Knowledge that Transforms

Franklin M. Fisher, Peter Temin, Regional Specialization and the Supply of Wheat in the United States, 1867-1914: A Reply, The Review of Economics and Statistics, Vol. 56, No. 1 (Feb., 1974), pp. 114-115

T HE stability of an economy may be upset by disequilibria in the commercial and financial relationships between the domestic economy and the economies of other nations. These disequilibria will either be stochastic or structural in nature. The expected value of a stochastic disequilibrium E(Q) is zero while the expected value of a structural disequilibrium is non-zero. In the latter case, there is a continuing tendency for the reserve stock either to increase or decrease. Since a structural disequilibrium can be removed by an alteration of the exchange rate, we will assume that E(Q)O and that the exchange rate is an equilibrium rate over the appropriate time horizon. The cost of adjustment to stochastic disequilibria may be defined as the change in national income required to alter temporarily the balance of payments by one monetary unit, thereby offsetting the disequilibrium by one monetary unit. That is, the cost of adjustment A is a ratio A -dYQdQ1) where dY is the change in real national income necessary to change the balance of payments by the amount dQ in order to offset an external disequilibrium of size Q. From a position of optimal employment and price levels, an expansion in real income to offset a positive disequilibrium (Q > 0) represents a cost in the form of greater price inflation, while a reduction in real income to offset a negative disequilibrium (Q < 0) represents a cost in the form of higher unemployment. It is to avoid or reduce these social costs that governments accumulate and hold buffer stocks of international reserves.' There is no agreement in the literature concerning the actual size of A for various countries. For example, Flanders (1971, pp. 14-15) argues that advanced countries possess adjustment costs higher than the less developed countries, while Hawkins and Rangarajan (1970, pp. 884-886) argue the opposite. What is agreed is that the cost will depend upon whether expenditure changing policies or expenditure switching policies are used for adjustment. Expenditure changing policies (monetary policy, fiscal policy, etc.) are believed to carry a higher cost than expenditure switching policies (tariffs, subsidies, quotas, exchange controls, etc.). Also, expenditure switching policies, being more direct, can be utilized more quickly to achieve balance of payment effects. Therefore, governments are presumed to usually employ expenditure switching policies. This supposition is supported in recent theoretical work by Britto and Heller (1973) and an empirical study by Gillespie and Rushing (1973). However, the tradition in the literature has generally been to measure the cost of adjustment in expenditure changing terms. On an a priori basis, the cost of adjustment under expenditure changing policies can be quantified as the reciprocal of the marginal propensity to import m. Thus, writers such as Heller (1966, pp. 297-301), Kelly (1970, pp. 657-658), and Clark (1970a, pp. 357-360) use (1/m) as a measurement of the size of the cost of adjustment A. In contrast, no convenient a priori measurement of A is available under expenditure switching policies. Thus the

use of this distinction. And when we are confronted with terms of trade gains or losses between individual industries, caused by changes in relative prices that have been created artificially by protectionism, it would be most unfortunate to give it up. It does involve a lot of nasty index ambiguities, but that is a difficulty we have to live with, if at all we want to break down real national income changes in terms of trade gains, changes in primary factor quantities, and changes in productivity of primary factors. se of this distinctio . And when we are confronted REFERENCES

Belsley, D. A., The Power of the r-Test: Furthering Comment, this REvIEw, LV (Feb. 1973), 132. Cramer, H., Mathematical Methods of Statistics (Princeton: Princeton University Press, 1946), 290. Geary, R. C., Relative Efficiency of Count of Sign Changes for Assessing Residual Autoregression in Least Squares Regression, Biometrika, 57 (1970), 123. Habibagahi, H., and J. L. Pratschke, A Comparison of the Power of the Von Neumann Ratio, DurbinWatson and Geary Tests, this REVIEW, LIV (May 1972), 179. White, J. S., and J. A. Tillman, A Zero Crossing Statistic for a Gaussian Markov Process (mimeograph).

centripetal, with a mode in the range D1 0.50 0.59, the corresponding NE being 2 1.69. The probable reason for this peak is that what had been relatively small specialist enterprises had entered one or other industry in which minimum efficient size is relatively large. As Gort (1962, p. 74) writes, successful entry (into such industries) will necessarily produce a high ratio of nonprimary to primary employment. This could lead to some bunching of values of D1 in the neighbourhood of 0.50. The evidence is consistent with this explanation, since enterprises classified to D1 0.500.59 had the lowest average size of enterprises in all of the size categories in table 5 except D1 1.