that signals obtained from note prices could contain considerable noise. The provision forcing the FDIC to sell notes regularly gives the financial markets a periodic opportunity to provide information about the quality of the insurance fund. Notes are issued at several maturities to provide a time profile of the risks facing the insurance fund. For example, suppose investors thought banks were taking excessive risks but believed these risks would not become apparent until the next recession. In this case, notes maturing in one year may show very little, if any, risk premium, but notes maturing in five years would trade at Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 substantially higher yields than comparable Treasury securities, reflecting concerns about future losses. Not only would requiring the FDIC to provide an evaluation of the fund’s status with each new issue help noteholders and independent analysts in assessing the adequacy of the fund; it would also provide a way to review the regulatory agencies’ outlook on the current condition of the banking system and suggest how they might handle problem banks. The capital notes are intended to measure the insurance fund’s solvency, but payment of interest on the notes would be suspended if the fund had to turn to the Treasury to obtain sufficient liquidity.21 Suspension of interest payments is designed to ensure that solvency problems are not hidden by the FDIC under the pretense that the only issue is the liquidity of the fund. The Treasury, as supplier of these funds, should charge interest for providing the required liquidity. However, capital noteholders would retain their claim to future interest in this case so long as Congress did not appropriate taxpayer funds to cover failed banks’ losses.22 Stipulating that directors of the FDIC invest in the capital notes would give them an added incentive to follow market signals.23 The optimal amount of investment for directors would balance their incentive to protect the taxpayers from deposit insurance losses with not giving them such a stake that they would avoid using the fund when failure to do so would have adverse consequences for the overall economy.24 This part of the proposal would expose the directors of the FDIC to the risk of loss and thereby reduce the expected value of their compensation. This reduction should be offset with higher salary or benefits unless some evidence indicates that FDIC directors are currently overcompensated. How the Proposal Would Reduce Taxpayers’ Risk key concern in issuing capital notes is whether they would actually reduce the risk that the deposit insurance fund would require taxpayer support. Three aspects of this question will be considered in the following discussion: First, how would the notes contribute to better decisions by the regulatory agencies? Second, how would the notes increase the probability that Congress would take appropriate oversight actions if the need arose? Finally, what limitations might adversely impact the informativeness of the pricing signal? Capital Notes and the Regulatory Agencies. Because the capital note plan lacks a mechanism to compel changes at the regulatory agencies, to what extent would the plan, in practice, change regulators’ actions? If regulators always use their discretion in a manner contrary to the interests of taxpayers, then a plan that fails to force changes in regulators’ behavior cannot help protect the taxpayers. However, existing theory does not indicate that regulatory agencies must always act contrary to taxpayers’ interests; theory merely states that conflicting incentives will sometimes produce suboptimal results. Regulators’ preferred policies will in many cases be optimal, especially if timely information allows them to prevent a problem from occurring. In other cases, while the A 17. Kane discusses two specific alternatives: “marketing cash flows from uninsured funding instruments [for example, subordinated debt] or FDIC coinsurance agreements” (1995, 454). These alternatives would provide information about the performance of the various individual banks or groups of banks that constitute the portfolio of exposures facing the deposit insurer. As such, they would require changes in the way losses are currently distributed when a bank fails and might entail changes in the way the closure decision itself is made. 18. This provision could be strengthened by including a stipulation that would automatically increase deposit insurance premiums when capital note rates exceeded those of investment grade notes. An increase in insurance premiums might not be the most efficient way to reduce the risk that taxpayer funds would be used to resolve bank failures. However, the prospect of higher premiums could help regulators and banks reach a consensus on which measures (such as higher capital requirements or implementation of least-costly resolution) would best reduce risk to the fund. 19. For example, the directors could sell the notes evenly over a four-year period, with 25 percent of the notes eligible for sale after one year, 50 percent eligible after two years, 75 percent after three years, and the entire investment eligible after four years. 20. Alternatively, the funds could be used to bolster the insurance fund if the plan were adopted in a country other than the United States. 21. The FDIC currently has the authority to borrow from the U.S. Treasury, but the insurance fund is obligated to repay the loans. A congressional appropriation is required for the FDIC to obtain funds from the Treasury that the insurance fund would not be obligated to repay. 22. For example, suppose the fund had $20 billion in assets and faced immediate liquidity needs from the failure of some depositories of $40 billion but would also acquire a claim on the failed institutions’ assets with a current market value of $35 billion. In this case the fund might need a loan of approximately $20 billion to resolve the depositories, but, after selling off the assets the fund should have approximately $15 billion. 23. The use of the regulators’ pension fund as a mechanism for generating financial accountability follows a recommendation by Kane (1996). 24. The claim that failures in the financial sector could spill over with adverse consequences for the real economy is controversial. For example, Benston (1995) and Kaufman (1996) argue that deposit insurance is not needed to protect against systemic risk. The link between FDIC directors’ compensation and note values could be strengthened if one did not believe such a spillover is possible. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 21 personal benefit of pursuing a suboptimal strategy will dominate under the current incentive system, it will be less likely to dominate under some other incentive structures. Capital notes are likely to influence regulatory actions both by providing the agencies with additional information and by changing their incentives. Risk premiums on the notes would serve as signal about a number of different problems that could be addressed by the regulatory agencies.25 An increase in risk premiums (a decrease in note prices) could be a signal that the riskiness of the banking system has increased. For example, changes in regulation or technology could allow banks to expand their product offerings into riskier financial activities. In this case regulators could seek to insulate the fund more effectively from the additional risk, or they could limit banks’ ability to take the additional The capital note proposal risk. The risk premiwould prompt investors to um might also increase in response to scrutinize the underlying a reduction in the economic realities facing market’s confidence the fund and ignore disinthat the FDIC would follow least costly formation proffered by resolution. In this those with a stake in a case, the regulatory misinformed public. agencies might wish to review the feasibility of their strategies to implement least costly resolution and then better advertise their plans to financial markets. A third possibility is that deposit insurance premiums or the insurance fund have become inadequate in relation to the level of risks facing banks. In this case, the FDIC could choose to increase its insurance premiums or, if the fund has reached its maximum permissible level (as has recently been the case), could seek permission from Congress to increase the fund’s size. Admittedly, these market signals to the regulatory agencies about problems facing the fund are unlikely to alert regulators to completely unrecognized problems. However, an increase in the risk premium would provide independent information about the severity of a threat to the insurance fund. This independent information could be useful both for setting priorities within the regulatory agencies and for helping to overcome political opposition to measures the regulators deem necessary to protect the fund. The plan changes the FDIC directors’ incentives to engage in forbearance in several ways. First, the proposal imposes costs on regulators while they are in office, especially FDIC administrators, by requiring reports from the FDIC and an examination by the GAO if the yield on notes exceeds that of comparable maturity investment22 grade securities. Few administrators like to be forced to discuss publicly possible problems that are occurring on their watch, and even fewer would want the GAO to second-guess their past policies and their plans for future action. Second, senior regulators, including FDIC directors, could not expect to be able to hide their mistakes until either they had found new employment, leaving their successors to clean up the problem, or banks had become healthy as a result of taking large gambles. The price or rate on capital notes would serve as a clear signal to Congress and potential employers of significant problems facing the insurance fund while the regulators were still in office, as suggested by Kane (1995). Thus, senior regulators who engaged in forbearance could not count on leaving their regulatory agency with their good reputations intact. Third, requiring directors of the FDIC to own capital notes would put part of their own wealth at risk if the FDIC engaged in forbearance. As noted above, the amount of notes held by the FDIC’s directors might be restricted to serve the social goal of not excessively discouraging the FDIC from invoking the systemic risk exception in severe cases, even if that action bankrupts the fund. Another limitation of this advantage is that the FDIC shares regulatory responsibility over banks with the OCC and the Federal Reserve. The FDIC has the unconditional right to enter any bank it insures and is thus able to protect the fund from forbearance by these other agencies. Further, the proposal creates subtle changes in regulators’ incentives to deal with emerging problems before they threaten to impose significant losses on the fund. The potential cost to regulators if notes should start trading at a substantially lower level becomes an incentive to prevent note prices from falling. One way regulators might address this incentive would be to deal more aggressively with potential threats to the insurance fund before economic losses occur. They might also try to reduce the risk premiums on the notes by assuaging market uncertainty about the state of the fund with more information. This impetus to disseminate information would be reinforced by the requirement that the FDIC report on the state of the insurance fund each time it issues new notes. The process of providing additional information will further encourage timely action by regulators: not only will it force them to recognize potential threats at an earlier stage, but it will also help Congress and taxpayers to monitor the regulatory process. Capital Notes and Congress. Issuing capital notes would give voters and their representatives a clear signal about the health of the deposit insurance fund. Banks in weak financial condition have a strong incentive to produce accounting numbers that give a misleading impression of good health. Bank auditors have mixed incentives but they are hired by the banks, so they have a strong Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 motivation to approve the rosiest picture possible of opaque financial information as long as the analysis is consistent with the legal standards for auditors’ performance. The three federal bank regulatory agencies also face conflicting incentives, but experience around the world shows that when serious financial problems arise, regulators often look for ways to defer taking action. In contrast, this capital note proposal would prompt investors to scrutinize the underlying economic realities facing the fund and ignore disinformation proffered by those with a stake in a misinformed public. Potential investors in the notes stand to suffer substantial losses if they underestimate the likelihood that payments will be deferred or eliminated. Thus, the price or rates set on these notes could serve as a more accurate signal about the true condition of the fund than reports from banking and regulatory systems might provide. The capital note proposal would encourage timely oversight and legislation by Congress because it would lower the cost of obtaining information and thus reduce congressional members’ incentive to remain uninformed (or underinformed) about the state of the deposit insurance fund. Members may choose to remain underinformed if obtaining information about the true condition of the fund is costly and the likely political benefit from obtaining the information is small.26 Capital notes could serve as a low-cost warning signal that would allow legislators to focus attention on the fund primarily during periods when gains from legislative changes would be the largest. The benefits to Congress from note price signaling are reinforced by the stipulation that would require mandatory reports by the FDIC, the Treasury, and the GAO should the notes be issued at rates exceeding those of comparable investment-grade securities. Congressional members’ incentives to monitor the condition of the deposit insurance fund is affected by the capital note plan in two ways. First, by reducing the cost of recognizing when the fund poses a threat to the taxpayers, the proposal makes it harder for legislators to plausibly deny their responsibility to take action to reduce the threat. The rates charged on capital notes send a signal that may be more easily understood by the electorate than the conflicting testimony of “experts.” In this way the rates may be used in political campaigns by challengers to bolster their chances of defeating members who shirk their obligations to taxpayers. Second, capital noteholders could also provide a political counterweight to lobbying depositories. In the early stage of a threat to the insurance fund, noteholders would have an incentive to lobby for more aggressive regulatory action to preserve the value of their claims. While appealing in principle, this advantage of the plan is only of second-order importance because banks would probably retain greater political clout and, if the insurance fund becomes sufficiently weak, noteholders’ primary aim becomes persuading Congress to bail them out along with the fund. The Pricing of Capital Notes. All of the arguments for the advantages of the capital note proposal depend on the prices of the notes sending a clear signal about the condition of the fund. But both the FDIC and Congress could take steps that would reduce the information content in the notes’ prices. The FDIC could mislead capital noteholders by withholding information to prevent the notes from signaling most potential problems. This scenario is unlikely, however. Regulators collect and publish a substantial volume of data on banks’ financial condition, and most of the banking system’s assets are in publicly traded banks that are already closely watched by stock analysts. Further, bond rating agencies currently evaluate the credit quality of many bank debt issues, including those of virtually all of the very large banking organizations, and this knowledge could be applied to rating notes issued by the bank insurer. Thus, the FDIC would have a very difficult time concealing unrealized losses to the banking system that would threaten the fund’s solvency.27 A more serious threat to the effectiveness of the plan is the potential for Congress to bail out the capital noteholders along with the deposit insurance fund. There is no way to prevent noteholders from petitioning for a bailout or to bind the hands of future Congresses. However, capital noteholders are being paid specifically to accept the risk that the FDIC fund could become impaired, and, in general, these noteholders would be capable of bearing that risk. Thus, as a group, the capital noteholders would not have a strong case for a bailout. Even given the potential for a bailout of noteholders, the price of the notes would remain depressed until the bailout became a certainty, so prices could still be used as a signal that problems remained. The primary limitation of such signals would be in interpreting changes in capital note prices and rates when the value of deposit insurance claims has a substantial probability of exceeding the fund’s assets. These changes could reflect variations both in the fundamental condition of the fund and in the probability of a bailout. 25. For a survey of the evidence that various market signals contain information about banks’ financial condition, see Gilbert (1990) as well as a more recent study by Flannery and Sorescu (1996). For a contrary opinion, see Simons and Cross (1991) and Randall (1993). 26. For example, in those states where no congressional action is appropriate, the political benefit of obtaining additional information about the state of the fund may be close to zero. 27. An example (perhaps the only example) of such a deception would be a large bank taking very substantial losses from undisclosed fraud or unauthorized risk taking that was recognized by the examiners but not the bank’s auditors. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 23 Potential Disadvantages of the Capital Note Proposal robably the biggest disadvantage of the plan outlined here is that the FDIC would bear a continuing responsibility to pay interest on the capital notes, implying a need for higher deposit insurance premiums for depositories. However, even if higher premiums were a pure loss to depositories, it could be argued that the gains to taxpayers from better regulatory decisions and improved congressional oversight would exceed the loss if banks passed the higher premiums to their customers. Moreover, the net cost to depositories from the issuance of capital notes is likely to be small. The proceeds from the note issue would go to pay down the FICO note issue and reduce insurance premiums on those notes. What cost, if any, the plan would impose on banks depends on the amount and timing of lower insurance premiums stemming from reduced FICO obligations, the amount and timing of payments by the capital notes, and the discount rate applied to those future payments (which should reflect banks’ marginal cost of funds). (Rates likely to be paid on capital notes are discussed in Box 2.) Another possible disadvantage of the capital note plan, it might be argued, is that if the notes indicated a high degree of risk to the deposit insurance fund, the public’s confidence in the deposit insurance fund could erode, thereby precipitating a systemic problem. However, the FDIC has a $30 billion line of credit at the U.S. Treasury, and Congress may appropriate additional funds as a loan or as a grant to resolve failed banks; thus, plunging capital note values would not necessarily indicate that the insurance fund would be unable to honor its obligations to insured depositors. Any public misperception that the note values measure the fund’s ability to honor its claims could be corrected, providing a low-cost solution to the problem. The capital notes are a only signal about whether the insurance fund is likely to need government help, not an indication of its ability to pay off insured depositors. P 24 Conclusion ank regulators and deposit insurers around the world have repeatedly failed to resolve foundering depositories in a timely manner. In the United States the risk that a financial breakdown could lead to a taxpayer bailout of the deposit insurance fund has been cited to justify current regulatory controls and the imposition of inefficient taxes for social welfare purposes. Despite some regulatory changes in the 1990s to protect taxpayers from future debacles, however, widespread failures could still expose taxpayers to losses. The proposal outlined in this article provides a way to monitor the deposit insurance fund—through capital notes issued by the FDIC—that would better serve the interests of both taxpayers and banks. Because the interest paid on capital notes would be suspended if the fund required a loan from the Treasury or eliminated if taxpayer funds were contributed to offset deposit insurance losses, noteholders would have more incentive to take action should the risk of loss to the taxpayers become substantial. Capital notes would provide taxpayers and their congressional representatives clear signals about the health of the fund and would change the incentive structure facing senior regulators. Banks should benefit under the capital note proposal because the receipts from issuing the notes could be used to reduce banks’ insurance payment obligations. In addition, by relieving some of the concerns policymakers have about the insurance fund, the capital note plan could lead to a more deregulated environment for banks. Finally, both regulators and Congress may profit from a better method for assessing the status of the insurance fund as they struggle to cope with safety and soundness questions arising from the integration of the banking industry with other finance-related industries. B Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 B O X 2 A Rough Calculation of the Likely Interest Rate on Capital Notes he capital notes proposed in this article are intended to be an effective, low-cost signal of the condition of the FDIC fund. Potential investors in capital notes, however, are concerned about more than the risk that the FDIC will have insufficient funds to honor its promises to the noteholders. Investors will demand that the note’s pricing reflect the current value of the default-free term structure. Investors will also demand compensation for risks other than economic default, such as the possibility that notes could become illiquid or the risk that payments could be suspended or terminated for political reasons rather than because of the FDIC’s inability to make timely payments. If the notes are to be an effective signal of the state of the FDIC fund, analysts need to be able to filter out most of the changes in the status of the fund from other reasons for rate changes. Further, if the notes are to be low-cost signals, then the nondefault risk premiums attached to the notes should be small. No one can say with certainty exactly how the notes will be priced because they currently do not exist. However, varieties of notes already in the market, such as municipal revenue notes and bonds as well as corporate notes and bonds, share many of the same characteristics as capital notes. This box compares the features of capital notes with those of other types of debt securities to obtain a rough estimate of likely prices if capital notes were to be issued. T Determinants of the Prices of Capital Notes The rate on capital notes may be reasonably approximated by the following function: Rate on notes = f{default-free term structure, tax status, default risk premium, premium for the risk of interest deferral, information cost premium, liquidity premium, risk of political interference}. Default-Free Term Structure. The default-free term structure represents the payment to investors for deferring consumption; it is an important element in the pricing of all fixed-income securities. The default-free term structure for obligations denominated in U.S. dollars is usually approximated by the term structure of U.S. Treasury securities. Capital notes and corporate notes of the same default risk would be expected to respond in a similar fashion to movements in Treasury securities. Municipal revenue notes also respond to movements in the Treasury rate, but their rate movements are dampened because they are not subject to federal taxation. Tax Status. Income from ordinary corporate notes and capital notes is subject to federal income tax. However, interest on municipal notes, which are obligations of state and local governments, are not subject to federal tax. This tax break allows municipal notes to trade at lower yields, imuni, than otherwise identical corporate notes and capital notes, itax. The formula for determining the lower yield is imuni = itax(1 – t), where t is the federal income tax rate of the marginal investor. Default Risk Premium. Municipal revenue debt issues, corporate debt issues, and capital notes are all subject to default risk, and numerous studies show that this risk is priced in the debt markets.1 Municipal revenue securities are used to finance a specific project and are backed solely by the revenue from that project. The debt issues are expected to default if the revenues are insufficient to repay the noteholders. Corporate debt is typically backed by the cash flow of the entire corporation, but these cash flows must pay the firm’s operating costs and its other debt issues. Corporate notes will default if the firm’s cash flow is insufficient. One difference among the three types of debt obligations is the potential for noteholders to hedge changes in credit risk. In theory, a corporate debtholder could perfectly hedge her exposure to changes in a firm’s credit risk by taking a short position in the firm’s stock. In practice such a hedge is unlikely to be perfect for a variety of reasons; for example, payments to the noteholders may depend on the decisions of a bankruptcy court (especially if the firm has multiple classes of debt outstanding), or the market value of the firm’s assets may be subject to discontinuous movements. Nevertheless, a large fraction of the risk could be hedged by taking appropriate short positions in the firm’s stock. 1. For example, Altman (1989) shows that bond ratings are generally negatively correlated with default probabilities and yields. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 25 B O X 2 ( C O N T I N U E D ) The default risk on capital notes depends on the magnitude of losses to the deposit insurer and on the insurer’s ability to recover those losses via higher premiums, both of which in turn depend on the performance of a portfolio of banks. Not all banks have publicly traded stock, but the FDIC’s biggest exposures are concentrated in banks with traded stocks. Thus, investors could hedge a significant portion of the default risk on capital notes. However, the fraction of risk covered by such a hedge would almost certainly be less than that for a typical corporate note of comparable risk. Most municipal projects tend to be specialized with a high degree of idiosyncratic risk. Some revenue bonds and notes, such as those associated with hospitals or power facilities, may be somewhat correlated with the stocks of firms in the same industry, but other revenue debt issues (such as those associated with toll roads and university dormitories) may have few, if any, natural hedges. In general, the proportion of municipal revenue note default risk that can be hedged is likely to be substantially lower than either corporate notes or capital notes. Risk of Deferral of Interest Payments. Even if investors ultimately receive full payment of interest and principal, these payments may be delayed if the issuer is in financial distress. The capital note proposal provides for a suspension of payments if the FDIC obtains a loan from the Treasury. Payments on a corporate note may be suspended if the firm enters bankruptcy proceedings. The bankruptcy court will typically place a stay on payments to the firm’s noteholders, at least until either an acceptable restructuring plan has been approved by the court or the firm is liquidated. Municipal revenue note payments are also subject to holds if the project enters bankruptcy. Information Cost Premium. Investors rarely, if ever, know the true probability of default on a debt security. Instead, they form their best estimate based on publicly available information and on the private information they collect. Each investor then determines the risk premium required to cover her private individual estimate of default. If the investor believes that other market participants (the issuer or other investors) have superior information, then she will demand a premium to cover the risk that she is being sold an overpriced security. If the competition among unaffiliated investors to produce information helps in accurately pricing debt obligations, those investors that have superior information will, at least in the long run, be the marginal buyers of an issue. However, the possibility that the issuer is acting on superior information may make prices less accurate measures of actual default risk. Issuers with 26 superior information may sell notes when debt markets overvalue them and defer selling notes when the market undervalues the securities. Investors recognize that issuers are likely to have superior information and will demand higher prices if they believe issuers are using their information advantage to sell overvalued notes. Corporate debt issuers have an incentive to issue overpriced notes because mispricing gains accrue to the shareholders. Furthermore, these issuers often have some discretion in the timing of their issues that would permit them to exploit mispricings. However, the value of most new corporate bonds and notes is relatively insensitive to inside information because the notes involve very low risk. Hence, the maximum possible mispricing gains are usually small.2 Municipal project managers may also gain some operational flexibility by issuing mispriced debt issues that have too low a promised interest payment given their risk; hence, they have some incentive to time their issues to coincide with market mispricings. However, the incentive for municipal projects to issue mispriced notes is likely to be lower since the persons responsible for the issue are in a weaker position to capture part of the mispricing gains. Further, municipal projects often require some legislative approval, a stipulation that could sharply reduce the managers’ scope for timing an issue. Investors in capital notes, by contrast, should have minimal concern about the FDIC manipulating the timing of its note issues, given that the proposal leaves little discretion about timing. The FDIC may have some discretion over the amount of each issue, but the directors of the FDIC have very little ability to capture any rents associated with mispriced notes. Liquidity Premium. An important component of note pricing is investors’ beliefs about their ability to sell the notes in the secondary market at a price that fairly approximates the notes’ true value. While some investors buy notes with an intention of holding them until maturity, others plan on selling their holdings before the note matures. Further, even those investors that plan on holding notes until maturity will place positive value on the option of being able to sell their holdings before they mature. Thus, if two notes differ by only their expected liquidity in the secondary market, the note that promises a more active secondary market will require lower yield. Notes and bonds tend to have far less active secondary markets than comparable stock issues, in large part because many investors follow a “buy and hold” strategy. The lack of a secondary market may be less important in short-term issues than in the longer ten- to thirty-year bonds. Shorter Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 maturity issues, such as with capital notes, allow “buy and hold” investors who want to cash out of the notes to receive their investment back from the issuer on a shorter time scale. Thus, the liquidity of capital notes is likely to be less important to their pricing than the liquidity of longer term securities would be. When financial economists are analyzing the pricing of various bonds, a common proxy for the probable depth of the secondary market for a bond is its issue size. Corporate issues generally range from tens of millions of dollars to over $1 billion (in rare cases involving longer-term bonds), with larger corporations often issuing amounts exceeding $100 million. The size of some municipal revenue bonds exceeds $100 million, but many revenue bonds have a par value under $10 million.3 The amount of capital notes to be issued will be determined by the FDIC, in large part based on liquidity considerations. The ability to issue up to $8 billion will give the FDIC considerable flexibility; for example, if the agency issued the full $8 billion, sold notes every six months with each sale consisting of five issues maturing annually over the next five years, then each issue could be over $250 million.4 Thus, based on the issue’s size, the liquidity premium on capital notes may not be any larger than on comparable corporate or revenue notes (after tax adjustment). Risks about the Amount and Timing of Payments. In most cases, payments received by noteholders fairly reflect borrowers’ ability to make full and timely payments in accordance with their debt contract. However, under certain circumstances, noteholders may not receive full and timely payments even though the note issuer has the economic capacity to make the payments. Conversely, in rare cases noteholders may receive larger and more timely payments than the economic capacity of the issuer would permit. Investors will demand a risk premium to cover the potential that the issuer will default for reasons other than economic capacity, and they will accept a lower rate to the extent that they anticipate a bailout should the note issuer become unable to pay. Both types of distortions will reduce analysts’ ability to use note rates to identify the changes in a note issuer’s economic ability to pay. Corporate notes have limited exposure to both distortions. Corporations have used bankruptcy proceedings to avoid honoring burdensome obligations (usually labor contracts or obligations arising from civil suits). Corporations also have a very small possibility of receiving a government (national or local) bailout if the political authorities are unwilling to accept the consequences of a bailout. However, the main source of distortion may be unpredictable deviations from the absolute priority rule in bankruptcy proceedings. Corporate obligations typically provide for varying degrees of seniority if the firm should fail, but this seniority is often not followed strictly; for example, equityholders may receive a payment even though the junior creditors are not fully repaid.5 Some deviations from absolute priority are likely to be anticipated ex ante by noteholders, but certain classes of creditors may receive unexpected gains or losses due to unexpected deviations from absolute priority. Municipal revenue bonds and notes may also suffer from both types of distortions. Local political authorities may change the ground rules that govern a project’s operation in a way that reduces the revenues (for example, by allowing more competition) or increases the expenses associated with a project. Conversely, they may bail the project out by using other sources of revenue. Capital notes are subject to both possible risks. The FDIC could suspend payments on the notes by borrowing from the U.S. Treasury when other resolution methods may have reduced the FDIC’s outlays for failed bank resolutions. Similarly, Congress could appropriate taxpayer money to cover FDIC losses even though the losses could have been covered by current and future bank insurance premiums. The one important difference between capital notes and the other two types of notes is that a suspension of payments on the other notes may trigger a loss of managerial control. The bankruptcy court will assume control over the major decisions made by a corporation or municipal project in bankruptcy, and the owners and managers of a corporation may lose total control of the firm in bankruptcy proceedings. Hence, the managers of corporations and municipal projects face potentially large costs if they enter 2. See Smith (1986) for a survey of empirical studies of corporate security issuance. 3. In a study of municipal general obligation bonds, Kidwell, Koch, and Stock (1987) examine the reoffering yield on bonds issued between 1978 and 1980. They find that the reoffering yield on issues of less than $15 million is influenced by state-specific factors but that larger issues seemed to be sold into a national market. 4. The total number of issues outstanding at any given time would be thirty. At any given time five issues would mature in six months, five issues in one year, four issues in one and a half years, four issues in two years, three issues in two and a half years, three issues in three years, two issues in three and a half years, two issues in four years, one issue in four and a half years, and one issue in five years. 5. For a recent analysis of absolute priority rule violations see Longhofer and Carlstrom (1995). Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 27 B O X 2 ( C O N T I N U E D ) into bankruptcy to defer, reduce, or eliminate payments to noteholders. The FDIC loses no control rights over the insurance fund if it borrows from the Treasury or receives a congressional appropriation to cover losses under the capital notes proposal. If the FDIC did borrow from the Treasury, it would be subject to some costs, and the automatic increase in bank insurance premiums would probably generate political heat for the FDIC; however, these costs are likely to be less than those faced by corporate and municipal note issuers. Similarly, Congress may face political opposition to an “unnecessary appropriation” of funds to cover FDIC losses. Thus, the odds of note payments being adjusted for noneconomic reasons appears to be greatest for capital notes. However, these risks would seem less likely to be priced in to the notes if the FDIC fund appears to be very strong and shows little probability of default for economic reasons. If the insurance fund faces minimal resolution costs relative to its existing fund, then the FDIC and Congress would have to manufacture a situation that would justify suspension or termination. However, as the cost of resolving failures rises relative to the size of the insurance fund, so does the ability of the FDIC and Congress to justify suspensions and terminations of note payments. If the fund is in sufficient financial distress, relatively small changes in the assumptions about liquidity needs and resolution costs may be sufficient to justify suspension or termination of interest payments. Thus, if default for noneconomic reasons has any significant impact on capital note pricing, it is most likely to be at a point when the probability of the FDIC needing a loan or congressional appropriation has become significant but such actions are not yet a certainty. The implication of this analysis is that if a significant premium is required for these nondefault risks, it will tend to accentuate the notes’ sensitivity to economic default risks.6 Expected Pricing of Capital Notes The above analysis suggests that the pricing of capital notes would probably be similar to that of similar corporate and municipal revenue notes after adjusting for tax differences, with comparable default risk ratings. According to this analysis, corporate notes seemed least subject to various nondefault risks. Corporate notes may have slightly higher risk premiums because of the risk that their issuers have superior information, but otherwise corporate notes have equal or lower risk levels (for any given note rating class). Capital notes may have more political risk than municipal revenue notes in some cases, but otherwise their risk premiums would appear to be equal or lower than comparably rated revenue notes. Thus, this analysis indicates that, to a first approximation, capital notes with low default risk ought to trade at rates somewhere between comparably rated corporate notes and revenue notes, after adjusting for their varying tax status. 6. An offsetting influence would be the potential for a congressional bailout of the capital noteholders. However, the noteholders’ case for a bailout is weakened by the fact that they are being paid to bear this risk. Further, noteholders’ prospects for a bailout are likely to be unclear until the need for a congressional appropriation to cover FDIC losses becomes apparent. Thus, if noneconomic risks are significant, the risk that noteholders will get less than they deserve will probably dominate when the FDIC fund’s condition first starts to deteriorate. This situation implies that capital notes will be sending the desired signal while there may still be time to reduce the fund’s losses. However, changes in the note rate may be a less reliable indicator of the FDIC fund’s condition in the later stages of financial deterioration when the fund becomes very weak. 28 Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 R E F E R E N C E S ACHARYA, SANKARSHAN, AND JEAN-FRANCOIS DREYFUS. 1989. “Optimal Bank Reorganization Policies and the Pricing of Federal Deposit Insurance.” Journal of Finance 44 (December): 1313-33. FLANNERY, MARK J. 1994. “Debt Maturity and the Deadweight Cost of Leverage: Optimally Financing Banking Firms.” American Economic Review 84:320-31. ALTMAN, EDWARD I. 1989. “Measuring Corporate Bond Mortality and Performance.” Journal of Finance 44 (September): 909-22. BENSTON, GEORGE J. 1995. “Safety Nets and Moral Hazard in Banking.” In Financial Stability in a Changing Environment, edited by Kuniho Sawanmoto, Zenta Nakajima, and Hiroo Taguchi. New York: St. Martin’s Press. BENSTON, GEORGE J., ROBERT A. EISENBEIS, PAUL M. HORVITZ, EDWARD J. KANE, AND GEORGE G. KAUFMAN. 1986. Perspectives on Safe and Sound Banking: Past, Present, and Future. Cambridge, Mass: MIT Press. BENSTON, GEORGE J., AND GEORGE G. KAUFMAN. 1988. “Risk and Solvency Regulation of Depository Institutions: Past Policies and Current Options.” Monograph Series in Finance and Economics, no. 1. Salomon Brothers Center for the Study of Financial Institutions and Graduate School of Business Administration, New York University. BERGER, ALLEN N., KATHLEEN KUESTER KING, AND JAMES M. O’BRIEN. 1991. “The Limitations of Market Value Accounting and a More Realistic Alternative.” Journal of Banking and Finance 15 (September): 753-83. CALOMIRIS, CHARLES W. 1990. “Is Deposit Insurance Necessary? A Historical Perspective.” Journal of Economic History 50 (June): 283-95. CARGILL, THOMAS F., MICHAEL M. HUTCHISON, AND TAKATOSHI ITO. 1995. “Lessons from Financial Crisis: The Japanese Case.” In The New Tool Set: Assessing Innovations in Banking, 101-109. Proceedings of the Thirty-First Annual Conference on Bank Structure and Competition, Federal Reserve Bank of Chicago. CHAN, YUK-SHEE, STUART I. GREENBAUM, AND ANJAN V. THAKOR. 1992. “Is Fairly Priced Deposit Insurance Possible?” Journal of Finance 47 (March): 227-46. COLE, REBEL, AND ROBERT A. EISENBEIS. 1996. “The Role of Principal-Agent Conflicts in the 1980s Thrift Crisis.” Real Estate Economics 24 (Summer): 195-218. DAVIES, SALLY M., AND DOUGLAS A. MCMANUS. 1991. “The Effects of Closure Policies on Bank Risk Taking.” Journal of Banking and Finance 15 (September): 917-38. DE KRIVOY, RUTH. 1995. “Lessons from Financial Crisis: Evidence from Venezuela.” In The New Tool Set: Assessing Innovations in Banking, 162-70. Proceedings of the ThirtyFirst Annual Conference on Bank Structure and Competition, Federal Reserve Bank of Chicago. DEMSETZ, REBECCA S., MARC R. SAIDENBERG, AND PHILIP E. STRAHAN. 1996. “Banks with Something to Lose: The Disciplinary Role of Franchise Value.” Federal Reserve Bank of New York Economic Policy Review 2 (October): 1-14. ELY, BERT. 1994. “Financial Innovation and Deposit Insurance: The 100 Percent Cross-Guarantee Concept.” Cato Journal 13 (Winter): 413-36. FLANNERY, MARK J., AND SORIN M. SORESCU. 1996. “Evidence of Bank Market Discipline in Subordinated Debenture Yields: 1983-1991.” Journal of Finance 51 (September): 1347-77. GALL, NORMAN. 1996. “Behind the World’s Biggest Bank Failure.” Institutional Investor 30 (September): 83-95. GILBERT, R. ALTON. 1990. “Market Discipline of Bank Risk: Theory and Evidence.” Federal Reserve Bank of St. Louis Review (January/February): 3-18. GREENBAUM, STUART I. 1996. “Twenty-Five Years of Banking Research.” Financial Management 25 (Summer): 86-92. JONES, DAVID S., AND KATHLEEN KUESTER KING. 1995. “The Implementation of Prompt Corrective Action: An Assessment.” Journal of Banking and Finance 19 (June): 491-510. KANE, EDWARD J. 1989a. “Changing Incentives Facing FinancialServices Regulators.” Journal of Financial Services Research (September): 265-74. ———. 1989b. The S&L Insurance Mess: How Did It Happen? Washington, D.C.: Urban Institute Press. ———. 1995. “Three Paradigms for the Role of Capitalization Requirements in Insured Financial Institutions.” Journal of Banking and Finance 19:431-59. ———. 1996. “Foundations of Financial Regulation.” In Rethinking Bank Regulation: What Should Regulators Do? 308-33. Proceedings of the Thirty-Second Annual Conference on Bank Structure and Competition, Federal Reserve Bank of Chicago. KANE, EDWARD J., JAMES C. HICKMAN, AND ALBERT E. BURGER. 1993. “A Plan for Private-Federal Deposit-Insurance Partnership.” Center for Credit Union Research, University of WisconsinMadison, Working Paper. KANE, EDWARD J., AND GEORGE G. KAUFMAN. 1993. “Conflict in Deposit-Institution Regulation: Evidence from Australia.” Pacific-Basin Finance Journal 1 (March): 13-29. KANE, EDWARD J., AND MIN-TEH YU. 1995. “Measuring the True Profile of Taxpayer Losses in the S&L Insurance Mess.” Journal of Banking and Finance 19:1459-77. KAUFMAN, GEORGE G. 1996. “Bank Failures, Systemic Risk, and Bank Regulation.” Cato Journal 16 (Spring/Summer): 17-45. KEELEY, MICHAEL. 1990. “Deposit Insurance Risk and Market Power in Banking.” American Economic Review 80 (December): 1183-1200. KIDWELL, DAVID S., TIMOTHY W. KOCH, AND DUANE R. STOCK. 1987. “Issue Size and Term-Structure Segmentation Effects on Regional Yield Differentials in the Municipal Bond Market.” Journal of Economics and Business 39 (November): 339-47. KUMAR, RAMAN, AND GEORGE E. MORGAN. 1994. “Bank Regulatory Triage: Optimal Closure Rules and an Economic Explanation for Forbearance.” Virginia Polytechnic Institute and State University, Working Paper 93-15, June. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 29 LITAN, ROBERT E. 1987. What Should Banks Do? Washington, D.C.: Brookings Institution. PIERCE, JAMES L. 1991. The Future of Banking. New Haven, Conn.: Yale University Press. LONGHOFER, STANLEY D., AND CHARLES T. CARLSTROM. 1995. “Absolute Priority Rule Violations in Bankruptcy.” Federal Reserve Bank of Cleveland Economic Review 31 (Fourth Quarter): 21-30. RAJAN, RAGHURAM G. 1996. “Why Banks Have a Future: Toward a New Theory of Commercial Banking.” Journal of Applied Corporate Finance (Summer): 114-28. MAILATH, GEORGE J., AND LORETTA J. MESTER. 1994. “A Positive Analysis of Bank Closure.” Journal of Financial Intermediation 3 (June): 272-99. NOE, THOMAS H., MICHAEL J. REBELLO, AND LARRY D. WALL. 1996. “Managerial Rents and Regulatory Intervention in Troubled Banks.” Journal of Banking and Finance 20:331-50. OSTERBERG, WILLIAM P. 1996. “The Impact of Depositor Preference Laws.” Federal Reserve Bank of Cleveland Economic Review 32 (Quarter 3): 2-11. PEEK, JOE, AND ERIC S. ROSENGREN. 1996. “The Use of Capital Ratios to Trigger Intervention in Problem Banks: Too Little, Too Late.” New England Economic Review (September/ October): 49-58. 30 RANDALL, RICHARD E. 1993. “Lessons from New England Bank Failures.” New England Economic Review (May/June): 13-38. SIMONS, KATERINA, AND STEPHEN CROSS. 1991. “Do Capital Markets Predict Problems in Large Commercial Banks?” New England Economic Review (May-June): 51-56. SMITH, CLIFFORD W., JR. 1986. “Investment Banking and the Capital Acquisition Process.” Journal of Financial Economics 15 (January/February): 3-29. WALL, LARRY D. 1989. “A Plan for Reducing Future Deposit Insurance Losses: Puttable Subordinated Debt.” Federal Reserve Bank of Atlanta Economic Review 74 (July/August): 2-17. ———. 1993. “Too-Big-to-Fail after FDICIA.” Federal Reserve Bank of Atlanta Economic Review 78 (January/February): 1-14. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 What’s Really New about the New Forms of Retail Payment? W I L L I A M R O B E R D S The author is a research officer and senior economist in charge of basic research in the Atlanta Fed’s research department. He thanks Gerald Dwyer, Robert Eisenbeis, Charles Kahn, Frank King, Bruce Smith, and Larry Wall for helpful conversations and comments on earlier drafts. T O MOST AMERICAN CHARGE.” THIS CONSUMERS, THE WORD PAYMENT IS SYNONYMOUS WITH “CASH, CHECK, OR FAMILIAR TRIAD IS NOW BEING AUGMENTED, HOWEVER, WITH A VARIETY OF ALTERNATIVE PAYMENT METHODS, INCLUDING DEBIT CARDS, REMOTE BANKING, STORED-VALUE OR “SMART” CARDS, AND “ELECTRONIC CASH.” There is much that is new about these alternative methods of payment, which have come about through the widespread availability of technologies that were unavailable even a decade ago. The apparent novelty of some of these new forms of payment has led some observers to conclude that the new forms will be different from the old not only in a technological but also in an economic sense. For example, one recent analysis offered readers the following warning: “Don’t think the differences between traditional currency and the coming electronic versions are as superficial as updating our economic lexicon. The changes underway in our monetary system will fundamentally alter how consumers interact with businesses and how businesses interact with one another” (Flohr 1996, 74). While such a prediction may hold true in some limited respects, it would be difficult to believe that the costly lessons of economic history are not relevant for electronic payments. This article examines the question of whether, from the standpoint of economic theory, there is or will likely be anything new about these new forms of payment. The discussion begins with a description of some of the conflicts of interest that confront all types of payment sys- 32 tems in market economies. The article then considers in some detail why traditional forms of payment, such as checks and banknotes, represent reasonable solutions to these conflicts of interest and outlines some shortcomings of the traditional forms. The article also analyzes the economic characteristics of the new forms of payment and explains why they differ little and in some cases not at all from more traditional forms. Finally, the article briefly considers some of the policy issues raised by the introduction of the new payment methods. The Conflict of Interest between Buyers and Sellers arious forms of payment have evolved as a means of resolving the natural conflict of interest between buyers and sellers of goods (or services). In developed economies, a buyer of a given commodity only rarely possesses a commodity that a seller wishes to consume. In the absence of a double coincidence of wants, a buyer must offer a seller a good that the seller believes can be used to purchase other goods. In market exchanges the natural self-interest of the seller is to provide goods to the buyer in exchange for V Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 something of equal or greater value. Since the seller is not directly consuming the good offered in payment, however, he or she may often have difficulty discerning the quality of this good. For example, a check written on insufficient funds looks exactly like a check written on good funds. This sort of problem, known by the term adverse selection, has been studied extensively by economists. Compounding the problem of adverse selection is the problem of moral hazard. In market exchanges moral hazard can occur when buyers undertake deliberate actions, unobservable by the seller, that would undermine the value of goods the buyer offers in exchange. For example, a buyer could have sufficient funds to cover a check written for a certain purchase but then remove the funds from the account before the check clears. In market settings the incentive problems of adverse selection and moral hazard are not necessarily one-sided. Buyers do not always know the quality of goods and services they are purchasing, and sellers can in many cases undertake unobservable actions to lessen the value of the items or services being sold. While these problems are quite serious in some markets (real estate, for instance), it can be argued that in most cases such problems are probably less severe for the buyer than for the seller. It is inherently easier to judge the quality of groceries, for example, than to judge the quality of the check used to pay for the groceries. For this reason, the discussion below will concentrate on the incentive problems faced by the seller and not the buyer. All payment systems must address such incentive problems by providing timely, accurate information concerning the value of the goods offered in payment. To the extent that modern electronic technology can improve the speed and accuracy of communication, such technology can provide less costly solutions to these incentive problems. The use of technology is unlikely to provide an automatic solution to these problems, however. For example, a company offering a news service over a computer network such as the Internet might require payment by some sort of funds sent over the Internet. The company needs to send the product (news) to its customers immediately in order for the product to have value. However, even if the customers can send electronic “checks” very quickly over the Internet, the company still needs to know with a reasonably high probability whether the checks are good. The presence of incentive problems leads sellers to prefer means of payment that provide them maximum assurance concerning the value of the assets received in exchange, even in cases where such assurance can be costly or inconvenient for the buyer. In some cases, the seller can gain such assurance by withholding delivery of the good until the value of the payment is verified. In other cases, such delays are either not feasible (as in the case of the hypothetical news service described above) or are uneconomical (as in the case of goods that have a very small value). Buyers’ preferences are in many ways opposed to sellers’. Buyers naturally prefer immediate use of the goods they have purchased with a minimum of cost and inconvenience to themselves. Dishonest buyers (those intentionally offering to pay with something worth less than the value of the purchased good) would prefer that the The presence of incentive seller know as little problems leads sellers to as possible about the value of the good prefer means of payment offered in exchange, that provide them maxiand in many cases, mum assurance concernas little as possible about their own idening the value of the assets tity and financial conreceived in exchange, even dition. Honest buyers in cases where such assur(those offering something worth at least ance can be costly or the value of the purinconvenient for the buyer. chased good) would prefer that sellers have access to enough information (but generally no more)to distinguish them from dishonest buyers. Cash, which today means government-issued currency, provides a time-honored if somewhat imperfect solution to the buyer-seller conflict. From the buyer’s perspective, cash is desirable because it is relatively convenient and almost perfectly anonymous.1 From the seller’s perspective, cash is desirable because it eliminates the need to evaluate the true worth of assets offered by the buyer. Despite its time-honored popularity as a transactions medium, cash carries with it its own disadvantages. Since cash does not bear interest it loses value over time as long as inflation rates are positive. Holders of cash also forgo the interest that would accrue if the cash were held as an interest-bearing asset. Hence, people using cash pay an implicit tax on their cash holdings as long as interest rates are positive. There are also substantial costs associated with handling large amounts of cash, and the anonymous nature of cash encourages theft, counterfeiting, and its use in illegal activities. Payment by Check Solves One Problem with Another. Since the Civil War, checks have been the principal alternative to cash for retail payments in the United 1. See, for example, Townsend (1989) or Williamson and Wright (1994) for a formal discussion of the informational role of cash in anonymous transactions. Cash also serves as a numeraire, that is, a good whose price is always equal to one and therefore can be used to determine the relative prices of other goods. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 33 States. While checks have been a successful payment mechanism, the use of checks to resolve the buyer-seller conflict of interest is in itself somewhat paradoxical. When paying for a good or service by check, a buyer transfers a claim on the assets of a bank or similar institution to the seller.2 The check itself represents the transfer of a fixed-value debt claim (deposits) on the assets of the bank on which the check is drawn. In most cases, the writer of the check does not know the “true” or market value of the bank’s assets. Thus, payment by check apparently compounds one adverse-selection/moralhazard problem (that between the buyer and the seller) with another (that between the depositor and the bank). The paradox of payment by check is resolved when a check is presented to the bank on which it is drawn and settled by transfer of some reserve commodity (cash or the Alternative forms of payment equivalent) from the bear their own costs resultbuyer’s bank to the ing from factors such as seller or the seller’s bank. The act of setrisks associated with delayed tlement is proof to settlement, the physical and the seller that the interest costs of clearing and buyer does have ownership of sufficient settlement, the costs of onvalue to pay for the line verification systems, and purchase and is proof the risks associated with to the buyer that the assets of the bank are counterfeiting. “good,” or sufficient to fund settlement. Under current U.S. banking law, the value of virtually all deposits of less than $100,000 is guaranteed by deposit insurance, so a bank’s ability to exchange a dollar in deposits for a dollar in cash is effectively guaranteed for small depositors. For these depositors the act of settlement carries with it no information concerning the liquidity of their own deposit claims. However, checkable deposits evolved before the establishment of the governmental banking safety net. One can argue that the informational value of settlement was more important prior to the advent of governmental guarantees. Also, the fact that many of the (proposed) new forms of payment involve exchanging uninsured claims on relatively unregulated institutions makes the informational value of settlement potentially critical for these new forms. The following discussion considers some of the informational aspects of check-based payment in more detail. Advantages of Payment by Check. Why should banks and similar financial intermediaries issue primarily short-term (demandable or zero-maturity) debt in the form of deposits, and why should people pay for goods and services by transferring these debt claims? Modern financial theory suggests a number of answers to the first question but has less to say on the second. 34 On the first question, Diamond (1991) offers one possible explanation. Diamond examines the effect of debt maturity on the adverse selection problem faced by the bondholders of a generic firm, who may have difficulty discerning the true worth of the firm’s assets. He argues that firms holding high-quality (higher-yielding) assets may wish to restrict their debt issue to short maturities. By showing a willingness to roll over its debt, a firm can signal its belief that future news about the firm’s earnings will be good. In some cases, however, Diamond suggests that the use of short-term debt as a signaling device can be too costly. If bondholders have difficulty obtaining accurate information concerning the quality of a firm’s assets, then the cost of rolling over short-term debt may force a firm into liquidation, even if the firm is fundamentally solvent.3 Despite this disadvantage, Flannery (1994) argues that the issue of short-term debt makes sense for highly leveraged financial firms (those with a high debt/equity ratio) such as banks. Flannery points out that a high degree of leverage can induce a sharp conflict of interest between a firm’s debt- and equityholders, even when debtholders (such as depositors) have good information concerning the quality of the firm’s assets. The issuance of short-term debt can help to ameliorate this conflict of interest, from the viewpoint of the debtholders, by limiting the firm’s ability to acquire assets that are too risky. In the case of financial firms, this feature of short-term debt is especially useful because financial firms’ assets (for example, bank loans) are subject to many risks that cannot be controlled by contracts or covenants. Another justification for banks’ issuance of shortterm debt is described by Calomiris and Kahn (1991), who emphasize the role of short-term debt in controlling the moral hazard risk faced by depositors in the absence of deposit insurance. In the event that a bank becomes insolvent and has to be liquidated, holders of bank debt face (again, absent deposit insurance) the risk that the bank’s management will undertake actions to dilute the value of the debtholders’ claims.4 By issuing debt that is redeemable in cash on demand at par (at some prespecified value), Calomiris and Kahn argue that banks give debtholders the option of forcing the bank into liquidation early, before the value of the debtholders’ claims can be diluted. This “put option” feature of bank deposits increases people’s willingness to hold bank debt in the presence of moral hazard risk. In summary, each of the analyses cited above provides a strong rationale for banks’ issuance of primarily short-maturity or puttable (demandable at par) debt. Issuance of demandable or short-maturity debt helps to ameliorate the conflicts between bank debtholders (including depositors) and equityholders (or bank management) resulting from leverage, adverse selection, and moral hazard. Other things being equal, these conflicts Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 can be minimized if depositors know they can walk into their bank at any time and exchange their deposit claims for a fixed amount of cash.5 In modern banking systems depositors do not have to rely solely on the demandable nature of bank debt for protection of their interests. Banks in most developed countries are supervised and regulated by governmental agencies, whose mission is to limit banks’ risk-taking behavior. While the par demandability of most bank deposits is required by law, the deposits themselves are usually backed by a governmental safety net of deposit insurance and, if necessary, by banks’ direct access to central bank credit via a discount window or a similar lending facility. It is worth noting, however, that, despite these more recent developments, the nominal form of the deposit contract has not changed for hundreds of years. This fact suggests that par demandability of deposits continues to be important in reassuring depositors that their interests are being protected.6 On the second question—why people pay for goods and services by transferring short-term debt claims— Calomiris and Kahn (1991, 509) suggest that the puttable feature of bank debt makes it a natural choice for use in transactions. For the reasons outlined above, bank debt must be essentially demandable at par. It would then seem there is little to be lost, and much to be gained, if bank debt claims are transferred from buyer to seller and then immediately redeemed. In other words, the convention of payment by transfer of bank debt that is demandable at par simultaneously solves two information problems by providing a high degree of assurance to depositors concerning the quality of bank assets and to sellers concerning the value of claims (for example, checks) they have received in exchange. Payment by transfer of bank debt therefore constitutes a “natural” solution to the two-dimensional conflict of interest between banks and depositors and between buyers and sellers. In this sense it is not surprising that the U.S. payments system historically evolved so as to emphasize transfers of claims on banks (such as checks) as an alternative to cash payment. However, the term natural in this instance does not mean inevitable. There are many examples, both historical and contemporary, in which leveraged financial firms have issued large amounts of par demandable debt without such debt being used as a transactions medium. Calomiris and Kahn (1991, 509) note that the debt of Roman banks was demandable but not accepted as a form of payment. Wall (1989) and Flannery (1994, 321) point out that the debt of modern finance companies is often puttable or contains put-option-like features designed to protect debtholders. However, such debt is not commonly accepted as a transactions medium. Economic history also provides many examples in which debt used as a medium of exchange was not shortmaturity or demandable. Longer-maturity notes known as bills of exchange were widely used as a form of payment among merchants until the twentieth century (see, for example, Braudel 1982, 138-48, or Cuadras-Morató and Rosés 1995). There were certain difficulties associated with this practice, however. The most critical problem was that, in the event of a default by the party on which the bill was drawn, the legal recourse of those parties who had accepted the bill as payment was in many cases quite limited. Consequently, in cases in which the bill issuer defaulted, parties (other than the issuer) using the bill as a means of payment were expected to provide payment by some other means. In practice this drawback led to such bills being used for payments only between parties who had long-standing business relationships or other grounds to trust one another’s ability to pay.7 2. Throughout the article the term bank will be used to indicate both banks and other depository institutions (such as thrifts and credit unions) that offer similar services. 3. In some cases, however, such liquidity problems can be overcome via a lender-of-last-resort arrangement. See, for example, Kahn and Roberds (1996). 4. In the vernacular, such actions are described by the phrase “take the money and run.” 5. A downside of par demandability of bank deposits is that it can lead to bank runs. However, in Calomiris and Kahn’s view, in the absence of regulation runs may be necessary in order to control moral hazard risk. Various other rationales have been offered for demand deposits. For example, Jacklin (1987) suggests that demand deposits can work as a sort of insurance contract against the risk of depositors having to consume earlier rather than later. Gorton and Pennacchi (1990) hypothesize that the demandable nature of deposit contracts may help to insulate depositors from adverse fluctuations in the market value of banks’ assets when such fluctuations result from inaccurate information about banks’ future earnings. However, in contrast to the theories described in the text, these theories do not suggest why bank debt would have a natural role as a transactions medium. 6. The idea that deposit contracts naturally tend to take the form of demandable debt is reinforced by recent experience with money market mutual funds, as described in Collins and Mack (1994). In theory, these funds differ fundamentally from banks because (1) they are required to hold a narrow class of short-term, liquid assets and (2), in contrast to bank deposits, the value of each share in the fund is marked daily to the market value of the fund’s assets. In practice, certain funds’ stated share values have at times diverged from the market value of their assets, causing their shares to be viewed more like debt deposit contracts. 7. Prepaid phone cards represent a modern-day example of a transactable debt instrument that cannot be converted to cash on demand. Naik (1996) recounts various problems that have been associated with the use of these cards. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 35 Traditional Alternatives to Cash and Check Payments ayment by check has many drawbacks, some of which are described below. Attempts to circumvent these difficulties have led to the use of other forms of payment. This section will describe some traditional alternatives to check payment. Credit Transfers. An obvious drawback to check payment is that checks do not constitute “good funds” unless they have been cleared and settled. The gap between payment and settlement poses a risk to a seller if the seller delivers goods or services before a check payment becomes final. In some countries this drawback of checks has contributed to the disuse of checks and the predominance of giro or credit transfers.8 In giro transactions the buyer of a good or service initiates payment by instructing her bank to arrange for the appropriate sum to be debited from her own account and credited to the seller’s bank account. Provided that the seller does not deliver a good or service until payment has been made, this form of transaction eliminates some of the risk the seller faces at a cost of less convenience to the buyer. Historically, giro transactions have not been widely employed at the retail level in the United States.9 Banknotes. Another disadvantage of payment by check is that the clearing and settling of checks entails substantial costs such as physical costs of clearing and settlement and costs associated with the use of non-interestbearing reserves in settlement. If each check transaction had to be settled one-for-one by transfer of non-interestbearing reserves, checks would bear the same implicit tax as cash. If some checks can be settled on a net basis or through correspondent arrangements, then the use of checks can economize on the use of reserves. However, the requirement that every check transaction be cleared and settled means that payment by check still imposes some implicit tax, though obviously less than if the same payments were made with cash.10 Some of the costs associated with clearing and settling checks can be abated by the use of privately issued banknotes. Although such banknotes are no longer used today, they were widely used in the United States during earlier periods. The term banknotes refers to bank-issued debt that is issued in circulating or “bearer” form and is convertible on demand into cash.11 From the standpoint of a seller of goods and services, the key distinction between checks and banknotes is that the latter is a debt claim issued directly by the bank and not a transfer of a debt claim initiated by a buyer. Banknotes also differ from government-issued fiat currency because their value derives from the value of the private issuer’s assets and not from the monetary authority of a sovereign government. As long as a banknote cannot be counterfeited and as long as a seller believes that the note-issuing bank is willing to exchange its notes on demand for cash, then banknotes can resolve the conflict between buyer and P 36 seller by making the buyer’s creditworthiness essentially identical to that of the issuing bank. And as long as other people are willing to accept a banknote in exchange at its par value, there is no need for each transaction to be settled by exchange of cash. This feature of banknotes makes them particularly useful for transactions in which the time or money cost of clearing and settlement makes payments by check impractical. In theory, banknotes can circumvent the conflict between buyer and seller by creating a form of deposit that does not have to be cleared and settled through the banking system in order to be useful for transactions. In practice, however, the use of banknotes as a transactions medium has been associated with at least two serious problems. First, the issue of banknotes does not in and of itself resolve the conflict (discussed in the previous section) between holders of the issuing institution’s debt (for example, noteholders and depositors) and the institution’s equityholders or managers. If banking laws, regulations, and customs insufficiently restrain the ability of equityholders and/or management to dilute the value of debtholders’ claims, then a payments system based on banknotes can be ineffective. However, history suggests that it is possible to create systems of banking practices, laws, and regulations that would provide noteholders with a high degree of confidence in the value of the banknotes.12 The effect of these restrictions has often been to place strict restraints on the types of assets that can be used to back banknote issues.13 Second, the relatively anonymous nature of banknotes also introduces a new dimension of risk into market transactions—the moral hazard associated with counterfeiting activities. If banknotes are issued in untraceable bearer form, then this anonymity provides strong incentives for counterfeiting.14 If counterfeit notes are accepted by sellers and presented to the issuing institution, then the institution faces a difficult choice. If the issuer fails to redeem the counterfeit notes, then it may undermine public confidence in the value of its legitimate notes. On the other hand, if the issuer redeems the counterfeit notes it must absorb the resulting loss, again possibly undermining public confidence in its notes. As is the case with cash, the anonymity of banknotes carries with it certain other advantages and disadvantages. The advantages include convenience and privacy during transactions, and the disadvantages include encouragement of theft and illicit activity. In the United States, banknotes circulated widely until the Civil War and continued to circulate until 1935. During the Civil War, banknotes issued by institutions other than national (federally chartered) banks were essentially taxed out of existence, and stringent regulations were placed on the issue of notes by national banks (see Timberlake 1993, 86-88; Friedman and Schwartz 1963, 20-23). The legal authority for issue of banknotes by Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 national banks expired in 1935 (Friedman and Schwartz 1963, 422).15 Debit Cards and Credit Cards. Credit cards and debit cards represent more recent alternatives to the use of checks and cash.16 Debit card transactions are functionally similar to check transactions.17 When a buyer pays for a good or service using a debit card, the buyer authorizes the seller to transfer funds from the buyer’s account to the seller. There is an important distinction, however, when debit card transactions take place on-line. In on-line transactions, funds are immediately deducted from the buyer’s account. This immediacy provides sellers with almost complete assurance against the moral hazard and adverse selection risks associated with check clearing and settlement.18 However, this assurance comes at a cost. The costs of constructing and maintaining a dedicated on-line verification system makes this form of payment inefficient for some small-value transactions. Caskey and Sellon (1994, 90) report that the direct cost of on-line debit card payments for small-value (grocery store) transactions is slightly less than the cost of check payments but still substantially greater than the cost of cash payments. And, because the on-line system directly accesses buyers’ bank accounts, each on-line debit card transaction has to be authorized by the buyer, typically by entering a PIN (personal identification number) at a retail terminal. Credit card transactions superficially resemble debit card transactions but are different in terms of their eco- nomic function. As with on-line debit cards, sellers of goods and services usually accept credit cards in payment only after the transaction has been authorized by an online verification system. The distinguishing feature of credit card transactions is that they do not represent a direct transfer of funds between buyer and seller. Rather, funds flow from the card-issuing institution to the seller. The card issuer is then responsible for collecting the debt incurred by the buyer. The problem of judging the creditworthiness of the buyer is thus transferred from the seller to the card issuer. While credit cards are a convenient and relatively secure means of payment, Caskey and Sellon (1994, 90) report that using a credit card is by far the most expensive method of payment for small-value transactions. New Forms of Payment he foregoing discussion suggests that there is much room for improvement in the area of retail payments. Cash is convenient and anonymous, but it bears an implicit tax and is subject to theft and illicit use. Various alternative forms of payment bear their own costs resulting from factors such as risks associated with delayed settlement, the physical and interest costs of clearing and settlement, the costs of dedicated on-line verification systems, and the risks associated with counterfeiting. These problems, combined with the advent of new computer and communications technologies, provide economic incentives for the creation of new methods of payment. T 8. This practice has been most notable in Germany. See, for example, Bank for International Settlements (1993, 161-62). 9. In recent years electronic credit transfers have been widely used for certain other types of payments, however, such as direct deposits of payrolls, government benefit payments, and corporate payments to vendors and contractors. See Bank for International Settlements (1993, 442). 10. Checkable accounts in the United States have also been subject to a legal reserve requirement. See Feinman (1993) for a historical summary of reserve requirements in the United States. Prior to the Federal Reserve System’s involvement in the check payments system, it was common for banks to pass along the costs of check clearing and settlement by discounting the value of checks drawn on other banks. Duprey and Nelson (1986) present a detailed description of this practice, known as nonpar banking. 11. In this article the term banknotes will refer only to circulating notes issued by commercial banks or other private institutions. Currency issued or backed by governments or central banks will be referred to as cash. 12. See, for example, Dwyer (1996) for examples of both types of regime from the U.S. Free Banking Period (1837-65). 13. White (1995) traces the pre-Civil War history of various devices employed by state governments to protect the interests of banknote holders, including restrictions on minimum denominations, state-sponsored insurance plans, and restrictions on asset holdings. Broadly speaking, placing restrictions on asset holdings seems to have been the most efficient mechanism. Williamson (1989) and Champ, Smith, and Williamson (1996) point out that the Canadian experience with banknote issue was quite different from the U.S. experience. In Canada, banks were historically able to issue banknotes against general assets. The value of these notes was backed by cooperative agreements among banks that would have been difficult to implement under U.S. banking laws. 14. The incentive to counterfeit also exists with government-issued currency. However, the likelihood of successful counterfeiting is greater if there a large number of private issuers of banknotes. 15. According to Lacker (1996), however, most of the Civil-War-era legal restrictions on banknote issue have been repealed by recent banking legislation. 16. In this section, the term debit card does not apply to “stored-value” or “smart” cards, which are discussed below. 17. That is, both checks and debit cards represent debit transactions as defined in the glossary. 18. Not all debit card transactions are on-line. See Caskey and Sellon (1994) for a discussion of different types of debit card transactions. On-line verification systems can also be used to guard against check fraud and reduce the risks associated with check payments. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 37 It is impossible to predict exactly which of the various new and proposed forms of payment will be successful in the marketplace. A combination of economic theory and historical experience suggests, however, that whatever the operational features of the new forms, these forms will function similarly to either checks or banknotes or perhaps some combination of these. The key economic attributes of these two traditional forms of payment are described below and summarized in Table 1.19 The Check Model for Retail Payments. First, the check model requires that the payment itself be a transfer from the buyer to the seller of a zero-maturity, parvalued debt claim on a financial institution’s assets. As discussed above, there are numerous theoretical reasons for payments to take this form. Second, as is the case with checkable bank deposits, the institution against which the payment is drawn holds a diversified portfolio of both short-term, liquid assets and at least some longer-term, illiquid assets.20 Third, the transactions instrument (check) is considered a liability of the buyer and not the institution on which it is drawn; the instrument is easily reproducible, relative to currency. Fourth, the value of the payment is verified as quickly as possible by clearing and settlement through the banking system. As discussed above, this step is necessary if claims can be easily reproduced. Fifth, the payment is not anonymous in the sense that the act of clearing and settlement reveals the identity of the buyer to both the seller and/or the bank against which the check is drawn. The Banknote Model. As with the check model, in the banknote model the payment itself consists of a transfer from the buyer to the seller of a zero-maturity, par-valued debt claim on a financial institution’s assets. However, the banknote model differs from the check model in the following ways. First, the historical experience in the United States has been that the composition of assets against which banknotes can be issued has been more tightly regulated than the assets that are used to back checkable deposits. For example, during the U.S. Free Banking Period, notes were generally issued only against certain types of bonds (see, for example, White 1995 or Dwyer 1996).21 Second, the transactions instrument is considered to be a liability of the issuing institution and is relatively difficult to counterfeit. Third, the seller receiving the payment has the option of verifying its value by presenting it to the issuing institution for redemption in cash. Fourth, the payment itself need not reveal the identity of the buyer to either the seller or the issuing bank. Charts 1 and 2 depict highly stylized examples of transactions under the two models.22 In Chart 1 a buyer has funds on deposit at bank A. The buyer purchases goods from the seller and pays by check. The seller deposits the check in an account at bank B. B presents the check to A, which debits the buyer’s account and transfers reserve funds to B. Finally, B credits the seller’s account for the amount of the purchase. In Chart 2 a buyer deposits funds with a bank A, which in turn issues banknotes. The buyer uses the notes to purchase goods from another party, the buyer/seller. This process is repeated potentially many times until a buyer/seller buys goods from a seller who wishes to exchange the notes for some other form of money. The seller does this by depositing the notes at bank B. B presents the notes to A, and receives reserve funds in settlement. T A B L E 1 Two “Model” Forms of Payment 38 Model Characteristics Check Model Banknote Model Form of Payment Transfer of zero-maturity, par-valued debt issued by a financial institution Transfer of zero-maturity, par-valued debt issued by a financial institution Backing Assets Historically, diversified asset portfolio Historically, less diversified, more liquid asset portfolio Liability of Check writer (buyer) Issuing institution Immediate Clearing and Settlement Required At the option of the seller Anonymity No Yes, at least for some transactions Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 The key differences between the two models are easily seen from the charts. The check model favors security over anonymity and convenience by involving the banking system in each transaction. The banknote model offers potential cost savings because not every transaction has to be routed through the banking system for clearing and settlement. Under the banknote model, transactions outside the banking system (for example, those depicted with dashed lines in Chart 2) are also potentially anonymous, with attendant advantages and disadvantages. Finally, the fact that not all transactions are cleared and settled potentially raises the risks associated with each unsettled transaction.23 The next section analyzes two of the most widely discussed new forms of payment using the two formal models as benchmarks. Payment with Stored-Value Cards. A stored-value card is a relatively new form of payment card.24 Storedvalue cards differ from traditional debit cards in the sense that the card does not provide access to the buyer’s bank account. Instead, the buyer purchases stored value with cash or bank funds, and the appropriate amount of stored value is placed on a card in the form of data on a magnetic strip or electronic chip. When the card is used to make a purchase, the amount of the purchase is deducted from the balance on the card, not from the buyer’s checking account. Merchants and other receivers (or their banks) of these stored-value claims then present the claims to the issuing bank (or other institution) for settlement. Stored-value cards thus can offer potential cost savings over on-line debit and credit card systems to the extent that they eliminate the need for costly on-line verification of each transaction. Stored-value cards resemble both checks and banknotes in the sense that the transfer of stored value repre- sents the transfer of a demandable, par-value debt claim from buyer to seller. Do these cards more closely resemble electronic checks or electronic banknotes? The answer depends on the manner in which the stored value is created and on what happens after the stored value is transferred from buyer to seller. If the stored value represents claims on bank assets, that is, on funds in a bank account, then in this respect the value placed on stored-value cards represents something closer to traditional checkable deposits than it does banknote claims. On the other hand, if the stored value represents a claim on a firm outside the safety net of the traditional banking system, then it is likely that a special pool of liquid assets will be maintained in order to back the stored value. In such cases, stored-value cards would more closely resemble banknotes.25 Stored-value cards also resemble banknotes to the extent that the stored value placed on the card represents a liability of the issuing institution. As discussed above, this feature of stored-value cards is advantageous in the sense that it can eliminate the need for on-line verification. However, this banknote-like feature of storedvalue cards makes them potentially subject to risks from counterfeiting.26 Various issuers of stored-value cards have proposed different rules for clearing and settlement of stored-value transactions. Some stored-value card systems require that each stored-value transaction be cleared and settled through the banking system. This first type of system more closely adheres to the check model in this respect. In other stored-value systems, stored value can be transferred from one card to another without clearing and settlement of the transaction; such transactions are known as peer-to-peer transactions. This second type of system 19. The “check model” and the “banknote model” correspond in a very rough way to the “account-based/notational” and “tokenbased” models of electronic money that have been discussed in the computer science literature. See the discussions in Wayner (1996, 210-11) or Camp, Sirbu, and Tygar (1995, 1-2). 20. In the history of economic thought there have been numerous theoretical arguments both for and against such maturity mismatches between assets and liabilities. This debate dates back at least to the “currency” and “banking” schools of early nineteenth century Britain. For some more recent contributions see, for example, Flannery (1994, 323-26), who argues that in the case of banks, such mismatches are likely to occur because of a combination of the effects of leverage and noncontractable risks associated with bank assets. On the other side, Gorton and Pennacchi (1992) argue that maturity mismatches are unnecessary for transactions accounts and that short-term transactions liabilities can be backed by short-maturity, liquid assets. 21. Again, it should be noted that in other countries banknotes have historically been issued without such restrictions. 22. The transactions shown in Charts 1 and 2 are meant to serve as examples. Other patterns of transactions are possible in each case. 23. Absent a reserve requirement, the ability to issue circulating banknotes could lead to an indeterminate increase in the aggregate quantity of outstanding bank liabilities. For a formal discussion of this effect, see, for example, Wallace (1983). 24. For an introduction to the economics of stored-value cards see Congressional Budget Office (1996). See Allen and Barr (1997) or Zoreda and Otón (1994) for an introduction to the operational aspects of stored-value cards. Stored-value cards that contain an electronic chip (as opposed to a magnetic strip) are commonly called smart cards. 25. McAndrews (1996, 22) argues that the most likely issuers of stored-value cards will be joint ventures involving both banks and nonbanks. McAndrews argues that one problem that will have to be resolved by such joint ventures is the question of exactly whose liability is represented by the stored value. 26. The need for protection against counterfeit risk has been underscored by recent experience with stored-value cards in Japan. There, widespread counterfeiting of stored-value cards led to losses reported to be as great as $500 million. See Glain and Shirouzu (1996) or Pollack (1996). Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 39 C H A R T 1 The Check Model of Payments Settles Bank A Bank B Presents Check Deposits Funds Debits Account Deposits Check Credits Account Pays by Check Seller Buyer Provides Goods or Services more closely approximates the hand-to-hand transfer of banknotes from buyer to seller. There is also a wide range of possibilities concerning the anonymity of transactions with stored-value cards. If buyers can purchase stored-value cards anonymously, then in this respect stored-value cards more closely resemble banknotes. However, this anonymity is compromised somewhat if each transaction made with a stored-value card can be traced back to an individual card or to a particular seller. Likewise, if each stored-value transaction has to be cleared and settled through the banking system, then this requirement limits the anonymity of stored-value transactions. On the other hand, if stored-value transactions can take place without the involvement of the banking system, these transactions could be almost completely anonymous, even if the original purchase of value can be traced back to a specific issuer or location. In summary, payment by means of stored-value cards mimics the banknote model in the sense that the payment instrument represents a liability of the issuing institution and not that of the buyer using the stored-value card. In other respects, payment by stored-value card can follow either the check or the banknote models, depending on exactly how the stored value is issued, transferred, and settled. If the stored-value claim is issued in a nonanonymous way against bank assets and each storedvalue transfer has to be cleared and settled through the banking system, then payment by stored-value card comes close to the check model.27 If stored value is issued anonymously against a specific pool of backing assets and can be transferred anonymously without the involvement of the banking system, then payment by stored-value card almost perfectly matches the banknote model. Other types of stored-value systems would probably fall somewhere between these two extremes. 40 Payment with Electronic Cash. A limitation of stored-value payment systems is their requirement for specialized cards, computers, and electronic networks in order to hold and transfer stored value. Another variation on the stored-value idea would go a step further and eliminate the need for such specialized equipment. Instead, stored value would be held on nonspecialized computers and transferred via widely accessible computer networks such as the Internet. This method of payment has been given a variety of names, such as electronic cash, digital cash, electronic currency, electronic coins, and electronic scrip.28 The discussion will use the term electronic cash, which seems to be the most commonly used. The term may be somewhat misleading, however, since electronic cash represents claims on the assets of private institutions and, unlike the paper cash in common use today, does not have governmental backing. As is the case with smart cards, electronic cash resembles traditional, privately issued banknotes in the sense that it represents a liability of the issuer and not of the buyer using the electronic cash to make a purchase. However, in other respects, payment via electronic cash may conform more closely to the check model than to the banknote model. For example, electronic cash issued as a claim on a firm outside the traditional banking system could be issued either against a specific pool of backing assets (as in the banknote model) or as a claim on bank assets (as in the check model). There are also at least two areas in which technological constraints pose significant challenges to the ability of electronic cash to conform to the banknote model. The first is anonymity. Transferring stored value from one computer to another ultimately involves transferring data from one computer to another. Since data on computers are readily copied and manipulated, some Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 C H A R T 2 The Banknote Model of Payments Settles Bank A Bank B Presents Notes Deposits Funds Issues Notes Deposits Notes Notes Buyer Goods Buyer/ Seller verification procedure is necessary in order to ascertain that the transferred data represent a legitimate claim to stored value. If the verification process involves a third party (other than the buyer and seller), anonymity could be compromised. Some innovative techniques have been proposed to circumvent this problem.29 At least in theory, these techniques should allow almost complete anonymity of electronic cash transactions while simultaneously providing verification of the stored-value claim. A second problem with electronic cash has to do with the issue of clearing and settlement. To perfectly emulate the banknote model, electronic cash should allow for peer-to-peer transactions between buyer and seller that do not require clearing and settlement through the banking system. If, however, the security of electronic cash is such that a seller cannot discern the legitimacy of an electronic cash transfer, then mandatory clearing and settlement of each transaction through the banking system may be necessary. As with stored-value cards, the bottom line for electronic cash payments is that they can follow the banknote model, come close to approximating the check model, or fall somewhere between these two extremes. It seems clear, however, that in economic terms neither Credits Account Notes Seller Goods stored-value cards nor electronic cash represent radical departures from traditional modes of payment. Old versus New Forms of Payments: Some Caveats hile most traditional forms of payment evolved in relatively unregulated environments, the contemporary use of these forms is governed by a large and well-established body of laws and regulations. The purpose of these laws and regulations is to protect the public interest, more generally, and often the rights of consumers and small depositors, more specifically. While in some cases these laws and regulations apply to some of the new forms of payment, in other cases their applicability is at best ambiguous. A complete discussion of potentially applicable banking laws and regulations is beyond the scope of this article. But the potentially large impact of banking laws and regulations on the new payment forms merits a brief survey of some of the relevant legal and regulatory issues.30 The first and perhaps most crucial question is whether entities other than banks have the legal right to issue transferable liabilities in the form of, say, storedvalue cards or electronic cash. Current U.S. law limits the ability of nonbanks to offer deposits and limits the ability W 27. Formally, such stored-value cards most closely resemble traditional travelers checks or cashier’s checks, which in contrast to ordinary checks are considered liabilities of the institution on which they are drawn. 28. On the details of various existing and proposed forms of electronic cash, see, for example, Chaum (1992), Congressional Budget Office (1996), Flohr (1996), and Wayner (1996). 29. The most prominent of these innovations is Chaum’s (1992) technique, based on the idea of “blind signatures.” Digital blind signatures allow both a buyer and an electronic-cash issuer to “sign” electronic cash in a way that is verifiable to the signer and to other designated parties but is unobservable and irreproducible by other parties. 30. The discussion below borrows heavily from U.S. Department of the Treasury (1996, apps. 14) and Congressional Budget Office (1996, chap. 4). The reader is referred there for more detailed discussions of legal and regulatory issues. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 41 of nonbank depository institutions to make commercial loans. If new forms of transactions liabilities were to be seen as deposits, these laws would also apply to nonbank firms offering these new types of liabilities. Whether or not various new forms of payment legally constitute deposits is not entirely resolved, although the Federal Deposit Insurance Corporation recently ruled that most types of stored-value cards are not deposits for insurance purposes (see FDIC 1996). From a regulatory viewpoint, the distinction between bank and nonbank issuers of new forms of transactions liabilities is also important. Banks and bank holding companies are subject to both specialized laws and regulatory oversight designed to limit their risk exposures. In return, banks and their depositors are protected from potential losses by the safety net afforded by federal deposit insurance and by banks’ access to the Fed’s discount window. In the case of bank-issued transactions liabilities, at least some of these restrictions and assurances could carry over to the newer forms. Transactions deposits at banks are also subject to a legal reserve requirement, which mandates that banks maintain a certain percentage of their transactions deposits as either cash or non-interest-bearing accounts at the Fed. As of this writing, it appears likely that balances on storedvalue cards issued by banks will be subject to reserve requirements (see Blinder 1995). By contrast, nonbank issuers of new types of transactions liabilities could or could not largely be free of the restrictions and oversight required by state and federal laws. Some exceptions to this statement might occur if a nonbank issuer were owned by a bank or bank holding company. It is also unlikely that the coverage of the federal safety net would extend completely to nonbankissued transactions liabilities. Another important question has to do with applicability of the rules governing the validity of electronic funds transfers. For retail payments, these rules are provided by the Electronic Funds Transfer Act of 1978 and the Federal Reserve System’s corresponding Regulation E. Regulation E also requires extensive disclosure of information to consumers regarding their rights and obligations when using various forms of electronic funds transfer. Currently, the applicability of Regulation E to various new forms of payment is uncertain. In the case of stored-value cards, for example, the Fed has proposed exempting from Regulation E all cards containing no more than $100 as well as all stored-value cards that are off-line and that do not track individual transactions (see Board of Governors 1996). A final area of regulatory ambiguity results from potential conflicts between the putative anonymity of 42 some of the new forms of payment and the reporting requirements of federal anti-money-laundering laws. These laws currently impose extensive record-keeping requirements on financial institutions for certain types of transactions, especially those that involve exchanging cash for other types of liabilities. The general applicability of these laws to the new forms of transactions liabilities is again uncertain. Conclusion: The More Things Stay (Virtually) the Same he advent of various new electronic forms of payment cannot be described as revolutionary. The new types of payments are better described as evolutionary adaptations of some older forms of payment— checks and banknotes—to modern communications technology. Since the new forms of payment do not really represent anything particularly new from the standpoint of economic theory, it seems likely that the same policy issues that apply to the creation of checkable deposits and to the issue of banknotes will apply to the creation of the new forms of payment liabilities. Among the most critical open policy questions are the following: First, should institutions not regulated as banks be able to offer the same types of transaction services as banks—that is, should there be “free” electronic banking? Second, if the answer to the first question is yes, what are the rights and responsibilities of nonbank providers of transactions services? In particular, to what extent should existing banking laws apply to these nonbank providers? And what should be the responsibility of the public sector toward these nonbank providers, particularly in the case of a failure of a provider or a more widespread liquidity crisis? Third, should banks and other providers of transactions services be allowed to create electronic liabilities with some characteristics of circulating banknotes? And what restrictions, if any, should apply to these liabilities? Fourth, is it necessary to impose a non-interestbearing reserve requirement on all transactions liabilities in order to maintain a stable overall level of prices?31 Aside from the occasional interjection of the word electronic, these are classical questions of monetary economics. These questions were widely debated in the nineteenth and early twentieth centuries, but by the mid-twentieth century they had been resolved, at least in a policy sense, in favor of the regulated form of banking that we are familiar with today. If the new types of payments become popular enough to force these same questions to be asked again, it will be interesting to see if the same answers emerge. T Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 Glossary Adverse selection: a condition that exists in economic situations when a meaningful attribute of one party (say, the creditworthiness of a borrower) is unobservable by another party (say, a lender). Banknote: for the purposes of this article, a debt obligation issued by a bank (or some other private institution) that the issuer promises to redeem in a prespecified amount of cash on demand and that is intended to circulate in bearer form. Bill of exchange: an order from one party (for example, a buyer of a good or service) to another party (often a bank) to pay a certain amount of money to a third party (often a seller of a good or service) on a certain date. In contrast to checks, bills of exchange are not demandable at par value. Cash: for the purposes of this article, either (a) governmentissued fiat money, as circulates in virtually all countries today, or (b) specie or government-backed claims to specie, as circulated in many countries prior to the 1930s. Check: an order from one party (for example, a buyer of a good or service) to another party (usually a bank) to pay a certain amount of money to a third party (often a seller of a good or service) on demand. Clearing: the process by which a payment order (such as a check) moves to the bank on which it is drawn, prior to settlement. Credit card: a card that indicates that the holder has access to a line of credit with a bank or other institution. The line of credit can be used to make transactions up to a limit; the balance on these transactions is then paid off by the card holder. Credit transaction or giro transaction: a transaction in which the order to pay moves from the bank of the buyer of a good or service to the bank of the seller. Examples of credit transactions include the giro transactions that are commonly used in many European countries and direct payroll deposits in the United States. Debit card: a card that allows the holder to make transactions by accessing funds on her account with a bank or sim- ilar institution. Differs from a credit card on which funds are first spent down and then paid off. Debit transaction: a transaction in which the order to pay moves from the bank of the seller of a good or service to the bank of the buyer. Examples of debit transactions include payments by check or by debit card. Demandable at par: a condition of debt claims that are puttable at any time at par (face) value. For example, today virtually all checks are demandable at par. Electronic cash: a par-valued debt claim on a bank or other institution designed to be used as a means of payment over the Internet or other nonspecialized computer network. Also called electronic scrip, electronic currency, and electronic coins. Moral hazard: a condition that exists in economic situations in which one person can undertake actions to her own benefit and to the detriment of other people without such actions being observed. Put option: an option contract that entitles its holder to sell or “put” an asset at a prespecified price. Puttable debt: debt that can be resold to its issuer at a prespecified price. Smart card: a type of stored-value card on which the relevant account information is stored on a computer chip. Stored-value card: a type of payment card on which the relevant account information is accessible from the card itself in the form of data stored on a magnetic strip or computer chip. Settlement: an act that discharges obligations between two parties. For example, when one bank presents another bank with a check drawn on a depositor’s account, the latter bank can settle this obligation by transferring an equal amount of reserve funds to the former. —————————————— Adapted from Bank for International Settlements (1993) and Congressional Budget Office (1996). 31. This issue has been raised by numerous theoretical studies—for example, Wallace (1983), Woodford (1990), and Smith (1991)—that suggest that some sort of non-interest-bearing requirement may be necessary. By contrast, Goodhart (1993) argues that non-interest-bearing reserve requirements are not necessary for the conduct of monetary policy, essentially because of private-sector demand for central bank liabilities as a transactions medium. The Federal Reserve System has strongly endorsed the latter viewpoint; see, for example, Blinder (1995) and Greenspan (1996). For an extended discussion of this issue, see Roberds (1994). Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 43 R E F E R E N C E S ALLEN, CATHERINE A., AND WILLIAM J. BARR, eds. 1997. Smart Cards: Seizing Strategic Business Opportunities. From the Smart Card Forum. Chicago: Irwin Professional Publishing. FEINMAN, JOSHUA N. 1993. “Reserve Requirements: History, Current Practice, and Potential Reform.” Federal Reserve Bulletin 79 (June): 569-89. BANK FOR INTERNATIONAL SETTLEMENTS. 1993. Payment Systems in the Group of Ten Countries. Basle. FLANNERY, MARK J. 1994. “Debt Maturity and the Deadweight Cost of Leverage: Optimally Financing Banking Firms.” American Economic Review 84 (March): 320-31. BLINDER, ALAN S. 1995. Statement before the Subcommittee on Domestic and International Monetary Policy, U.S. House Committee on Banking and Financial Services, October 11. BOARD OF GOVERNORS OF THE FEDERAL RESERVE SYSTEM. 1996. “Electronic Funds Transfers.” Federal Register 61 (May 2): 19696-705. BRAUDEL, FERNAND. 1982. The Wheels of Commerce. Translated by Sîan Reynolds. New York: Harper and Row. CALOMIRIS, CHARLES W., AND CHARLES M. KAHN. 1991. “The Role of Demandable Debt in Structuring Optimal Banking Arrangements.” American Economic Review 81 (June): 497-513. CAMP, JEAN, MARVIN SIRBU, AND J.D. TYGAR. 1995. “Token and Notational Money in Electronic Commerce.” Paper presented at the Usenix Workshop on Electronic Commerce, New York, New York, July 11-12. Available on the Internet at http://www.cs. cmu.edu/user/jeanc/www/usenix.html. CASKEY, JOHN P., AND GORDON H. SELLON JR. 1994. “Is the Debit Card Revolution Finally Here?” Federal Reserve Bank of Kansas City Economic Review 79 (Fourth Quarter): 79-95. CHAMP, BRUCE, BRUCE D. SMITH, AND STEPHEN D. WILLIAMSON. 1996. “Currency Elasticity and Banking Panics: Theory and Evidence.” Canadian Journal of Economics 29 (November): 828-64. CHAUM, DAVID. 1992. “Achieving Electronic Privacy.” Scientific American 267, no. 2:76-81. COLLINS, SEAN S., AND PHILLIP R. MACK. 1994. “Avoiding Runs in Money Market Mutual Funds: Have Regulatory Reforms Reduced the Potential for a Crash?” Board of Governors of the Federal Reserve System, Finance and Economics Working Paper 94-14. CONGRESSIONAL BUDGET OFFICE. 1996. Emerging Electronic Methods for Making Retail Payments. Washington, D.C. CUADRAS-MORATÓ, XAVIER, AND JOAN R. ROSÉS. 1995. “Bills of Exchange as Money: Sources of Monetary Supply during the Industrialization in Catalonia, 1844-74.” Universitat Pompeu Fabra, Economics Working Paper 111. DIAMOND, DOUGLAS W. 1991. “Debt Maturity Structure and Liquidity Risk.” Quarterly Journal of Economics 106 (August): 709-37. DUPREY, JAMES N., AND CLARENCE W. NELSON. 1986. “A Visible Hand: The Fed’s Involvement in the Check Payments System.” Federal Reserve Bank of Minneapolis Quarterly Review 10 (Spring): 18-29. DWYER, GERALD P., JR. 1996. “Wildcat Banking, Banking Panics, and Free Banking in the United States.” Federal Reserve Bank of Atlanta Economic Review 81 (December): 1-20. FEDERAL DEPOSIT INSURANCE CORPORATION. 1996. “Notice of FDIC General Counsel’s Opinion.,” No. 8, FR Doc. 96-19697, filed August 1, 1996. 44 FLOHR, UDO. 1996. “Electric Money.” Byte (June): 74-84. FRIEDMAN, MILTON, AND ANNA J. SCHWARTZ. 1963. A Monetary History of the United States, 1867-1960. Princeton, N.J.: National Bureau of Economic Research. GLAIN, STEVE, AND NORIHIKO SHIROUZU. 1996. “How Japan’s Attempt to Slow Nuclear Work in North Korea Failed.” Wall Street Journal, July 24, A1. GOODHART, CHARLES A.E. 1993. “Can We Improve the Structure of Economic Systems?” European Economic Review 37:269-91. GORTON, GARY, AND GEORGE PENNACCHI. 1990. “Financial Intermediaries and Liquidity Creation.” Journal of Finance 65:49-71. ———. 1992. “Financial Innovation and the Provision of Liquidity Services.” In The Reform of Federal Deposit Insurance, edited by James R. Barth and R. Dan Brumbaugh Jr. New York: Harper. GREENSPAN, ALAN. 1996. Remarks at the U.S. Treasury Conference on Electronic Money and Banking: The Role of Government, September 19, Board of Governors of the Federal Reserve System. JACKLIN, CHARLES J. 1987. “Demand Deposits, Trading Restrictions, and Risk Sharing.” In Contractual Arrangements for Intertemporal Trade , vol. 1 of Minnesota Studies in Macroeconomics, edited by Edward C. Prescott and Neil Wallace. Minneapolis: University of Minnesota Press. KAHN, CHARLES M., AND WILLIAM ROBERDS. 1996. “On the Role of Bank Coalitions in the Provision of Liquidity.” Unpublished manuscript, June. LACKER, JEFFREY M. 1996. “Electronic Money and Monetary Policy.” Paper presented at the Conference on the Foundations of Policy toward Electronic Money, Federal Reserve Bank of Minneapolis, December 3-4. MCANDREWS, JAMES J. 1996. “Banking and Payment System Stability in an Electronic Money World.” Paper presented at the Conference on the Foundations of Policy toward Electronic Money, Federal Reserve Bank of Minneapolis, December 3-4. NAIK, GAUTAM. 1996. “Sorry, Your Prepaid Phone Card Has Been Deactivated.” Wall Street Journal, July 16, B1. POLLACK, ANDREW. 1996. “Counterfeiters of a New Stripe Give Japan One More Worry.” New York Times, June 20, D1. ROBERDS, WILLIAM. 1994. “Changes in Payments Technology and the Welfare Cost of Inflation.” Federal Reserve Bank of Atlanta Economic Review 79 (May/June): 1-12. SMITH, BRUCE. 1991. “Interest on Reserves and Sunspot Equilibria: Friedman’s Proposal Reconsidered.” Review of Economic Studies 58:93-105. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 TIMBERLAKE, RICHARD H. 1993. Monetary Policy in the United States. Chicago: University of Chicago Press. WHITE, EUGENE N. 1995. “Free Banking, Denominational Restrictions, and Liability Insurance.” In Money and Banking: The American Experience, 99-118. Fairfax, Va.: George Mason University Press. TOWNSEND, ROBERT M. 1989. “Currency and Credit in a Private Information Economy.” Journal of Political Economy 97 (December): 1323-44. WILLIAMSON, STEVE. 1989. “Restrictions on Financial Intermediaries and Implications for Aggregate Fluctuations: Canada and the United States.” In NBER Macroeconomics Annual, edited by Olivier J. Blanchard and Stanley Fischer. Cambridge, Mass.: MIT Press. U.S. DEPARTMENT OF THE TREASURY. 1996. “An Introduction to Electronic Money Issues.” Paper prepared for the conference Toward Electronic Money and Banking: The Role of the Government, September 19-20. WALL, LARRY D. 1989. “A Plan for Reducing Future Deposit Insurance Losses: Puttable Subordinated Debt.” Federal Reserve Bank of Atlanta Economic Review 74 (July/August): 2-17. WALLACE, NEIL. 1983. “A Legal Restrictions Theory of the Demand for ‘Money’ and the Role of Monetary Policy.” Federal Reserve Bank of Minneapolis Quarterly Review 7 (Winter): 1-7. WAYNER, PETER. 1996. Digital Cash: Commerce on the Net. Boston: Academic Press. WILLIAMSON, STEVE, AND RANDALL WRIGHT. 1994. “Barter and Monetary Exchange under Private Information.” American Economic Review 84 (March): 104-23. WOODFORD, MICHAEL. 1990. “The Optimum Quantity of Money.” Chap. 20, vol.2, in Handbook of Monetary Economics, edited by Benjamin M. Friedman and Frank H. Hahn. Amsterdam: NorthHolland. ZOREDA, JOSÉ, AND JOSÉ MANUEL OTÓN. 1994. Smart Cards. Boston: Artech House. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 45 The Buck Stops Where? The Role of Limited Liability in Economics T H O M A S H . N O E A N D S T E P H E N D . S M I T H Noe is an associate professor and Smith holds the H. Talmage Dobbs Jr. Chair of Finance in the College of Business Administration at Georgia State University. Both authors are also visiting scholars in the Atlanta Fed’s research department. If you come to grief, and creditors are craving (for nothing planned by mortal head is certain in this Vale of Sorrow—saving that one’s liability is limited), do you suppose that signifies perdition? If so, you’re but a monetary dunce; you merely file a winding-up petition, and start another company at once! —Gilbert and Sullivan, Utopia Limited L IABILITY, OR THE LACK THEREOF, HAS LONG PLAYED AN INTERESTING ROLE IN THE FIELDS OF ECONOMICS, FINANCE, AND LAW. FROM EARLY TIMES SOCIETIES HAVE DEBATED WHEN TO SHARE LOSSES ARISING FROM BAD ECONOMIC OUTCOMES, WHETHER THESE OUTCOMES ARE DUE TO BAD DECISIONS ON THE PART OF INDIVIDUALS, EVENTS INDEPENDENT OF INDIVIDUAL ACTIONS, OR SOME COMBINATION OF THE TWO. In modern societies personal limited liability is the norm, given such conditions as finite wealth and the elimination of debtors’ prisons. In fact, over the last few centuries, many societies have taken this principle further by passing laws that allow investors in banks and other business enterprises to limit their losses to either their initial investment (pure corporate limited liability) or some multiple of their initial capital contribution. This latter liability structure might call for an additional infusion of funds on the part of investors up to some maximum (say, two times the investment) should an enterprise fail to meet its obligations from available resources. Bank shareholders, for example, were once routinely 46 required to post at least some additional funds in the event of a bank failure. This practice ceased only after the substitution of public capital, in the form of government deposit insurance, for the private capital formerly used to support the system.1 Overall, changes in liability provisions, by many accounts, have been among the major influences on both the level and distribution of contemporary economic output as well as the allocation of financial resources in today’s financial markets. This article reviews a large and growing literature on the role of personal and corporate limited liability in the economy. As early as Adam Smith’s ([1776] 1994) criticism of the emerging joint stock corporations of the eigh- Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 teenth century and Walter Bagehot’s ([1873] 1991) analysis of the reasons for and consequences of the incorporation of the Bank of England in the seventeenth century, economists have been aware that liability structures, almost by definition, influence decisions made by households, businesses, and government agencies. This review attempts to provide a more thorough understanding of incentive structures under alternative liability regimes and, in doing so, should help policymakers better understand the possibly unintended effects of certain policies and programs. Workers, investors, managers, and policymakers confront limited liability considerations every day. It is therefore useful to look at examples that cover the import of limited liability on activities ranging from investment, labor, and financing decisions made by individuals and corporations to the implementation of discretionary government policies that are intended to promote growth or redistribution of wealth in the economy. This examination begins with an illustration of some conflicts that may arise in labor markets because of certain rights, such as personal limited liability, held by providers of human capital, or the fact that the floor of zero wealth generally associated with personal limited liability may not be sufficient to sustain productive work. Next is a discussion of how liability rules influence the incentives of debtors and creditors at the level of individual corporations and of how liability structures are important in the investment and financing decisions of managers, acting as agents for shareholders. An outline of the role of limited liability in the relationship between government and private institutions as it relates to economic growth and the provision of liquidity to the banking system rounds out the article. Labor Contracting, Limited Liability, and Subsistence Levels otential problems arise in labor contracting when individuals have limited liability and cannot be forced to work. Another factor to consider with regard to labor contracting is that the “real” lower bound for labor income might not be zero but some positive subsistence level. Limited Liability and the Inalienability of Human Capital. Limited liability, combined with other basic rights, can provide those who supply labor an incentive to “hold up” the owners of a firm. Consider, for example, an individual whose only wealth exists in the form of human capital, in particular an idea that may generate future P cash flows if he or she expends the required labor input. This “entrepreneur” might choose to sell the right to future cash flows generated by this idea to individuals with current wealth. Since the price of a security represents the present value of potential future cash flows, an “idea person” needed to make an ongoing contribution may well have an incentive to attempt to negotiate an additional share of future output after starting a project even after having sold the rights to all future cash flows at the outset. Leverage in such a situation is based on the facts that a person cannot be forced to work and that he or she possesses limited liability.2 Furthermore, any threat by disgruntled shareholders to confiscate assets may be met with a “take the money and run” response on the part of the entrepreneur.3 Hart and Moore Changes in liability provi(1994) and Noe and sions have been among Smith (1994) argue the major influences on that these problems can to some extent be both the level and distrimitigated by simply bution of contemporary arranging financial economic output as well transactions that do not transfer the total as the allocation of finanvalue of a project to cial resources in today’s an idea person immefinancial markets. diately. This arrangement seriously blunts entrepreneurs’ incentives to hold up other claimants for a larger share of output. If the amount held back is large enough, the negative incentive effects of “no-forced” work and limited liability can be eliminated. In other words, investors can solve a potential hold-up problem by holding up the transfer of part of the value of a project to the entrepreneur. This idea is very much in the spirit of venture capital arrangements and certain relationships with builders, whereby compensation is passed along in a piecemeal fashion, conditional on the completion of certain measurable outcomes. Such a “carrot-and-stick” approach can be used to induce “good” behavior. In fact, in some cases a simple labor contract— whereby an entrepreneur is paid a fixed wage immediately and, conditional on performance, a fixed wage in the future—will solve the hold-up problem. Destitution and Positive Limited Liability. This discussion has so far been based on the premise that the 1. For a discussion of the history of multiple liability provisions and a rationale for their application to U.S. banking, see Wilson and Kane (1996). 2. Hart and Moore (1994) define the inability to force labor as the “inalienability of human capital.” 3. One might think that in a world of repeated contacts the loss of reputation on the part of the idea person would be sufficient incentive to “behave” (put forth effort). However, there is an end-game problem here. For example, if there are a finite number of times the idea person needs funds, he or she may well deliver as promised at the beginning of a relationship but have no incentive to produce after a certain point. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 47 worst that can happen to an individual is that he or she will “go broke,” that is, reach zero income or wealth. Realistically, however, a very low (or zero) level of income may be insufficient to allow a worker to produce in the future. Dasgupta (1993) argues, for example, that a certain level of caloric intake is necessary if a laborer is to have the physical strength necessary to engage in agricultural or other economic production. This argument suggests that, for purposes of relevance to economic output, there is a strictly positive subsistence level below which certain economic resources such as food cannot fall.4 While the cost of a minimally nutritious diet is relatively trivial to most households in the more developed world, the need for positive subsistence levels is a large factor in the functioning of many economies throughout the less developed world. Moreover, the lack of serious consideration of these constraints might have important implications for economic theory, particularly so-called general equilibrium approaches that purport to model an entire economic system, albeit in an abstract and simplified setting. Dasgupta and others have argued that these fixed costs arising in the theory of consumer demand raise important questions regarding employment, wages, and the distribution of nonhuman capital, such as land. Consider, for example, a situation in which $400 worth of calories per year is needed in order for a human to be productive in some economic activity (see footnote 4). This cost is fixed, and it can be assumed that individuals who have sources of wealth that allow them to achieve this intake will be extremely efficient relative to workers without this level of wealth. In particular, their production will, at some levels, display increasing returns to scale (that is, a small increase in labor input will produce a more than proportional increase in output once the fixed cost of $400 has been covered). Those unable to clear this nutritional hurdle will simply not be able to compete in the labor market; they may be able to survive, but they will not be productive in the conventional sense of the term. Thus, the distribution of nonhuman capital sources of production, such as land, becomes important. Even if there is sufficient aggregate wealth (and Becker 1993 notes that most countries have per capita incomes far exceeding subsistence levels) the distribution of that wealth may be such that it simply does not benefit individuals with wealth to hire workers who have access to no capital other than their own labor. Hiring an already healthy worker, even at a slightly higher wage, is simply more efficient than paying the fixed nutritional cost of hiring a malnourished one. In a more developed economy, even someone who goes bankrupt will typically have access to income sufficient to cover his or her basic needs, either through labor or through a social safety net. However, since, conservatively, 300 million to 600 million people worldwide are in economic circumstances below the subsistence level, it is not meaningful to speak of personal limited liability as 48 being simply a non-negative wealth position. It is in this sense that subsistence, and not zero, levels of income are the relevant ones for some of the analyses in the area of economic theory and practice. Liability Rules and the Incentives of Debtors, Creditors, and Managers here is a large body of work that seeks to analyze the importance of liability rules at the individual firm level in terms of the relationship between borrowers and creditors as well as potential conflicts between stockholders and managers. Major focal points in this area include investment and financing decisions of firms and the distortionary bargaining power generated by liability provisions in bankruptcy. Here, this liability structure will be examined in light of what conflicts induced by limited liability may arise with outside claimants, even when managers and owners have nonconflicting goals, and problems that arise when managers have their own, potentially separate, objectives and possess liability protection. Creditors versus Owners. Consider the problem faced by creditors with multiple potential borrowers, some with relatively lower-risk ventures available for investment and others offering higher-risk projects. As noted by Stiglitz and Weiss (1981), a creditor is going to be unable, without further analysis, to distinguish among these potential borrowers and their associated projects. The borrowers, on the other hand, know that, should a lender grant them a loan, their payoff will be the larger of the difference between the value of the project less what they owe on the loan or zero. This limited liability associated with being the residual claimant, or equityholder, is well known. A creditor faces the problem of choosing between (1) simply setting a loan rate that reflects what potential borrowers believe to be the average risk, based on, say, previous experience in the field for such a group of projects; (2) asking that potential borrowers post collateral (which they may or may not have); or (3) expending money on further investigation of potential projects (credit analysis). In all cases, the lender faces a problem brought on in large measure by the personal or corporate limited liability of borrowers. If simply setting a loan rate, a lender could charge a relatively low rate, in which event almost all potential borrowers would seek funds. But lender profits would be low or negative in this case, providing lenders the incentive to charge higher loan rates. Eventually, lower-risk borrowers, offering lower-risk investments, will find it unprofitable to seek bank financing, leaving banks with a relatively higher-risk pool of potential applicants. As an extreme example, consider two potential borrowers, one with a project offering a certain 10 percent rate of return and the other with a project with an equal chance of paying 20 percent or 0 percent. If the lender T Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 charges 5 percent, both borrowers will seek funds, while at 11 percent the borrower with the lower-risk project will obviously be unable to make a profit and will drop out. However, the borrower proposing the higher-risk project will, precisely because she has limited liability, continue to seek funds, since the best she can do is 9 percent (20 – 11 percent), while the worst is no profit (the project pays 0 percent and the borrower cannot pay off the loan). This so-called adverse selection problem (at higher prices for credit, only the risky apply) may lead lenders to limit the supply of credit below that demanded by borrowers. Indeed, in this simple example there is no interest rate at which the creditor can expect to make a profit. As a result, the supply of credit will be zero while the demand for credit will be positive for any interest rate between 0 percent and 20 percent. Another option open to lenders to help mitigate the “heads I win, tails you lose” advantage of the borrower is to require all potential borrowers to post collateral. But this solution poses its own problems. Requiring full collateral is likely unrealistic since, if borrowers could finance their own projects, they probably would not seek outside funding in the first place. And while partial collateral may alleviate the net effect of borrower-limited liability, the essential conflict between borrowers and creditors remains. Obviously, lenders also engage in credit analysis. However, as long as risk assessment is less than a perfect science, there may remain groups of borrowers to whom banks are unwilling to lend at any interest rate. Clearly, none of the three proposed options is generally sufficient to eliminate the incentive problems associated with limited liability. A problem closely related to the adverse selection issue involves the fact that limited liability provides, holding other factors constant, an incentive for a borrower to choose a relatively higher-risk project as opposed to a lower-risk project. Consider again the extreme numerical example, now supposing that a single entrepreneur has a choice between undertaking one or the other of the two projects. Since the expected return on the two projects is the same—that is, 10 percent = (0.5)(20 percent) + (0.5)(0)—the expected return to the borrower is actually higher for the riskier project. That is, the borrower is better off (in terms of the expected pecuniary reward) by taking the higher-risk and not the lower-risk project because, and in this case only because, the borrower has limited liability. For example, at an interest rate of 15 percent, the borrower would expect to receive either 5 percent (20 – 15 percent) or no profit by taking on the higher-risk project and nothing for taking on the lowerrisk project. This story would be different, however, if it were possible to impose a large enough nonpecuniary cost on the borrower in the event that she failed to repay the loan. Debtors prisons in earlier centuries exemplify such a cost. But of course this approach essentially begs the question since, for all intents and purposes, borrowers would no longer have limited liability. It is sometimes argued that borrowers will not exploit the default option in repeated contacts in order to avoid reducing their reputation. Still, unless individuals derive some nonmonetary benefit from being thought well of in terms of meeting their obligations, entrepreneurs may continue to have an incentive to exploit their limited liability by taking on higher-risk projects. This situation is not unique to the borrower/creditor As well as a direct effect relationship. While on investment policy, corpersonal limited liaporate limited liability has bility alone is an important factor in an indirect effect on corrisk-taking decisions, porate capital structure the existence of cordecisions because shareporate limited liability creates an addiholder gambling incentives tional layer of incenare anticipated by rational tive problems in the creditors and priced into relationship between equityholders in cordebt contracts. porations and bondholders (debtholders in the firm). One striking illustration of shareholders attempting to exploit limited liability occurred when the owners of Tri-State Paving, a small California contracting firm, responded to the threat of imminent bankruptcy by driving to Las Vegas to gamble with the company’s liquid assets. A good day at the tables could yield a payoff sufficient to let the owners retain control of the firm; a bad day would lead to financial ruin—a situation no worse than it already faced. From the perspective of the owners, the Las Vegas investment was a no-lose proposition. The firm’s creditors, of course, felt differently, as evidenced by the legal actions they took in response to Tri-State’s innovative investment strategy. Fortunately, most financially distressed firms do not gamble as blatantly as Tri-State did. In fact, debt covenants frequently preclude such blatant speculation. The problem in establishing such covenants is that often the outcomes of decisions made by firms are nonverifiable in that they are difficult or impossible for third parties to monitor. For example, a firm may decide to save 4. For example, Stigler (1945) estimates that a diet primarily consisting of dried beans, cabbage, and rice, at a cost of $400 per year in 1993 dollars (Becker 1993), is the least-cost diet consistent with the nutritional needs of the typical person. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 49 money now by hiring a mediocre team of risk managers who can handle routine situations but may not be able to cope with complex problems that may arise. In most situations, this strategy will increase firm profits, but it increases the potential of generating huge losses when exceptional circumstances occur. Thus, this hiring policy increases risk. However, since employment policies are notoriously hard for third parties to second-guess, it would be very difficult to write debt covenants precluding such actions. Corporate limited liability gives the shareholders of financially weak firms an incentive to gamble to the potential detriment of societal welfare. In the classical economic paradigm, the total value of a project includes all the benefits and costs associated with the project. Introducing a consideration other than total value (in this case, the riskiness of project returns) into the stockholders’ investment calculus serves only to distract them from the objective of value maximization, a result that is socially harmful. As pointed out by John, John, and Senbet (1991), a consequence of this relationship is that even fairly priced deposit insurance for banks will not eliminate the incentive of banks to gamble. At the same time, the incentive to gamble may actually counter other distortions related to investment externalities. For example, risky projects frequently are more innovative than safe projects. However, because of externalities (such as insufficient enforcement of patent laws) firms may underinvest in innovative projects. In such cases, then, corporate limited liability, by providing a countervailing incentive to choose more innovative risky projects, may actually increase social welfare. Indeed, Zha (1995) has shown that providing exceptions in bankruptcy can, in some circumstances, increase social welfare precisely by encouraging risk-averse entrepreneurs to invest in risky but socially valuable projects. As well as a direct effect on investment policy, corporate limited liability has an indirect effect on corporate capital structure decisions because shareholder gambling incentives are anticipated by rational creditors and priced into debt contracts. This pricing effect can lead firms either to eschew debt financing in favor of equity or to reject profitable investment options altogether. Box 1 provides a numerical analysis of this problem that can be summarized as follows: fully informed, rational potential bondholders recognize that equityholders can switch investment policies in a way that is detrimental to bondholders’ interests in a manner analogous to the borrower/bank example already mentioned. These potential bondholders will incorporate this factor into their decision regarding the promised payments they demand from equityholders or, equivalently, the yields they require in order to hold the bonds. While equityholders may pledge personal assets as collateral, residual incentive problems, analogous to the borrower/banker case, may remain a problem. Thus, equityholders may be forced 50 to forgo new debt financing even though the combined wealth of bond and stockholders may have been increased by an investment on the part of the firm. An even bigger problem arises if shareholders have better information than bondholders do. This asymmetry of information creates a classic “lemons” problem. Issuers of new securities with bad information have an incentive to flood the market with overvalued securities. The profits from this strategy are proportional to the informational sensitivity to the claim being offered. Financial market participants recognize this fact and therefore react skeptically to the prospect of the issuance of information-sensitive securities such as equity. This reaction drives firms toward issuing information-insensitive securities to outsiders. The most informationally insensitive claim is, of course, a claim that pays a fixed amount regardless of the firm’s value. However, given corporate limited liability, such a claim is not feasible when firm value is less than the minimum stipulated payment on the claim required to finance the investment. Thus, limited liability on the part of corporate investors increases the mispricing of corporate securities. When mispricing becomes large relative to the potential profits from the new investments funded by the security issuance, firms may, in fact, forgo profitable investment opportunities when forced to finance them on the capital market. Because, absent limited liability, firms can issue informationally invariant claims on firm cash flows and such claims entail no mispricing costs, corporate limited liability restrictions are necessary for informational asymmetries to lead to the mispricing of claims or underinvestment. Nachman and Noe (1994) provide a formal treatment of these issues. Limited liability also affects the incentives of the financially distressed firm through its effect on bankruptcy negotiation. The key to this relationship is that bankruptcy is costly. For example, bankruptcy entails large legal and administrative expenses and may make it more difficult for a firm to market its product line.5 Thus, both debtors and creditors can gain from avoiding bankruptcy. The division between both stockholders and creditors of the gains from avoiding bankruptcy will depend on their respective bargaining power. In the United States, bankruptcy law grants stockholders a number of clear advantages. Perhaps the most important of these is the exclusionary period, a 180-day period after the filing for bankruptcy (which can be extended by the court) in which only shareholders can file reorganization plans. Because of corporate limited liability and the fact that corporate value is less than promised debt payments, all costs associated with remaining in bankruptcy during the exclusionary period are borne by creditors. The upshot of this legal arrangement is that, even if after the exclusionary period ends and creditors can finally obtain the most favorable settlement possible from Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 B O X 1 Liability and Financing Decisions he effect of corporate limited liability on financing decisions is illustrated by the example of Gamblers-Dream Enterprises, a company with a very simple structure of corporate cash flows and liabilities. All of its cash flows and debt obligations will accrue in the next period, called “time 1.” The possible time 1 cash flows under its current operating policies are as follows: T Calculating the firm value and the value of the debt, under the new cash flow distribution the value of the firm is 100(0.4) + 500(0.3) + 900(0.3) = $460, while the value of the debt is 100(0.4) + 500(0.3) + 700(0.3) = $400. Cash Flow $100 $500 $900 Probability .20 .60 .20 To simplify calculating values, assume that market values are synonymous with expected cash flows. Under current operating polices, the value of Gamblers-Dream is 100(0.2) + 500(0.6) + 900(0.2) = $500. Suppose that Gamblers-Dream has debt outstanding with a promised payment of $700. The market value of this debt at time 0, under the current operating policies, is 460 = min{F, 100}(0.4) + min{F, 500}(0.3) + min{F, 900}(0.3), 100(0.2) + 500(0.6) + 700(0.2) = $460. The value of the equity is the difference between total firm value and the debt value $500 – $460 = $40. This firm appears to be in financial distress since the promised payment exceeds the firm’s market value. Suppose that Gamblers-Dream can change its operating policies to produce the following distribution of time 1 cash flows. Cash Flow $100 $500 $900 Probability 0.40 0.30 0.30 Hence the value of equity is $60. It is clear that it is in the shareholders’ interest to undertake this shift since their equity claim increases in value by $20, even though the value of the firm declines by $40. The decline in value of creditors’ claims of $60 exceeds the decline in firm value by $20, which is the shareholders’ gain. Suppose that instead of having debt outstanding Gamblers-Dream needs to issue debt in order to finance its operating policies. In this case the $40 loss in market value is factored into the pricing of the debt. For example, suppose that the firm needs to raise $460 in external financing to undertake the project. If the firm borrows the $460, the promised payment, F, will have to satisfy the equation where min(x, g) denotes the minimum of x and g. This equation reflects the fact that the firm will choose the riskier operating policy. Guessing that a solution would have to be such that F > 700, this equation reduces to $460 = 190 + 0.3F. Solving this gives F = 900. Thus, the only way the firm can obtain debt financing is to promise its entire cash flow to debtholders. Therefore, the shareholders would no doubt either switch to equity financing or eschew investing in the project entirely. 5. An interesting example of this effect is the problem that Wang Computers faced in trying to market its word processors to corporations when it faced financial distress. Corporations were understandably reluctant to commit to computer systems produced by a firm that might soon be defunct and unable to support its product line. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 51 their perspective (absolute creditor priority), they still must bear the costs of the bankruptcy process. Limited liability protects the penurious shareholders from these costs, allowing them to use the threat of dumping the costs of bankruptcy on bondholders to force concessions in prebankruptcy negotiations. Shareholders may thus be able to negotiate a settlement in which the costs of bankruptcy are avoided and debt is written down sufficiently to ensure that they receive a fraction of firm value proportional to the costs of bankruptcy. Therefore, corporate limited liability, in the presence of costly bankruptcy options, not only protects shareholders from contributing their private capital in the event of financial exigency but also provides them with a cushion of residual value proportional to the costs of the bankruptcy reorganization. While rational potential claimants may factor this option to equityholders into their required returns, limited liability may result in actions like those described above once the firm is in financial distress. Box 2 provides a graphic example of this phenomena. Owners versus Managers. For the purpose of this discussion, managers so far have been treated as agents acting in the best interest of shareholders. However, in almost all cases, save that of the sole proprietor who also manages the enterprise, there is an important potential conflict, fueled in part by the personal limited liability of the manager, between residual stakeholders (owners) and agents (managers), who actually implement policies and procedures. Indeed, Anderson and Tollison (1982) argue that this agency problem was the primary concern motivating Adam Smith in his much-discussed critique of the then relatively new limited liability corporations.6 B O X 2 Limited Liability and Bankruptcy he following example illustrates the use of the exclusionary period to hold up creditors. Consider a firm that currently owes $1.4 million to creditors: $0.8 million on a senior debt issue and $0.6 million on a junior issue. The current value of the firm is $1.3 million. If the firm enters bankruptcy, the costs of bankruptcy and financial distress will eat up $0.3 million in firm value. If the shareholders’ initial formal offer made after filing for bankruptcy is rejected, the delay caused by shareholders drawing out the automatic stay period will result in an additional $0.1 million loss. The actions taken by shareholders and creditors will depend upon what these agents predict about what will occur in the next stage of the reorganization. For this reason, it is useful to first consider what will happen if the final stage of negotiations is reached—the stage after the exclusionary period has elapsed, when creditors can make a proposal. At this point, firm value is T 1.3 – 0.3 – 0.1 = $0.9 million. At this stage, creditors can force a division based on absolute priority. The payoffs would be Shareholders: Junior debt: Senior debt: 52 $0 $0.1 million $0.8 million Thus, before the final court-ordered disposition of assets, shareholders might prefer to offer an acceptable alternative. Shareholders: Junior debt: Senior debt: $0.05 million $0.15 million $0.8 million If shareholders declare bankruptcy without delay, they will obtain $0.05 million, senior debt will be unimpaired receiving $0.8 million, and junior debt will lose $0.45 million. The cost of bankruptcy gives shareholders bargaining power. Also, in lieu of declaring bankruptcy, shareholders can make a first and final offer to restructure debt by lowering the promised payment to junior debt to $0.15 million and threaten bankruptcy if the offer is rejected. Junior debt, if it finds the threat credible, will accept the offer. The threat of forcing bankruptcy is credible because of corporate limited liability. Absent renegotiations, the shareholders have nothing to lose. From this example we see that corporate limited liability exerts an effect upon the allocation of value, even in firms that have failed. Limited liability comes into play because shareholders can threaten to use limited liability in bankruptcy negotiations. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 A manager has every incentive to engage in activities that maximize his or her welfare so long as it is costly for shareholders to monitor the effort put forth or output produced by the manager. Much of this compensation may take the form of perquisites such as limousines, deluxe furnishings, and extra staff. Personal limited liability limits the damages that owners can extract from their agents, absent criminal plundering of the firm, so the efficacy of shareholders’ monitoring of management becomes a key determinant of corporate efficiency. However, since corporate limited liability also limits the losses of shareholders to their initial investment, it clearly attenuates the incentives of shareholders to monitor managers. For this reason, firms where monitoring is particularly important, such as Lloyds of London, might not provide owners with limited liability protection. Given the adverse effect of limited shareholder liability on owner monitoring of managers and creditor monitoring of owners, it would appear, at first glance, that unlimited shareholder liability would generate welfare improvements relative to limited liability. However, as Winton (1993) shows, this conclusion does not hold uniformly. Unlimited liability, combined with the transferability of shareholdings, means that the expected payoff from holding shares depends on the identity of their owner. A wealthy investor knows that he will be forced to pay when a firm incurs large losses. A poor investor, on the other hand, knows he is protected by personal limited liability from being forced to make large payments to the firm. Because the expected payoffs on shares are higher for poor shareholders than they are for rich shareholders, this disparity generates an incentive, other things held constant, for rich shareholders to sell to poor shareholders. Thus, in a situation with unlimited corporate liability, the rich may end up owning bonds while the poor own stock. The result can be a decline in efficiency: since the poor have small liquid balances to finance unexpected capital needs of the firm in the event that investment opportunities look particularly rewarding, external funds may have to be raised to finance growth. The resulting need for new investors implies increased flotation and other related costs that may more than offset the reduced monitoring engendered by granting shareholders the protections of limited liability. Governments, Intermediaries, and the Structure of Financial Markets s previously shown, limited liability has the potential to distort investment and financing decisions at the individual firm level. Moreover, it has been argued that increasing liability beyond initial investments creates socially inefficient risk sharing because, A under these circumstances, the value of shares will be inversely related to investors’ private wealth. A solution to this problem would be to allow outside claimants to seize the assets of a firm while it is still “alive,” although it is notoriously difficult to determine the market value of many assets, particularly those that are not actively traded in financial markets. Given such measurement problems, alternative contracts have evolved that specify that shareholders must either have sufficient liquid assets on hand to meet current obligations or be able to borrow the necessary funds from an outside entity. Under these circumstances, the ability to access liquidity works as a signal (albeit an imperfect one) that a firm is economically solvent. Moreover, in modern times, central banks have emerged as the ultimate providers of liquidity (that is, the lender of last resort). It is therefore useful to review the role of corporate limited liability in the context of central bank provision of liquidity through the banking system and other government insurance programs. Bank Charters and Limited Liability. It is generally thought that the first institution to be granted corporate limited liability status was the Bank of England (BOE), then a private institution. The monarchy, in need of financing, struck a deal with the BOE and, in the process, granted it sole authority to act as an agent to the government in terms of the circulation of currency. Included in this arrangement was a provision that BOE “shareholders” would not be held personally liable for any debts incurred by the new corporation (Bagehot [1873] 1991). Before long, any number of commercial firms were appealing to the crown for similar liability protection and monopoly rights to trade either certain goods or in certain areas of the world. Historically there has been a great deal of variation in the liability rules regarding individuals engaged in the business of brokering money and credit. In Europe, financial institutions other than central banks were often denied corporate limited liability long after this status (often accompanied by monopoly trading and governing rights) was allowed to commercial firms. Even after the advent of essentially universal access to corporate limited liability, financial institutions, including twentieth-century investment banks as well as accountants and lawyers, continued to operate under the partnership structure, that is, without the protection of limited liability. Banking, as much or more than almost any other industry, has been the subject of extensive debate and policy discussions concerning liability rules. For example, Evans and Quigley (1995) provide a lively and insightful discussion of shareholder liability regimes as well as an analysis of some data from nineteenth-century Scottish 6. The manager/owner conflict can be viewed as a special case of the principal/agent problem discussed by Adam Smith and formalized by more recent writers such as Ross (1973). Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 53 banking, where corporate limited liability and unlimited liability (personal limited liability only) institutions competed for deposits and investors. They suggest that the dominance of one liability structure over another will depend on whether it is cheaper to monitor the quality of the assets or the personal wealth of the investors providing guarantees. If the former is less costly, corporate limited liability organizations will dominate, and vice versa if asset values are more costly to verify than personal wealth. This argument is consistent with the proposal that, should a banker be able to post sufficient personal collateral, the value of which is verifiable, depositors could eliminate the potentially negative influences of limited liability. It is also consistent with the idea that as a sophisticated system of “monitoring the monitors” (the professional auditing industry) emerged, the other advantages associated with corporate limited liability caused this organizational structure to become dominant in banking in the twentieth century. These advantages primarily involved the easing of risks associated with transferring ownership (Woodward 1985). Indeed, it is difficult to imagine the set of advanced financial markets and contracts that have developed over the past two centuries without some restrictions on the seizure of the personal assets of existing and potential shareholders. Interestingly, however, multiple liability provisions, requiring some additional capital infusions by shareholders in bankruptcy, were not eliminated in the United States until around the time the Federal Deposit Insurance Corporation (FDIC) was formed in 1933. In essence, at least for small depositors, the FDIC substituted a guarantee from the government for private guarantees by bank shareholders.7 Limited Liability and the Lender of Last Resort. Other governmental or quasi-governmental agencies established by efforts to improve the functions of the economy may cause liability-induced behavioral changes in many areas of commerce and the delivery of financial services as well. Consider, for example, the development of the central bank. The central bank, acting as a lender of last resort, works through a subset of financial institutions in order to provide for an elastic currency. Practically, this means it provides liquidity to the financial system during times of stress. These times of need arise in situations when market participants, because of less than full information, have trouble distinguishing economically solvent from insolvent firms. In times when there is great uncertainty investors often seek liquidity. Absent a central bank, “corners” on the provision of liquidity might arise where its provision is left solely to private institutions (Donaldson 1988). Thus, the existence of a lender of last resort allows private institutions of all types, to some degree or another, to hold fewer liquid assets and increase investments in riskier but, on average, more profitable ventures. 54 In the United States, banks and thrift institutions pay a positive tax for privileged access to the discount window by, among other things, holding zero-interestbearing accounts at the Federal Reserve. At the same time, other firms also tend to benefit from the lender of last resort actions by being able to draw their short-term liquidity directly or indirectly through the banking system. To the extent that not all claims on these corporations or partnerships are counteractable (Grossman 1995), there may be welfare effects as these other institutions hold less liquidity than they would in the absence of a lender of last resort (Calomiris 1989; Smith 1993). Limited liability plays a role in this situation because, with less than a full array of contracts to cover all contingencies, managers of these institutions, acting in the interests of shareholders, rationally have an incentive to exploit the optionlike characteristic of their residual claim on the cash flows from production. Of course, a similar argument can be made for institutions with explicit or implicit government guarantees. They too may have an incentive to hold less liquid positions than they would in the absence of backing by the government. In short, the brute force of limited liability can potentially tend to exacerbate the risks faced by policymakers concerned with stabilizing financial markets during times when, for whatever reason, liquidity is in short supply. There is also a potentially positive view of this provision of emergency liquidity. Bernanke (1983), for example, has provided evidence to support the idea that much of the economic damage in the Great Depression was caused by the failure of banks and their customers, eliminating valuable information concerning whether firms were fundamentally solvent but in financial distress or essentially bankrupt. In this sense, bankruptcies can be viewed as socially costly and the public provision of liquidity through the central banking system can be viewed as a way of minimizing these costs, particularly during times of stress in markets and the banking system (Holmstrom and Tirole 1996). That is, if the social costs of bankruptcies are high, then the provision of liquidity to temporarily weak but fundamentally sound economic units may actually improve the welfare of the society despite the partially offsetting effects of limited liability. Conclusion his article has reviewed some of the effects of liability structure on the actions of individuals in financial and labor markets. With roots stretching back at least to the early days of the Bank of England, corporate limited liability has had a strong influence on the development of modern capitalism. Resulting improvements regarding transferability of ownership have greatly enhanced the flow of financial capital and encouraged riskier ventures than might have been taken T Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 if stockholders were personally liable for more of an endeavor. However, these same risk-taking incentives may cause conflicts of interest at both the macro- and microeconomic level. From central banks providing an elastic supply of liquidity in times of financial crises to government guarantees, limitations of liability have at times resulted in risks taken at the firm level that may not be socially optimal. Policymakers, in their quest to balance social goals with excessive risk taking, should be aware that decisions made in the name of safety and soundness may result in unintended consequences. Corporate liability risks can even drive a wedge between contracting parties in the sense that the rational reaction by outsiders to the option value of limited liability held by stockholders may result in investment, financing, and security design decisions that would not be optimal if liability provisions were less generous. Even in bankruptcy negotiations, the limited liability option is potentially valuable to stockholders who may hold up creditors with the threat that they can walk away, leaving claimants to pay the bankruptcy costs associated with dissolving a firm. Limited liability has also proven to be an important factor in the adverse selection (only the risky apply) and moral hazard (taking on riskier ven- tures) problems often encountered in borrower/bank relationships. Absent nonpecuniary costs (such as debtors prison), complications due to limited liability may be sufficiently severe to cause some or even all borrowers to be credit rationed. Personal limited liability may also act as an incentive in labor contracting arrangements, whereby individuals sell their ideas and then hold up outside claimants for additional compensation for their needed labor efforts. Or, given the fact that, realistically, a positive amount of income is needed in order to engage in productive effort, the assumption that personal limited liability is zero, and not a subsistence value, may result in a cycle of unemployment and malnutrition for at least a portion of a population. Although this article has covered but a portion of the issues arising from the existence of corporate and/or personal liability, these and other examples are sufficient to show that alterations in liability provisions have changed the nature of contracting in ways that require us to remember that the buck does stop somewhere, and where it stops is not irrelevant from either a private or public perspective. 7. While the system of public insurance has been established in the United States for more than sixty years, it is not surprising that many of the problem financial institutions in the 1980s were those that were allowed to remain open even though their liabilities exceeded their assets. With limited liability and, for all intents and purposes, no investment in the firm, managers of these institutions had every incentive to take on high-risk projects in the hopes of growing out of their problems. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997 55 R E F E R E N C E S ANDERSON, GARY M., AND ROBERT D. TOLLISON. 1982. “Adam Smith’s Analysis of Joint-Stock Companies.” Journal of Political Economy 90, no. 6: 1237-56. NACHMAN, DAVID C., AND THOMAS H. NOE. 1994. “Optimal Design of Securities under Asymmetric Information.” Review of Financial Studies 7, no. 1: 1-44. BAGEHOT, WALTER. [1873] 1991. Lombard Street: A Description of the Money Market. Reprint, Philadelphia: Porcupine Press, Orion Editions. NOE, THOMAS H., AND STEPHEN D. SMITH. 1994. “Contractual Opportunism, Limited Liability, and the Role of Financial Coalitions.” Federal Reserve Bank of Atlanta, Working Paper 94-17, November. BECKER, GARY S. 1993. “George Joseph Stigler: January 17, 1911December 1, 1991.” Journal of Political Economy 101, no. 5: 761-67. BERNANKE, BEN S. 1983. “Nonmonetary Effects of the Financial Crisis in Propagation of the Great Depression.” American Economic Review 73, no. 3: 257-76. CALOMIRIS, CHARLES W. 1989. “The Motivations for Loan Commitments Backing Commercial Paper: Comment.” Journal of Banking and Finance 13, no. 2: 271-77. DASGUPTA, PARTHA. 1993. An Inquiry into Well-Being and Destitution. New York: Oxford University Press. DONALDSON, R. GLEN. 1988. “Panic, Liquidity, and the Lender of Last Resort: A Strategic Analysis.” Board of Governors of the Federal Reserve System, International Finance Discussion Paper 332. EVANS, LEWIS T., AND NEIL C. QUIGLEY. 1995. “Shareholder Liability Regimes, Principal-Agent Relationships, and Banking Industry Performance.” Journal of Law and Economics 38, no. 2: 497-520. GROSSMAN, SANFORD J. 1995. “Dynamic Asset Allocation and the Informational Efficiency of Markets,” Journal of Finance 50, no 3: 773-88. HART, OLIVER, AND JOHN MOORE. 1994. “A Theory of Debt Based on the Inalienability of Human Capital.” Quarterly Journal of Economics 109, no. 4: 841-79. HOLMSTROM, BENGT, AND JEAN TIROLE. 1996. “Modeling Aggregate Liquidity.” American Economic Review 86, no. 2: 187-91. JOHN, KOSE, TERESA A. JOHN, AND LEMMA W. SENBET. 1991. “Risk Shifting Incentives of Depository Institutions: A New Perspective on Federal Deposit Insurance Reform.” Journal of Banking and Finance 15, nos. 4-5: 895-915. 56 ROSS, STEPHEN A. 1973. “The Economic Theory of Agency: The Principal’s Problem.” American Economic Review 63, no. 2: 134-39. SMITH, ADAM. [1776] 1994. An Inquiry into the Nature and Causes of the Wealth of Nations. Reprint, New York: Modern Library Press. SMITH, STEPHEN D. 1993. “Private versus Public Efficiency in the Securities Industry.” Journal of Banking and Finance 17, no. 2-3: 563-65. STIGLER, GEORGE J. 1945. “The Cost of Subsistence.” Journal of Farm Economics 27:303-14. STIGLITZ, JOSEPH F., AND ANDREW WEISS. 1981. “Credit Rationing in Markets with Imperfect Information.” American Economic Review 71, no. 3: 393-410. WILSON, BRENT D., AND EDWARD J. KANE. 1996. “The Changing Value of Double Liability to Large-Bank Stockholders during 1927-33.” Photocopy. WINTON, ANDREW. 1993. “Limitation of Liability and the Ownership Structure of the Firm.” Journal of Finance 48, no. 2: 487-512. WOODWARD, SUSAN E. 1985. “Limited Liability and the Theory of the Firm.” Journal of Institutional and Theoretical Economics 141, no. 4: 601-11. ZHA, TAO. 1995. “Bankruptcy Law, Capital Allocation, and Aggregate Effects: A Dynamic Heterogenous Agent Model with Incomplete Markets.” Federal Reserve Bank of Atlanta, Working Paper 95-8, October. Federal Reserve Bank of Atlanta E C O N O M I C R E V I E W First Quarter 1997