Printer Friendly

A positive model of reserve requirements and interest on reserves: a clearinghouse interpretation of the Federal Reserve System.

1. Introduction

The Federal Reserve Act required member banks to hold reserves at the Fed, but it did not authorize pecuniary interest payments on these reserves. There has been considerable debate on the welfare justification of these provisions (Sellon and Weiner 1996). One view is that the provisions imposed an unnecessary burden on the banking system in the form of a reserve requirement tax. The welfare implication is that either reserve requirements should be eliminated or reserves should pay interest. An opposing view is that, while burdensome, reserve requirements are necessary for monetary control and should be set as high as is feasible. The question of how the government actually chooses a long-run level for the reserve requirement has received little attention in the literature.

The related issue of what determines short-run fluctuations in reserve requirements has received somewhat more attention. A traditional textbook monetary policy approach suggests a stabilization motive: The government lowers the reserve requirement when real economic activity is below its long-run trend and raises the requirement when it is above trend. With respect to interest payments on reserves, the presumption is that they have been absent throughout the Fed's history.

This paper develops a positive model of Fed reserve requirements and interest payments on reserves, which differs significantly from previous approaches. For one thing, the model views the Fed first and foremost as a clearinghouse. Second, the model replaces the stabilization theme with one that is rooted in public finance considerations - in particular, optimal tax considerations. At the center of the model is a government whose objective is to satisfy a revenue requirement by raising revenue from a number of sources, including a levy on money issue.

The clearinghouse model starts with the observation that, in establishing the Fed, the founders granted the Fed the right to issue its own currency. Although the founders' stated intent was to produce an elastic currency (Miron 1986), the side effect was a profit flow for the Fed. These funds could be used to subsidize member banks and to finance general government outlays. The model allows bank subsidies to take the form of in-kind payments, related to a bank's reserve holdings, or low interest rate discount loans. Contrary to the conventional approach, therefore, the absence of pecuniary payments does not necessarily imply a zero interest rate on reserves.

Bank subsidies are more than a theoretical possibility. Historically, the Fed has spent a major portion of its money creation revenue on bank services associated with the payments system and has set its discount rate below market rates. There was one exceptional period centered around 1940, however, when subsidies were insignificant. One challenge facing a positive model is to explain this exceptional period. A second challenge pertains to reserve requirements. After staying constant from 1918 to 1935, reserve requirements doubled over the next two years (1936 and 1937) and fluctuated thereafter. According to the clearinghouse model developed here, subsidy payments are closely tied to market interest rates on bank assets, and reserve requirements are tied to the government's financing needs.

2. Origins of the Federal Reserve

A major purpose of the Federal Reserve Act was to nationalize the private clearinghouse system in the form of the Federal Reserve System (Gorton 1985). Legal restrictions on currency issue had interfered with the ability of private clearinghouses to provide liquidity to the rest of the banking system (Spahr 1926; White 1983). Relieving the Fed of such restrictions would lower its operating costs, thereby freeing up revenue for funding a government revenue requirement and for subsidizing banks.

A crucial issue in the design of the Fed as a national clearinghouse was its relationship to the rest of the government. At one extreme, clearinghouse functions (setting reserve requirements and subsidy levels) could be subsumed within the general government in the form of a budget-funded Fed. At the other extreme, some or all functions could be "farmed out" to an independent Fed. The choice would largely depend on the cost of overseeing (monitoring) the operation of a budget-funded Fed versus the loss of control associated with independence.(1)

The Federal Reserve Act resolved the trade-off by assigning some but not all clearinghouse functions to an independent Fed. The precise terms of the Act were shaped by the founders' desire to balance the competing claims of three groups - the government, Fed management, and member banks - on the revenue freed up by the elimination of the legal restriction (Willis 1923). The Act in its final form institutionalized revenue transfers to the government (Federal Reserve Act, section 7) and made membership in the System voluntary by allowing banks to choose state or national charters, with only the national banks required to become Fed members (Federal Reserve Act, sections 2 and 9). Whether a bank would find Fed membership attractive would depend on the extent to which the Fed subsidized its members by passing along the profits from its legal exemption. Although explicit interest payments on reserves were not authorized, the Act allowed Federal Reserve Banks to waive fees for services provided to member banks and to set discount rates below market levels.

3. A Clearinghouse Model of the Fed

The Setup

This section outlines a clearinghouse model of the Fed that is consistent with the idea that in addition to an elastic currency, the Federal Reserve Act provided a source of revenue for general government outlays.(2) The model is based on a government that has a seigniorage requirement and a financial sector that consists of the Fed, a private clearinghouse, and banks. Banks can obtain clearinghouse services from private sources or by choosing to become members of the Fed.

The Fed generates revenue by backing Fed money (currency and reserves of member banks) with income-earning assets. Revenue can be transferred to the government or rebated to member banks through subsidies that are tied to a bank's holding of reserves. The higher the subsidies, the lower will be the burden of membership. The burden per dollar of bank deposits can be given by k(i - s), where k is the reserve requirement, i is the market rate of interest, and s is the subsidy rate on reserves. Conceptually, s may be thought of as an in-kind payment to banks per dollar of reserves, thus representing an implicit interest rate.

The model assumes that the government's objective is to satisfy its seigniorage requirement in a way that minimizes the burden on the banking sector. Figure 1 depicts the government's decision problem where the vertical axis plots the reserve requirement and the horizontal axis plots the spread, (i - s), between the loan rate and the subsidy rate on reserves. There is a family of rectangular hyperbolas with points on a hyperbola giving the various k and (i - s) combinations which result in a particular burden, k(i - s). The burden rises with movements from one curve to another, [B.sub.1] to [B.sub.2].

There is also a family of iso-net-revenue ellipses. Each ellipse shows revenue net of subsidies and therefore represents the amount, G, that can be transferred to the government. The slope of an iso-net-revenue ellipse shows the rate at which the spread can be substituted for the reserve requirement with no change in net revenue. One point, A, gives the unique k and (i s) combination that maximizes net revenue. From A, an increase in k or (i - s) will decrease the Fed's net revenue because of a relatively large fall-off in membership and therefore in reserves held at the Fed.

Assume that the government requires revenue equal to [G.sub.1] and that the government does not have to worry about the Fed complying with its wishes. To obtain [G.sub.1], the government would instruct the passive Fed to set the reserve requirement at [k.sub.2] and the subsidy rate at [s.sub.2]. The corresponding burden is indicated by the hyperbola labeled [B.sub.1], which passes through point D. Any other k and s combination would result in either too little revenue or else in an unnecessarily high burden for the [G.sub.1] requirement.

As long as the government head can signal preferences to the Fed at zero cost and can monitor the Fed's activities at zero cost, then solution D in Figure 1 will emerge. If agency costs are positive, however, then the government head has no assurance that the desired k and s will be forthcoming. One way of introducing agency costs is to assume that the government cannot readily control one or both of the clearinghouse instruments, k and s. When subsidies are paid implicitly, as was the case with the Federal Reserve, control of s may be particularly problematic. It may be difficult to verify the quantity and, perhaps more important, the quality of services provided. The government may deal with agency costs by establishing a budgetary relationship with a dependent Fed or by "farming out" clearinghouse responsibilities to an independent Fed.

A Dependent Fed

This section assumes that the government head deals with the problem of observing the subsidy rate by establishing a budgetary relationship with the Fed. The government grants a budget in exchange for revenue and for promised levels of the subsidy rate as well as the reserve ratio. Presumably, the budget covers subsidy payments, which will be recouped when the Fed transfers revenue back to the government.

There are advantages and disadvantages to this type of budgetary tie-in with a dependent Fed. A disadvantage is that it may provide the Fed management team with the opportunity to engage in what might be referred to as "bureaucratic slack" (Meone 1986). An advantage is that the budget process gives the government a method of intervening within and between periods to influence the subsidy rate (Weingast and Moran 1983). To highlight these advantages and disadvantages, assume that the budget process gives the government head complete control over the subsidy rate and reserve ratio but not over slack. Because the Fed's budget would have to rise to cover slack, less Fed revenue would be left over for the government to satisfy a seigniorage requirement such as [G.sub.1].

Figure 1 illustrates the effect on the reserve requirement and the subsidy rate. If, say, [G.sub.2] must be raised in order to have [G.sub.1] left over, then the best the government can do is to choose the k and s combination corresponding to point D[prime]. The burden ([B.sub.2]) on member banks is higher at D[prime] than at D, where there are no agency costs.

An Independent Fed

An alternative way of dealing with the agency cost problem would be to farm out the selection of a subsidy rate to an independent agency. We shall call this new type of clearinghouse the independent Fed. In this case, the government could be viewed as entering into a contractual relationship with the Fed.(3) The contract could specify a particular value for the readily observable variable, k, but delegate the selection of s to the Fed.(4) The contract also could be made contingent on an up-front payment, which would constitute the seigniorage requirement, leaving the Fed with rights to residual revenue (after subsidy payments) produced in its capacity as a clearinghouse.

Again, assume that the government requires revenue equal to [G.sub.1]. Also assume that Fed payoffs are positively related to its residual. If the government were to set a reserve requirement of [k.sub.2] in Figure 1, the independent Fed would move to point I[prime] on reaction curve MN, leaving the Fed with residual revenue after transferring [G.sub.1] and paying subsidies. Anticipating this outcome, the government would reduce the reserve requirement to [k.sub.1] and the Fed would respond with [s.sub.1]. Comparing point i with D, [s.sub.1] is lower than [s.sub.2] and [B.sub.2] is higher than [B.sub.1]. The higher burden is unavoidable given that the independent Fed has a built-in incentive to chisel on subsidy rates as a way of increasing retained earnings. The [B.sub.2] burden is lower, however, than if the government had delegated control to the winning bidder to set both k and s. At point A, too much net revenue is generated and an unnecessarily high burden is imposed on the banking system.

A Dependent or an Independent Fed?

The clearinghouse model offers a new perspective upon which to view the timeworn debate between those advocating central bank dependence versus independence. The crucial issue within the context of this model is whether dependence or independence provides the government with the means of raising a given amount of revenue while imposing the smaller burden on the banking sector. Figure 1 illustrates the case where the government would be indifferent between the two. As constructed, dependence (point D[prime]) and independence (point I) result in the same burden, [B.sub.2]. Higher monitoring costs would induce the government head to choose an independent Fed, and lower monitoring costs would lead to the selection of a dependent Fed.

The founders of the Fed chose neither a strictly dependent nor a strictly independent structure. Dependence was rejected in that the Fed was not set up as an ordinary executive agency whose chair serves at the discretion of the President. Of even more significance, the Fed is not subject to the ordinary congressional appropriation process - operating expenses are financed from money creation revenue. On the other hand, the Fed is not a private firm with shareholders exercising a residual claim to revenue. Commercial banks own Fed stock but do not possess voting rights normally accruing to shareholders. Nor are there secondary resale markets for shares.

Which structure represents the closer fit? While the contracting motif underlying independence should not be taken as literally descriptive, it is not too farfetched to imagine that a President would offer to a Fed appointee employment conditions that are akin to those implied by the hypothetical contract of the section An Independent Fed:

The administration wants you to serve as the next Fed chair. I ask only two things from you: enforce the prevailing reserve requirement and turnover $x to be dispersed in annual lump sum payments during your four-year term. The administration respects the Fed's independence. Other than these two things you are free to conduct the Fed's business as you choose. Do you accept these terms?

Because the government has formulated the contract so that there is only one viable subsidy rate - the one that leaves the Fed with zero surplus - the contract need not be backed with threats or with more subtle means of enforcement. If the Fed attempted to chisel (undercut) the subsidy rate, then the attractiveness of membership would fall, and the Fed would be unable to survive after making the required transfer payment. The Fed would literally go bankrupt.(5)

One criticism of applying the contracting motif to the present-day Fed might be directed toward the modeling decision that the Fed controls the subsidy rate but not the reserve requirement. In defense of this assumption, consider that the Federal Reserve Act granted the Fed broad discretion over the scope of subsidies to the private banking system but gave only Congress the authority to change reserve requirements. Although the Banking Acts of 1933 and 1935 increased the scope of the Fed's authority, the Act of 1935 did so within the context of prescribing minimum and maximum limits on reserve requirements. Periodically, Congress has adjusted these limits and has debated proposals that, although not ultimately enacted, would have imposed specific requirements on banks. The observation that reserve requirements have typically been at or near their congressionally mandated upper limits and that Congress has been more involved in overseeing reserve requirements than in subsidy rates provides a rationale for modeling the government as effectively controlling reserve requirements, even after 1935, while delegating subsidy rates.

4. Comparative Static Analysis

Figure 1 can be used to illustrate how the government and an independent Fed would respond to shocks in G or in i. An increase in the revenue requirement from [G.sub.1] to [G.sub.2] would require the government to recontract with the Fed for a higher reserve requirement of [k.sub.2], and the Fed would respond by choosing I[prime], where the subsidy rate is somewhat higher than at I. An increase in the market interest rate tends to shift the iso-revenue map in the northeast direction with maximum net revenue now higher than before because of increased earnings from issuing currency to the public.(6)
Table 1. Model Predictions

 (A)
 Clearinghouse Approach

 Shock
Model G i

Independent Fed

k + -
s + 9

Dependent Fed

k + -
s - +

 (B)
 Traditional Approach

 Shock
Model GNP G i

Stabilization

k + -(a) +(a)
S NR(b) NR(b) NR(b)

a Conditional on negative and positive correlation of G and i
with economic activity.

b NR: not relevant.


The new reaction path is M[prime]N[prime], with maximum revenue at point A[prime], and with [G.sub.1][prime] now designating the ellipse that generates revenue of [G.sub.1]. The government would decrease the reserve requirement from [k.sub.1] to [k.sub.0], and the independent Fed would set a subsidy rate that results in position I[double prime]. Whether s increases or decreases is indeterminate. Whereas the higher loan rate encourages a higher s, the lower reserve requirement pulls in the opposite direction. The government's and the Fed's optimal choices can be summarized by [k.sup.*] = k(i, G) and [s.sup.*] = s[i, k(i, G)] with the signs shown in Table IA.

The budget-funded, dependent clearinghouse would respond differently. In Figure 1, an increase in G leads to an increase in k and a decrease in s (a movement from D' to a point northeast on the expansion path DA). With an increase in i, k falls and s rises.

These clearinghouse approaches can be contrasted with a more traditional central bank stabilization model. Loungani and Rush (1995) show that decreases in reserve requirements tend to increase real economic activity. If the government has a stabilization objective, then it will change the reserve requirement according to the economy's position in the business cycle. The government will systematically respond to movements in interest rates and in government spending only to the extent that these variables are correlated with economic activity. For instance, if the government's revenue requirement rises and interest rates fall during a period of low economic activity (see section Tests), then the stabilization model reverses the outcome produced by an independent Fed. The stabilization model predicts that reserve requirements fall with increases in the government's revenue requirement and decreases in the market interest rate.(7)

Table 1 compares the predictions of the models. There is at least one point of contention between the model of an independent clearinghouse and the other models. The model of a dependent clearinghouse disagrees on the response of the subsidy rate to a change in the government's revenue requirement. The stabilization model conditionally predicts different responses of the reserve ratio to the government's revenue requirement and to the market interest rate.

5. Empirical Analysis

Overview

The first empirical issue is measurement of the clearinghouse variables, particularly the subsidy rate. The subsidy rate is the sum of a spending rate and a monetary rate. The spending rate stems from Fed outlays on bank services (for example, check collection, coin and currency pickup and delivery, wire transfers, and safekeeping securities) made available free of charge. The monetary rate stems from discount loans extended at a rate that is below the market loan rate.

Because a consistent time series on individual bank services is not available, I take a highly aggregated approach in measuring the spending rate. The spending rate for any year is the sum of the two reserve bank outlay categories "net expenses" and "dividends paid" (Board of Governors 1987, their Table 7) divided by end of the year member bank reserves (Board of Governors 1987, their Table 14).(8) The monetary rate is measured by (i - b)[q.sub.B] + [iq.sub.F], where (i - b) is the spread between the paper rate and the discount rate, [q.sub.B] is discount loans divided by bank reserves, and [q.sub.F] is float divided by bank reserves.(9)

Figure 2 plots the log of the reserve requirement on demand deposits for country banks, and Figure 3 plots the log of the subsidy rate. The series start in 1917 when consistent reserve requirement data were first available and end in 1980 with passage of the Monetary Control Act.(10) Reserve requirements over this sample period have varied from a low of 7% to a high of 16%, and measured subsidy rates have varied from close to zero to over 5%.

In the independent clearinghouse model, profits that are not used to support the banking system are transferred to the government. During the 1917-1922 war and peacetime transition years, Fed transfers averaged 25% of Fed revenue; from 1923 to 1946 about 1%; and from 1947-1979 about 90%. As a percentage of federal government receipts, transfers have grown in importance from roughly 0.5% in the 1950s to close to 2% by 1979 (Goodfriend and Hargraves 1983).

A crucial claim of the clearinghouse model is that the transfers are not at the discretion of the Fed but are instead determined by the government's revenue needs. Recorded transfers, however, are at times a poor measure of the "true" money creation revenue accruing to the Treasury (Toma 1985). For this reason, the empirical work focuses on a measure of permanent government spending and assumes that the seigniorage requirement changes with this measure.

Figure 2 plots a permanent government spending variable from Barro (1986) alongside the reserve requirement.(11) With movements in Barro's variable reflecting movements in G, the independent clearinghouse model would attribute the absence of variation in the reserve requirement from 1917 to the mid-1930s to the low volatility (post-World War I) of G. Once it became evident that G was increasing and had become more volatile in the aftermath of the Great Depression, Congress reacted by passing the Banking Acts of 1933 and 1935, which changed the k-setting mechanism in a way that would be expected to generate more flexibility. In the near term, the government's financing objective and the Fed's earning objective called for the same policy - an increase in the reserve requirement. This is precisely what occurred with the doubling of reserve requirements in 1936 and 1937 to their new statutory upper limits.

From a traditional stabilization perspective, reserve requirement changes are undertaken in order to dampen the business cycle. The presumption is that there are significant deviations in output from a deterministic long-run trend. Using a traditional decomposition method, Figure 4 plots detrended real gross national product (GNP) and, for comparison purposes, the change in reserve requirements.(12) The 1936 and 1937 increase in reserve requirements occurred during a period when economic activity was significantly below trend. This represents a clear-cut piece of evidence against the stabilization model. Indeed, whereas the independent clearinghouse model would interpret this episode as indicative of maximizing behavior on the part of the government, the traditional interpretation is that it represented a misguided attempt by the Fed in the midst of a depressed economy to reduce the excess reserves held by the banking system (Barro 1993).

Turning to the Fed's subsidy rate policy, Figure 3 highlights the relationship between the overall subsidy rate, s, and the market loan rate, i, as represented by the commercial paper rate. Consistent with the independent clearinghouse model, the pattern in the subsidy rate roughly corresponds to the pattern in the commercial paper rate. This suggests that the pre-World War II fall and the post-World War II rise in the subsidy rate can be explained most simply as a reaction of the Fed to the fall and rise of market interest rates. In contrast, the stabilization model is silent on the issue of subsidy rate movements.

Of course, Figures 2-4 provide only a preliminary indication of the explanatory power of the independent clearinghouse model versus the stabilization model. Figure 2 does not indicate the potential role of the market interest rate in reserve requirement movements, and Figure 3 does not indicate the potential role of the government's revenue requirement in subsidy rate movements. Casual inspection suggests the importance of taking these other factors into account. For instance, the post-World War II rise in interest rates may help explain the post-World War II fall in reserve requirements in Figure 2; the rise in the 1930s in permanent government spending may help to explain the reduction in the gap between the market interest rate and the subsidy rate in Figure 3.

Also, it would be premature to reject the stabilization model on the basis of one piece of evidence. Traditionally, the 1936-1937 episode has been viewed as a learning experience for the Fed. Having learned the lesson, the Fed would lower reserve requirements when confronted with a period of low economic activity in the future. The next section turns to formal tests of the independent clearinghouse model versus the stabilization model.

Tests

Testing will be organized as follows. The primary focus will be on the subsidy rate and reserve requirement predictions of the independent clearinghouse model and its chief rival, the stabilization model. A finding that the government's revenue requirement is negatively and the market interest rate is positively correlated with economic activity over the sample period (19171980) leads to a head-to-head test between the two theories, because they differ in their predictions of how the reserve ratio responds to G and i shocks under these circumstances. This test, however, is a relatively weak one in that it is conditional on the above-mentioned properties of the G and i series. A stronger test of the stabilization model will focus directly on the short-run adjustment of the reserve ratio during periods of abnormally low and high economic activity. The performance of the independent clearinghouse model also will be compared to the dependent model's performance.

An important issue in time series data is whether variables are stationary or nonstationary (Nelson and Plosser 1982; Hamilton 1994). A nonstationary series would imply that changes are permanent. Previous research has found that 20th-century nominal interest rates approximate a unit root (Barro 1989). Also, if reserve requirements are adjusted on the basis of long-ran considerations, then a unit root process may best characterize them.

This fact suggests a particular testing strategy. After transforming the variables into natural logs, test whether unit-root processes generate the subsidy rate, the paper rate, and the reserve ratio. A finding that the variables are nonstationary leads next to a test for cointegration in order to determine whether there is a stable long-run relationship among the variables. The Johansen method is particularly well suited for the problem at hand because it indicates the number of cointegrating vectors.(13) A finding of cointegration with the expected signs provides strong evidence for the independent clearinghouse model. Other approaches do not predict a relationship, or else they predict signs that disagree with the independent clearinghouse model.

Dickey-Fuller tests indicate that all of the variables are nonstationary. Given these results, Table 2 reports findings from a cointegration test.(14) The trace and maximum likelihood eigenvalue statistics indicate the number of cointegrating vectors. Both sets of statistics point to the conclusion that there is one cointegrating vector with a coefficient of 1.28 on the loan rate and 1.25 on the reserve requirement. Interpreting ln(s,) = 1.28 ln(i,) + 1.25 ln([k.sub.t]) as the long-run Fed subsidy rate equation gives a relationship that is broadly consistent with the independent clearinghouse model.(15)

Based on the finding of cointegration, the short-run dynamics of the system can be represented by an error-correction model. The Johansen procedure gives error-correction equations for [Delta]ln([s.sub.t]), [Delta]ln([i.sub.t]), and [Delta]ln([k.sub.t]) as left-side variables. Section c of Table 2 reports the error-correction coefficients for each equation. Generally, the coefficients indicate the extent to which that variable adjusts in the short run to bring the system back into long-run equilibrium. The insignificance of the error-correction coefficient in the interest rate equation is consistent with the model's treatment of i as an exogenous variable. The significance of the coefficients in the other two equations indicates that in the short run, s and k respond to deviations from long-run equilibrium. Moreover, the magnitude of the subsidy rate coefficient points to s as the primary adjustment variable.

Next, consider the government's decision as given by [k.sup.*] = k(i, G). Table 3 presents the basic cointegration results. The Johansen procedure indicates one cointegrating vector. As predicted [TABULAR DATA FOR TABLE 3 OMITTED] by the independent clearinghouse model, the coefficient on the loan rate is negative, and the coefficient on the revenue requirement is positive. Both are significantly different from zero according to likelihood ratio test statistics.

The error-correction coefficient for the k equation differs significantly from zero, implying that the gap between the actual reserve requirement and its long-run equilibrium value in one period is reduced in a subsequent period. The other error-correction coefficients are not significant, which is consistent with viewing i and G as exogenous. Taken together, the results in Tables 2 and 3 are broadly consistent with the model's characterization of the Fed's and the government's decision problems.(16)

These results cannot be easily explained within the context of the competing models. In particular, the stabilization view would deny the importance of the subsidy rate as a Fed control variable. For the reserve ratio, the Fed's objective would be to move economic activity toward a long-run trend. If the paper rate covaries positively and the revenue requirement negatively with economic activity, then the error-correction coefficient signs in Table 3 should be the opposite of the ones reported. Simple correlation tests (not reported) suggest this type of covariation. Therefore, Table 3 provides indirect evidence against the stabilization model.

A more direct test of the stabilization view would focus on the short-run adjustment of the reserve ratio during periods of abnormally low and high economic activity. Casual inspection of Figure 4 indicates no obvious relationship between GNP and the reserve requirement. More formally, Granger-causality tests for the change in the log of the reserve requirement and for the detrended log of real GNP were run for different Fed sample periods. Table 4 reports the results.(17) For the full sample, the tests indicate bidirectional Granger causality. In the GNP-causes-k [TABULAR DATA FOR TABLE 4 OMITTED] regression, the sum of the GNP coefficients is not significantly different from zero, which is contrary to the stabilization prediction.

Because the stabilization approach treats the Fed as the stabilizing agent, and because the Fed had neither statutory nor effective reserve requirement powers before 1935, I reran the tests for the sample period exclusive of 1917-1934. The results do not qualitatively differ from the full sample results. Finally, I considered the possibility that the Fed did not understand how to conduct a stabilization policy until after WWII. The 1947-1979 results again indicate bidirectional Granger causality and lead to the same overall conclusion: There is little evidence for the hypothesis that the Fed responded to deviations in economic activity around trend by manipulating reserve requirements.

With respect to the dependent clearinghouse model, the evidence, provided in Tables 2 and 3, also is not supportive. The dependent model posits opposing movements in the subsidy rate and the reserve ratio in response to exogenous shocks to the revenue requirement. The positive coefficient on In(k) in section b of Table 2 provides strong counterevidence.

6. Conclusion

This paper represents a first attempt to construct and test a theory of how reserve requirements and Fed subsidies to the banking system are set. While positing a degree of independence, the theory highlights that the Fed, like any other government firm, must in the long run be treated as an agent of the general government. The driving force behind the theory is that the government in 1914 wanted to create a national clearinghouse that could be called upon in the future to generate revenue for general funding purposes.

One novel and perhaps controversial feature of the paper is the conjecture that the clearinghouse model continued to be an apt characterization of the Federal Reserve throughout its existence. The clearinghouse approach stands in stark contrast to more traditional approaches that emphasize monetary policy. At a fundamental level, monetary policy presumes a decisionmaker that is able to exercise discretion. The presumption behind the clearinghouse model is that conditions external to the Fed limit discretion. The government's revenue demands and the banks' membership options impose tight constraints on the clearinghouse activities of a central bank.

The clearinghouse model could be modified to allow for the possibility that the Fed controls a traditional monetary policy variable, such as the market interest rate, in addition to the subsidy rate. As long as the clearinghouse setup was retained, however, the basic implications of the simple model concerning movements in reserve ratios and subsidy rates would tend to carry over for this version. Ultimately, the clearinghouse model of an independent Fed must be judged on the same standard used to assess the performance of any other economic model - "how well does it predict?" In this regard, the implications of the basic model provide a reasonably accurate account of short- and long-run movements in reserve requirements and Fed subsidy rates up to 1980.

An interesting topic for future research is how the Monetary Control Act of 1980 altered the constraints on the Fed and the relationships among its clearinghouse instruments. The Act impinged on Fed decision making in two ways. First, it reduced the exit options of banks by imposing the System's reserve requirements on all federally insured depository institutions. Second, it gave the Fed the opportunity to extend its services to nonmember banks and thrift institutions. The traditional rationale for these provisions is a monetary policy one - to tighten the Fed's control over the money supply. In contrast, the model developed in this paper would emphasize the membership decline in the previous two decades. If not reversed, this trend would have all but eliminated Fed revenue as a source of general government financing. By providing the Fed with a more captive banking audience, the Act slowed the erosion of the monetary tax base.

1 Toma and Toma (1992) discuss this issue with respect to the collection of taxes.

2 The formal model underlying this section is available upon request.

3 See Walsh (1995) for a formal analysis of central bank contracts.

4 The contract also could specify a subsidy rate; however, with the assumption that in-kind subsidies are not readily observable, this more complex contract would result in the same type of monitoring problem that arose with the dependent Fed.

5 Although the Federal Reserve Act provided for transfers from the Fed to the government, it did not provide for transfers in the opposite direction (Goff and Toma 1993).

6 I assume that currency is exogenous, neither entailing collection costs nor paying interest.

7 A completely autonomous Fed that controlled both the reserve ratio and the subsidy rate would increase the reserve ratio in response to an increase in the market interest rate.

8 The measured spending rate is an overestimate in that some outlays (for example, spending on research activities) do not necessarily provide benefits targeted to member banks.

9 Although the spending rate is the major component, the monetary rate has increased relative to the spending rate since World War II.

10 From 1960 to 1979 the reserve requirement is measured as the statutory rate multiplied by a factor of 0.77. This adjustment factor reflects the change in the definition of reserves to include vault cash after 1959.

11 Barro's variable is the ratio of normal federal spending to GNP.

12 Detrended real gross national product is the residual from regressing the log of GNP against a constant, time, and time-squared.

13 Implementation of the Johansen methodology is described in Dickey, Jansen, and Thornton (1991) and in Hamilton (1994).

14 Changes in the statutory reserve requirement were not continuous over the sample period. As a check on the reported cointegration results, I follow Barro (1993) in using the ratio of required reserves to checkable deposits as a continuous measure. Replacing the statutory with the continuous measure does not significantly affect the results in Tables 2 and 3. The cross-correlations between the measures are 0.92 and 0.86 in levels and differences.

15 Likelihood ratio test statistics indicate that the elasticities differ significantly from zero.

16 The model also predicts that the proportion, m, of banks that are members falls with increases in i and k. Testing indicated cointegration among the logs of m, i, and k with the predicted signs.

17 Table 4 uses the detrended variable described in note 12. I also ran the tests using the Hodrick and Prescott filter with similar results. See Nelson and Plosser (1982) for a critique of the traditional stabilization approach, which decomposes output into deterministic trend and stationary cyclical components.

References

Barro, Robert. 1986. U.S. deficits since World War I. Scandinavian Journal of Economics 88:195-222.

Barro, Robert. 1989. Interest rate targeting. Journal of Monetary Economics 23:3-30.

Barro, Robert. 1993. Macroeconomics. 4th edition. New York, NY: John Wiley.

Board of Governors of the Federal Reserve System. 1987. Annual report. Washington, DC: U.S. Government Printing Office.

Dickey, David, Dennis Jansen, and Daniel Thornton. 1991. A primer on cointegration with an application to money and income. Federal Reserve Bank of St. Louis, Review March/April:58-78.

Goff, Brian, and Mark Toma. 1993. Optimal seigniorage, the gold standard, and central bank financing. Journal of Money, Credit, and Banking 25:79-95.

Goodfriend, Marvin, and Monica Hargraves. 1983. A historical assessment of the rationales and functions of reserve requirements. Federal Reserve Bank of Richmond, Economic Review May:3-21.

Gorton, Gary. 1985. Clearinghouses and the origins of central banking in the U.S. Journal of Economic History 42:277-84.

Hamilton, James. 1994. Time series analysis. Princeton, NJ: Princeton University Press.

Loungani, Prakash, and Mark Rush. 1995. The effect of changes in reserve requirements on investment and GNP. Journal of Money, Credit, and Banking 27:511-26.

Meone, Karl. 1986. Types of bureaucratic interaction. Journal of Public Economics 29:335-45.

Miron, Jeffrey. 1986. Financial panics, the seasonality of the nominal interest rate, and the founding of the Fed. American Economic Review 76:125-40.

Nelson, Charles, and Charles Plosser. 1982. Trends and random walks in macroeconomic time series. Journal of Monetary Economics 10:139-62.

Sellon, Gordon H., and Stuart E. Weiner. 1996. Monetary policy without reserve requirements: Analytical issues. Federal Reserve Bank of Kansas City, Economic Review 81:5-24.

Spahr, Walter. 1926. The clearinghouse and collection of checks. New York, NY: Bankers Publishing Company.

Toma, Eugenia, and Mark Toma. 1992. Tax collection with agency costs: Private contracting or government bureaucrats. Economica 59:107-20.

Toma, Mark. 1985. A duopoly theory of government money production: The 1930s and 1940s. Journal of Monetary Economics 15:363-82.

Walsh, Carl. 1995. Optimal contracts for central bankers. American Economic Review 85:150-67.

Weingast, B. R., and M. Moran. 1983. Bureaucratic discretion or congressional control: Regulatory policy-making by the FTC. Journal of Political Economy 91:765-800.

White, Eugene. 1983. The regulation and reform of the American banking system, 1900-1929. Princeton, NJ: Princeton University Press.

Willis, Henry Parker. 1923. The Federal Reserve System. New York, NY: The Ronald Press Company.
COPYRIGHT 1999 Southern Economic Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1999, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Author:Toma, Mark
Publication:Southern Economic Journal
Date:Jul 1, 1999
Words:6616
Previous Article:On-the-job training, establishment size, and firm size: evidence for economies of scale in the production of human capital.
Next Article:Virtual patent extension by cannibalization.


Related Articles
Income and expenses of the Federal Reserve Banks.
Competition and Monopoly in the Federal Reserve System, 1914-1951: A Microeconomic Approach to Monetary History.
PRELIMINARY FIGURES AVAILABLE ON OPERATING INCOME OF THE FEDERAL RESERVE BANKS.
The Demand for Excess Reserves.
PRELIMINARY FIGURES AVAILABLE ON NET INCOME OF THE FEDERAL RESERVE BANKS FOR 2000.
The Federal Reserve Banks as Fiscal Agents and Depositories of the United States.
PROPOSED ACTIONS.
Announcements.
Revision and interpretation of Regulation K. (Announcements).
Reserving clash: the life industry is divided over reserve requirements for certain UL policies with secondary guarantees.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters