Published online by Cambridge University Press: 19 October 2009
Models developed to explain variations in cash balances of firms have generally postulated forms of rational choice for the decision maker. Two examples of these kinds of models are (1) Baumol's inventory-type model where the choice of the initial balance is made in terms of a planning period in which outflows of cash, but no inflows, are considered; (2) Miller and Orr's model wherein inflows and outflows occur randomly and a decision is triggered to increase or reduce cash balances when an upper or lower threshold is passed. Statistical tests of the inventory-type model have had limited success, particularly in attempts to identify the increasing efficiency in the use of cash balances as a function of the size of the firm. The apparent linear double logarithmic relationship between cash balances of firms and their sales volumes has cast doubt upon the increased efficiency proposition that derives from the Baumol model. Meltzer has modified this model to demonstrate that it implies linearity. The Meltzer tests will be challenged in this paper, and we shall establish that his results, as well as Baumol's conclusions, are particular outcomes that can be better explained in another type of model.
1 Baumol, William, “The Transactions Demand for Cash: An Inventory Theoretic Approach,” Quarterly Journal of Economics (November 1952)CrossRefGoogle Scholar. A similar model was developed by Tobin, James, “The Interest Elasticity of Transactions Demand for Cash,” Review of Economics and Statistics (August 1956).CrossRefGoogle Scholar
2 Miller, M. H. and Orr, D., “A Model of the Demand for Money by Firms,” Quarterly Journal of Economics (August 1966)CrossRefGoogle Scholar. The authors introduce stochastic net cash flows and two action bounds: an upper cash balance bound that when reached causes management to transfer some cash to investments, and a lower bound that causes the reverse action. The optimal values of these bounds depend on the parameter values of the Bernoulli function generating the flows of cash.
Other typical models with modifications are found in the following:
Frazer, William J. Jr., “The Financial Structure of Manufacturing Corporations and the Demand for Money: Some Empirical Findings,” Journal of Political Economy (April 1964)CrossRefGoogle Scholar. Frazer emphasizes the precautionary motive to explain the lower cash to sales ratios of larger firms. However, he does not consider the low cash to sales ratios of smaller firms.
Mathews, R. C. O., “Expenditure Plans and the Uncertainty Motive for Holding Money,” The Journal of Political Economy (June 1963).CrossRefGoogle Scholar
3 Meltzer, A. H., “The Demand for Money: A Cross-Section Study of Business Firms,” Quarterly Journal of Economics (August 1963).CrossRefGoogle Scholar
Whalen, Edward L., “A Cross Section Study of Business Demand for Cash,” The Journal of Finance (September 1965)Google Scholar. Whalen reformulates the Baumol-Tobin model as
(MT = Transactions Cash, S = sales; a′ = 0 and b′ < 1). He finds that the results of this equation differ little from the linear, given the range of the data.
Alan W. Heston, “An Empirical Study of Cash, Securities and Other Current Accounts of Laye Corporations” (a doctoral dissertation submitted in 1961) did not find much support for Baumol's formulation. Published in Yale Economic Essays.
4 Budin, M. and Eapen, A. T., “Cash Generation in Business Operations: Some Simulation Models,” Journal of Finance (December 1970).CrossRefGoogle Scholar
5 In recent years large firms have also turned to the long-term market to cover cash needs. This can be included in (▵L) as an additional potential source for firms that find their long-term market interest rates are satisfactory.
6 The α′f is based on longer period market conditions; it is probably not influenced by short period small changes in interest rates.
7 In this model α1 and α2 are independent of the asset size of the firm. This assumption is maintained for among-finn-size analysis within an industry. In reality α1 and/or α2 may vary by size of firm reflecting differences in the choices of rule-of-thumb. However, the slopes of αi = αi (A) are probably very small, so we maintain the simplification of slopes equal to zero. Although the α functions are not directly observable, the empirical evidence during periods of high interest rates and good access to credit in the late 1960s appears to substantiate their (near) horizontality.
8 See Appendix I for statistical evidence directly testing the cash-sales ratios of individual firms as a function of interest rates.
9 The implications of this model of demand (equations 5, 6, 7, and 8) might also be derived from Baumol or the Miller-Orr models. These give negative slopes to , and would cause αd = f(A) to decline and flatten with a rise in the interest rate . The former effect is due to the decrease in the variance of daily cash flows relative to sales as sales increase. The latter effect is due to the assumption of constant elasticity of cash holding to the interest rate.
The evidence and argument provided by C. M. Sprenkle in Effects of Large Firm and Bank Behavior on the Demand for Money of Large Firms (1971 - American Bankers Association) leads us to doubt the applicability of cash demand studies based upon assumptions of rational and optimizing behavior. As he indicates (pp. 13–15), large firms should hold extremely small overnight balances because of the availability of various overnight capital markets. The balances of corporations have been far in excess of these uninvestible blocks of cash.
As indicated in the Budin-Eapen article (footnote 4), the balances needed for planning periods through their simulations are also much smaller than those held generally.
10 In Section III the liqudity position of the firm is defined as α* = (C + N + ▵L)/S. The supply function αs differs only by the possibly different views of credit worthiness taken by the borrowing firm and the lending banks. For α* the ▵L is the firm's view of what it believes it can borrow; for ot the ▵L2 is the actual lending provided by the banks when the need arises.
In a full knowledge situation the two are equal.
11 We have not included the regressions in the paper but these are available upon request.
12 See Appendix I for interest rates. The 1957–58 period had higher rates, but liquidity was lacking in the recession of 1958. Furthermore, adjustments in rules of thumb seldom respond to changes that are short-lived.
13 Ibid.
14 Average Annual Prime Interest Rates (1947–68):
1947 - 0.594
1948 - 1.040
1949 - 1.102
1950 - 1.218
1951 - 1.552
1952 - 1.766
1953 - 1.931
1954 - 0.953
1955 - 1.753
1956 - 2.658
1957 - 3.267
1958 - 1.839
1959 - 3.405
1960 - 2.928
1961 - 2.378
1962 - 2.778
1963 - 3.157
1964 - 3.549
1965 - 3.954
1966 - 4.881
1967 - 4.321
1968 - 5.339