The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
IS IT MONEY OR CREDIT, OR BOTH, OR NEITHER?' Credit, Money, and Aggregate Demand By B en S. Bernanke and Most standard models of aggregate de mand, such as the textbook IS /L M model, treat bank assets and bank liabilities asym metrically. Money, the bank liability, is given a special role in the determination of aggre gate demand. In contrast, bank loans are lumped together with other debt instruments in a “ bond market,” which is then conve niently suppressed by Walras’ Law. Much recent research provides reasons to question this imbalance. A growing theoreti cal literature, based on models with asym metric information, stresses the importance of intermediaries in the provision of credit and the special nature of bank loans. Empiri cally, the instability of econometric moneydemand equations has been accompanied by new interest in the credit-GNP relation ship (see especially the work of Benjamin Friedman). We have developed several models of ag gregate demand which allow roles for both money and “ credit” (bank loans). We pre sent a particularly simple one, a variant of model, in this paper. Though it has a simple graphical represen tation like IS /L M , this model permits us to pose a richer array of questions than does the traditional money-only framework. A lan S. B l in der * credit are viewed as perfect substitutes for auction-market credit (“ bonds” ), and finan cial markets clear only by price. Models with a distinct role for credit arise when either of these assumptions is abandoned. Following James Tobin (1970) and Karl Brunner and Allan Meltzer (1972). wc choose to abandon the perfect substitutability as sumption and ignore credit rationing .1 Our model has three assets: money, bonds, and loans. Only the loan market needs explana tion. We assume that both borrowers and lenders choose between bonds and loans according to the interest rates on the two credit instruments. If p is the interest rate on loans and / is the interest rate on bonds, then loan demand is: L d * U p , v). The dependence on GNP ( y ) captures the trans actions demand for credit, which might ansc. for example, from working capital or liquid ity considerations. To understand the genesis of loan supply, consider a simplified bank balance sheet (which ignores net worth) with assets: re serves, R\ bonds, B h\ loans, L*\ and liabili ties: deposits, D. Since reserves consist of required reserves, rZ), plus excess reserves. £ , the banks’ adding-up constraint is: B h + L s + E * D (\ - t). Assuming that desired portfolio proportions depend on rates of re turn on the available assets (zero for excess reserves), we have L* ** \ ( p . i)D ( 1 - r). with similar equations for the Shares of B h and E. Thus the condition for clearing the loan market is I. The Model The LM curve is a portfolio-balance conition for a two-asset world: asset holders oose between money and bonds. Tacitly, oans and other forms of customer-market (1) L ( p , f, .y) * A(p, i)Z)(l - T) Cu.riiSaf sants: Charts Freedman, Bank of Canada; Plosscr, University of Rochester; Robert H. Michigan State University. *l*ri?ccton University, Princeton, NJ 08544. We are *™eful to the NSF for supporting this research. ‘Blinder (1987) offer* a model in which there i& rationing and no substitute for bank credit 435 AEA PAPERS AN D PRO CEEDING S 436 M A Y 1988 The money market is described by a con ventional LM curve. Suppose banks hold excess reserves equal to c ( /) £ )(l- r ) .2 Then the supply of deposits (we ignore cash) is equal to bank reserves, R , times the money multiplier, m (i) = [f(/)( 1 - r ) + t ) ] -1. The demand for deposits arises from the transac tions motive and depends on the interest rate, income, and total wealth, which is con stant and therefore suppressed: D (i,y). Equating the two gives (2) D (i,y ) = m ( i) R . Implicitly, /)(/, y ) and L ( p J , y ) define the nonbank public’s demand function for bonds since money demand plus bond demand minus loan demand must equal total finan cial wealth. The remaining market is the goods market, which we summarize in a conventional IS curve, written generically as 3 (3) y = Y(L>e). II. Graphical Representation Use (2) to replace D( 1 - t ) on the righthand side of (1 ) by (1 - r )m (i)R . Then ( 1 ) can be solved for p as a function of i, y, and R :4 (4) P = <#>(', ^,-R ). Finally, substitute (4) into (3) to get (5) y = Y (iM i> y , R ) ) , which, in deference to Don Patinkin (1956), 2 F or simplicity we assume that only i, not p, influences the demand for excess reserves. 3 The interest rates in (3) should be real rates. But a model of aggregate demand takes both the price level and inflation as given; so we take the expected inflation rate to be constant and suppress it. P is an increasing function of / as long as the interest elasticity of the money multiplier is not too large. F ig u r e 1 we call the CC curve (for “ commodities and credit”). It is easy to see that the CC curve is negatively sloped like an IS curve, and for much the same reasons. However, it is shifted by m onetary policy ( R ) and by credit-market shocks that affect either the L(*) or A( ) functions, while the IS curve is not. The CC and LM curves are shown together in Figure 1 . O ur C C curve reduces to the IS curve if loans and bonds are assumed to be perfect substitutes either to borrowers ( L p - * - 00) or to lenders (A p -> o o ), or if commodity dem and is insensitive to the loan rate ( Yp = 0 ) —which would make the loan market irrelevant to I S /L M . This clarifies the spe cial assum ptions implicit in the money-only view. The opposite extreme, or credit-only view, would arise if money and bonds were perfect substitutes (D i -> - oo), which would make the LM curve horizontal. Keynes’ explana tion for the liquidity trap is, of course, we known. We think of high substitutability as more likely to arise from financial innova tions which create new money substitutes. However, even with a liquidity trap, mone tary policy still m atters because it influences the CC curve. Now let us turn to the intermediate cases represented by Figure 1. VOL. 78 NO. 2 I S I T M O N E Y OR CREDIT\ OR BOTH, OR N E ITH E R ? III. Comparative Statics5 Most conventional shocks work in our model just as they do in IS /L M . For exam ple, an expenditure shock shifts the CC curve along a fixed LM curve, and a moneydemand shock shifts the LM curve along a fixed CC curve. The effects are familiar and need not be discussed. The only note worthy difference is that a rise in bank re serves might conceivably raise the rate of interest in the credit model. Graphically, the ambiguity arises because an increase in R shifts both the CC and LM curves outward. Economically, the credit channel makes monetary policy more expansionary than in IS /L M and therefore raises the transactions demand for money by more than in the conventional model. Greater interest attaches to issues that elude the IS /L M model. An upward shift in the credit supply function, X(-) (which might correspond, for example, to a decrease in the perceived riskiness of loans) shifts the CC curve outward along a fixed LM curve, thereby raising i and y. The interest rate on loans, p, falls, however. An upward shift in the credit dem and function, L (-), which might correspond to a greater need for work ing capital, has precisely the opposite effects. We find it difficult to think of or identify major shocks to credit demand, that is, sharp increases or decreases in the dem and for loans at given interest rates and GNP. But shocks to credit supply are easy to con ceptualize and to find in actual history. For example, Bem anke’s (1983) explanation for the length of the G reat Depression can be thought of as a downward shock to credit supply stemming from the increased riski ness of loans and banks’ concern for liquid!ty m the face of possible runs. According to t- ^ comparative statics results require no assumpns °ther than the ones we have already made. But, in we ^ c o u n te r theoretical ambiguities that tinn r^ ved by invoking certain elasticity assump_ i . spelled out in a longer version of this paper. If bv p . 1S on the supply side, y would be replaced w Figure 1 and in the text discussion that follows. 437 the model, such a shock should reduce credit, GN P, and the interest rate on government bonds while raising the interest rate on loans. Another notable example with the same pre dicted effects is the credit controls of March-July 1980. In this instance “ tight money” should, and apparently did, reduce interest rates on government bonds. IV. Implications for Monetary Policy We turn next to the traditional target and indicator issues of monetary policy. The socalled monetary indicator problem arises if the central bank sees its impact on aggregate dem and only with a lag but sees its impacts on financial-sector variables like interest rates, money, and credit more promptly. W hat does our model say about the suitabil ity of money or credit as indicators? Table 1 shows the qualitative responses of GNP, money, credit, and bond interest rates to a wide variety of shocks, assuming that bank reserves is the policy instrument. Col umns 1 and 2 display a conclusion familiar from IS /L M : money is a good qualitative indicator of future GNP movements except when money demand shocks are empirically im portant. Columns 1 and 3 offer the corre sponding conclusion for credit: credit is a good qualitative indicator except when there are im portant shocks to credit demand. If money demand shocks were indeed more im portant than credit demand shocks in the 1980’s, credit would have been a better indi cator than money. W hat about the target question, that is, about the choice between stabilizing money vs. stabilizing credit? Rather than try to con duct a complete Poole-style (1970) analysis, we simply ask whether policymakers would respond “correctly” (i.e., in a stabilizing way) to various shocks if they were targeting mon ey or targeting credit. Consider first an expansionary IS shock. Table 1 (line 5) shows that both money and credit would rise if bank reserves were un changed. Hence a central bank trying to stabilize either money or credit would con tract bank reserves, which is the correct stabilizing response. Either policy works, at least qualitatively. A similar analysis applies AEA PAPERS AND PROCEEDINGS T able 2 — S im pl e C o r r e la tio n s of G rowth Ra m o f G N P w it h G r o w t h R ates of F in a n c ia l A g g r e g a t e s , 1973-85* h T a b u 1 E m m o f Shoc ks o n O b s e rv a b le V a r ia b le s (1) Income Ri*r tn B«nk R c « n c » M onc> D em and C red it Supply C red it D em and C o m m o d ity D em and (2) Money (3) Credit 4 •f ♦ -f + 4 -f * (4) Interest R ate - + ♦ + *(>n bond* to shocks to the supply of credit or to the money multiplier. But suppose the demand for money in creases (line 2 ), which sends a contractionary impulse to GNP. Since this shock raises Af, a monetarist central bank would contract reserves in an effort to stabilize money, which would destabilize GNP. This, of course, is the familiar Achilles heel of monetarism. Notice, however, that this same shock would make credit contract. So a central bank try ing to stabilize credit would expand reserves. In this case, a credit-based policy is superior to a money-based policy. The opposite is true, however, when there are credit-demand shocks. Line 4 tells us that a contractionary (for GNP) creditdemand shock lowers the money supply but raises credit. Hence a monetarist central bank would turn expansionary, as it should, while a creditist central bank would turn contrac tionary. which it should not. We therefore reach a conclusion similar to that reached in discussing indicators: If monev-demand shocks are more important than credit-demand shocks, then a policy of targeting credit is probably better than a policy of targeting money. V. Empiric*] Evidence The foregoing discussion suggests that the case for credit turns on whether credit de mand is, or is becoming, relatively more stable than money demand. We conclude with some evidence that this is true, at least since 1979.6 6In follows, “ money” is A/1, “ credit” is an aggregate invented by one of us: the sum of intermedi ated borrowing by households and businesses (derived MA Y l m Period 1 9 5 3 :1 -1 9 7 3 :4 1 974:1-1979:3 1 9 7 9 :4 -1 9 8 5 :4 With Money With Credit .51,.37 .50,.54 .11,.34 .17. 11 .50, 51 .38. 4" “ G ro w th ra te s are first differences of natural loga rithm s. b C o rrelatio n s in nom inal term s come first; correla tions in real term s com e second. Table 2 shows the simple correlations be tween G N P growth and growth of the two financial aggregates during three periods Money was obviously much more highly cor related with income than was credit during the period of stable money demand, 1953- 73 But the two financial aggregates were on a more equal footing during 1974:1-1979:3. Further changes came during the period of unstable money demand, 1979:4—1985:4; m oney-G NP correlations dropped sharply while money-credit correlations fell onl> slightly, giving a clear edge to credit.7 M ore direct evidence on the relative magnitudes of money-demand and creditdem and shocks was obtained by comparing the residuals from estimated structural money-demand and credit-demand functions like D ( ) and L ( ) in our model. We used the logarithmic partial adjustment model, wit adjustm ent in nominal terms, which we are not eager to defend but which was design to fit money demand. Hence, our procedure seems clearly biased toward finding re a tively larger credit shocks than money shoe s Unsurprisingly, estimates for the enure 1953-85 period rejected parameter stabiM across a 1973:4-1974:1 break, so we con centrated on the latter period .8 Much to ou from Flow -of-Funds data). F or details and anahwthe latter, see Blinder (1985). y 7Similar findings emerged when we c001™ m any variables via a v e c t o r - autoregression and at correlations betw een VAR residuals. in$tn»8Estim ation was by instrum ental ^ m ents were current, once, and twice lagged ^ government purchases, real exports, bank a supply shock variable which is a weighted a .^ the relative prices of energy and agncultural p IS I T M O N E Y OR CREDIT, OR BOTH, OR NEITH ER? VOL. 78 NO. 2 amazement, we estimated moderately sensi ble money and credit demand equations for the 1974:1—1985:4 period on the first try (standard errors are in parentheses): logM = - .06 + ^ l o g M . i - .2 2 2 / (.34) (.059) (.089) + .0831ogP+ .0121og>> (.052) (.059) SE E = .00811 D W = 2.04, log C = - 1.75 + .885 log C_ x— .424p (0.63) (.076) (.285) 439 measure, the variance of money-demand shocks was much smaller than that of credit-demand shocks during the first sub period but much larger during the second. The evidence thus supports the idea that money-demand shocks became much more im portant relative to credit-demand shocks in the 1980’s. But that does not mean we should start ignoring money and focusing on credit. After all, it is perfectly conceivable that the relative sizes of money-demand and credit-demand shocks will revert once again to what they were earlier. Rather, the mes sage of this paper is that a more symmetric treatm ent of money and credit is feasible and appears warranted. + .514/ + .0751ogP+ .292log y (.389) (.086) (.107) REFERENCES SE E = .00797, D W = 2.44. Here y is real GNP, P is the G N P deflator, p is the bank prime rate, and i is the threemonth Treasury bill rate. Although the inter est rate coefficients in the credit equation are individually insignificant, they are jointly significant, have the correct signs, and are almost equal in absolute value— suggesting a specification in which the spread between p and i determines credit demand. Notice that the residual variances in the two equations are about equal. Since the sample was too short to test reliably for param eter stability, we examined the residuals from the two equations over two subperiods with these results: period 1^740-1979:3 1979:4-1985:4 variance of money residual variance of credit residual .2 6 5 x 1 0 '4 .687 X l O 4 .888 XlO-4 .435 XlO-4 The differences are striking. By this crude Bemanke, Ben S., “ Nonmonetary Effects of the Financial Crisis in the Propagation of the Great Depression,” American Eco nomic Review, June 1983, 75, 257-76. Blinder, Alan S., “ Credit Rationing and Effec tive Supply Failures,” Economic Journal, June 1987, 97, 327-52. ______ , “ The Stylized Facts About Credit Aggregates,” mimeo., Princeton Univer sity, June 1985. Brunner, Karl and Meltzer, Alan H., “ Mon ey, Debt, and Economic Activity,” Journal o f Political Economy, September/October 1972, 80, 951-77. Patinkin, Don, Money, Interest, and Prices, New York: Harper and Row, 1956. Poole, William, “ Optimal Choice of Monetary Policy Instruments in a Simple Stochastic Macro Model,” Quarterly Journal of Eco nomics,, May 1970, 2,197-216. Tobin, James, “ A General Equilibrium Ap proach to Monetary Theory,” Journal of Money, Credit and Banking, November 1970, 2, 461-72.