¹ Keynes, J. M., The Economic Consequences of the Peace (London, 1920), 1971 ed., pp. 15–16.Google Scholar

² Barnes, D. G., A History of the English Corn Laws (New York, 1930).Google Scholar

³ Isaacs, A., International Trade, Tariff and Commercial Policies (Chicago, 1948), p. 360.Google Scholar

⁴ Ashley, P., Modern Tariff Policy: Germany, United States, France (London, 1920), pp. 60–109.Google Scholar

⁵ Ibid., pp. 269–355.

⁶ Isaacs, International Trade, p. 359.

⁷ Benham, F., Great Britain Under Protection (New York, 1941), p. 25.Google Scholar

⁸ Isaacs, International Trade, p. 369.

⁹ Ibid., p. 374.

¹⁰ Chalmers, H., World Trade Policies (Berkeley and Los Angeles, 1953), p. 12.Google Scholar

¹¹ Beveridge, W. H. et al. , Tariffs: The Case Examined (London, New York, Toronto, 1931).Google Scholar

¹² Benham, Great Britain, pp. 29–30.

¹³ Ibid., p. 31.

¹⁴ Snyder, R. C., The Most-Favored-Nation Clause (New York, 1948).Google Scholar

¹⁵ League of Nations, Monthly Bulletin of Statistics (Geneva, Jan. 1928-Dec. 1939).Google Scholar

¹⁶ Isaacs, International Trade, pp. 369–70.

¹⁷ Condliffe, J. B., The Reconstruction of World Trade: A Survey of International Economic Relations (New York, 1940)Google Scholar, Intro.

¹⁸ Benham, Great Britain, p. 12.

¹⁹ League of Nations, World Economic Survey, 1931–1939, 8 vols. (Geneva, 1940), 1: 77–82.Google Scholar

²⁰ Ibid., pp. 31–34.

²¹ Condliffe, Reconstruction, pp. 200–08.

²² Ibid., p. 212.

²³ Isaacs, International Trade, pp. 645–46.

²⁴ Liepmann, H., Tariff Levels and the Economic Unity of Europe (London, 1938), p. 384.Google Scholar

²⁵ Friedman, P., “The Welfare Costs of Bilateralism: German-Hungarian Trade, 1933–1938,” Explorations in Economic History, 13 (Jan. 1976).CrossRefGoogle Scholar

²⁶ League of Nations, World Economic Survey, 1931–1939, 2: 200–02.

²⁷ League of Nations, International Economic Commission Report (Geneva, 1927), pp. 112–62.Google Scholar

²⁸ Leamer, Edward E. and Stern, Robert M., Quantitative International Economics (Boston, 1970)Google Scholar. Leamer and Stern devote a major portion of this publication to the problems of trade function. One conclusion is that the errors suggested by Orcutt are more endemic to price elasticities than income coefficients (pp. 34–35). They also assure us that careful specification can improve upon the dismal projection offered by Orcutt. Leamer and Stern, however, are also guilty of the specification errors of previous researchers by omitting the tariff, re-expenditure, policy and repercussion elements of import demand from their, suggested modeling. Their major contribution in offering lagged structures is inappropriate to the data-scarce interwar period.

²⁹ Benavie, A., “Imports in Macro Economic Models,” International Economic Review, 14 (June 1973), 530–32CrossRefGoogle Scholar. M = F(Y) where Y = C + G + A + (X-M) and A = all other exogenous components. The adjustment in Benavie is identical to M = F(Y + M).

³⁰ Leamer and Stern include a dummy variable to take account of changed structures or a crisis date in the data. In this case the “dummy” will be a vector of zeros before, and income data after, the depression crisis date for each country. This specification will allow the MPM coefficient to change, reflecting change under crisis, and will not be restricted to a simple intercept shift.

³¹ The two sources of tariff data are the League of Nations, International Economic Commission Report, which has limited data for 1925, and Leipman, Tariff Levels, which has more coverage for both 1927 and 1931. Furthermore, tariff rates, aggregated by nation, by goods, are only available for a subset of countries and for a few years during the period.

³² Exchange rate data are available as a continuous series throughout the period in the form of an index of the current price relative to par of any country's currency (League of Nations, Monthly Bulletin of Statistics [Geneva, Jan. 1928-Dec. 1939]). While normally thought of as a market measure, this series can serve as a policy variable. During the period before 1931, exchange rates were fixed and fluctuations were minor, moving to gold points only. After the sterling crisis, when currencies were floated after revaluation, the changes in rate, especially measured annually, were dominated by devaluation rather than by market fluctuation. Moreover, exchange control hid market forces which otherwise might have been reflected in measured rates.

³³ For both the narrative history and an extended discussion of the transfer to a ranking of the historical evidence, see P. Friedman, The Impact of Trade Destruction upon National Incomes, chs. 2 and 4. The choice of nations was restricted due to paucity of income data. The twelve included are representative, being industrial and non-industrial in trade.

³⁴ For changes in trade balance where dX = dM^o, dY/dL is positive when τ is greater than M₁/(1 + M₁).

³⁵ The exact choice of a method for choosing instruments is still a subject of debate in econometrics. We will use a procedure which attempts to isolate those exogenous variables at hand. In this study exogenous variables are available among two sets of data, commercial policy and income data of countries that have had large trade flows with the nation in question and crisis data dummies and trade magnitudes from the rest of the world. See Appendix.

³⁶ There is no analysis of specific policy changes, behavioral shifts away from foreign goods, export demand creating import leakages, revenue implication of non-prohibitive tariffs, repercussions or the possibility of income endogenicity for the entire system, in the extant literature. Chang, T. C., Cyclical Movements in the Balance of Payments (Cambridge, 1951)Google Scholar; Neisser, H. and Modigliani, F., National Incomes and International Trade, A Quantitative Analysis (Urbana, Ill., 1953)Google Scholar; and Polak, J. J., An International Economic System (London, 1954)Google Scholar; studying interwar Europe, all use simple import propensity calculations to determine the trade responses to exogenous declines in income. The results of the above specifications are to calculate import propensities which suffer several types of bias, all of which operate to inflate the estimates.

³⁷ This is for values of dY/dL where

however, even in unbalanced level changes, the direction is the same.

³⁸ This criteria was met throughout the period by seven of the nine countries for which we have complete tariff data. The two counter-examples were Belgium and Switzerland which had extremely low ad valorem rates.

³⁹ The break date choice was determined by that year (1930 or 1931) in which the crisis was effected according to League of Nations, World Economic Survey, 1931–1939, cycle data. The rise in Y_L for Netherlands is the result of a downward biased tariff measure and a rise in the estimated value of M₁1.

⁴⁰ Intra-European trade stood at 80.3% and 71.5% of European exports and imports including the U.K. which had a sizable trade with the rest of the world, 1928–1935. See “Flow of Trade Statistics,” in P. Friedman, The Impact of Trade Destruction upon National Incomes, and League of Nations, Network of World Trade (Geneva, 1942).Google Scholar

⁴¹ The weights use the 1928 and 1935 proportion of imports in country j from i relative to i's total export to sampled countries.

⁴² The aggregate balance of trade in Europe improved from a deficit of —6.841 billion dollars to 2.590 billion dollars between 1928 and 1935, an improvement relative to exports of 62.1% in the balance of trade. Twenty of twenty-four European nations improved their trade balance during the same period. League of Nations, Balance of Payments, 1930–1945 (Geneva).

⁴³ I have chosen to limit the period of the simulation from the break-date until 1936 for several reasons. The quality of the estimated values of the model, in terms of goodness of fit, are superior inside the time range of the estimation. Also, our major interest is in policy changes that occur just after the depression break-date in most countries.

⁴⁴ The choice of actual income as the set of values with which to compare an alternative income series is not optimal. As with any statistical model, there is a portion of the difference between actual income levels and the levels implied by the model which are due to statistical error terms and random variations. If we were to compare the changed income series with the actual income series, part of the difference observed would be attributable to statistical noise. This condition would be tolerable if we could assign a value to the part of the observed differences that results from noise and to the part that results from changed exogenous variables. That assignment, however, is impossible. Therefore, we need another income series for each country that will have all of the same conditions of exogenous variables as those in actual income. The statistical properties of the series, however, are to be the same as for the alternative income series. The set of income values which possesses such properties is the generated value of income with no changes in exogenous variables (Y).

⁴⁵ The constructed set of policy variables are equal to the actual level of policy variables up to the break-date.

⁴⁶ The choice of the division of countries into subgroups will of necessity have arbitrary elements. Two sets of criteria were available, one based upon a division of countries by industrial production, and a second one derived from compositional trade flow data. These alternatives lead to different classifications of countries into subgroups. Both sets of subgroups have been tested by the equations. The estimation of these regressions indicates that the industrial subgroup experienced, as a group, changes in its MPM's that were greater than the changes experienced by the non-industrial subgroup.

There were two possible categorizations of our group countries into industrial subgroups.

When the grouping was determined by industrial production, the null hypothesis that MPM's were equal was rejected by F test at 3.705 (5% level). The same test for a subgrouping determined by trade flows produced an F statistic of 3.915 (10%). The changes in the average coefficient unadjusted by leakage and tariff for each subgroup were as follows:

Both sets of subgroups produced a significant F statistic with respect to differential change. As the results show, the change in MPM's for subgroup (a), industrial countries, was much larger than for the (b) group. In addition, the significance levels of the income coefficients were much higher for the (a) group. The contraction was centered in industrial Europe and the destruction of trade was most severe there.