1
On W. Arthur Lewisâs Contributions to Economics
RONALD FINDLAY
A major concern of economics in our time has been the problem of economic development, the achieving of a sustained increase in per capita incomes of the millions of people living in the so-called Third World. The urgency of this issue has led economists in the past three decades to return to the âinquiry into the nature and causes of the wealth of nationsâ that they had abandoned during the previous century for the investigation of the allocation of given resources and the stabilization of employment and income. Among the many distinguished economists who have contributed notably to this recently revived field the preeminent figure is Professor W. Arthur Lewis of Princeton University. The purpose of the present chapter is to describe and evaluate the nature and significance of his contribution.
I
Lewis was born on the island of St Lucia in the British West Indies in 1915. He studied at the London School of Economics, obtaining his doctorate in 1940, and taught there for several years after the war. His first three books were written during this period. Economic Survey, 1919â1939 (1949) began as a course of lectures to provide students with some background knowledge of the period but, as he says in the preface, it grew from this humble origin into âan over-ambitious attempt to interpret the inter-war years in the setting of world economic historyâ. This concern with âworld economic historyâ has remained with Lewis throughout his career, as witnessed by the fact that his last book, published three decades later, is devoted to the evolution of the world economy from 1870 to 1913. One of the main strengths of his work on development economics is that he has never considered the problems of the LDCs (less developed countries) in isolation but always in the context of the world economy as a single interdependent system.
Also dating from this period is Overhead Costs (1949), a collection of essays dealing with the application of price theory to problems of industrial organization and public utilities. The chapter on the âtwo-part tariffâ for the pricing pf electricity and telephone services provided a definitive analysis of this scheme. There is also a very interesting model of âcompetition in retail tradeâ which considers the question of the optimal number of retail stores. The solution proposed, in terms of balancing economies of scale against customer convenience in having a larger number of outlets to purchase from, anticipates the work done only very recently by Lancaster and others on optimal product variety. The third book of this London period was Principles of Economic Planning (1949), an early discussion of the subject.
The time Lewis spent at the London School of Economics can be thought of as preparatory to his work on development economics, which began after he moved to Manchester, where he occupied the Stanley Jevons Chair of Political Economy. With Harry Johnson and Ely Devons there at the same time the bleak northern city, home and symbol of the Industrial Revolution, must have been a stimulating environment, as one can gather from the many exciting papers that appeared in Manchester School during the mid-1950s. His two most famous works, the celebrated article on âEconomic development with unlimited supplies of labourâ (Manchester School, May 1954) and the magisterial treatise on the Theory of Economic Growth (1955) were written during this decisive period of his career.
The treatise is much broader in scope than the title indicates. It not only discusses capital accumulation, technical progress and population growth, but also the bearing upon these of geography, social structure, religious and cultural values, political institutions and the psychology of human motivation. While much of the attention is concentrated on the LDCs, the book is concerned with the problems of economic growth and social change at such a basic level that it has relevance for societies at all stages of development. Since the purpose of the book was to provide a synthesis of existing knowledge bearing on the subject, rather than to put forward any novel theories or views, we shall not attempt to summarize and comment upon its contents. Attention should be drawn, however, to the fact that in the appendix Lewis asks and effectively answers the profound question âis economic growth desirable?â long before critiques of âgrowth maniaâ became fashionable. He says quite bluntly that the benefit of economic growth is not that it increases happiness, since there is no evidence at all for this, but that it increases the scope for choice by giving man greater control over his environment, thereby increasing his freedom. Also, well before the rise of âwomenâs liberationâ, he wrote these words: âIt is open to men to debate whether economic progress is good for men or not, but for women to debate the desirability of economic growth is to debate whether women should have the chance to cease to be beasts of burden, and to join the human race.â
Since 1963 Lewis has been at the Woodrow Wilson School of Public and International Affairs at Princeton University, USA, where he now occupies the James Madison Chair of Political Economy. He has also held several important administrative positions in the West Indies and has been an advisor to many African countries and to the UN and World Bank. While at Princeton, he published an authoritative textbook on Development Planning (1966) and edited a volume of essays by his students on Tropical Development, 1880â1913 (1970). He has also published three sets of public lectures, each notable for the manner in which a wide-ranging subject is treated in a brief space. The 1969 Wicksell Lectures and the 1978 Janeway Lectures will be considered below in relation to the main line of his research on trade and development.1 The Whidden Lectures on Politics in West Africa (1965) deal with the crucial issues of the relations between democracy and development in the West African context. In addition to being a valuable contribution to political science, these lectures are memorable for the unequivocal commitment to liberal values displayed by the author and his scorn for the pretensions of the one-party state. It should be required reading for the many advocates of âdictatorship for developmentâ among Western intellectuals.
II
In terms of Isaiah Berlinâs well-known classification of thinkers into âhedgehogsâ and âfoxesâ Lewis is indubitably a hedgehog. His one big idea is set forth in the 1954 article on âEconomic development with unlimited supplies of labourâ. His own subsequent work, and in fact a large part of the literature of development economics, can to a large extent be seen as an extended commentary on the meaning and ramifications of this central idea. Few other instances come readily to mind of an entire field being so dominated by a single paper. In view of its importance, and the controversy with which it is still to some extent surrounded, it will be necessary to devote considerable space to an examination of its contents.
There have been, broadly speaking, three alternative approaches to economics. These are the Classical, the Keynesian and the Neoclassical. The Marxist approach is, for present purposes, similar to the Classical and so will not be considered separately. The Keynesian situation is one in which output is constrained not by any supply factor, either labor or capital, but by a lack of effective demand. The remedy for this, especially in the midst of a deep depression, is clear, though these days we are not so confident of what to do when the economy is closer to full employment. At full employment both labor and capital are scarce, and the Neoclassical approach determines the rewards to both factors (or any number, if different types of labor and capital are distinguished) on the basis of the marginal productivity principle. Lewis regards this approach as the relevant one for developed industrial economies, with integrated labor markets, operating under conditions of reasonably full employment and utilization of capacity. For LDCs, however, he considers it more appropriate to return to the Classical method of postulating a real wage rate for unskilled labor that is determined by social convention, not âsubsistence requirementsâ in any biological sense. He thus conforms to the Sraffa or Cambridge view of the real wage as being exogenous for the LDCs but not for the developed countries.
The LDC he regards as a âdual economyâ, divided into a âcapitalistâ and a âtraditionalâ sector. In the âtraditionalâ sector labor is considered as being essentially self-employed, as in peasant family farms or petty trade, or as consisting of service occupations of various sorts such as domestic servants, feudal retainers, and the like. The motive for employment, whether of oneâs self or family members, or in hiring others, is essentially consumption. In the âcapitalistâ sector, on the other hand, the motive for employment is to generate profit, which is the source of investment and, thus, of greater employment in the future. Lewis here explicitly invokes Adam Smithâs distinction between âproductiveâ and âunproductiveâ labor and his interpretation of Smith on this point corresponds to that of the important doctrinal study by Myint (1948, ch. 5), who argues that Smithâs preferences for the former arose not from any âmaterialistic fallacyâ, but from a desire to promote growth over stagnation. Needless to say, the âcapitalistâ employer in this sense can be the state itself and the model can apply quite well to the USSR in the 1920s and 1930s as well as to the Asian communist countries of today.
The exogenously given real wage in the âcapitalistâ sector exceeds alternative earnings available in the âtraditionalâ sector, so that employment in the former sector is constrained by demand and not by supply. It is in this sense that he speaks of âunlimited suppliesâ of labor. Given the wage and technology, profit maximization determines the capitalâlabor ratio and the rate of profit, while the size of the capital stock determines the level of employment in the âcapitalistâ sector. Lewis is also Classical in assuming that all wages are consumed and that profits are the only source of savings. The rate of growth in his model is, therefore, equal to the product of the rate of profit and the propensity to save out of profits, just as in the von Neuman model or the KaldorâRobinson Cambridge theory. The demand for labor can expand faster than the growth of population without driving up the real wage, because peasant agriculture, the urban underemployed and increasing female labor-force participation form a labor reservoir which can be drawn on for a considerable time. It is only when this reservoir is exhausted that there is an integrated labor market with a uniform wage determined endogenously by demand and supply, at which point the economy is developed and the Neoclassical approach comes into its own.
The economic history of England is consistent with this model, since the real wage rate was virtually constant for two generations from 1780 to 1840, the classic era of the Industrial Revolution. It is possible that other instances can be found. In contemporary developing economies, however, real wage rates generally appear to have been rising. For skilled labor of various kinds this is only to be expected, since the supply is limited in the short run. The capital stock, fixed at each instant but expandable over time, is interpreted to include both âhumanâ and âphysicalâ capital. The differentials obtained by skilled labor, just like the quasi-rents of particular types of physical capital, provide the inducement to expand the supply of these inputs by education and training. The average wage will therefore rise if the demand for skilled labor increases faster than for unskilled labor, as one might expect to occur during the process of economic development. Thus, it is only if the real wage rate for unskilled labor rises that a problem appears of whether the Lewis or the Neoclassical view of the labor market in developing countries is the appropriate one.
It is consistent with the spirit of Lewisâs approach to have a perfectly elastic supply of labor over the relevant range at each instant at a given real wage, while allowing this wage to rise over time as a result of exogenous forces such as union pressure or government intervention. For the Neoclassical approach to be valid, the rise in the real wage must be due to excess demand at the original level of wages. This is difficult to reconcile with the widespread existence not only of rural and urban underemployment, but of open urban unemployment throughout most of the less-developed world, which has been increasing not only absolutely, but relatively to the size of the labor force.2 A plausible hypothesis is that the real wage rises at the same rate as productivity increases, leaving the rate of profit constant. The extensive literature associated with the popular Harris Todaro (1970) model of urban unemployment follows Lewis in postulating an exogenously fixed real wage in the urban sector, above alternative rural earnings, which induces a flow of migration equilibrated by the probability of unemployment in relation to the urbanârural wage differential.
Lewis regards the purpose of his model as attempting to explain âthe central problem in the theory of economic developmentâ, which for him is âto understand the process whereby a community which was previously saving and investing 4 or 5 per cent of its national income or less, converts itself into an economy where voluntary saving is running at about 12 or 15 per cent of the national income or moreâ. His model accounts for this fact not by a rise in the propensity to save out of any given type of income, but as a result of an increase in the share of profits brought about by the relative expansion of the âcapitalistâ sector. As Chenery and Syrquin (1975) point out, the econometric results of a number of studies have tended to confirm this basic hypothesis, that âsaving rises relatively to the national income because the incomes of the savers increase relatively to the national incomeâ.
The origins of the âcapitalistâ sector are outside the scope of the model, as are the determinants of the rate of technological change and the propensity to save out of profits. These would obviously depend upon the particular historical circumstances of each country and upon broader social and cultural forces that are beyond the scope of economics as a discipline.3 Once this process has started, for whatever reason, growth in the economy as a whole accelerates even if the growth rate of the âcapitalistâ sector is constant, since it is higher than in the rest of the economy. Under these conditions, the growth rate would be continually increasing but at a decreasing rate, since it is bounded from above by the growth rate of the âcapitalistâ sector itself.
Eventually, however, the conditions of unlimited supply of labor must come to an end. Growth can then be regarded as being governed by a Neoclassical-type model, with an integrated labor market and a fixed saving rate. Initially, the growth rate of capital exceeds the natural rate of growth (population growth plus labor-augmenting technical change). The growth rate therefore decelerates in this phase, asymptotically approaching the natural rate by which it is bounded from below.
The two models in combination therefore yield the prediction of a âlong swingâ in the rate of growth, rising from a level of near zero toward a limit set by the growth rate of the âcapitalistâ sector and then falling toward a lower bound determined by the natural rate, the Lewis model governing the first phase and the Neoclassical model the second. The cross-section data conforms to this prediction, since growth rates plotted against per capita levels show an inverted U-shaped pattern, with low growth in the very poor countries such as Bangladesh, Burma and Nepal, very high growth rates in middle-income countries such as Brazil, Korea and Taiwan and growth at around 3 or 4 per cent in the rich countries such as Germany and the USA. On a time-series basis the experience of several mature countries may also correspond to this pattern, but the picture is not so clear since there has probably been a secular increase in the rate of technological progress itself.
III
As we have seen, the relationship between Lewis and the Neoclassical approaches is one of complementarity in accounting for the characterist...