The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1979 V o l. 61, N o . CONTENTS mrzm'pmw® Money Stock Control Under Alternative Definitions of M o n e y .......... 3 Federal Agency Debt: Another Side of Federal Borrowing ................. 10 11 The R e v i e w is published monthly by the Research Department of the Federal Reserve Rank of St. Louis. Single-copy subscriptions are available to the public free of charge. Mail requests for subscriptions, back issues, or address changes to: Research Department, Federal Reserve Rank of St. Louis, P.O. Rox 442, St. Louis, Missouri 63166. Articles herein may be reprinted provided the source is credited. Please provide the Rank’s Re search Department with a copy of reprinted material. Money Stock Control Under Alternative Definitions of Money JOHN A. TATOM I n recent years, existing definitions of the monetary aggregates have come under increasing attack. Ini tially, this assault stemmed from allegations that money demand had shifted in 1974 and that the con duct of monetary policy required new measures which were more closely related to the concerns of policy, such as total spending and prices.1 It has also become apparent that changes in technology, regulations, and financial market institutions have had a significant effect on the payments process and, perhaps, the link between existing aggregate measures and economic activity. An important example of such a change was the introduction of automatic transfer services (A TS) and the extension of NOW accounts to the state of New York on November 1, 1978.2 Other innovations include the increasing volume of repurchase agree ments, money market mutual funds, and regulatory changes that now allow corporations and state and local governments to hold savings accounts. In response to these concerns, the staff of the Board of Governors of the Federal Reserve System proposed new definitions for M l, M l-f-, M2, and M3 monetary aggregates in the January 1979 Federal Reserve Bulle tin.3 The principal criteria underlying the redefinitions were to improve the ability of the Fed to control out put and inflation and to combine deposits that are close substitutes for each other. The staffs proposal has generated considerable com ment both within and outside the Federal Reserve 1See S. M. Goldfield, “The Case of the Missing Money,” Brook ings Papers on Economic Activity (3 /1 9 7 6 ), pp. 683-789, and Jared Engler, Lewis Johnson, and John Paulus, “ Some Prob lems of Money Demand,” Brookings Papers on Economic Activity (1 /1 9 7 6 ), pp. 261-280. 2The problems of controlling a broader measure of M l, includ ing new ATS and N O W account balances, with existing meas ures and policy procedures are discussed in John A. Tatom and Richard W . Lang “ Automatic Transfers and the Money Supply Process,” this Review (February 1979), pp. 2-10. 3See “ A Proposal for Redefining the Monetary Aggregates,” Federal Reserve Bulletin (January 1979), pp. 13-42. System. Most critics are in favor of the effort to rede fine the aggregates but find fault with the specific measures proposed. The criticisms center on (1 ) the exclusion of certain means of payment from the pro posed M l measure, (2 ) the questionable improvement of the relationship of the proposed measures to spend ing, or of the stability of the demand for money, and (3 ) disagreements over whether the proposed meas ures adequately meet the staff’s criterion of combining deposits that are close substitutes. The actual redefinitions that will be forthcoming, if any, are still unknown. Nonetheless, it is useful to examine the issue of money stock control with the measures initially proposed in January 1979. The pri mary criticism of these measures, from the viewpoint of control, is that timely data from thrifts and other financial institutions have been unavailable. T o the extent that the Federal Open Market Committee uses the proposed measures instead of existing meas ures for targeting aggregate growth, the relationship between the instruments of monetary policy and the different aggregate measures is of considerable impor tance, regardless of timely data availability. More over, any definitions of monetary aggregates that ultimately will be chosen are unlikely to deviate sig nificantly from those examined here. The issue of controllability is especially important in view of the October 6, 1979, announcement of a Federal Reserve System policy change to improve control over the growth of monetary aggregates by placing greater emphasis on the supply of bank re serves in day-to-day operations. This action represents both a fundamental change in the focus of monetary policy and a clearer recognition of the link between Federal Reserve actions that affect bank reserves and the monetary aggregates which it seeks to control. This article examines the proposed definitions of mone tary aggregates as an example of the type of control consideration required by this shift in policy. The rePage 3 FEDERAL RESERVE BANK OF ST. LOUIS suits indicate that the proposed aggregate measures are less controllable than existing aggregates, although only slightly less so for proposed M l.4 EXISTING AND PROPOSED MONETARY AGGREGATES The major changes in the proposed redefinitions of monetary aggregates are designed to account for demand-deposit type accounts that are not classified as demand deposits, and to aggregate assets by type, irrespective of the institution involved in the creation of such assets. With regard to the first change, pro posed M l would include NOW accounts, demand de posits at thrift institutions, credit union share drafts, and savings accounts at commercial banks that are subject to automatic transfers to demand accounts. This change is especially important after November 1978, when NOW accounts were extended to New York State and ATS was introduced nationwide. Be fore then, NOW accounts at commercial banks and thrifts, as well as credit union share draft balances, were relatively small. Also, as recommended by the Bach Committee, deposits held by foreign institutions at domestic banks are excluded from the proposed M l measure.5 The attempt to aggregate similar monetary assets regardless of issuing institution is especially important in the proposed measures of M2 and M3. Currently, M2 is equal to M l plus other deposits at commercial banks — including NOW accounts, ATS savings ac counts, other savings accounts, time deposits, and CDs 4Other studies have shown that money demand estimates and the link between intermediate monetary aggregate measures (M l, M2, M 3) and GNP are not improved by the proposed measures. See, for example, the staff study, “ A Proposal for Redefining the Monetary Aggregates,” and Laurence H. Meyer and Murray L. Weidenbaum, “Fed’s Proposed Redefinition of Monetary Aggregates Seen Falling Short of Goal,” The M oney Manager (M ay 7, 1979). Together with the results here, it can be concluded that the link between the mone tary base and GNP will be worse under the proposal, as long as intermediate targeting is used. This is significant for the conduct of monetary policy. Andersen and Kamosky have shown that the mean ancf variance of forecast errors of GNP using the monetary base are not significantly worse than occurs using existing M l or M2 measures. Adoption of the proposed measures would therefore increase the desirability of targeting on the monetary base instead of intermediate monetary aggregates. See Leonall C. Andersen and Denis S. Kamosky, “ Some Considerations in the Use of Monetary Ag gregates for the Implementation of Monetary Policy,” this Review (September 1977), pp. 2-7. 5See Advisory Committee on Monetary Statistics, “ Improving the Monetary Aggregates,” Federal Reserve Board of Gover nors, June 1976. M 1 + is not revised in the proposal except for the exclusion of foreign balances. Currently, M 1 + is the same as proposed M l plus other savings accounts at commer cial banks. This measure is not discussed here. Page 4 NOVEMBER 1979 (except large CDs at weekly reporting banks). The proposed M2 (PM 2) would add to proposed M l (P M l) savings accounts at both commercial banks and thrifts. Consequently, it would differ from the cur rent measure of M2 primarily in its exclusion of time deposits at commercial banks and its inclusion of de mand, NOW, and other savings balances at thrifts, and credit union share drafts. M3, by existing definitions, differs from M2 in its inclusion of time and savings deposits at thrifts and credit unions. Since all but time deposits at these in stitutions are in PM2, proposed M3 (PM3) is intended to reflect the distinction between savings and time ac counts. Thus, PM3 is PM2 plus all time deposits and CDs at commercial banks and thrifts. PM3 differs from M3 in its inclusion of large CDs at weekly re porting banks and demand deposits at thrifts ($864 million in June 1978). Except for the latter difference, PM3 is essentially the same as the existing aggregate M5. Existing measures M4 and M5 will be dropped according to the proposal. Table 1 summarizes these differences.® The proposal for aggregating over similar types of deposits rather than similar institutions is not without shortcomings. The rationale for the change is based upon an increase in substitutability of deposits among institutions.7 It is unclear, however, whether the sub stitutability of these deposits has increased. Barnett has shown, for example, that there is no significant substitutability between small time deposits at com mercial banks and savings and loans, before or after 1974. Also, his evidence shows increases in substituta bility between deposits within institutions, making the M2 aggregate a more justifiable measure than before on this criterion.8 The proposed redefinitions also ignore the question of whether other assets should be included in the mon etary aggregates, or where they might be likely candi dates for inclusion. For example, Wenninger and Sive6This table is adapted from one originally used by John W en ninger and Charles M. Sivesind, “ Defining Money for a Changing Financial System,” Federal Reserve Bank of New York Quarterly Review (Spring 1979), pp. 1-8. "See William A. Barnett, “ Substitutability, Aggregation and Superlative Quantity Indices,” Memorandum, Federal Reserve Board of Governors, April 17, 1979, processed, and “ A Fully Nested System of Monetary Quantity and Dual User Cost Price Aggregates,” (Board of Governors of the Federal Reserve System, Division of Research and Statistics, Econometric and Computer Applications Section, November 1978; processed). 8These points have been made by Kenneth C. Froewiss, John P. Judd, Michael W . Keran and John L. Scadding, “ Comments on Redefining, the Monetary Aggregates,” (Federal Reserve Bank of San Francisco, July 5, 1979, processed). FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1979 T a b le 1 C o m p a riso n o f Current an d Proposed Definitions of the M o n e ta ry A g g r e g a t e s Ml C o m p o n e n ts C urren cy in circulation M2 M3 C urrent P rop ose d C urren t P rop ose d C urrent P rop ose d X X X X X X X X X X X X X X X X X X X X X X X X X X A t Com m ercial B a n ks: D e m an d d e p o sits1 NOW accounts S a v in g s subject to autom atic tran sfer O th e r s a v in g s accou nts2 S m a ll time d ep o sits X L arge time d e p o sits3 X X X X X X C D s4 A t Thrift Institutions: D e m a n d dep o sits NOW accounts X X X X O th e r s a v in g s accou nts5 O th e r time d ep o sits C redit union sh a re drafts X X X X X X X X X X X X irThe definition o f demand deposits differs between the current and proposed aggregates fo r technical considerations such as the exclusion o f de posits held by foreign institutions at dom estic banks in the proposed definitions. P recise definitions and historical data m ay be found in the Federal R eserve Bulletin. 2E xcluding negotiable order o f withdrawal (N O W ) accounts and savings subject to autom atic transfer. 3$100,000 or m ore. 4N egotiable certificates o f deposit in denom inations o f $100,000 o r m ore issued by large weekly rep ortin g banks. 5E xcluding N O W accounts. sind question the omission of repurchase agreements, (R Ps), money market mutual funds (M M M Fs), and new savings balances of state and local governments and corporations, and suggest a broader definition of the money stock, M l.9 Meyer and Weidenbaum also argue that RPs and MMMFs should probably be in cluded in a new measure of M l and/or M2.10 The major tests of the usefulness of the proposed aggregates have consisted of studies of the compara tive predictive performance of the proposed vs. exist ing aggregates in money demand equations and in the estimation of GNP in reduced form equations (both within sample and out-of-sample). Generally, the evi dence does not indicate that the proposed measures are an improvement over existing measures.11 CONTROLLABILITY Andersen and Kamosky have provided a useful ana lytical framework for the choice of a monetary aggre gate target.12 They argue that Federal Reserve actions 9See Wenninger and Sivesind, “ Defining Money.” 10See Meyer and Weidenbaum, “ Fed’s Proposed Redefinition.” n See, for example, Meyer and Weidenbaum, “ Fed’s Proposed Redefinition” or the staff study, “A Proposal for Redefining the Monetary Aggregates.” 12See Andersen and Kamosky, “ Some Considerations in the Use of Monetary Aggregates.” determine the adjusted monetary base which, in turn, affects monetary aggregates. Intermediate monetary aggregates influence spending decisions and, conse quently, are an indicator of nominal GNP. In order to determine the selection of an intermediate aggre gate, the forecasting accuracy of reduced-form equa tions for GNP can be compared. However, to deter mine the Fed’s ability to influence GNP, they argue, analysts must consider not only the variance of GNP estimates, given an intermediate money aggre gate target, but also the relative size of errors in achieving the intermediate monetary aggregate target.13 This framework is useful in considering the desir ability of conducting policy by controlling the pro posed measures. There is no evidence that the pro posed measures represent an improvement over 13They conclude that, even if there is zero control error in the achievement of intermediate monetary aggregate targets, control of the base itself results in no worse an ability to in fluence nominal GNP. This conclusion is reinforced by the results below, if the proposed measures are adopted. It is conceivable that the relationship between total spending and existing aggregate measures could have worsened in recent years because of the introduction of new means of payments or other financial assets. In this case, the proposed measures might represent an improvement for policy purposes, despite the lack of evidence using past data. Andersen and Karnosky performed tests for structural change in the relation ship between spending and M l, M2, and the monetary base for the period after 11/1971 in their equations which are estimated over the period I/1952-IV /1975. The tests rejected such a change in the relationships. Page 5 FEDERAL RESERVE BANK OF ST. LOUIS existing measures as intermediate targets in controlling spending. Thus, the question arises whether the pro posed measures might improve policymaking by reducing the control errors linking Federal Reserve ac tions and intermediate aggregate measures. If the pro posed measures are controllable with less error, they would represent an improvement over existing meas ures for policy purposes, and the evidence would strengthen the case for intermediate targeting. On the other hand, if the proposed aggregates do not exhibit more precise control, the case for using the pro posed measures as intermediate targets is seriously weakened.14 The evidence below shows that the pro posed measures are less controllable. A simple model linking equilibrium money stocks to the adjusted monetary base can be used to assess the controllability of monetary aggregates. In equilibrium, a monetary aggregate M* may be thought of as the product of the adjusted monetary base (M B ) and a money multiplier (k ). Converting this relationship to logarithms (In) results in the expression: In M " = In k -|- In MB. Thus, changes in a monetary aggregate are related to changes in the adjusted monetary base and/or the money multiplier for this aggregate meas ure. Federal Reserve actions determine the adjusted monetary base, but the money multiplier is influenced by the decisions of households, businesses, and finan cial institutions. Consequently, the ability to control a monetary aggregate requires that the money multiplier be predictable. Variations in the money multiplier cause control errors in achieving a given amount in a monetary aggregate through actions affecting the adjusted monetary base. A model to assess the variance of monetary aggre gates (or the money multiplier), given the monetary base, may be written as (1) In M* = |£ + p ! In MB, + £ t where t is included to allow for a time trend in the money multiplier. Since changes in the adjusted mone tary base may not result in instantaneous adjustment of the equilibrium money stock, an adjustment process can be specified as (2) In M , - In M t_, = X ( In M, - In M t-i) which states that actual changes in the monetary ag gregate are some proportion X of the discrepancy be tween equilibrium and past levels of the monetary 14This also implies that the adoption of the proposed measures would reinforce the case for targeting on the monetary base instead of M l or a higher order M. 6 Digitized forPage FRASER NOVEMBER 1979 aggregate. Combining equations 1 and 2, the model may be estimated in the form15 (3) A In M, = Po + P, A In MB, + p2 A In M ,-, where the parameters of the model, excluding |30, may be obtained from (4) X = 1 - Ps This model is estimated for the period 1/196011/1978 for existing aggregates M l, M2, and M3 and the proposed aggregates PM1, PM2, and PM3.18 D if ferences in the logarithms of monetary aggregates and the adjusted monetary base are multiplied by 400 in order to express annual growth rates. A summary of the estimated equations is given in table 2, where |30 is omitted when it is insignificant. The equations for the growth rates of M l and PM1 do not exhibit sig nificant autocorrelation (at the 1 percent level). The equations for the higher aggregates, both existing and proposed, are estimated using the Hildreth-Liu tech nique to control for the significant first-order autocor relation of growth rates. The Durbin-Watson h-statistic indicates the absence of remaining autocorrelation. The significant autocorrelation coefficient for higher 15Estimating the model using equation 3 instead of its counterpart in level-form is motivated primarily by policy makers’ interest in controlling aggregate growth. In addition, significant autocorrelation exhibited in the level equations cannot be removed for the aggregates M2, PM2, M3, and PM3 using a first-order autocorrelation adjustment. Thus, in the level-form, the results and experiments below would be biased. Whether the money multiplier for each aggregate is better explained by a model other than that implicit in equations 1-3 is beyond the scope of this paper. The model is not intended to represent the best means for fore casting money. For a recent work on modeling the money multiplier which is more useful for operational purposes, see James M. Johannes and Robert H. Rasche, “ Predicting the Money Multiplier,” Journal of Monetary Economics (July 1979), pp. 301-325. 1(iSince some analysts expect the federal funds rate to be important in money stock control, the logarithm of this vari able was added to equation 1, resulting in the addition of the first-differences of the logarithm of the rate to the esti mated equation 3. The resulting estimates for the six equations failed to reveal a significant impact of the federal funds rate on the equilibrium stock of each aggregate. Since some of the newer deposits such as N O W accounts did not begin until later in the sample period, a check of the rela tionship of the series was conducted by computing cor relation coefficients over the first and last half of the sample. Existing and proposed measures have correlation coefficients of .9 9 + in both subperiods. Comparing cor relations of A Ins fails to reveal any substantial deterioration in the relations, as well. For the period I/1960-IV /1968, and I/1969-II/1978, the correlation coefficients of A in M l and A in PM1 are 0.98 and 0.97, respectively. For M2 and PM2, comparable coefficients are 0.38 and 0.79; for M3 and PM3, the correlation coefficients are 0.96 and 0.89, respectively. FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1979 Table 2 M onetary A g g re g a te s and the M o n etary Base’ A ggre g a te (A I n ) & — Ml — PM1 M2 PM 3 X p d/hb _P L 0 .5 5 7 0 .3 1 3 0 .6 8 7 0 .8 1 1 (7 . 1 1 ) (3 .2 9 ) (7 .2 3 ) (1 9 .0 9 ) se / r! — 1.78 1.61 — 1 .6 3 0 .6 2 0 .5 8 2 0 .2 8 0 0 .7 2 0 0 .8 0 8 — 1 .7 7 1.68 (7 . 2 6 ) (2 . 8 6 ) (7 .3 5 ) (1 9 .1 2 ) — 1 .8 5 0 .5 9 2 .6 1 4 0 .4 6 9 0 .2 8 1 0 .7 1 9 0 .6 5 3 0 .2 8 7 1 .9 2 1 .8 6 (2 .9 9 ) (3 . 5 1 ) (2 .7 1 ) (6 .9 3 ) (3 .9 9 ) (2 .5 6 ) 0 .7 7 0 .5 6 — PM2 M3 k 0 .5 0 7 0 .4 0 9 0 .5 9 1 0 .8 5 8 0 .5 5 1 1.98 2 .1 9 (4 . 7 5 ) (4 .1 8 ) (6 .0 3 ) (6 .4 2 ) (5 .6 4 ) 0 .1 3 0 .6 4 3 .3 5 3 0 .3 1 6 0.3 8 1 0 .6 1 9 0 .5 0 9 0 .4 8 7 1.85 1 .6 9 (3 .2 6 ) (2 .5 3 ) (3 .6 6 ) (5 .9 5 ) (2 .9 5 ) (4 .7 6 ) 1 .4 3 0 .6 7 2 .7 8 3 0 .4 3 4 0 .3 8 6 0 .6 1 4 0 .7 0 6 0 .5 2 7 1 .8 4 1 .8 4 (2 .4 5 ) (3 .2 5 ) (3 .8 7 ) (6 .1 6 ) (3 .2 5 ) (5 .3 0 ) 1 .2 7 0 .7 0 *The numbers in parentheses are t-statistics. bThe D urbin-W atson d and h statistic, respectively. order M’s may be due to omitted variables or simply the structure of the error process. In either case, control of these aggregates via adjusted monetary base targeting is more difficult because it requires finding and forecasting the omitted variables or ac counting for the autocorrelated errors. In all six equa tions, the model fits the data quite well judging by the R2 and the significance of (3! and |32.17 The growth rate of the existing aggregate measure is more controllable — as shown by its smaller standard error — than the proposed measure for each of the three M’s. Both M l and PM1 are more con trollable than the higher order M’s. For M l, PM1, and M2, the equilibrium adjustment process for a change in MB is 90 percent complete within two quarters. While the M3 equations have smaller stand ard errors than those for M2, the lagged adjustment process is longer (smaller X) for M3 than for M2, but slightly shorter for PM3 than for PM2.18 Of course, the more important test of controllability is whether the equations in table 2 forecast well in dynamic simulations. Table 3 presents the results for within-sample dynamic simulations of the six equa 17Formulas for computing the variance of the restricted param eters ( P“ ) may be found in Jan Kementa, Elements of Econ ometrics, (N ew York: The Macmillan Company, 1971), p. 444. 18Similar equations were estimated using the net source base instead of the monetary base as the control variable. The results from the comparisons above were the same, but the standard error of the equations was higher in each case. The most striking result of those estimates is that Pi and Pi are not significantly different from zero for M2, PM2, M3, or PM3. Thus, control of the source base alone has no impact on aggregates other than M l or PM1. tions reported in table 2. The simulations of the quar terly growth rates (annualized percentage-point dif ferences in logarithms) perform remarkably well for all of the measures except PM2, according to the mean error over the whole sample period. PM2 growth is apparently underestimated on average. The root-meansquared error (R M SE ) compares favorably to the standard errors reported in table 2 for M l, PM1, and M2. The RMSE is substantially higher than the stand ard error for PM2 and PM3, as well as for the existing M3 measure. Control of these aggregates via the mone tary base is poor according to the dynamic simulations. The controllability of the growth rate of the existing measures is greater than that of the proposed meas ures, as judged by the error statistics reported in table 3. According to the RMSE and mean absolute errors, the controllability of existing aggregates deTable 3 Dynam ic Sim ulations of A g g r e g a t e Growth Rates1 1/1 960-11/1 978 A ggre g a te Ml R o o t-M e a n S q u a re d Error 1 .7 0 % M e a n A b so lu te Error Mean Error 1 .3 1 % -0 .0 3 % PM 1 1.75 1 .3 7 -0 .0 4 M2 2 .1 3 1.63 0.00 PM2 3 .5 3 3 .0 6 0 .5 9 M3 2 .4 7 1 .9 0 0 .0 2 PM3 2.81 2 .0 8 0.00 1F irst differences o f logarithm s at annual percentage rates. Page 7 FEDERAL RESERVE BANK OF ST. LOUIS teriorates moving from M l to M2 to M3. For the pro posed measures, however, the growth of PM3 is more controllable than PM2. The results indicate that the adoption of these redefinitions would worsen mone tary control, and that, except for PM1, intermediate targeting of quarterly growth rates would be subject to substantially larger control errors. The dynamic simulations can also be used to com pare the accuracy of control of the levels of the vari ous aggregates on a quarterly basis. The results of these simulations are given in table 4. The equations track the quarterly level of the aggregates quite well over the full period, with an average error of less than $1 billion. The simulations of M l, PM1, and M2 track the level the best, with mean errors of $30 million or less. The RMSE and mean absolute error of the quar terly level simulations are $1 billion or less for both M l and PM1, with M l control proving slightly supe rior again. The RMSE and mean absolute error of the dynamic forecasts for M2 and M3 levels are several times larger than those for M l under either the ex isting or proposed definitions, but, as above, existing measures are generally superior to the proposed meas ures of M2 and M3. While the results for one-quarter growth rates and quarterly levels from the simulations are compelling, policymakers also concern themselves with growth of aggregates over a longer period. Currently, intermedi ate targets for M l and M2 are announced for fourquarter periods. Over such a span, the quarterly errors in growth rates tend to average to a smaller level. To investigate the extent of control over a four-quar ter period, the dynamic simulations of the table 2 equations can be used to provide four-quarter growth rate estimates for the period from I/1961-II/1978. The results of comparing the predicted annual growth rates to the actual annual growth rates for each aggregate are summarized in table 5. For annual periods, M l and PM1 are substantially more controllable than the corresponding higher order M’s. Control of existing aggregates deteriorates moving from M l to M2 to M3; PM3, however, is more con trollable than PM2 as indicated in table 3. The most startling result in table 5 is that control of M3, PM2, and PM3 fails to improve sufficiently when the control horizon moves from a one-quarter to a four-quarter period so that the RMSE is larger than the standard error of the respective equation in table 2. The vari ances of errors in annual growth control for M l, M2, and PM1, however, are reduced by more than 30 per cent of the RMSE for one-quarter forecasts. Digitized for Page FRASER 8 NOVEMBER 1979 T able 4 D yn a m ic Sim ulation Errors o f The Level o f M o n etary A g g r e g a t e s 1 1/1960-11/1978 A ggre g a te R o o t-M e a n S q u a re d Error M e a n A b so lu te Error Mean Error Ml $ .9 7 $ .73 $ -0 .0 3 PM1 1.01 .7 6 -0 .0 3 M2 2 .4 7 1 .7 9 -0 .0 1 PM2 4 .5 9 3 .7 6 0 .5 5 M3 4 .8 0 3 .4 9 0 .2 6 PM3 4 .8 2 3 .5 7 -0 .1 5 *Billions o f Dollars. T a b le 5 C ontrol of A g g r e g a t e G ro w th Rates For F our-Q uarter Periods 1/1961 - l l / l 9 78 A g g re g a te Ml R o o t-M e a n S q u a r e d Error M e a n A b so lu te Error Mean Error 1 .1 2 % 0 .9 1 % -0 .0 6 % PM 1 1.11 0 .91 -0 .0 8 M2 1 .4 6 1 .4 6 0 .0 2 PM2 3 .0 3 2 .7 3 0 .6 0 M3 1 .9 9 1 .5 3 0 .0 2 PM3 2.31 1 .6 7 -0 .2 0 T a b le 6 Controllability o f a Broader Asset M easu re 11/1969-11/1978 GROW TH A ggre g a te A R o o t-M e a n S q u a re d Error RATES M e a n A b so lu te Error Mean Error 1 .8 3 % 1 .3 8 % PM 1 .0 0 3 % 1 .7 2 1 .3 9 - .0 7 Ml 1 .5 8 1 .2 6 -.0 6 LEVELS (B illio n s o f D o lla rs ) $ 1 .2 9 $ .98 $ .02 PM1 1 .2 0 .95 -.0 3 Ml 1 .0 9 .8 7 -.0 4 A FEDERAL RESERVE BANK OF ST. LOUIS Finally, since questions have been raised about the omission of other assets that are close sub stitutes for demand deposits from M l, it is useful to examine the controllability of such a broader aggre gate. The Wenninger and Sivesind measure (referred to as A ) consists of the sum of: current M l, corporate and state and local government savings deposits, NOW deposits, ATS savings deposits, credit union share drafts and demand deposits at thrifts, assets of money market mutual funds, repurchase agreements (RPs) at nonbank government securities dealers with nonfinancial corporations, and RPs at 46 large commercial banks. This measure was constructed for the period IV/1968-I/1979. When the model above (equations 1-4) is estimated using this enlarged definition of “monev” for the pe riod II/1969-II/1978, the results are p0 = 0, p, = .662 (5.08 ), pT = 858 (16 .6 9), X = .771 (5 .1 6 ), R2 = .32 and a standard error of 1.84, where t-statistics are given in parentheses. For a comparison of controlla bility, the model was estimated and dynamically simulated over the same period for M l and PM1.19 The resulting comparisons of simulations over the pe riod II/1969-II/1978 are presented in table 6 for 19When the M l equation is estimated over the same period, the estimate of X rises to 0.797 (t = 5 .9 5 ). The point estimate of P? (0.785) is essentially the same and its t-statistic is 18.27. The standard error of the equation is 1.54. A Chow test for structural change in the subperiods I/1960-I/1969 and II/1969-II/1978 rejects the structural change hypothesis. Similar results are obtained for PM1, where X. rises to 0.804 (t = 4 .4 9 ) and (3? is 0.784 (t= 1 7 .3 5 ). The standard error for the PM1 equation is 1.69. NOVEMBER 1979 the quarterly growth rate equations as well as the aggregate level simulations. The mean errors of the simulations of growth rates are quite small. The liquid asset measure, however, is substantially less controllable than M l for both growth rates and levels. The comparisons generally indicate that PM1 is also more controllable than A, although not by as large a difference. The money measure, A, is an inferior measure by which to con duct monetary policy.20 SUMMARY AND CONCLUSION The Board of Governors is currently considering revising the definitions of the monetary aggregates. One important criterion that should influence the process of redefining these aggregates is the con trollability of these measures through Federal Reserve actions. This criterion is especially crucial if the ag gregates are to be used as intermediate targets of monetary policy. Given the framework developed in this article for assessing Federal Reserve control of the monetary aggregates, the evidence indicates that the measures proposed by the Board’s staff in January 1979 are sub ject to greater control errors than current aggregate measures, except for proposed M l. -"Since the Wenninger and Sivesind study, the Board of Gov ernors has released more comprehensive measures of repur chase agreements at commercial banks. See Norman N. Bowsher, “ Repurchase Agreements,” this Review ( September 1979), pp. 17-22, for a description of this data. When the Wenninger and Sivesind A measure is adjusted by taking out RPs at 46 large commercial banks and adding in RPs at all commercial banks, the controllability of the resulting aggregate deteriorates further. A detailed comparison is not reported here because data for the comparison is only avail able for the period since the fourth quarter of 1974. Page 9 Federal Agency Debt: Another Side of Federal Borrowing DAVID H. RESLER and RICHARD W. LANG I DISAPPOINTMENT in recent domestic economic performance has sparked increasing interest in the role of the federal government in the U.S. economy. As it is widely believed that government deficits con tribute to inflation, much of the public concern focuses on the size of the federal budget deficit and the growth of federal indebtedness. necessary background and perspective, the article first examines the general nature and function of the agencies. Unlike most discussions of these agencies, which focus primarily on their microeconomic effects, this article considers the macroeconomic implications of agency debt operations. Since the government finances most of its spending in excess of tax receipts by issuing new debt, con tinued budget deficits enlarge the amount of federal debt outstanding. For instance, the large budget defi cits of the last four years have contributed to about a $250 billion increase in Treasury debt outstanding from December 1975 to September 1979. FEDERAL AGENCIES AND SPONSORED AGENCIES While the federal deficit is coming under closer public scrutiny, a substantial portion of federally re lated programs and their associated debt has escaped much of this attention. Specifically, the debt of fed erally owned and federally sponsored agencies is often overlooked.1 Like the debt of the Treasury (or, for that matter, any other debt), agency debt has im portant effects on capital markets. This article focuses on this additional source of federal influence on capital markets. To provide the 1Federally sponsored agencies are, technically, independent private enterprises that have been created by congressional legislation. Despite this independent status, these agencies remain subject to broad policy guidance from the federal government. 10 Digitized forPage FRASER The federal government conducts its business through various departments and agencies, which for the most part receive their authorization to spend (appropriations) through the budget process. De cisions about the level and allocation of federal spend ing are reflected in the budget of the U.S. govern ment.2 This authorized budget spending is primarily financed by the Treasury’s tax receipts and by sales of Treasury debt, although some on-budget agencies are authorized to issue their own debt ( panel A of ex hibit 1). Spending of some federally owned agencies, however, is placed outside the budgetary process and is designated as “off-budget.” Panel B of exhibit 1 lists these federally owned, off-budget agencies. 2Although the unified budget concept represents the official budget of the federal government, the national income ac counts ( NI A) concept is more frequently used in evaluating economic activity. For a description of these two budget con cepts, see David J. Ott and Attiat F. Ott, Federal Budget Policy, 3rd ed. (Washington, D .C.: The Brookings Institution, 1977), pp. 4-23. FEDERAL RESERVE BANK OF ST. LOUIS In addition, the federal government sponsors a num ber of other “private” agencies, whose spending is also excluded from the federal budget. These spon sored agencies are listed in panel C of exhibit 1. NOVEMBER Exh ib it 1 A. O n - B u d g e t Entities A u th o riz e d to Federal D ebt Lim itations to Issu e D ebt G o ve rn m e n t N a tio n a l M o r t g a g e A sso c ia tio n E xp o rt-Im p o rt B a n k o f the U nite d States Functions of Government Agencies While some agencies, such as the Postal Service or the Tennessee Valley Authority, mainly provide cer tain services, the primary function of many federal agencies is to allocate credit to particular sectors of the economy. For instance, the Federal Housing Ad ministration (F H A ) was established in the 1930s to mitigate the increase in mortgage foreclosures that accompanied the Depression. Similarly, a variety of other federal agencies has been established to allocate credit to particular sectors such as housing (e.g., Fed eral National Mortgage Association), agriculture (e.g., Federal Land Banks), small businesses (e.g., Small Business Administration), and international trade ( e.g., Export-Import Bank). The justification for directing agency aid to particu- • lar sectors generally relies on allegations that capital market imperfections prevent resources from flowing naturally to certain socially desirable activities. These alleged imperfections include monopolistic elements in lending markets, economies of scale enjoyed by some borrowers but not others, and external benefits to society in excess of those capturable by the bor rower (and ultimately the lender). The belief that such imperfections discriminate “unfairly” against par ticular sectors or classes of borrowers within the econ omy has prompted Congress to authorize various sector-specific credit programs.3 Federally Owned Agencies Though the off-budget agencies listed in panel B of exhibit 1 are wholly owned by the U.S. government, their spending is not reflected in the unified budget totals, despite the recommendation of the President’s Commission on Budget Concepts that the unified budget “ . . . include all programs of the federal :iSometimes, however, imperfections are confused with economic differentiation. It is not an imperfection of the capital market when an unrestricted market assigns a higher risk premium to certain activities, since some projects entail more risk than others. If riskier projects are to succeed in attracting financial capital, they must offer investors a higher potential return. The primary function of a competitive capital market is to allocate credit toward the most promising among alter native projects for a given level of risk. In a free market, lenders bear the risk inherent in a given project. Government loan guarantees, loan participations, and other subsidies that are channelled through government agencies do not eliminate the inherent risk of making loans to certain groups; they simply spread the risk over a larger segment of society, the taxpaying public. 1979 Not Subject (G N M A ) (E x -lm B a n k ) T en n e sse e V a lle y A u th o rity (T V A ) F a m ily H o u s in g P rogra m o f the D e fe n se Departm ent B. F e d e ra lly O w n e d , O ff-b u d g e t A g e n c ie s Rural Electrification a n d T e le p h o n e R e v o lv in g Fund Rural T e le p h o n e B a n k P en sion Benefit G u a ra n t y C o rp o ra tio n Postal Service Fed eral F in a n c in g B a n k (FFB) R e g io n a l Rail R e o rg a n iz a tio n P rogra m of the U.S. R a ilw a y A sso c ia tio n C. F e d e ra lly S p o n so re d A g e n c ie s Federal N a tio n a l M o r t g a g e A sso c ia tio n ( F N M A ) 1 Stu d en t Loan M a r k e tin g A sso c ia tio n ( S L M A ) * B a n k s fo r C o o p e ra tiv e s3 Federal Interm ediate Credit B a n k s ( F IC B ) 3 Federal Land B a n k s ( F L B ) 4 F ed eral H om e Loan B a n k s ( F H LB ) 5 Federal H om e Loan M o r t g a g e C o rp o ra tio n ( F H L M C ) 6 'C onverted to p rivate ow nership in Septem ber 1968. 2Created as private enterprise in June 1972. 3Converted to private ownership in December 1968. •Converted to p rivate ownership in 1947. 5Converted to private ownership in 1951. 6Created as p rivate corporation in July 1970. government and its [wholly owned] agencies.”4 Some of these agencies’ activities are, nevertheless, subject to congressional and presidential review. Both on-budget and off-budget federal agencies allocate credit primarily through the administration of loan programs directed toward particular sectors of the economy. The agencies may grant loans either directly, by lending to specific borrowers, or indirectly, by purchasing loans initiated by private lenders but guaranteed or insured by the federal government. Loans are financed either by the Treasury or by agency borrowing. The Treasury-financed portion of the agencies’ activities is, like all Treasury debt, sub ject to statutory debt limitations.3 4Repurt of the President’s Commission on Budget Concepts (Government Printing Office, 1967), p. 7. 5The statutory debt limitation refers to the legal ceiling of the debt of the federal government, as set by Congress. Only the House of Representatives may initiate changes in this statutory debt limit, which was $830 billion as of June 30, 1979. While constituting a legal ceiling on outstanding debt, the debt limitation has been changed by Congress whenever gov ernment financing need's have nudged against the ceiling. Page 11 NOVEMBER FEDERAL RESERVE BANK OF ST. LOUIS Debt issued directly by off-budget agencies, how ever, is free from these statutory limitations, although it is reported by the Treasury as part of gross federal debt. In addition, there are some on-budget agencies that can issue debt which is not subject to statutory limitation (see panel A of exhibit 1). Consequently, federal debt subject to statutory limitation, a fre quently used measure of the overall government debt, understates the full extent of direct federal govern ment borrowing. For instance, at the end of June 1979, outstanding Treasury debt was $804.9 billion while federally owned agency debt was $7.3 billion, for a total of $812.2 billion. Of this total, only $806 billion was subject to the prevailing statutory limit of $830 billion. Since 1973, outlays of federally owned, off-budget agencies have increased rapidly (table l ) . 6 This rapid growth has largely been associated with the activity of the Federal Financing Bank (F F B ), an intermedi ary that merits special attention. The Federal Financing Bank In December 1973, Congress established the Fed eral Financing Bank as an independent agency of the U.S. government.7 The FFB acts as an intermediary by coordinating the federally owned agencies’ fund raising activity in U.S. capital markets. Of the fed erally sponsored agencies (panel C of exhibit 1), only the Student Loan Marketing Association (SLM A) is eligible for FFB financing.8 The FFB facilitates the funding of various agencies’ programs in three ways. First, it acquires new or out standing debt from federally owned agencies. This effectively reduces competition among the agencies and the Treasury for the existing supply of loanable funds in capital markets. Testimony given in the congressional hearings on the creation of the FFB clearly reveals this to be the primary function in tended for the FFB. The declining volume of offbudget, federally owned agency debt provides evi dence that the FFB has succeeded in reducing the independent debt operations of these agencies. Second, it acquires loan assets from federal agen cies and third, it acquires loans that have been guar6Outlays, in general, refer to the expenditures and loans of an agency. 7The actual operations of the FFB are carried out in an office within the U.S. Treasury. 8Even though the SLMA is an independent, federally spon sored agency, its loan assets are fully guaranteed by the gov ernment and are, therefore, eligible for FFB financing. 12 Digitized for Page FRASER 1979 T a b le 1 O u tla y s o f O ff-B u d g e t, Federally O w n e d A ge n c ie s ( M illio n s of D o lla rs ) Fiscal Y e a r 1973 O u t la y s $ 60 1974 1 ,4 4 7 1975 8 ,0 5 4 1976 7 ,2 8 5 TQ* 1 ,7 8 5 1977 8 ,6 8 4 1978 1 0 ,3 2 7 1 9 7 9 (Est.) 1 1 ,9 9 0 1 9 8 0 (Est.) 1 1 ,9 5 6 *TQ = T ransition Quarter (J u ly 1976-September 1976). S O U R C E : U .S. Office o f M anagem ent and Budget, The B udget o f the U.S. G overnm ent, 1980. anteed by other federally owned agencies.9 As dis cussed below, both of these FFB transactions can be used by agencies as alternatives to debt issuance in financing agency programs. FFB acquisition of agen cies’ loan assets and guaranteed loans in 1978 had risen to about two-thirds of its financing of agency activity. In all three cases, the FFB finances its activity either by issuing its own debt or by borrowing di rectly from the Treasury. Though the FFB is author ized to issue up to $15 billion of its own securities, it has raised virtually all of its funds through Treas ury borrowing. On the only occasion when it issued its own securities, its market-determined borrowing costs were considerably higher than anticipated. The rationale for creating the FFB was to lower the cost of marketing agency debt by effectively consoli dating the debt of several agencies and by coordinat ing its placement. The differential between the rate paid on the FFB’s borrowings and that earned on its holdings of various agencies’ debt covers its operating expenses. The FFB currently acquires agency debt at a yield 12.5 basis points (Vs percentage point) above the interest rate on Treasury securities of comparable maturity. The FFB itself borrows from the Treasury at this latter rate. The balance sheets of a typical agency, the FFB, and the Treasury help to illustrate the method of 9For simplicity, guaranteed loans are defined here to include certificates of beneficial ownership ( CB O s), which some agencies issue. These CBOs are essentially ownership claims on a pool of loans, which themselves remain in the agency’s possession. The CBOs are then guaranteed by the issuing agency. Under present accounting procedures, CBOs are not treated as agency debt. FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1979 Exhibit 3 E xh ib it 2 Agency FFB Agency A sse ts A sse ts Liabilities A sse ts Liabilities $ 1 0 M C a sh (P roceed s from sa le of note to FFB) $ 1 0 M N o te $ 1 0 M Agency N ote $ 1 0 M Loan from T re asu ry Loan A sse ts D ebt $ 2 0 M illio n A sse ts Liabilities Loan A ssets N o te from A $ 2 0 M illio n B o rro w in g s from T re a su ry $ 2 0 M illio n — N o te from A $ 2 0 M illio n -f“ N o t e from B $ 2 0 M illio n T re asu ry A ss e ts Liabilities $ 1 0 M Loan to FFB $ 1 0 M T re asu ry N o te FFB debt intermediation (exhibit 2 ). Suppose the agency has issued a new $10 million note which the FFB purchases. The FFB, in turn, finances this transaction by borrowing $10 million directly from the Treasury. The Treasury issues $10 million of its securities to finance its loan to the off-budget FFB. This final transaction becomes part of the debt sub ject to the congressionally imposed statutory debt limitation. In the case of on-budget agencies, the discipline imposed by the budget process may be compromised when the FFB acquires loan assets from agencies or loans guaranteed by agencies. This may occur because of the way in which spending authorizations are deter mined. In calculating the outlay totals for budgetary purposes for some on-budget agencies, repayments of past loans to the agency and the sale of existing loan assets are deducted from new loans. The resulting net new loan figure is the basis for determining the agen cy’s budgeted outlays. For example, suppose an agency is budgeted to make outlays of $10 million and currently has outlays totalling $10 million, includ ing $7 million of loans. If it sells this $7 million of loan assets to the FFB, its outlays for budgetary pur poses then would amount to $3 million. Similarly, sales by an agency of guaranteed loans are treated as an offset to the agency’s outlay totals for budgetary purposes. In effect, these loan-asset and guaranteed-loan sales become alternative means of financing the agencies’ programs. To illustrate this type of FFB intermediation, exhibit 3 again presents the balance sheets of an agency, the FFB, and the Treasury. Suppose the agency grants a loan of $20 million to a borrower (A ) and finances this loan with a sale of its own debt. If this debt is purchased by the FFB, the example parallels exactly the case developed in exhibit 2. The agency, however, can sell this loan (the loan asset) ~f- N o te from A $ 2 0 M illio n FFB Liabilities T re a su ry A sse ts Liabilities L o a n s to the FFB T re a su ry D ebt $ 2 0 M illio n $ 2 0 M illio n to the FFB if the loan is guaranteed by this or another agency. Suppose the agency conducts such a sale and uses the proceeds to issue a new loan to another borrower ( B ). The net result of this trans action is that the agency’s balance sheet remains un changed. The FFB’s balance sheet now shows a $20 million loan asset (the note from A ) and a $20 mil lion liability in the form of its borrowings from the Treasury. Finally, the Treasury’s balance sheet shows a $20 million loan to the FFB and $20 million in newly issued debt. The net result of this transaction is that the Treasury has indirectly financed the exten sion of an agency loan. This process could, in prin ciple, continue repeatedly until the agency reached its lending limit. Only the most recently granted loan would appear on the agency’s balance sheet as a loan asset (or note). The FFB, however, would be holding notes on all previously granted loans while the Treas ury’s debt would expand to accommodate these transactions. When the FFB acquires a guaranteed loan from an agency, the transaction effectively transforms a contingent liability of the government (in the form of a loan guarantee) into a direct loan by the Treasury. This transformation occurs because the FFB finances its acquisitions with loans from the Treasury, which, in turn, finances the loan to the FFB by issuing new debt. Thus, the FFB’s acquisition of guaranteed loans or loan assets tends to distort the budget process by lowering the outlay totals for budgetary purposes of those agencies authorized to make such sales. Never theless, sales of loan assets and guaranteed loans to the FFB, as well as the budgeted outlays, all ulti mately affect the Treasury’s indebtedness.10 10The dramatic increase in guaranteed loans and their exten sive purchase by the FFB has led to several, as yet unsuc cessful, legislative initiatives to limit their use. Page 13 FEDERAL RESERVE BANK OF ST. LOUIS A hypothetical example illustrates how these trans actions could distort what many people consider the meaning of a “balanced” federal budget. Though it is widely believed that a balanced budget implies no ad ditional federal borrowing, this is not necessarily true. Suppose that all authorized and budgeted out lays of agencies are fully financed by tax receipts. While the budget would be balanced, the Treasury or the FFB would still issue additional debt if at least one agency, whether on- or off-budget, grants loans that are then sold to the FFB in a loan asset sale. In this case, the net outlays of the agency remain un changed, but the Treasury’s debt rises when it fi nances the FFB’s acquisition of the loan asset. The FFB’s operations have at least two additional effects on capital markets. First, the agencies’ access to funds via the FFB lowers their cost of funds rela tive to low-risk private borrowers. While govern ment insurance or loan guarantees for various pro grams have always given such programs a competitive edge in financial markets, the method of placement used before the FFB’s creation did involve an implied market assessment of their riskiness. When different agencies issued debt, the market implicitly made a rel ative evaluation of the various programs. Furthermore, the cost of funds to the agencies prior to the FFB’s creation was higher. This is evident from the fact that in 1974 the FFB initially was lending to the agencies at a 37.5 basis-point (% percentage point) premium over the new-issue Treasury bill rate. This rate was at or below the prevailing interest rate on agency secu rities at that time. This spread was reduced in two steps (in November 1974 and May 1975) to the pres ent 12.5 basis points. In congressional hearings, these reductions were said to have primarily reflected a nar rowing in the market spreads between yields on agency and Treasury issues as the general level of in terest rates declined during the 1974-75 recession.11 Although such yield spreads tend to be cyclical, the FFB’s lending rate has maintained the same 12.5 basispoint spread over Treasury yields as interest rates have risen during the 1975-79 expansion. Thus, agencies can obtain funds at only a slight premium above the Treasury’s own rates, a fact which may have the additional long-run effect of encouraging agencies to place more debt than otherwise. A second effect the FFB has on capital markets re sults from the transformation of guaranteed loans into direct loans. Before the establishment of the FFB, insured agency loans were sold to the public in prin Loan Guarantees and the Federal Financing Bank, Hearings before the Subcommittee on Economic Stabilization, 95 Cong. 1 Sess. (G PO, 1977), p. 180. Digitized forPage FRASER 14 NOVEMBER 1979 T a b le 2 A n n u a l A v e ra g e s o f M o n th ly Levels o f O u tsta n d in g Debt (B illio n s o f D o lla rs) Year A g g re g a te Fed e ral Debt T re a su ry Debt Agency Debt 1955 $ 2 8 0 .8 $ 2 7 7 .8 1956 2 8 0 .0 2 7 6 .1 4 .0 1957 2 7 9 .6 2 7 4 .3 5 .3 6 .5 $ 3 .0 1958 2 8 3 .7 2 7 7 .2 1959 2 9 5 .5 2 8 7 .5 8 .0 1960 2 9 8 .8 2 8 9 .1 9 .6 1961 3 0 2 .0 2 9 2 .0 1 0 .0 1962 3 1 1 .7 29 9 .5 1 2 .2 1963 3 1 9 .2 3 0 5 .6 1 3 .6 1 5 .6 1964 3 2 8 .3 3 1 2 .7 1965 3 3 6 .1 3 1 8 .5 1 7 .6 1966 3 4 6 .1 3 2 3 .5 2 2 .6 1967 3 6 1 .4 3 3 4 .1 2 7 .2 1968 3 8 6 .2 3 5 2 .2 3 4 .0 1969 4 0 1 .1 3 6 0 .7 4 0 .4 1970 4 2 4 .1 3 7 5 .6 4 8 .6 1971 4 5 3 .6 4 0 3 .4 5 0 .2 1972 4 8 5 .4 4 3 2 .5 5 2 .9 1973 5 2 3 .1 4 5 9 .6 6 3 .5 1974 5 5 6 .1 4 7 7 .7 7 9 .4 1975 6 2 3 .1 5 3 5 .5 9 3 .2 1 0 1 .5 1976 7 1 1 .0 6 2 0 .0 1977 7 7 8 .3 6 8 2 .2 1 0 8 .4 1978 8 6 4 .9 7 5 4 .3 1 2 4 .9 1979* 9 2 4 .9 7 9 7 .6 1 4 4.1 ♦Average based on first six m onths o f 1979. S O U R C E : F ederal R eserve Bulletin. vate capital markets. Since its formation, however, the FFB (and ultimately the Treasury) has financed an increasing portion of these transactions. By effectively underwriting loans to the private sector in this way, the FFB channels loanable funds to relatively highrisk borrowers who otherwise would have acquired the loans only at a higher cost, if at all. Federally Sponsored Agencies In addition to federally owned agencies, Congress has also established several federally sponsored agen cies to allocate credit to selected sectors of the econ omy. Because these sponsored agencies either were converted to private ownership or were initially es tablished as private enterprises, their activity falls completely outside the budgetary process. Like fed erally owned agencies, they channel funds to spe cialized sectors within the economy, either through NOVEMBER FEDERAL RESERVE BANK OF ST. LOUIS AGGREGATING FEDERAL DEBT Table 3 Percent C h a n g e s in A n n u a l A v e ra g e s o f O u tsta n d in g Year A g g re g a te Fed eral D ebt Debt T re a su ry Debt Agency Debt 3 0 .5 % 1956 -0 .3 % -0 .6 % 1957 -0 .2 -0 .6 3 3 .0 1958 1.5 1.0 2 4 .3 1959 4.1 3 .7 2 2 .5 1960 1.1 0 .6 20.1 19 6 1 1.1 1.0 4 .0 1962 3 .2 2.6 2 1 .9 1963 2 .4 2.0 11.1 1964 2.9 2.3 15.1 1965 2 .4 1.8 1 2 .9 1966 3 .0 1.6 2 8 .5 1967 4 .4 3 .3 2 0 .4 1966 6 .9 5 .4 2 4 .8 1969 3 .9 2.4 1 8 .7 1970 5 .7 4.1 2 0 .3 1971 6 .9 7 .4 3 .4 1972 7 .0 7.2 5 .4 1973 7.8 6 .3 2 0 .0 1974 6 .3 3 .9 2 5 .0 1975 1 2 .0 12.1 1 7 .4 1976 14.1 15.8 8.9 19 7 7 9 .5 1 0 .0 6.8 1978 11.1 1 0 .6 15.1 6 .9 5 .7 1 5 .4 1979* 1979 ‘ Based on first six months o f 1979. direct loans to individual borrowers or through the purchase of loans that were initiated in the private sector. These government-sponsored agencies are listed in panel C of exhibit 1. The programs and debt of sponsored agencies are not part of the unified budget, are not subject to the debt limitation of Congress, and, except for the SLMA, do not involve indirect FFB financing by the Treas ury. Many of these programs were converted to pri vate ownership and thereby removed from the bud getary process soon after Congress adopted the new unified budget concept in 1967. Though these agen cies’ budgets are not directly subject to congressional review, their activity is not completely independent of the federal government. For instance, FNMA has a guaranteed emergency access to Treasury loans up to specific limits. Despite the formal distinctions usually made between the Treasury and the sponsored agen cies, the debt of these agencies constitutes a source of federally related credit demand in U.S. capital markets. The sum of Treasury debt, federally owned onand off-budget agency debt, and federally sponsored agency debt is a more appropriate measure of the federal government’s full impact on U.S. credit mar kets. Since the FFB essentially converts agency debt into Treasury debt, that debt which the FFB inter mediates must be deducted from the total. An exam ple will clarify this calculation. When an agency issues debt to finance its programs, that debt is counted in the total of agency debt out standing. When the FFB purchases this debt, it bor rows from the Treasury which, in turn, issues new Treasury debt, thereby adding to the total Treasury debt outstanding. A simple total of these two debt categories essentially would count the agency-initiated debt twice. For example, this double counting would occur if, in exhibit 2, the debt of the agency and that of the Treasury were added together. Since the FFB essentially passed the agency debt through to the Treasury, the debt should be counted only once. Sub tracting the FFB-financed debt eliminates this double counting.12 Table 2 reports the annual outstanding debt of the Treasury, of all agencies, and the sum of these two categories (aggregate federal debt) after netting out FFB debt financing. These data reveal a substantial increase in the outstanding debt of the agencies. (From here on, federally owned agencies and fed erally sponsored agencies will be referred to as “agen cies.” ) From 1955 to 1978, outstanding agency debt grew at an annual rate of 17.5 percent while Treas ury debt grew at the much slower rate of 4.4 percent. In 1955, agency debt constituted only about 1 percent of all debt raised under federal auspices. By 1978, it totalled more than 14 percent of all fed erally related debt. Table 3 reports annual growth rates for the three debt categories. Comparing the growth rates of Treas ury debt and aggregate federal debt reveals that ag gregate federal debt grew faster in every year except 1971, 1972, and 1975-77.13 '-O n ly FFB intermediation of agency debt needs to be netted out of the total, since FFB loan asset acquisitions, in effect, transform agency assets into Treasury indebtedness. In ex hibit 3, adding agency debt and Treasury debt together would entail no double counting. 13A one-tailed t-test was performed on the between the two growth rates, calculated 1956-78. The tests confirmed, at the .995 that the growth of aggregate federal debt faster than the growth of Treasury debt. mean difference over the period confidence level, was significantly Page 15 FEDERAL RESERVE BANK OF ST. LOUIS NOVEMBER 1979 C h a rt 1 C h a n g e in Debt S h a d e d a r e a s r e p r e s e n t p e r io d s o f b u s i n e s s r e c e s s io n s . L a t e s t d a t a p lo t t e d : 1 9 7 8 The reversal of the growth trend from 1975-77 re quires some explanation. The sharp deceleration of agency debt growth from 1975 to 1977 was accompa nied by a sharp acceleration of Treasury debt growth. The annual growth rate of Treasury debt after 1974 is more than four times its growth rate for the entire period 1955-73. Since the post-1974 period spans the life of the FFB, the data suggest that the FFB has, in fact, transferred the debt financing operations from the agencies to the Treasury. However, the sharp ac celeration in the growth rates of both Treasury debt and aggregate federal debt since 1974 can be only partially attributed to the effect of FFB financing, since even budgeted programs have required unusu ally large debt financing in this period. When the impact of government debt is measured only by Treasury debt, a substantial portion of the Digitized for Page FRASER 16 effect of federal programs on credit markets is over looked. The potential consequences of this oversight can be illustrated with an example. The administra tion’s recent energy proposal calls for the establish ment of a “private” government-sponsored corporation to develop synthetic fuels. Estimates of the additional debt this corporation would issue run as high as $80 billion over the next decade. The debt of this federally sponsored corporation alone would increase off-budget agency debt by more than 60 percent over present lev els, yet would neither be included in commonly used measures of the federal debt nor be reflected in the official federal budget. Because the fund-raising ac tivities of all programs under federal auspices affect capital markets, however, it is appropriate to examine the total of federally related debt when analyzing the government’s impact on credit markets. FEDERAL RESERVE BANK OF ST. LOUIS C h a rt 2 Fiscal Measures ( + ) S u r p l u s ; ( -) D e f ic it NOVEMBER 1979 and since fiscal actions re garding federal expenditures and tax receipts are reflected in the budget deficit or sur plus, changes in Treasury debt could be expected to vary countercyclically. An important distinction between the activity of the Treasury and the agencies must be noted, however. Federally owned and spon sored agencies’ activities fre quently are directed toward allocating credit to specific sectors of the economy. In particular, a large portion of total agency debt is used to moderate cyclical fluctua tions in the housing sector. S o u r c e s : U.S. D e p a r t m e n t o f C o m m e r c e a n d F e d e r a l L at e s t d a t a pl ott ed: 3 r d q u a r t e r IMPLICATIONS OF AGGREGATING ALL DEBT RAISED UNDER FEDERAL AUSPICES Federal Borrowing and the Business Cycle When cyclical fluctuations in GNP growth are as sociated with similar fluctuations in private credit demands, the government can avoid amplifying the resulting interest rate cycle with its own debt opera tions. If increases in government debt are kept lower during business expansions than during contractions, pressures on interest rates from federal debt opera tions would be moderated over the business cycle. Since a major objective of federal fiscal policy is to counteract cyclical fluctuations in economic activity, Near the peak of business cycle expansions, interest rates generally have risen to levels that approach or ex ceed regulatory ceiling in terest rates on time and sav ings deposits. Consequently, financial intermediaries such as savings and loan associ ations and mutual savings banks have experienced diffi culty in attracting the funds needed to maintain mort R e s e r v e B a n k of St. L o u i s gage loan activity. Agencies such as the FHA, GNMA, and FNMA have attempted to offset this effect by in creasing their lending (or by purchasing loan assets) in the housing or mortgage markets. This activity has been financed by increasing the agencies’ debt at or near business cycle peaks. This implies a tendency for agency debt to behave procyclically, in contrast to the countercyclical behavior of Treasury debt. Changes in Treasury debt outstanding (chart 1) indicate that Treasury borrowing tends to follow changes in the budget deficit (chart 2 ), rising during recessions and declining near business cycle peaks. Agency borrowing, on the other hand, tends to rise just before the business cycle peak and to decline after the onset of a recession. Since Treasury borrow ing has generally been substantially larger than agency Page 17 FEDERAL RESERVE BANK OF ST. LOUIS borrowing, changes in aggregate federal debt have tended to mirror changes in Treasury debt. Neverthe less, ignoring the behavior of agency debt leads to an overstatement of the degree to which government bor rowing has been countercyclical. The effect of agency borrowing on cyclical move ments of interest rates is also important in examin ing the interaction of the federal government and the Federal Reserve System. Through open market purchases of government securities, the Federal Re serve can attempt to counteract the upward pressure on interest rates that can occur when the Treasury borrows. The extent to which Federal Reserve policy accommodates Treasury debt operations is a critical issue in the debate over the impact of government deficits on inflation. Note, however, that the Federal Reserve can undertake open market operations using both Treasury and agency securities. One might ex pect then, that when the Treasury and agencies in crease their debt, the Federal Reserve would tend to increase its holdings of both types of debt. In fact, Federal Reserve holdings of agency securities have in creased substantially in recent years, rising from less than $1 billion in early 1972 to more than $8 bil lion by August 1979.14 Consequently, studies of Fed eral Reserve responses to fiscal actions — such as the issue of whether larger government deficits are asso ciated with higher money growth — should consider the expenditures and debt operations of both the Treasury and the agencies. If off-budget agencies, especially the federally spon sored “private” corporations, continue to proliferate, agency borrowing as a proportion of all debt raised under federal auspices will become increasingly larger. Consequently, the behavior of aggregate federal bor rowing over the business cycle could change signifi cantly if the federal government continues to rely on these agencies to change the allocation of the econ omy’s resources while at the same time keeping their outlays out of the budget process. The High-Employment Budget Concept Since the early 1960s, many economists have em phasized that simple budget concepts overlook the impact of economic activity on the budget. Specific14On February 15, 1977, the Federal Open Market Committee amended its guidelines for the conduct of Federal Reserve System operations in federal agency issues to take account of the FFB. Federal Reserve purchases of agency securities were limited to those agencies that are not eligible to borrow funds from the FFB, although securities issued by the FFB itself may be purchased. See Board of Governors of the Federal Reserve System, 64th Annual Report, 1977 (1 9 7 8 ), pp. 189-90. Digitized for Page FRASER 18 NOVEMBER 1979 ally, during recessions the budget tends to fall deeper into deficit as unemployment insurance coverage and other social programs expand. Some economists argue, therefore, that the relevant measure of the budget’s impact on the economy is a full-employment or highemployment budget concept. This concept estimates the size of the budget deficit or surplus that would result if the economy were at a high level of employ ment. The high-employment budget deficit then meas ures the impact on the economy of the government sector alone. Chart 2 plots both the actual and the high-employ ment budget measures. These budget measures are calculated on a national income accounts ( N I A ) basis rather than the unified budget basis. The chief dif ference between the two methods is that the NIA budget nets out all loan activity while the unified budget does not. The NIA high-employment budget concept is an unofficial measure that is used by the Council of Economic Advisors and others for assess ing the impact of the government sector on economic activity.15 The rationale for netting out government loan activity is that such loans constitute a contingent liability and are not truly expenditures. This ap proach, however, overlooks allocational effects of such loans. Further, since the volume of such loans may vary over the business cycle, the NIA budget con cept overlooks some important aspects of the rela tionship between government spending and economic activity. Measures of the high-employment budget ignore the off-budget expenditures and receipts of both govemment-owned and -sponsored agencies. Consequently, the high-employment budget concept understates the economic stimulus attributable to the federal govern ment. Since a major part of off-budget agency ac tivity involves loans, which the NIA budget concept nets out, placing these agencies on budget would have litde effect on the currently constructed highemployment budget. Any additional economic stimu lus generated by off-budget programs still would not be taken into account when evaluating a high-employ ment budget measure. Chart 2 shows that during the recent expansion (I/1975-II/1979) the high employment budget was, like the actual NIA budget, considerably in deficit. 15Though the idea of the high-employment budget has a long history, it was first given prominence in a policymaking context during the Kennedy Administration. For a discussion of full-employment budget concepts and their application, see Alan S. Blinder and Robert M. Solow, “ Analytical Foun dations of Fiscal Policy,” in The Economics of Public Finance, (Washington, D .C.: The Brookings Institution, 1974), pp. 3-115. FEDERAL RESERVE BANK OF ST. LOUIS During the latter part of this period, accelerating in flation accompanied a rapid growth of agencies’ ac tivity, as reflected in their debt. Thus, at least during the past year, the conventional high-employment budget measure is likely to have understated the effect of the government’s overall impact on the economy. Balanced Budget Proposals Political pressure to contain government spending, especially deficit spending, has increased during the past few years. The most familiar proposals call for a constitutional amendment either to balance the budget, to limit federal spending to a specific percent age of GNP, or to limit tax revenues. Debate on these issues has centered on a few key arguments about the desirability and practicality of such limitations. Both sides in this debate, however, have often overlooked the federally owned and fed erally sponsored programs that are not part of the budget. The existence of these programs and their potential for expansion raises serious questions about the effectiveness of constitutional amendments or leg islation directed at containing government spending. For instance, Congress could satisfy requirements for balancing the budget by removing some agencies and their programs from the budget or by redesignating them “private” institutions, as was done with the Fed eral National Mortgage Association in 1968. Either action would improve Congress’ ability to balance the budget, but would violate the amendment’s intent. Ignoring these potential congressional actions reduces the likelihood that constitutional reforms, if adopted, will achieve their supporters’ objectives. NOVEMBER 1979 SUMMARY AND CONCLUSIONS While the function of federally owned and fed erally sponsored agencies primarily involves ques tions of microeconomic importance, these agencies also generate effects that are macroeconomic in na ture. When fiscal policies are examined, the actions of these independent, off-budget agencies are fre quently ignored. Aggregating these agencies’ debt with Treasury debt is necessary to assess the full im pact of federal programs on the economy, especially on credit markets. The growth of off-budget spending, especially that financed indirectly by the Treasury through the Fed eral Financing Bank, underscores the importance of these agencies. In practice, the FFB permits some pro grams to be funded without undergoing congressional review through the budget process. When analysts evaluate the government’s cyclical impact on capital markets and the economy, they usually examine only the behavior of Treasury debt. This approach, however, could produce misleading conclusions since agency debt behaves differently over the cycle than Treasury debt. Agency debt tends to fluctuate procyclically, thereby dampening the coun tercyclical effects of Treasury debt operations. Disre garding agency activity could also lead to incorrect measurement of the fiscal impact contained in any given “full-employment” budget measure. Agency debt activity also has important implica tions for recent proposals to balance the budget. If the alternative of financing federal programs through off-budget agencies is overlooked, proponents of a balanced budget may find that adoption of their pro posals will fail to achieve their objectives. Page 19