¹ But certain implicit assumptions make the relevance of these conclusions narrower than it was formerly thought to be. F. Y. Edgeworth, at least, was well aware of some of these limitations ( Papers Relating to Political Economy, London, 1926, vol. II, p. 7 ff.Google Scholar). More recent discussion of the welfare aspects of trade policy may be found in: Samuelson, Paul, “Welfare Economics and International Trade” (American Economic Review, vol. XXVIII, 06, 1938, pp. 261–6)Google Scholar; Kaldor, Nicholas, “A Note on Tariffs and the Terms of Trade” (Economica, vol. VIIn.s., 11., 1940, pp. 377–80CrossRefGoogle Scholar), and “Welfare Propositions in Economics” (Economic Journal, vol. XLIX, 09, 1939, pp. 549–52)Google Scholar; de Scitovszky, T., “A Reconsideration of the Theory of Tariffs” (Review of Economic Studies, vol. IX, summer, 1942, pp. 89–110)CrossRefGoogle Scholar; and Hicks, J. R., “The Foundations of Welfare Economics” (Economic Journal, vol. XLIX, 12, 1939, pp. 696–712).CrossRefGoogle Scholar

² This may involve a welfare judgment of no greater particularity than that an increase in welfare to all individuals (or the possibility of an increase in everyone's welfare, in case the previous distribution of income is not immediately restored) is to be preferred to the status quo. See the articles cited above, fn.1, and below, p. 526.

³ See Slichter, Sumner, “Postwar Boom or Collapse” (Harvard Business Review, vol. XXI, autumn, 1942, pp. 5–42)Google Scholar; also, Benjamin Higgins, “Post-war Tax Policy,” Part II (in this Journal, pp. 532-56).

⁴ Professor J. B. Brebner's phrase is likely, on his own admission, to remain his chief contribution to the debate!

⁵ It must suffice to note here that this qualification is not unimportant. Professor Haberler has drawn out some of the less comforting implications of assuming relative stability of the propensities involved in multiplier analysis (see his Prosperity and Depression, Geneva, 1941, chap. 13Google Scholar). Of course, Lord Keynes did not assume a constant marginal propensity to consume throughout his discussion of the multiplier effect of increased investment, but this relaxation of the conditions occurs by way of qualification and applies to “changes of a substantial amount” where “we must be guided by the average value of the multiplier based on the average marginal propensity to consume over the range in question” ( Keynes, J. M., The General Theory of Employment, Interest, and Money, New York, 1936, pp. 120–1Google Scholar). In any rigorous mathematical formulation the marginal propensity to consume and any other propensities which may be integrated into the multiplier itself, must, of course, be considered stable.

⁶ Thus, if Y = the national income in a given period of time,

C = consumption expenditure, and

I = investment for the same period (i.e., both C and I are ex post):

where C′ designates the marginal propensity to consume (dC/dY) over the period. The right hand side of the equation is the multiplier; dY is the product and dI is the multiplicand. The equation may, of course, be written:

.

The investment multiplier, therefore, is the reciprocal of the marginal propensity to save (1 − C′). The formula may also be derived from the series dY = dI [1 + C′ + (C′)² + (C′)³ …]. But this more “realistic” derivation may be misleading. The original equation focuses attention on the ex post nature of the simple investment multiplier; the latter introduces a false sense of movement through time (“rounds of spending”) into a relationship which is timeless, instantaneous, or purely formal (see Samuelson, Paul, “Fiscal Policy and Income Determination,” Quarterly Journal of Economics, vol. LVI, 08, 1942, p. 576.Google Scholar).

⁷

The consumption multiplier, therefore, is the reciprocal of the marginal reluctance to invest. The concept of “rounds of investment” in this case is just as misleading as the concept of “rounds of spending” in the case of the investment multiplier (see remarks above, fn. 6).

⁸ See Angell, J. W., Investment and Business Cycles (New York, 1941), pp. 189–90Google Scholar; also Samuelson, Paul, “Interactions between the Multiplier Analysis and the Principle of Acceleration” (Review of Economic Statistics, vol. XXI, 05, 1939, pp. 75–8)CrossRefGoogle Scholar, and his “Fiscal Policy and Income Determination,” p. 576.

⁹ See Angell, , Investment and Business Cycles, p. 196 Google Scholar; and O. Lange, review of Schumpeter's Business Cycles ( Review of Economic Statistics, vol. XXIII, 11., 1941, p. 191 Google Scholar); also Samuelson, , “Fiscal Policy and Income Determination,” p. 578 Google Scholar; and Hansen, Alvin, Fiscal Policy and Business Cycles (New York, 1941), p. 276.Google Scholar

¹⁰ Let the original increase in investment equal dI _o; and the original increase in consumption equal dC _o.

The investment multiplier then derives from the series,

That is,

.

Similarly the consumption multiplier equation may be written

The cumulative multiplier is the same whether the original injection of income comes from new investment or from new consumption ( Samuelson, , “Fiscal Policy and Income Determination,” pp. 599–601 Google Scholar).

Stability conditions require that C′ + I′ be less than 1. A discussion of this point along with its implications for international trade may be found in Metzler, Lloyd A., “Underemployment Equilibrium in International Trade” (Econometrica, vol. X, 04, 1942, pp. 97–112).CrossRefGoogle Scholar See also fn. 24.

¹¹ We assume provisionally a two-country world.

¹² In all cases there are other possible consequences if either country were unstable in isolation (i.e., if its C′ + I′ were greater than unity). These are noted by Dr.Metzler, , “Underemployment Equilibrium in International Trade,” p. 107 ff.Google Scholar

¹³ It is sufficient to note here that constancy of tastes, prices, and exchange rates is for convenience only and the absence of this condition does not necessarily affect the multiplier principle, so long as there is a dependable saving-consumption pattern. It is clear, however, that prediction on the basis of a simplifier analysis may be untrust-worthy because of further multiplier effects generated by some alteration in the factors impounded in ceteris paribus. (We confine ourselves to multiplier effects, without pre-judice to the variously esteemed power of flexible prices alone to affect the volume of employment, not through more or less fortuitous disturbances on the income side, but through substitution among factors and/or expansion of output in response to cost-price reductions.) For a fuller consideration of the assumptions involved see Metzler, , “Underemployment Equilibrium in International Trade,” p. 98 Google Scholar, and Machlup, Fritz, International Trade and the National Income Multiplier (Philadelphia, 1943), pp. 20–2.Google Scholar

¹⁴ If taxes are considered one of the leakages in the multiplier itself, the multiplicand need be adjusted for changes in government expenditure only. This complication would serve little purpose in the present connection; but it is worthwhile to draw attention to the fact that we have here a form of the foreign trade multiplier applied to two non-competing groups within the national economy: viz. the Treasury and the General Public. Taxes, like imports, depress incomes, ceteris paribus, and may be treated as a leakage, provided government expenditures are treated as income-creating “exports” on the part of the public. Thus the public may be considered, rather artificially, as importing government services through the payment of taxes and exporting services (on the W.P.A. or elsewhere), payment for which arises through government expenditure. The international analogy is not destroyed if an indefinite amount of the purchasing power transferred by either party is considered as a gift or tribute. An autonomous increase in the public's “exports” (government spending) is income-creating, but the multiplicand must be adjusted for induced changes in private (“home”) investment and for leakages due to the public's marginal propensity to “import” (at constant tax rates). If the Treasury's marginal propensity to “import” (or spend back) is constant and government revenue is insufficient to maintain the desired amount of expenditure (“exports” by the public), it can be maintained either by government borrowing (which need not depress incomes), or temporarily by an increase in taxes (i.e., an increase in the public's “imports” and a fall in their incomes). In either case, an analogous situation could easily arise in international trade. (This generalization of the foreign trade multiplier owes much to discussion with Professor Lange.)

¹⁵ Thus we may write

where X = exports, M = imports, C″ = the marginal aggregate propensity to consume, and the other symbols have their usual meaning (see previous footnotes).

Note that leakages due to induced imports do not appear in the multiplier 1/1 – C″, since C″ is the marginal propensity to consume domestic and foreign goods. If C′, the marginal propensity to consume domestic goods only, were substituted, the marginal propensity to import would be implicit in the expression 1 – C′, and we could write,

¹⁶ I.e. the multiplicand, dI + dX − dM would be initially greater by the autonomous decline in M. The cumulative multiplier (fn. 10) is not necessarily invalidated because I′ is less stable than C′, since C′ + I′ (or total expenditure) may be considerably more stable than I′ alone.

¹⁷ This is the normal effect of a deliberate expansion of exports or contraction of imports, i.e. the rise in domestic income occurs at the expense of the foreigner. The larger the foreigner's marginal propensity to import, the greater, ceteris paribus, will be the increase in foreign imports with a rise in foreign income, and conversely (as in this case) the greater will be the reduction in expenditure on imports (our exports) with a fall in foreign income.

¹⁸ If dC/dY or C′ in an open system is the marginal propensity to consume domestic goods, then 1 − C′ must include all the “leakage” propensities, including in an open system the marginal propensity to import (see fn. 15).

¹⁹ In a sense, of course, “practical interest” attaches to the simplest multipliers only, since these alone are measurable at all and even these are so difficult to assess statistically that the task appears to be hopeless. But if we cannot measure the multiplier anyway (except roughly and for, say, the past decade), we may as well have the best theoretical construction possible against the day when our techniques of measurement improve. Already attempts to measure the multiplier, imperfect as they are, have led to attempts to clarify the concept. See, for example, Robertson, D. H., “Mr. Clark and the Foreign Trade Multiplier” (Economic Journal, vol. XLIX, 06, 1939, pp. 354–6CrossRefGoogle Scholar); and Mr. Clark's reply (ibid., pp. 356-8); also other literature on the same or related points reviewed in detail by Villard, H., Deficit Spending and the National Income (New York, 1941), pp. 171–80.Google Scholar

J. Steindl has estimated the Canadian multiplier for the period 1930-7 by correlating gross savings, by which he means the sum of home investment, the foreign balance, and the budget deficit, with gross income produced. The latter figure includes depreciation, value added in services, and the government's payroll. It includes also interest paid by Canadians to foreigners but does not include an allowance for the annual rental value of owned homes. The figures are of course deflated, by the use of 1930 indices, to rule out the influence of price changes during the period. Assuming a time lag between investment and income of one-half year, Mr. Steindl estimates that one-half of the increment in gross income was saved; that is, the multiplier is 2. See Steindl, J., “Income and War Finance in Canada” (Bulletin, Institute of Statistics, Oxford, vol. III, 02 22, 1941, pp. 37–44 Google Scholar).

²⁰ See above, fn. 10.

²¹ For example,

If C + I = E, or total expenditure, this equation may be simplified by substituting E′, or the marginal propensity to spend on domestic goods, for C′ + I′; thus,

²³ In moving toward equilibrium, the 1 − E′ of our cumulative multiplier represents the aggregate marginal propensity to hoard, i.e. the marginal propensity not to spend on home investment or consumption goods. The part of received income not spent may be available for investment in the foreign balance, and we designate by H″ that part of the aggregate marginal propensity to hoard (H′) which frees funds for this purpose. Now, since in a two-country world the export surplus of one country must be the import surplus of the other, we may write,

This is the same as Professor Machlup's equation (International Trade and the National Income Multiplier, p. 78). Professor Machlup uses S for the marginal propensity to save, but he defines saving very narrowly so that S is identical with H″, i.e. all “savings” are invested in the foreign balance.

²⁴ Professor Machlup has derived what looks like a cumulative foreign trade multiplier from the ex post identity of expenditure and income (ibid., p. 78). This is, of course, the same identity from which Lord Keynes deduces the inevitable equality of savings and investment; but this becomes savings plus imports equal exports plus investment when the economy is open to trade. The multiplier which seems to be most compatible with this analysis in the closed economy is the simple (Keynesian) investment multiplier. But Professor Machlup gets rid of induced changes in the multiplicand by assuming that the domestic marginal propensity to invest is zero. Thus 1 − (C′ + I′) becomes 1 − C′ where I′ is zero, and the expression represents leakages due to the marginal propensities to save and to import. However, the existence of these leakages is strictly speaking incompatible with the original identity, since, according to the savings-investment equation (applied to foreign trade), income-reducing savings plus imports must equal income-creating exports (plus investment, if any). A computation ex post of the terms in the original identity, whether applied to the closed economy or to the economy open to trade, would show dY = dC + dI, or C′ + I′ = 1 (where I′ = 0, C′ alone must equal 1), In the sense appropriate to the cumulative multiplier, on the other hand, C′ + I′ must be less than 1, if the system is to be stable.

²⁵ I.e., C′ + I′ ⪚ 1, in the ex ante sense appropriate to the cumulative multiplier.

For example,

where dY _t−1 represents disposable income in a Robertsonian “day” (t) which has been created and received on the previous day ((t − 1).

In terms of the symbols used previously, Professor Machlup's equation derives from the identity (ignoring the incremental designation “d”),

The equation may also be written with the aggregate marginal propensity to hoard and the marginal propensity to import substituted for 1 − E′,

.

This equation follows from the original identity if H′ = H″, but an identity could be written where these two magnitudes differ. A true cumulative multiplier would derive from the series,

… where F′ represents the (usually negative) foreign repercussions, or the foreign “marginal propensity to spend back” to use Professor Lange's phrase. This becomes, after an infinite number of rounds,

which can be transformed immediately into Professor Machlup's multiplier if we are interested in dY _a∞. But, except in equilibrium where both derivations give the same result, the above series is a more useful expository device, and is of course implicit in Professor Machlup's tables.

²⁶ See Metzler, , “Underemployment Equilibrium in International Trade,” p. 107.Google Scholar As Dr. Metzler points out in this article, if both countries are unstable in isolation then the whole (two-country) world will be unstable also, whether international trade takes place or not.

²⁷ See Machlup, , International Trade and the National Income Multiplier, pp. 93 ff.Google Scholar

²⁸ Where the corresponding propensities in the original pair of countries, a and b, are equal or proportional, the propensities in the rest of the world will not matter in the final adjustment; in fact, where the propensities in a and b are equal, the propensities in the rest of the world will not matter at all (ibid., p. 95). The formula (fn. 24) becomes

where “n” is the number of foreign countries in addition to b, and where each foreign country shares equally in a's export trade, and a in theirs, so that the relative importance of each country trading with each of the others is the same.

²⁹ See, for example, Clark, J. M., Economic Planning of Public Works (Washington, 1935), pp. 80–104 Google Scholar; Machlup, , International Trade and the National Income Multiplier, pp. 49–51 Google Scholar, and “Period Analysis and Multiplier Theory” (Quarterly Journal of Economics, vol. LIV, 11., 1939, especially pp. 11–27)Google Scholar; also J. W. Angell, Alvin Hansen, and Paul Samuelson, passim.

³⁰ Oscar Lange has worked out equations under various assumptions concerning time lags in an MS now in preparation. The “truncated” multiplier (Samuelson's phrase) is used by Machlup (International Trade and the National Income Multiplier, pp. 49-51). Professor Machlup's time lags are simple and uniform and the truncated multiplier effect (but not, of course, the corresponding formulae) appears period by period in an ingenious series of tables throughout the book. It may be added that simplicity in the application of time does not relieve Professor Machlup's analysis of all complexities (see p. 88, and Appendix A). Like all good concepts, the implicit existence of the truncated multiplier antedates its explicit definition. Professor Plumptre seems to have a very similar idea in mind in his analysis of the probable effect of an increase in Canadian exports over the period of one year (see his article, “The Distribution of Outlay and the ‘Multiplier’ in the British Dominions,” Canadian Journal of Economics and Political Science, vol. V, 08, 1939, p. 367 Google Scholar).

³¹ Using our previous symbols, the marginal propensity to spend would appear as and the marginal propensity to import as, . In both cases dY _t−1 refers to the small increase in income received in t − 1, but which cannot be spent (i.e., is not disposable) until the following period (see above, fn. 24). If we let the first expression (marginal propensity to spend on home goods) equal E′ (see above, fn. 21), and ignore foreign repercussions (but not the propensity to import, which must be implicit in (1 − E′), the effect of the original increase in exports dX _o at the end of n units of time will be [1 + E′ + (E′)² + (E′)³ … (E′)ⁿ⁻¹] dX _o , and .

³² Where and

See Professor Machlup's table and detailed explanation, International Trade and the National Income Multiplier, pp. 50-1. The formulae are difficult to arrive at when foreign repercussions are involved, but tables such as appear in Professor Machlup's book would give approximate readings.

³³ The qualifications necessary are discussed at length by Professor Machlup, ibid., chap. 10.

³⁴ I.e. the elasticity of the supply of money may not be infinite in the real world, in the absence of appropriate banking policy, or ultimately, in the absence of a perfectly elastic supply of central bank reserves. See Plumptre, A. F. W., Central Banking in the British Dominions (Toronto, 1940), pp. 356–61Google Scholar, for the impact of monetary factors. These must, of course, be taken into account as soon as we drop the convenient assumption of the infinite elasticity of the monetary supply.

³⁵ Imre de Vegh estimates it to have been .0726 over the period 1926-40. See his “Imports and Income in the United States and Canada” (Review of Economic Statistics, vol. XXIII, 08, 1941, p. 133).Google Scholar

³⁶ See Machlup, , International Trade and the National Income Multiplier, p. 202fn.Google Scholar; and Salant, W. A., in Public Policy (edited by Friedrich, C. J. and Mason, E. S., vol. II, Cambridge, Mass., 1941), p. 220.Google Scholar

³⁷ According to de Vegh (“Imports and Income in the United States and Canada,” p. 136), Canada's marginal propensity to import over recent years has been .36.

³⁸ For example, according to de Vegh (ibid.), the United States had a marginal propensity to import from all countries of .0726, or 7¼ cents out of every dollar increase in the national income over the period 1926-40. But the marginal propensity to import from Canada was even lower: 2 cents out of each dollar increase from 1919-22 and only ¾ of one cent from 1922-37. This rose to 2.7 cents for each dollar decline in the United States national income in 1937-8; and the ratio was of one cent per dollar change over the period 1929-40 (ibid., pp. 139-40). The United States marginal propensity to import from Canada is accordingly lower than her marginal propensity to import in general, but the unstable nature of the magnitude historically must discourage any nice calculations as to the effect on American imports of a projected change in Canada's trade policy. Great Britain's marginal propensity to import is placed at .17 by Clark, Colin (The Conditions of Economic Progress, London, 1940, p. 479).Google Scholar Repercussions arising from the marginal propensity to import in Canada's chief foreign markets will not, therefore, on this showing, be very great.

³⁹ This seems at any rate to be borne out historically. When the relief of unemployment became the focus of policy, the Canadian response was the Bennett tariff. The existence of a variety of motives is to be taken for granted in any movement for a higher tariff, but in this case the reduction of unemployment was one of the avowed ends. For an evaluation of the principles of Mr. Bennett's practice, see Bladen, V. W., “Tariff Policy and Employment in Depression” (Canadian Journal of Economics and Political Science, vol. VI, 02, 1940, pp. 72–8).CrossRefGoogle Scholar

⁴⁰ Note that the effect is registered in the multiplicand, not in the multiplier, except in the case of the cumulative or total spendings multiplier where the propensity to invest appears in the multiplier itself. However, if we admit the possibility of a significant alteration in the inducement to invest as a result of government spending, the propensity to invest appearing in the multiplier might become a decreasing function of the rate of government spending, although the propensity to spend (C′ + I′) would probably exhibit a greater degree of independence and stability.

⁴¹ Paul Samuelson has some interesting remarks on this point, “Fiscal Policy and Income Determination,” p. 596.

⁴² The figure above corresponds to a gross need of approximately $2.5 billions of new investment (given in Benjamin Higgins's paper “Taxation Aspects of Post-war Fiscal Policy” read at the meeting of the Canadian Political Science Association, May, 1943, which was an earlier draft of the paper published in this number of the Journal). We assume that the major part of a multiplier effect of 2 occurs during the course of one year, with negligible foreign repercussions. (See fn. 32, where the assumed conditions are similar.)

⁴³ The twelve-year average value of items appearing in Canada's Balance of Payments on Current Account, 1926-37, were: credit items, $1.4 billions; debititems, $1.35 billions. See Canada, Dominion Bureau of Statistics, The Canadian Balance of International Payments (Ottawa, 1939), p. 185.Google Scholar The corresponding figures for commodity trade were: exports .9 billions; imports, .79 billions (ibid., p. 186).

⁴⁴ We are not excluding the possibility that either measure may be supported in a given instance by arguments which take considerations of equity and welfare into account.

⁴⁵ This would depend, of course, upon the factors mentioned parenthetically in fn. 13.

⁴⁶ British Information Services, International Clearing Union (New York, 1943).Google Scholar

⁴⁷ The rise in real income here is at the expense of real income abroad. Dr. Benham considers the possibility that a shift of this nature in the terms of trade may entail unemployment in the export industries in so far as their factors are specific. But this argument would apply also to a proposal for lowering the tariff if factors are specific in tariff-sheltered industries. Applied to the hypothetical economy contemplated by classical free trade doctrine, the case for a tariff which improves the terms of trade is exceptionally strong. Unlike classical doctrine, however, the argument is concerned with national welfare, not the welfare of the world as a whole. See Benham, F. H., “The Terms of Trade” (Economica, vol. VIIn.s., 11, 1940, p. 364)Google Scholar; also Nicholas Kaldor, “A Note on Tariffs and the Terms of Trade” (ibid., pp. 377-80).

⁴⁸ “Increasing the community's real income,” or “increasing the community's output of goods and services” are unambiguous in highly simplified analysis appropriate to real costs (efforts, sacrifices, or opportunity costs, etc.) and to yields in terms of satisfaction (real value on the income side). This is sufficient if one is content to discuss the doctrine of comparative costs on its own level of abstraction. But the terms may also be defined for general equilibrium analysis (see Samuelson, Paul, “The Gains from International Trade,” Canadian Journal of Economics and Political Science, vol. V, 05, 1939, pp. 195–205 CrossRefGoogle Scholar). In either case the definition of welfare given above is appropriate, and if real costs are retained their ratios only are important.

⁴⁹ The boondoggling effect of increased commodity exports was observed by Mandeville with apparent satisfaction: “A hundred bales of cloth that are burnt or sunk in the Mediterranean, are as beneficial to the poor in England as if they had safely arrived at Smyrna or Aleppo, and every yard of them had been retailed in the Grand Signior's Dominions” ( Fable of the Bees, Kaye edition, Oxford, 1924, vol. I, p. 364 Google Scholar).

⁵⁰ This does not require any absolute equality (or inequality) of marginal costs in different countries, since it is the ratios, not the magnitudes, which are relevant. Thus no comparison of costs at home with costs abroad need be made.

⁵¹ This is true even for constant costs, although in this case it may appear that cost ratios for all countries cannot be equalized. As Mill pointed out, however, with free trade and constant costs “A country would make nothing for itself which it did not also make for other countries” ( Principles of Political Economy, III, xviii, 3 Google Scholar). In other words, there would be complete specialization by the country with the lowest costs as long as constant costs mean infinite capacity (as Mill seemed to intend, III, iii, 1). That is, marginal (equals average) costs would not only be proportionate to prices in the producing country, but identical with prices there (in the absence of non-competing groups) and proportionate to prices abroad after trade.

Where limited capacity exists in each country producing under constant costs, we might imagine first a horizontal and then, abruptly, a perpendicularly rising supply curve for the domestic industry, and steps between constant cost plateaus as less efficient sources of supply are brought into use in one country or another. If the steps are sufficiently small there will be a gradually rising world supply curve with the country on the “extensive margin” determining the relevant costs. This is, of course, a summary of the Ricardian theory of rent which is itself a theory of “quasi-international trade.” If we cease to ignore the intensive margin, cost will rise more or less gradually as firms of equal efficiency fail to enter the industry because of the shortage of some factor. This is a case of immobility (of factors) into the firm although entry (of entrepreneurs) into the industry remains free. Old firms may receive rents, but the rents will be capitalized, and in any case they will disappear as soon as replacements are needed, and constant costs will rule once more. Such external dis-economies of large-scale production for the industry leading to rising supply price would probably be accepted as consistent with perfect competition so long as the supply of (hired) factors to each firm were perfectly elastic. Otherwise we should have imperfect competition.

Far from being an exception, therefore, the case of constant costs (with either unlimited or limited capacity) is the only one to which the doctrine of comparative costs can be said in all strictness to apply.

⁵² See Hotelling, Harold, “The General Welfare in Relation to Problems of Taxation and of Railway and Utility Rates” (Econometrica, vol. VI, 07, 1938, p. 242).CrossRefGoogle Scholar The proposition advanced by R. Frisch (ibid., vol. VII, April, 1939, p. 145), and admitted by Professor Hotelling, that proportionality between marginal costs and prices is sufficient for the optimum allocation, is necessary to the validity of free trade doctrine, as well as of laissez-faire doctrine applied to the national economy when non-competing groups exist. See also de Scitovszky, , “A Reconsideration of the Theory of Tariffs,” p. 90.Google Scholar

⁵³ This is emphasized by Samuelson (“Welfare Economics and International Trade”), as well as by Edgeworth, (Papers, vol. II, p. 7).Google Scholar

⁵⁴ All arguments for the tariff based on increasing returns apply to an economy operating under imperfect competition, unless falling supply price is due to the right kind of external economies, i.e. external to the firm but internal to the national industry (see Viner, J., Studies in the Theory of International Trade, New York, 1937, pp. 479–81Google Scholar). Graham's, F. D. article, “Some Aspects of Protection Further Considered” (Quarterly Journal of Economics, vol. XXXVII, 02, 1923, pp. 199–227 CrossRefGoogle Scholar), started a controversy in which many of these points were clarified. But even if one adopts the extreme position of Professor Knight (“Some Fallacies in the Interpretation of Social Cost,” ibid., Aug., 1924, pp. 582-606), that bona fide external economies are non-existent under competition, Graham's argument must receive a measure of “unearned” validity if his competitive assumptions are dropped. Anderson's stout defence of the doctrine of comparative costs as a guide to policy, even under imperfect competition, seems to minimize unduly the possibilities (but not perhaps the dangers) of a tariff to induce a fuller realization of economies of scale ( Anderson, K., “Tariff Protection and Increasing Returns,” Explorations in Economics, New York, 1936 Google Scholar).

⁵⁵ This possibility was noted by Higgins, Benjamin in “Elements of Indeterminacy in the Theory of Non-perfect Competition” (American Economic Review, vol. XXIX, 09, 1939, p. 475 n.).Google Scholar

⁵⁶ Such altruism is not unknown in the history of tariffs; but claims on behalf of the suffering consumers have not as yet become a source of international friction.