Against the Lewis Model

The work of Arthur Lewis is one the best entry points for an engagement with classic development thinking. In his biography of the Caribbean economist, Robert Tignor argued that “Economic Development With Unlimited Supplies of Labor,” Lewis’s classic 1954 article, “galvanized the new field of development economics, providing it with a legitimacy that it had not previously enjoyed.” In the years in which I taught a course on development for postgraduate students at the University of São Paulo, I used to present the key works from the 1950s and 1960s—those by Rosenstein-Rodan, Nurkse, Furtado, Hirschman, Myrdal, Kuznets, and Pinto—as extensions and/or critiques of the “Lewis model.” 

This is not to suggest that Lewis started the conversation. In fact, many of the ideas he articulated in the piece had been explored by others in the preceding decade.  One critical plank of his argument, that there existed unlimited supplies of labor in peripheral economies, was the topic of extensive research, usually described as “disguised unemployment.” His contribution was to provide a unifying perspective on the process of development, combining key insights about the realities of the global periphery with a rigorous analysis of the interaction between income distribution, capital accumulation, and structural transformation. As Hirschman put it, “he managed—almost miraculously—to squeeze out of the simple proposition about underemployment a full set of ‘laws of motion’ for the typical underdeveloped country, as well as a wide range of recommendations for domestic and international economic policy.”

The influence that the 1954 article had—together with a few subsequent works—was such that Lewis was awarded the Nobel Prize in Economics, making him the first Black person and the first (and, so far, only) Latin American to achieve that recognition. Today, his work remains a key reference point, mobilized by economists from very different perspectives. Yet abundant evidence from different countries and different periods indicate that Lewis’s framing of development should be turned on its head. His prediction that capital accumulation would transform labor-abundant countries into labor-scarce ones has rarely been borne out by the actual experience of development.

For a start, labor abundance has seldom been a feature of peripheral economies before they embark on the development path. Besides, in countries where capitalist social relations become widespread, labor scarcity is observed only transitorily, even in the cases of nations that achieve high levels of income per capita. This is because, as Marx indicated, in societies where capital becomes dominant, a surplus population relative to the needs of capital accumulation tends to be constantly reproduced. Thus, the story has not been one of a transition from labor abundance to scarcity, but in fact the opposite, from labor scarcity to truly unlimited supplies of labor.      

A classical model

The opening of the 1954 article is memorable: “This essay is written in the classical tradition, making the classical assumption, and asking the classical question. The classics, from Smith to Marx, all assumed, or argued, that an unlimited supply of labour was available at subsistence wages.” In a way, Lewis was using the same argumentative strategy that Keynes had used less than two decades earlier, when he claimed that the theory he was challenging was not wrong, but simply not applicable to the cases at hand: in Lewis’s view, neoclassical theory held true for economies in which labor was scarce, but this was not the case for many peripheral economies. So, to study the latter, one had “to work right back to the classical economists,” who put forward theories assuming labor abundance, supposedly in line with the realities of the beginning of the Industrial Revolution.

Such a claim revealed a linear view of history that was at odds with some development thinkers working at that time, such as the Latin American structuralists. As Celso Furtado put it, the situation of “underdeveloped” economies looked similar to that of the core countries in the first phase of their capitalist development, but this similarity was superficial. For him, “underdevelopment . . . is a particular process, resulting from the introduction of modern capitalist firms in archaic structures” and thus required “an autonomous theorization effort.” In much of Lewis’s writing—as in his discussion of the open economy model in the 1954 article—he was of course aware of this historical specificity and took it onboard. But the framing of his main model assumed it away. As he put it in a later piece, “the chief historical example on which the [1954] model was based” was Great Britain in the period between 1780 and 1870—as if mid-twentieth century peripheral economies would retrace Britain’s earlier path.

What was this path? Lewis’s model included two “sectors,” a capitalist one and a subsistence one, “heavily developed patches of the economy, surrounded by economic darkness.” Development meant increasing those patches, so that they could eventually encompass the whole economy. The main constraint was the lack of capital to productively employ workers in the capitalist sector, and overcoming it required profits to be saved and reinvested in the production process.

Lewis therefore claimed that the “central problem” of development was raising the savings rate from around 4 or 5 to about 12 to 15 percent of national income. Unlimited supplies of labor made this possible, as it anchored wages close to the average income obtainable in the subsistence sector and, thus, assured that workers would not squeeze profits. Therefore, as the capitalist sector grew, the share of income appropriated as profits would increase, concentrating income in the hands of the “saving class.” “The central fact of economic development,” he wrote, “is that the distribution of incomes is altered in favour of the saving class.”

Lewis and Malthus

Was this really a classical model? One can certainly find in the work of some of the classical economists the idea that wages are anchored at some sort of subsistence level—what Lewis termed the “classical assumption.” In its clearest formulation the idea is grounded on Thomas Malthus’ discussion of the “principle of population.” This principle holds that a general increase in the level of wages above basic subsistence would improve access to food, reducing infant mortality. The ensuing increase in population would, however, tend to outpace food production, leading to an increase in food prices that would bring wages back (in real terms) to the subsistence level. These reduced incomes would reduce access to food, restoring higher infant mortality and rebalancing the size of the population with food production.

This macabre theory, in which wage increases lead to food shortages and a rise in infant mortality, was part of Malthus’s attempt to dismiss the hopes created by the French Revolution that the lot of the poor could be improved—a typical example of the rhetoric of reaction. Yet even economists who did not share such reactionary politics, such as David Ricardo, ended up borrowing his “iron law of wages”—if in a weakened way. As Lewis himself put it, the “Malthusian law of population” provided one of the “cornerstones of Ricardo’s system.”

Lewis’s interpretation of the position of the two classical economists on this issue was, however, peculiar. In his view, Malthus had proved that an “increase of population is caused by economic development.” Given that development historically reduced the death rate to around twelve per thousand, Lewis argued that “in any society where the death rate is around forty per thousand, the effect of economic development will be to generate an increase in the supply of labour.” Falling mortality would spur population growth, guaranteeing an unlimited labor supply.

Then, he referred to findings from “modern population theory” about the demographic transition—that is, the fact that the fertility rate also tends to drop as income per capita levels increase, compensating for declining mortality and reducing the growth rate of population. That was, according to him, what Ricardo and Malthus overlooked, and as a result they “over-estimated the rate of growth of population.” Thus, they could not see that “if conditions are favourable for the capitalist surplus to grow more rapidly than population, there must come a day when capital accumulation has caught up with labour supply.” On that day, labor supplies would no longer be unlimited, and one could revert to neoclassical theory.

Instead of adopting the Malthusian (and Ricardian) view that the interaction between population dynamics and food production determines the level of wages, Lewis reduced their contribution to an argument that, in the course of development, the growth of population increases the labor supply. But it does so only transitorily, before the decline in the fertility rate catches up with falling mortality. To be sure, Lewis argued in 1954 that as long as capital could be exported to other countries with surplus labor, or immigration from those countries encouraged, the neoclassical case grounded on labor scarcity would remain invalid, as in the “neoclassical world labour is scarce in all countries.” Yet subsequent literature all but ignored this caveat and the open economy model he derived from it, focusing instead on identifying empirically the so-called Lewis turning-point, that is, the moment when labor abundance is exhausted and wages start to catch up with productivity. In doing so, subsequent scholars consolidated a “Malthusian” vision of development in which wages are crucially dependent on demographic trends.

Marx against Malthus

Such a vision had been strongly criticized by Marx in Capital. In line with his usual procedure regarding classical political economy, Marx examined in detail how Malthus’s “principle of population” was not ahistorical, as suggested, but in fact a historically specific phenomenon rooted in capitalism. In Marx’s words, “the natural law of population” was not rooted in the “eternal laws of nature” but in the “historical laws of the nature of capitalist production.” 

Such historical laws assured that a surplus population would be constantly reproduced to keep wages in check, guaranteeing the continuation of capital accumulation. The reproduction of labor abundance was independent of demographic trends, resulting from both the cyclical dynamics of capitalist economies—constantly absorbing and recreating an industrial reserve army through the recurrence of booms and crises—and the structural tendency of capital to repel labor, replacing it with machines. If the conditions of class struggle are such that it is possible to push upwards the historically accepted subsistence wage, individual capitalists are encouraged to accelerate mechanization to reduce costs and maximize profits.

With this argument, Marx gave an alternative explanation to the “classical assumption,” that is, that wages gravitate around a subsistence level. (His understanding of the subsistence level was significantly different from Malthus’s, admitting a historical and moral component, which opened the way for class struggle to influence wages and the rate of exploitation.) This alternative explanation is hard to square with Lewis’s claim that unlimited labor supply would tend to be exhausted in the process of demographic transition, as mechanization could accelerate to preserve the surplus population even if population growth decelerated.

Lewis was dismissive of Marx’s argument, referring to it as a “curious model” and claiming that it should be rejected “on empirical grounds.” “It is clear that the effect of capital accumulation in the past has been to reduce the size of the reserve army,” he explained, “and not to increase it.” Marx’s point was not that the reserve army would keep on growing, but simply that it would be continuously reproduced. But regardless, is the evidence really on Lewis’ side in this regard? At least two bodies of literature point in the opposite direction, vindicating the Marxian view.

First, many scholars have built on Marx’s argument and empirically investigated so-called “induced technical change,” that is, the fact that wage pressure tends to spur technical change that leads to the replacement of workers. A variety of statistical methods and datasets all lead to the conclusion that technological adoption is often driven by distributive conflict. Second, a vast number of studies, from many perspectives, show that wages have become decoupled from productivity in most rich countries, which were supposed to be characterised by labor scarcity.

Lewis was extrapolating from a very singular period of capitalism’s history. For a short time, in the global North, it seemed that the “standard employment relationship”—full-time, formal, conferring a variety of rights—had become the norm. During what is sometimes called the “golden age”—the two and half decades after the end of the Second World War—labor markets in these countries remained relatively tight and wages followed productivity closely, so much so that one of the “stylized facts” from the period was the stability of the wage share of income.  Strengthened trade unions and collective bargaining combined with the political effects of the mass unemployment of the 1930s to temporarily constrain capital and reduce the surplus population at the capitalist core. With hindsight, it is clear that it was an exceptional and transitory situation.

At least since the 1980s, labor abundance has been restored, not as a multitude of subsistence workers—as Lewis would have it—but as a deepening subordination of the working classes. In the United States, for instance, it is well established that the average wage of production workers (i.e., those in non-supervisory roles) in the private sector has remained almost flat since 1979: while it almost doubled between 1948 and 1979, it increased only 11 per cent in the subsequent four decades. The wage squeeze has not been as dramatic in the other rich countries, but it has been strong enough to push the wage share of income down in most of them.

Meanwhile, labor market precarity took hold across the capitalist core, leading some analysts to suggest that the global North’s labor markets increasingly resemble those of global South’s countries. If some kind of global economic convergence is to be expected, it will most likely result from the deterioration in the center rather than catching up from below. The surge in gig work and involuntary part-time employment are two examples of this broader phenomenon. Even conventional macroeconomists noticed the restoration of labor abundance, as their Phillips curve started to look increasingly flat (that is, the impact of unemployment on inflation has been weakened). Workers are no longer able to push for substantial real wage increases, not even during the expansion phases of business cycles when unemployment falls, as their bargaining power has been structurally reduced.

From labor scarcity to mass marginalization

All this suggests, contra Lewis, that the neoclassical assumption of labor scarcity is very far from being observed anywhere, even in the rich countries. Lewis, however, was less interested in the realities of the global North than in the process of development taking place in the South. What should we make of his analysis in this regard?

Many have argued that he overlooked the brutality that characterized processes through which labor abundance was established in the first place, forcing populations to seek wage employment via expropriation, displacement and political coercion. As John Sender and Christopher Cramer put it, Lewis “paid insufficient attention to the role of violence, coercion and forceful state intervention in explaining labour market outcomes.” Violent processes of primitive accumulation were hidden in a narrative that suggested that capitalists simply attracted labor from the subsistence sector by offering a small premium on the income they could obtain there. More often than not—as Lewis himself mentions in passing in the 1954 piece—that subsistence income was deliberately squeezed such that it gave the population no other choice but to sell their labor force to capitalist firms.

In an article from 1970, Giovanni Arrighi was probably the first one to make this point systematically, both theoretically and empirically challenging a Lewisian reading of development in Zimbabwe. According to Arrighi, between 1903 and 1920, the country was unambiguously characterized by labor scarcity and the “Lewis model” became relevant only for a short time in the 1920s, after a long-running effort to expropriate indigenous workers and proletarianize them. A similar story has been told about Brazil, where widespread proletarianization was observed only from the 1960s onwards. Proletarianization was the result of a profound transformation of social relations in the countryside, involving the growth of mechanized capitalist agriculture and the displacement of peasants to smaller and less-fertile lands. The state-sponsored “modernization” of agriculture was key for primitive accumulation in Brazil.

But once labor abundance is established, whatever the means, can we finally accept Lewis’s argument that capital accumulation gradually eliminates it? Not quite. As many development economists—such as Furtado and Aníbal Pinto—noticed already in the 1960s—the pattern of accumulation observed in the parts of the periphery where growth was relatively fast very soon shifted towards capital-intensive industries and technologies, absorbing less and less labor. The mass of expropriated workers had little hope of finding employment in the capitalist sector, giving rise to the urban informality that characterises the metropolises of the global periphery.

Revisiting the topic in 1979, Lewis was of course aware of the fact that labor abundance ended up being much more persistent than he predicted. However, instead of engaging with the sophisticated analysis of the Latin Americans on structural heterogeneity, he reverted once more to Malthusianism. Labor abundance was persistent, in his view, because population was growing too quickly. In his words, “population growth is at the core of labour abundance,” and “there is no way that a population can grow at 3 per cent per annum without experiencing an abundance of labour in its modern sector.” Marx’s comment about Malthus applies to Lewis perfectly: “he explains [the surplus population], in his narrow fashion, not by saying that part of the working population has been rendered relatively superfluous, but by referring to its excessive growth.”

Beyond Lewis

The point is not to question Lewis’s commitment to fighting for a fairer world, improving the standards of living of the great majorities in Latin America and in Africa—the two regions on which he was most focused. His effort to open space for original reflections about the economic situation of the global periphery was commendable, and his writing contains a series of valuable insights. Yet, in the end, the model that is his main legacy is an obstacle to critical reflection about development.

In assuming labor abundance to be the reality in most of the South and disregarding the brutality involved in bringing it about, the model ended up justifying these processes of primitive accumulation as necessary evils in the path to development. Lewis’s language often revealed such a bias—for instance, when he argued that workers in the subsistence sector needed to be convinced to adjust to “the more regimented and urbanised environment of the capitalist sector.” Moreover, the mechanism of development itself that he formulated provided legitimacy not only to expropriation but also to continuous concentration of income, as faster capital accumulation depended, according to him, on shifting income towards the euphemistically named “saving class.”

All this was demanded in exchange for the promise of a future of labor scarcity, which produces “magical” results—the words are his. Such a promise, however, hinged on a questionable Malthusianism that failed to grasp the actual dynamics of capitalist economies. When the exhaustion of the labor reserves was revealed to be elusive, many sharpened their critical analysis about the actual outcomes of peripheral capitalist development. Lewis, in contrast, blamed southern populations for reproducing too quickly. Soon after, it became clear that labor abundance was not only a scourge of peripheral development but the norm of global capitalism, having reasserted itself even in the rich countries.

Lewis was right that eliminating labor abundance is key to improving the standard of living of workers, North and South, as it means challenging workers’ subordination. But we can’t count on either population dynamics or capital accumulation for achieving it, as he did. The task of reimagining development beyond capitalist modernization and forced proletarianization should begin by focusing on ways to actually bring about and sustain labor scarcity, such as empowering the majorities through alternative ownership structures in agriculture, services and manufacturing. It requires at least some democratic control over the shape of the production processes and the direction of technology in order to weaken the mechanism that continuously produces a surplus population. Rejecting Lewis’s Malthusianism allows us to recognize that labor scarcity cannot come from accumulation outpacing population growth, but rather from constraining capital’s impulse to produce a reserve army of labor.