Place premium

The concept of the place premium refers to the increase in earnings (usually PPP-adjusted) of a given person when that person migrates from one place to another, without any increase in the person’s skills or human capital. Roughly speaking, it addresses questions like: if a construction worker crossed the border from Mexico to the United States and took a job with comparable hours and using the same skill set, how much would the person’s pay increase? Research suggests that in most cases, a person crossing the border to do the same job would instantaneously double their income, if not more.

TL; DR

  1. The concept of the place premium was introduced in a 2008 working paper by Michael Clemens, Claudio E. Montenegro, and Lant Pritchett for the Center for Global Development.
  2. Place premium is related to the concept of “productivity” but is not (necessarily) about how much you produce in a literal sense, as it is about the economic value of what you produce. For people providing services, the economic value of the services may be higher in some regions than others because of the higher opportunity cost of time of the people to whom the services are provided.
  3. The magnitude of the place premium depends on the origin, destination, and type of worker.
  4. Selection effects make place premia hard to measure precisely. However, there are creative experimental designs and clever ways of slicing historical data that help us with some estimation of place premia. Despite considerable uncertainty, it is quite clear that there are many contexts where place premia are significantly greater than 1.
  5. Place premia in spatially integrated labor markets (i.e., under open borders) are generally not greater than 2.
  6. Place premia describe the economic gains from migration at the current margin. They directionally indicate huge potential gains from open borders, but cannot be used to directly estimate those gains, because the marginal gain could change as more people migrate. Also, a full estimation exercise would require estimating effects on wages and assets for non-migrants in both the sending and receiving countries.
  7. The place premium does not account for the narrowing of skill gaps between the descendants of migrants and those in the receiving countries. In other words, the place premium accounts only for the gain to the migrant’s own income, rather than the (usually) larger gain the migrant’s child experiences relative to the counterfactual scenario where the parent hadn’t migrated.

Clemens, Montenegro, and Pritchett paper on place premium

A good starting point for understanding the place premium is a working paper titled The Place Premium: Wage Differences for Identical Workers across the U.S. Border – Working Paper 148 (link to webpage, PDF linked to from there) by Michael Clemens, Claudio E. Montenegro, and Lant Pritchett in 2008. From the online description:

Are your wages determined by what you know, or where you live? This paper compares the wages of workers inside the United States to the wages of workers outside the United States. Comparing wages alone isn’t enough, because workers in (say) Bolivia could differ from workers in the U.S. in many ways—some of them easily observed, such as their level of education, and others less easily observed.

The concluding paragraph of the online description states:

The implications of these enormous differences are profound: (1) these gaps represent one of the largest remaining price distortions in any global market; (2) for many countries, the wage gaps caused by barriers to movement across international borders are among the largest known forms of wage discrimination, typically much larger than wage discrimination based on ethnic group or gender within spatially integrated labor markets; and (3) these gaps imply that simply allowing labor mobility can reduce a given household’s poverty to a much greater degree than most known antipoverty interventions inside developing countries.

Nancy Birdsall also discusses the paper in a blog post titled Clemens’s Place Premium Tells Incredible Tale of Global Non-Market in Labor.

Place premium and “productivity”

The place premium is observed both for jobs producing tradable goods (such as software or widgets in a factory) and non-tradable goods (such as haircuts or concierge services). For tradable goods, the place premium arises primarily from greater productivity in a very obvious sense, such as workers operating better physical machinery, or in more effective collaborative environments or better institutional infrastructure to support production (such as reliable electricity or fast Internet).

For non-tradable goods, there is an additional explanation for the place premium, even controlling for the amount produced: the people served in some regions may value the services more. Note that this is specific to non-tradable goods, i.e., goods that need to be produced close to the place of final consumption. As John Lee explains in his blog post Understanding the place premium, or; building the economic intuition behind open borders, this is because the people served, being more productive, value the service more because of the higher opportunity cost of their time:

Here’s the classic example of the place premium: the bus driver. You take the bus driver in Yemen and transplant him to the US. Same guy, doing the same job, driving the same bus, even. Just in a different place. You’ve just boosted that Yemeni’s income by about 15 times over, because now he’s ferrying highly-productive Americans instead of relatively unproductive Yemenis. He’s driving them between Boston and New York instead of between Sana’a and Aden. And yes, that 15x multiplier is real — it’s the actual point estimate from the seminal paper on the place premium, calculating the premium between Yemen and the US in terms of real wages, adjusted for all statistically-identifiable characteristics of the worker and the job.

Now, you might contend that the Yemeni bus driver isn’t himself being more productive. After all, he’s doing exactly the same job he was before. The extra income he earns now doesn’t represent any increase in his productivity; it just represents some income he’s siphoned away from Americans, who pay a Yemeni bus driver the wages of an American bus driver, even though he’s plainly less productive than the American who might otherwise drive that bus.

[…]

Take an American bus driver and drop him down in Yemen. Would he still really deserve his American wages in Yemen? Sure, it’s no fault of his own that we’ve forced him to live and work in Yemen instead of the US. But the fact is that the Yemenis he’ll be driving around are less productive than his old American passengers. He used to drive software engineers and Starbucks baristas (you laugh, but baristas save productive doctors and executives plenty of time and money) in the US; in Yemen, he drives shepherds and hotel cooks. Given that the entire productivity of being a bus driver comes from ferrying valuable people around, it’s unsurprising that when the economic productivity of the people he transport drops, his productivity drops as well. So vice-versa, taking a Yemeni bus driver and giving him a bus in the US to drive makes him incredibly more productive. He used to support a small, relatively undeveloped economy; now he supports a much more prosperous economy. His taking that American job has directly lowered the transportation costs for quite a few highly-productive people.

Place premium depends on the source, target, and type of worker

The place premium is generally specified in terms of a source country (where the worker is from), target country (where the worker goes to), and the skill type and level of the worker. For instance, the place premium for construction workers might differ from the place premium for programmers, and both might differ from the place premium for fashion models.

To keep the estimation exercise somewhat manageable, the Clemens/Montenegro/Pritchett paper considers place premia for unskilled workers as a whole, based on the idea that workers who can work in one unskilled labor sector can usually work in another (even though each job has its own training, this is usually on-the-job training rather than relying on extensive sector-specific experience). This allows us to compute overall place premia for unskilled labor for country pairs.

Place premium measurement issues: selection versus treatment

One method of empirical determination of the place premium is comparing PPP-adjusted wages and skill levels for a given occupation in different countries. For instance, we might compare salaries for software engineers in the United States and in India. The problem with this approach is that it fails to adequately account for hard-to-measure human capital differences. Perhaps software engineers command higher wages in the United States than in India because the quality of software engineers is higher. Controlling for years of education may also be insufficient because the relation between years of education and skill level could vary by country.

Another approach is to compare the wages of people who actually did migrate, before and after they migrated. (To control for the effect of experience on wages, we might look at people who actually did migrate, and others who were in the same occupation and earning similar wages but didn’t migrate, and then compare the future trajectory of their wages).

  1. Selection by person: Those who actually move are unrepresentative of the populations of their source country, for reasons that would hold even under open borders. If others in the country were required to make a move, they wouldn’t do so well in the target country. (For more on selection effects of this sort, see this blog post).
  2. Selection by context: The situations in which people move are unrepresentative of somebody just randomly deciding to move. Perhaps, for instance, people move only if they expect to see a huge wage increase. Thus, existing migration might be accompanied by a huge wage increase, but one might not be able to increase one’s wage simply by migrating.
  3. Selection by migration policy: Migration policy is restrictive. It may effectively allow only the forms of migration that are highly likely to be accompanied by wage increases.

There is no way of satisfactorily controlling for all these for most instances of migration. However, lottery-based migration offers a way of partly overcoming the selection problems. Namely, consider people who applied for a migration lottery (such as the US Diversity Visa, or the migration lottery from Tonga to New Zealand) and look at those who won it (and moved) versus those who didn’t. Lotteries are as close to a controlled experiment as we can get from historical data. There is limited research on the use of lotteries as an approximation for unselected migration. For an example, see How Important is Selection? Experimental VS. Non-Experimental Measures of the Income Gains from Migration by David McKenzie, John Gibson, and Steven Stillman.

The Clemens/Montenegro/Pritchett paper uses all the approaches (wage levels in countries controlled for educational attainment, wage gains by movers, wage comparisons for people who win migration lotteries and people who don’t) to obtain estimates of the place premium. The estimates are not identical, but are all sufficiently far off from one that one can say with reasonable confidence that many more workers at the margin could increase their wage simply by moving. Exact numbers, however, are up for dispute.

Place premia in spatially integrated labor markets (aka under open borders)

Place premia can be observed even across spatially integrated labor markets (labor markets that have mutual open borders) but the magnitudes of place premia are smaller, usually no more than 2, in contrast with place premia across closed borders, that can in some cases be 10 or higher. The phenomenon of migration in an integrated labor market resulting in the place premium coming close to 1 is part of what Clemens calls labor market converge.

Place premium and the marginal versus average distinction

Empirical estimates of the place premium are based on migration at the current margin. As discussed above, open borders would likely cause the place premium to fall under 2. But any significant liberalization of migration, even one that falls well short of open borders, would change the values of the place premium, for a variety of reasons. Briefly:

  • Wages in sending countries might change (they could rise because of a decline in the supply of labor, they could fall if the demand for that kind of labor dries up).
  • Wages in receiving countries might change (see the discussion at suppression of wages of natives).

Thus, we cannot use place premium estimates to directly compute the global economic impact of open borders. Place premia do, however, allow us to compute the wage gains to potential migrants from migration liberalization at the currrent margin. As Nathan Smith wrote in his draft paper on the global economic impact of open borders:

What I propose to do here is different, namely, to describe as best I can what a world of open borders would look like. Because current policy is very far from open borders, constructing such an estimate involves large feats of extrapolation, and a heavy reliance on economic theory to sort these out. It is well-known that international migrants can enjoy very large increases in wages and living standards. One of the best contributions to this literature, “The Place Premium: Wage Differences for Identical Workers across the US Border,” by Michael Clements, Claudio E. Montenegro, and Lant Pritchett (December 2008), finds, for selected developing countries, that the “place premium”—the ratio of what workers could earn at home relative to the US—ranges from 1.99 in the Dominican Republic, to 6.25 for India, to 11.92 for Egypt, to 14.85 for Nigeria, to 15.45 for Yemen. But these results apply at the margin and under the status quo. They cannot legitimately be interpreted as indicative of what would happen to all Egyptians, or Nigerians, if controls on migration were removed worldwide. An educated guess about that can only be made in the context of a comprehensive theory of how the world economy works, fitted to the data as well as possible, but able to be solved for equilibrium when policy is changed.

For more discussion on the global economic impact of open borders, see our double world GDP page.

Place premium does not account for narrowing of skills gaps between descendants

The place premium estimation exercises attempt to control for the human capital of migrants. Many of these migrants did not have access to educational opportunities and early work experience opportunities similar to natives of the countries they migrate to, and this lack of access could be part of the reason for the lower human capital of migrants. Due to the place premium, the migrants see a wage increase, but their wages are still less than those of natives due to the human capital gap.

The children of the migrants, however, suffer less of this disadvantage, because their formative educational and work experience years are used more fruitfully. Thus, relative to the counterfactual where the parents had stayed in the home country, the children not only benefit from the place premium but also benefit from higher human capital. Place premium-based calculations may therefore cause us to understate the long-run gains from allowing more migration at the margin. As Carl Shulman writes:

The descendants of migrants from very poor countries will have more education and other human capital
The place premium estimates match workers by education levels, so if the education levels or other skills of workers systematically changed, it would throw off the impact estimates. In particular, levels of education, host country language skills, and health would be higher in later generations after a massive wave of low-skill migration.

Education levels in many developing countries are far below those in developed countries, and migrants close most of the extreme gaps in a single generation (or leapfrog the native population, in some cases).

In the longer term, taking this into account should boost GWP impacts, but with a number of decades of lag for migration, childbearing, and propagation through the work force.

Blog posts and articles

See Open Borders: The Case blog posts on the place premium, including:

The photograph featured in the header of this page is of a man commuting on bicycle across the Netherlands-Belgium border in the village of Putte.

"The Efficient, Egalitarian, Libertarian, Utilitarian Way to Double World GDP" — Bryan Caplan