5. EFFECTIVE RESERVES, CREDIT, AND CAUSALITY IN THE BANKING SYSTEM By GEORGE HORWICH I. TWO HYPOTHESES THE most serious challenge to American monetary policy oc curred in the 1930's when member bank excess reserves rose to astronomical levels-five to six billion dollars, compared with customary holdings of 100 to 200 millions in the preceding decade. This has been widely cited as prima facie evidence that the banking system of the thirties was in a "liquidity trap." Bankers were said to be indifferent between the holding of cash and noncash assets. This resulted in a zero marginal rate of lending and investing with respect to reserve changes. The supply of bank credit responded only to outside forces, particularly to movements in the demand for loanable funds by eligible borrowers.1 An alternative point of view holds that the excess reserves were the result of a low, but nevertheless positive, response of bankers to reserve increases. The excess reserves were functional, serving as a voluntary additional reserve over the legally required minimum. This was merely a reflection of the high, but not necessarily infinite, * This paper was begun at the Institute on Training in Monetary and Credit Policy, sponsored by the Social Science Research Council and the Board of Governors of the Federal Reserve System in the summer of 1957. A first draft was read at the meetings of the Econometric Society in December, 1957; an abstract appeared in Econometrica, October, 1958, pp. 602-3. The author wishes to acknowledge many helpful suggestions, particularly those of the late Edward J. Kilberg, whose tragic death in August, 1958, terminated a promising career; R. I. Robinson, R. G. Thomas, and E. T. Weiler. 1 The most consistent proponent of this view is E. A. Goldenweiser. See Monetary Management (New York: McGraw-Hill, 1949), pp. 57-59, and American Monetary Policy (New York: McGraw-Hill, 1951), chap. ix. liquidity preferences characteristic of the decade. In the words of Paul Samuelson, "They [excess reserves] were felt to be necessary in a world where uncertainty dictates a diversification of portfolios." On this interpretation the excess reserves were not idle "surpluses," to which bank lending was unresponsive. On the contrary, when the Federal Reserve sought to remove excess reserves by doubling reserve requirements in 1936-37, member banks sold government securities as a sharp reaction to their reduced "effective" position.3 The policy implications of each hypothesis are clear. If a liquidity trap prevails, bankers ignore reserve changes in either direction, while passively supplying loanable funds in accordance with shifts in market demand. Monetary policy, acting through the banks, is totally ineffective. We shall refer to this as the Keynesian view. On the other hand, if bank earning assets and the money supply are in any degree causally dependent upon reserves, then, barring widespread liquidity traps elsewhere in the economy, the member banks may serve as a medium for altering national income and employment. We shall call this the Wicksellian view of banks and monetary policy. A test of which hypothesis most accurately describes the behavior of the banking system in the 1930's will be made with data obtained from member bank call reports. We shall try to establish whether reserves were causally influential in determining both the level and composition of member bank earning assets. However, in order to study these relationships, it is necessary to adjust the raw data on reserves for lack of comparability in the capacity to use them. This is done by means of an "effective" reserves series, which is described in the appendix to this essay. Effective reserves express any change in the legal capacity to use reserves as a change in the volume of reserves, with the capacity held constant. Thus, a lowering of the reserve requirement (r.) appears as that increase in reserves, which, under the fixed requirement, finances the same maximum purchase of 2 P. A. Samuelson, "Fiscal Policy and Income Determination," Quarterly Journal of Economics, August, 1942, pp. 594, n. 3, and 594-95. 3 See L. H. Seltzer, "The Problem of Our Excessive Banking Reserves," Journal of the American Statistical Association, January, 1940, p. 28, n. 7; E. S. Shaw, Money, Income, and Monetary Policy (Chicago: Richard D. Irwin, Inc., 1950), p. 443; Steiner, Shapiro, and Solomon, Money and Banking (4th ed.; New York: Holt, Rinehart & Winston, Inc., 1958), p. 596; K. Brunner, “A Case Study of U.S. Monetary Policy: Reserve Requirements and Inflationary Gold Flows in the Middle 30's," Schweizerische Zeitschrift für Volkvirtschaft und Statistik, March, 1958, pp. 160-201; and M. Friedman, A Program for Monetary Stability (New York: Fordham University Press, 1959), pp. 45-46. earning assets that the lowered requirement makes possible. Effective reserves, as described in the appendix, also incorporate several factors other than the reserve requirement that affect reserve availability. The following section summarizes the author's earlier banking study of the 1950's for comparison with the thirties. Section III presents the data on total earning assets and effective reserves for 1930– 39. Section IV is an initial test of the Keynesian hypothesis, using money income as a proxy variable for investment demand. Sections V and VI examine the behavior of the loan and investment components, respectively, of total earning assets. Section VII discusses the impact of interest rates and possible lags in bank responses. Section VIII is a summary. II. THE FIFTIES1 Total member bank earning assets were related to effective reserves for the period December 31, 1952, to December 31, 1955. Thirtyseven monthly observations on member bank effective reserves, R", were obtained by the method outlined in the appendix. Whereas unadjusted reserves declined during this period from $19.95 billion to $19.00 billion, R", expressed in terms of a "standard" requirement of .1557, rose from $21.38 billion to $25.11 billion. Statistical measures gave evidence of a constant and sensitive response of total earning assets, E, to effective reserves of the same month. Out of 36 monthly increments of R", 26 increments of E were in the same direc tion, and seven of those that were not were clearly a delayed reaction of one or two months to a sudden reversal of a trend in R". Out of 35 corresponding second differences of R" and E, 30 were of the same sign. The agreement in the signs of second differences was especially pronounced (all but one were in agreement) when the first differences of the same month were of opposite sign. This tempered the divergence of the two series by giving them the same direction of concavity (thus if R" is rising by rising amounts, while E is falling, the agreement of second differences implies that E falls by declining rather than increasing amounts). The linear correlation coefficient between E and R", TER", was .95; that between their first differences, 4 This section summarizes my paper, "Elements of Timing and Response in the Balance Sheet of Banking, 1953–55," Journal of Finance, May, 1957, pp. 238-55. The specific results reported here differ somewhat from those of the article in that they are based on the adjustments described in the appendix, rather than in the original paper. ΔΔΕΔΔΗ TABAR", was .63; and between their second differences, TAAEAAR", .72. The equation of the least-squares regression line, in billions of dollars, was E 10.34 + 4.86 R". Whenever interest rates and business activity were rising, the observations tended to lie above the regression line. Periods of falling interest rates and activity were characterized by points below the line. III. THE THIRTIES: FOUR PHASES Figure 1 presents the time series of member bank earning assets and effective reserves on call report dates from December 31, 1929, to December 30, 1939. There were three call reports in 1932, 1933, 1936, and 1937, and four in each of the remaining six years. These furnish 37 observations on the variables at an average interval of 3 months. Earning assets are defined as all loans and investments, 2 1930 1931 1932 1933. 1934 1935 1936 1937 1938 1939 gross of valuation reserves. Reserves are deposits at the Federal Reserve plus vault cash. The standard requirement for effective reserves is .077. Earning assets fall from $35.39 billion in December, 1929, to $24.79 billion in June, 1933, and rise to $33.94 billion by December, 1939. Unadjusted reserves rise from $2.93 billion at the beginning of the period to $12.44 billion at the close. Most of this increase is due to the upward revaluation and tremendous inflow of gold that began in 1934. However, R" is $3.50 billion at the start, and only $6.10 billion at the close of the decade. The substantially smaller rise in effective than in unadjusted reserves is due chiefly to the increase in the required reserve ratio from .074 to .153. Periods of gold inflow and changing reserve requirements are indicated in Figure 1 between the call report dates nearest to these events. The National Bureau's upswings and downswings of economic activity are similarly marked off at the top of the chart. Figure 2 is the scatter diagram of corresponding values of R" and E. Although 24 out of 36 first differences of the two time series in Figure 1 have the same sign, TER" is only .04 and TAEAR" is .17. However, the data in Figure 2 seem to arrange themselves into four distinct subperiods or phases as follows: I. From December, 1929, to June, 1933, the least-squares line has a high positive slope (the "b" value) and a comparatively low verti cal-axis intercept. The equation of the line is E = −15.35 + 13.93 R", and r = .85. The observations move in zig-zag fashion from the top of the line to the bottom, making this a period of considerable liquidation of earning assets and some decline in effective reserves. II. This phase extends from June, 1933, to March, 1936. The fitted regression line has a relatively gentle positive slope, a high vertical intercept, and an extremely narrow dispersion of points about it―r is .99. The equation of the line is E = 19.09 + 1.80 R′′. Over time the points start at the lower end of the line and rise along it monotonically to its upper reaches. III. This phase, for which all observations are dated, begins when earning assets in the second quarter of 1936 rise sharply by almost $2 billion, while effective reserves decline $148 million. This vertical movement is located in the right central portion of the figure. The next point, that for December, 1936, lies well above the line in the upper central portion of the figure. It reflects a further increase in |