Immigration, technological progress and assimilation
Whereas, in some political circles, many fear that migrants are substituting for the native population, the real economic question on immigration concerns the degree of substitution between skilled and unskilled workers regardless of their origin. Despite a considerable number of studies, there is very little empirical evidence or confirmation of the impact of immigration on wages. This is sometimes explained by the fact that migrants do not compete directly with native workers, as the jobs for which the former apply are often shunned by the latter. If the impact on wages is smaller, it is primarily because firms adjust their equipment according to changes in the average skill level of the working-age population.
The impact of immigration on an economy depends on that economy’s capacity to absorb new labor, which is to a large degree unrelated to linguistic or cultural issues. The effect that new arrivals have depends in particular on their level of educational attainment. A marked difference with the native population will modify the proportions of skilled and unskilled individuals in the labor force. At first sight, this should accentuate or reduce the pay gap between the highest-educated and the rest. But an economy – and especially its production system – is not inert. Adaptation of production techniques often reduces the distributive effects resulting from the entry of new migrants. It thereby smooths their integration into the economy without the distributive effects being too negative for the native population or for the descendants of previous immigration waves.
One of the biggest fears associated with migrant flows is their impact on the wage gap between highly educated individuals and the rest. Migration is often studied with the help of the old Stolper-Samuelson model (1941). In its simplest form, it suggests that the specialization associated with opening the economy to foreign trade reduces the incomes of the least abundant production factor, which, in advanced economies, is made up of the least-skilled individuals. According to this approach, there exists a similarity between opening up to foreign trade and receiving immigrant workers. The economic and distributive consequences are the same: non-skilled labor is the big loser, whether the country imports labor-intensive goods or the immigrants have a low level of education. The fears held by low-skilled workers in the recipient countries are therefore understandable. These fears are often exacerbated when the country in question has a generous and universal social safety net. Not only is it likely to attract the least-skilled migrants, it may also be associated with a higher unskilled labor cost due to high social security contributions.
Fall in wages or adaptation by the production system?
However, this overlooks the fact that firms also adapt to the available supply of labor by tailoring their production techniques to local skills. In a study on American agglomerations with high levels of immigration, Ethan Lewis shows that the adoption of new technologies between 1988 and 1993 was much slower in cities where the ratio of non-graduates to graduates increased the fastest. The impact of the increase in the number of unskilled workers on their pay was at least partially offset by an increase in jobs offered. This example has been corroborated by numerous empirical studies suggesting that low-skilled immigration is sometimes a substitute for faster automation of production processes.
Mechanization and labor shortage
History abounds with examples of more or less capital-intensive production methods being adopted in response to demographic changes or constraints to migrant flows. A shortage of labor, regardless of its cause, increases capital intensity.
In 1934, the economist Harry Jerome inquired into “the extent to which the effects of immigration restriction upon the supply of labor were likely to be offset by an increasing use of labor-saving machinery. […] As industry began to recover and approach the 1923 peak of activity, questions began to arise concerning a possible or actual shortage of labor. Some urged that the bars on immigration should be lowered. Others voiced the opinion that labor-saving machinery and other improvements in efficiency would fill the gap. […] With high wages, restricted immigration, relatively low prices for capital goods and an abundant and easy capital market, the conditions for rapid mechanization were unusually favorable in that decade.”
It is a commonly held view that a shortage of labor is a driver of industrialization. This idea can be found in the works of historian John Habakkuk, for whom a shortage of labor laid the foundations for American industrial progress by forcing industrial firms to seize any opportunity to introduce capital-intensive equipment. The introduction of the “color bar” in South Africa in the early twentieth century also brought about a change of mining techniques. Fearing the negative effects of medium-skilled black immigrants on their wages, the white workers imposed a color bar that barred blacks access to a number of occupations. This code also sought to reduce immigration of semi-skilled workers sponsored by mining companies in search of cheaper labor. The lack of well-trained workers led to a gradual substitution of capital for skilled labor.
“Endogenous” changes in production techniques
History also brims with episodes in which techniques have responded to changes in the average skill level of labor. It has long been held that the arrival of migrants with a different skill level to that of the local population could bring about a change in the sectoral distribution of production. This is known as the “Rybczynski” effect. It suggests that the arrival of new low-skilled workers should have led the United States to specialize in the production of more goods intensive in unskilled labor (textiles as opposed to steel for the sake of simplicity).
It is true that the American regions that received the most unskilled immigrants in the early twentieth century recorded the slowest rates of agricultural mechanization. They generally continued to produce hay and corn instead of growing wheat, which was more automated.
However, in most cases and in the manufacturing sector in particular, it was not the nature or composition of the goods produced that changed but production methods. Changes in the average skill level of the population shaped production techniques, not the relative weight of such or such an industry. In response to migrant flows, firms invested in equipment that was more or less complementary with the type of labor in abundance. Techniques were shaped or even directed by structural population changes.
This point is crucial, since it suggests that in the medium term it is the production techniques and methods that make it possible to dampen the impact of immigration on wages. Immigration does not modify the nature of the goods produced, nor the sectoral distribution – or type of good – of production in a country or region. It mainly transforms the use that is made of the different types of labor – skilled or unskilled – in production processes.
Free trade of goods and services or free movement of people?
The adjustment of the producing sector toward the most abundant type of labor explains why its predominance in the labor force does not necessarily bring about a fall in its relative remuneration. This point is crucial to understanding two major episodes in economic history.
The evolution of the relationship between immigration policies and protectionism has been particularly instructive over the past two centuries. In the late-nineteenth century, immigration was open but foreign trade was restricted by tariffs and quotas. Today the opposite is the case: immigration is tightly controlled and goods and services are freely traded. Beyond religious or ethical criteria, forces exist that can explain how the transition took place from free movement of people to the exclusive liberalization of goods and services trade.
This transition can be explained by the combination of a changing average skill level in the recipient countries, the development of public social welfare systems and the adaptation of production techniques.
During the 70 years preceding the First World War, customs duties accounted for a significant share of fiscal revenues. For example, they provided 90% of the total American federal budget in the late nineteenth century. A large share of these revenues was spent on defense and on financing infrastructure projects associated with the westward expansion of the territory. Immigration was fiscally neutral: in the absence of a universal social welfare system it led to no public transfers.
The low average educational level of new migrants between 1880 and 1930 did not lead to a significant increase in the education premium. To be sure, the waves of immigration were not without conflict, as illustrated by the Know Nothing movement, but the negative effect on the wages of the least skilled members of the population was not as pronounced as initially expected.
Classical economists in the same vein as Stolper and Samuelson would have explained this development by a change of the type of goods produced. But the integration of the new arrivals did not overly change the distribution of American production by product. Production techniques adapted and limited the impact of immigration on low-skilled wages.
The second half of the twentieth century, often called the “Second Globalization”, saw a complete turnaround in the immigration-protectionism couple. The expansion of the welfare state, the necessary diversification of fiscal revenues and the impact of the highest-skilled on economic policy combined to reverse the order of the paradox: trade liberalization and tighter control of immigration.
A first explanation of this reversal is sociological and boils down to the relative political weights of the skilled and unskilled (the former generally benefits from trade liberalization). In the terminology of economics, the skill level of the median voter (he who splits the population into two equal parts) gradually evolved from low- to mid or even very highly qualified. Individuals well below this skill level have remained the big losers of trade liberalization or migrant flows, while their political influence has declined sharply.
The highest skilled, meanwhile, benefit from free trade in goods, services and capital. But they are more resistant to the arrival of new migrants – especially the least skilled – and fear that their economic dependence might weigh on the social welfare equilibrium. Graduate natives, who are the main contributors to public transfer systems, appear much more concerned by a potential increase in social security contributions than by a hypothetical fall in their wages. This is what drives their choice to restrict migrant flows.
Aside from the fiscal argument, however, one could imagine that in the absence of a threat to their wages or jobs, the most highly skilled individuals would show a greater tolerance toward unskilled immigrants. Yet it is the reverse that is observed. Immigration policies now tend to give preference to the most highly skilled individuals.
Does this mean that the education premium is going to fall? That the most highly skilled will end up being negatively affected? That they are shooting themselves in the foot? In OECD countries, the share of migrants with a tertiary education is now higher than that of natives of the recipient countries (30.3% versus 23.6%).
Are capital and skills complementary?
Technological progress and skilled individuals have been increasingly complementary since the early 1970s. The fall in the prices of new capital goods accelerated investment in information technology. This trend came hand in hand with an increase in demand for skilled people. Unlike the lessons of classical economics, the growing share of the number of graduates in the labor force has not driven down the education wage premium. To understand why an increase in the percentage of individuals with a given type of qualification does not bring about a fall in the relative wage of this population, two effects must be distinguished:
a. The “elasticity of substitution” effect. The elasticity of substitution measures the response by the wage gap between graduates and non-graduates to a change in the weight of each of these types of labor in the population (1). This definition of “skills” is very narrow and does not take into account individuals’ experience, non-cognitive predispositions or ability to adapt. In the case of immigration, whether or not the migrant speaks the language of the recipient country could alone constitute a significant threshold. But the elasticity of substitution makes it possible to simplify and summarize in one figure how a change in the skills make-up of a population – whether through the education system or immigration – will affect relative wages. The higher this elasticity, the less the qualification premium will decrease in response to an increase in the percentage of well-trained individuals. In other words, the relative productivity of the unskilled would need to increase more for firms to use more of this type of labor.
Although the value of this elasticity and therefore the degree of sensitivity of the wage gap to the relative weights of skilled and unskilled workers are the subject of debate, the effect of an increase in the proportion of the highest skilled is always negative. It may be larger or smaller but, to put it naturally, it tends to reduce the education premium.
The persistence or even increase in the wage gap according to years of study is therefore primarily explained by changes in the relationship between capital and labor.
b. There exists a “bias” effect in technological progress in favor of the most highly trained individuals. Considering only the elasticity of substitution overlooks technological progress. And it is seldom neutral. It benefits either capital at the expense of labor or skilled labor at the expense of unskilled labor (2). Over the past 50 years, technological progress has tended to be complementary with skilled labor. Capital and skills are therefore complementary and not substitutable. Technological progress benefits the highest skilled despite the increase in their proportion in the labor force (3).
We therefore live in a world where an increase in the number of skilled individuals skews innovation in favor of this abundant “production factor”. Their wage does not suffer as a result of the absence of any real possibility to substitute them with less-trained individuals and especially because this new equipment, once adopted, can be used by a growing number of people, themselves often skilled. This calls to mind SAP-type software, the functionalities of which are constantly improved by a growing number of (skilled) developers and are complementary with the work of their (skilled) users.
Implications for immigration policy
Contemporary economies function completely differently to those in the middle of last century. In particular, the skills make-up of the population has a decisive effect on the type of innovation that takes place.
Technological progress is induced – or “directed” in the words of economist Daron Acemoglu – by the skills make-up of the population. An increase in the percentage of graduates favors technologies that are intensive in skilled labor. It is because firms have an interest in introducing technologies that increase the productivity of the highest skilled that the negative price effect on wages (increase in the relative share of graduates) is offset by a positive volume effect (dissemination of technology).
Based on this observation, one can only hold a favorable view of open immigration for the highest skilled, which combines several benefits: net positive contributions to social welfare systems, innovation and dynamism in the business community. Moreover, the impact on the education premium is minor.
This optimism should nevertheless be tempered somewhat by the fact that the increase in the ratio of skilled to unskilled workers skews the national income distribution toward profits and raises the more general question of redistribution. Targeting immigration policies toward skilled individuals does not generate direct distributive effects but does amplify the effects that the skills-favored bias of contemporary technologies may have on inequality and on the share of wages in GDP.
(1) Technically, it is the calculation of how much the productivity of skilled labor relative to unskilled labor changes in response to a change in their relative proportions. As the wage is supposed to follow changes in productivity, we arrive at the definition used in the text. It is not uncommon for the elasticity of substitution to be used to estimate change in relative demand for labor (trained or otherwise) according to changes in its relative price. This is the case for example in studies on the minimum wage.
(2) Economists use several sets of terminology to describe the same phenomenon and it is crucial to clarify what they mean. Hicks looks from the perspective of the production factor that technological progress economizes. When it is “capital saving”, it is because it makes it possible to increase the use of labor by increasing its productivity. An invention that is “factor saving” increases the productivity of the other factor, which is generally but not always cheaper. Uzawa, meanwhile, uses the term “labor augmenting” to describe the same phenomenon: more labor, less capital.
(3) An increase in the percentage of skilled workers brings about an increase in the production of goods that use their skills. It can be expected to exert downward pressure on the price of the products or services produced and end up reducing the relative wage of skilled workers. This is the substitution effect. It is relatively modest and above all offset by the high degree of complementarity between the highest skilled and capital. This is because innovation in recent decades has been of an endogenous nature, that is dictated by local production conditions: an increase in the number of graduates increases the use of tools that are “biased” in their favor, which increases their productivity and explains the lack of a contraction in the education premium.