This article is a revised version of a paper which I presented at the 1969 winter meetings of the Econometric Society in New York, on December 30, 1969. I wish to acknowledge a great debt owed Thomas Saving for directions, suggestions, criticisms, and encouragement on the numerous drafts of this paper.

¹ Even the stock of base money is to some degree endogenously determined due to the effects of business conditions on international payment flows and gold movements (see Friedman, M. and Schwartz, A., “Money and Business Cycles,” Review of Economics and Statistics, XLV (Feb. 1963), 46)Google Scholar. In general, however, the stock of base money can be considered exogenous to the economic system and dependent on Treasury and Federal Reserve policy.

² Pesek, B. P. and Saving, T. R., Money, Wealth and Economic Theory (New York: Macmillan, 1967), pp. 79–102Google Scholar.

³ Saving, T. R., “Outside Money, Inside Money and the Real-Balance Effect,” Journal of Money Credit and Banking, II (Feb. 1970), 83–100CrossRefGoogle Scholar.

⁴ Ibid., pp. 87–91. Even if , so long as the size of the real-balance effect is not affected by the presence of costs. That is, if maintenance costs depend only on nominal units but not prices, maintenance costs have no effect on the real-balance effect.

⁵ Cagan, P., Determinants and Effects of Changes in the Stock of Money 1875–1960 (New York: National Bureau of Economic Research, 1965), p. 123Google Scholar.

⁶ Friedman, M. and Schwartz, A. J., “The Definition of Money: Net Wealth and Neutrality as Criteria,” Journal of Money Credit and Banking, I (Feb. 1969), 2–8Google Scholar, and Patinkin, D., “Money and Wealth: A Review Article,” Journal of Economic Literature, VII (Dec. 1969), 1140–60Google Scholar would apparently agree with the above formulation of the real-balance effect on demand deposits although they have argued that demand deposits are debt and the net wealth they generate is attributable to the value of the bank charter. As Saving has demonstrated the real-balance effect generated by various forms of money depends only on the capitalized value of the income flows. In particular, the size of the real-balance effect is not affected by a decision to attribute the income flow emanating from bank assets to the value of the bank charter or to the demand deposits themselves.

⁷ See Friedman, and Schwartz, , “Money and Business Cycles,” p. 47 and A Monetary History of the United States 1867–1960 (Princeton: N.B.E.R., 1963), pp. 117Google Scholar, 342, 345, 684–85, and Cagan, Determinants and Effects …, pp. 134–50, 219–33.

⁸ See Friedman and Schwartz, “Money and Business Cycles.”

⁹ In the empirical analysis below it will be helpful to make even more restrictive assumptions about bank assets.

¹⁰ See Cagan, Determinants and Effects …, pp. 134–36, 211, and Friedman and Schwartz, A Monetary History …, pp. 177, 342, 685.

¹¹ See Saving, T. R., “Transactions Costs and Monetary Theory,” American Economic Review, LXI (June 1971), 418–19Google Scholar.

¹² Prior to 1934 interest was paid on demand deposits. In the 1929–1933 period such payments were small and probably largely offset by service charges so that r_d « 0. The Banking Act of 1933 prohibited interest payments on demand deposits by member banks, and positive service charges produced a negative rd (Cagan, Determinants and Effects …, pp. 317–21.

¹³ Ibid., p. 143. As Pesek has pointed out, the Cagan type approach neglects the impact of the operations costs of banks on service charges (“Banks' Supply Function and the Money Stock,” Canadian Journal of Economics, III (Aug. 1970), 357–85Google Scholar, and r_d should be gauged as reflecting both cost and return factors facing commercial banks. While both cost and return factors affect r_d, since real bank costs are not generally assumed to vary systematically over the reference cycle, cyclical movements in r_d can be attributed largely to return factors.

¹⁴ Friedman and Schwartz, A Monetary History …, p. 299. Even these figures understate the severity of the banking crisis of the period. Counting voluntary liquidations, mergers and consolidations the number of commercial banks fell by over one-third. Further the collapse entailed banking holidays in many states and a nationwide banking holiday that closed even the Federal Reserve Bank. None of these events had any precedent in U.S. history (Ibid.).

¹⁵ See Cagan, Determinants and Effects …, pp. 147–50.

¹⁶ The most famous theory of currency demand has evolved from the works of Mitchell, Wesley (Business Cycles (New York: N.B.E.R., 1927)Google Scholar) and Hawtrey, R. G. (Currency and Credit (4th ed.; London: Longmans, Green, 1950))Google Scholar. The Mitchell-Hawtrey theory may be summarized into two assumed relations. (1) Retail purchases are more currency intensive than other transactions. (2) The relative income of wage earners, who use most of their income to purchase in retail outlets, fluctuates directly with the level of economic activity. Thus, the use of currency tends to vary with business conditions, and presumably the currency-deposit ratio rises in an expansion and falls in a contraction.

While no evidence has been presented to contradict the assumption that retail trade is procyclical, the amplitude of variation has been found to be rather small and varies from cycle to cycle (Creamer, D., Personal Income During Business Cycles (Princeton: N.B.E.R., 1956), pp. xvii–xviii)Google Scholar. The second proposition regarding the stability of the ratio of currency to consumer expenditures over cycles is not consistent with statistical evidence (Cagan, Determinants and Effects …, pp. 145–47).

An alternative theory of cyclical movements of the currency-deposit ratio is what might be called a Permanent Expenditures Theory. This theory is based on two empirically observed conditions. Currency is used more in transactions involving nondurable consumer goods than in transactions involving durable goods which are generally larger in per unit value and are more often transacted by check. Nondurable goods purchases tend to remain fairly stable over the reference cycle while durable goods purchases fluctuate greatly with the level of economic activity (Ibid., pp. 149–50). Given these two conditions, rises in the level of economic activity would tend to lower the currency-deposit ratio and declines in the level of economic activity would tend to raise it.

Using data on the currency-money ratio adjusted to eliminate trend, Cagan has observed that for the fifteen panic and non-panic cycles from 1897–1954, excluding the two war cycles and the 1927–1933 and 1933–1938 cycles, the currency-money ratio has tended to rise during the recession phase of the average reference cycle (Ibid., pp. 134–36). While there is some question of whether or not the currency money ratio leads or lags cycle turning points, the general relation between the level of economic activity and the currency-money, ratio appears to be an inverse one (Ibid., p. 137). Friedman and Schwartz found the same general relation using the ratio of deposits to currency (Friedman and Schwartz, A Monetary History …, pp. 177, 342, 685).

¹⁷ See Ibid., p. 685, Cagan, Determinants and Effects …, p. 221 and Meigs, A. J., Free Reserves and the Money Supply (Chicago: University of Chicago Press, 1966), pp. 42–50Google Scholar.

¹⁸ Cagan, Determinants and Effects …. pp. 209–10.

¹⁹ Friedman and Schwartz, A Monetary History …, p. 685.

²⁰ The desired reserve-deposit ratio of banks, while primarily determined by the probability distribution of redemption by depositors into base money, is affected by banks' general solvency conditions. Even if the realized distribution of redemption remains constant, an impairment to the solvency of banks might lead them to bolster cash reserves. As Cagan has noted: “Every effort is made to bolster cash reserves, not only to meet heavy withdrawals but also to attain sufficient liquidity to allay depositors' suspicions of financial weakness” (Determinants and Effects …, p. 224).

²¹ The impact of cyclical fluctuations in the price level on the deposit ratios affects only bank owners because the non-bank owning public trades off in nominal terms after the price decline (increase) and thus receives a reduction (increase) in its bonded indebtedness in return for the reduction (increase) in its money stock. However, the wealth of the bank owners is directly affected by the reduction (increase) in the money stock because they face a reduction (increase) in bond holdings not offset by a concomitant increase in money holdings.

²² ibid., p. 42.

²³ See Friedman and Schwartz, A Monetary History …, pp. 308–32.

²⁴ Ibid., p. 786.

²⁵ Pesek and Saving, Money, Wealth …, pp. 103–37.

²⁶ Cagan, P., “The Demand for Currency Relative to the Total Money Supply,” Journal of Political Economy, LXVI (Aug. 1958), 313–15Google Scholar.

²⁷ It should be noted that the proxy of for r_d and τ as well as δ_d does not significantly alter the results obtained. When income (y) is proxied for τ as is often done, is proxied for δ_d and r_d is employed, the estimated equation for Δγ for the 1929–1939 period is

.

²⁸ Cagan, Determinants and Effects …, p. 16.

²⁹ In the above analysis banks are assumed to hold either base money or homogenous bonds. In reality banks hold a range of assets with varying degrees of liquidity. A rise in the probability distribution of redemption by depositors or an impairment of bank solvency would tend to lead banks to shift to assets of greater liquidity. Thus banks' desired liquidity positions would be reflected in the overall liquidity of their portfolios as wefl as in their holdings of base money. While portfolio shifts not involving base money might be considered irrelevant so far as the determination of the real-balance effect on demand deposits is concerned, the lower average yield on liquid financial assets, however, produces a lower yield than society's alternative return in the absence of banks' convertibility requirement considerations, and the net wealth of society falls. Thus the existence of a relation between changes in the price level and changes in the optimum liquidity position of banks can produce an alteration in society's wealth both by forcing banks to hold more base money as reserves and by forcing banks to use their resources less efficiently than society would be capable of doing in its absence. Breaking bank asset holdings down into reserves, investments, and loans, changes in the composition of bank portfolio balances can be related to changes in the price level by regressing changes in the level of prices on changes in the ratio of investments to loans. Since loans are on the whole less liquid than investments, a rise in the investment-loan ratio represents a movement toward a more liquid bank portfolio (see Friedman and Schwartz, A Monetary History … 449–62; and Morrison, G. R., Liquidity Preference of Commercial Banks (Chicago: University of Chicago Press, 1966), p. 42)Google Scholar. For the 1919–1933 period the simple correlation of first differences between the price level and the ratio of investments to loans assets is —.716, significant at the 1% level. The period 1934–1966 yields a simple first difference correlation value of —.335, significant at the 5% level. These results appear to indicate that there is a distinct and negative relation between changes in the price level and changes in the investment-loan ratio or general liquidity of bank portfolios.

³⁰ See Saving, “Outside Money, Inside Money …,” pp. 93–97.

³¹ Patinkin, D., “Price Flexibility and Full Employment,” American Economic Review, XXXVIII (1948)Google Scholar, reprinted in Readings in Monetary Theory, Lutz, F. A. and Mints, L. W. (eds.), (Homewood, Illinois: Irwin, 1951), p. 32Google Scholar.

³² See Pesek and Saving, Money, Wealth …, pp. 281–83.