VoxEU Column Migration Taxation

Immigration and the welfare state

Immigration policy has become a hot-button issue in both Europe and the US, with questions concerning optimal policy as well as the welfare state dominating discussions. This column revisits the idea of the immigration surplus, exploring a number of possible scenarios in terms of how policymakers should address the challenge. Correctly configuring fiscal policy so as to capture the benefits of both high- and low-skill immigrant (and native) workers is at the heart of optimal policy design and may help to address the swelling anti-immigrant sentiment that continues to exist in many countries today. 

In a televised address aired in 1977, Milton Friedman discussed the differences in US attitudes toward immigration before and after the advent of the welfare state: “free immigration in the same sense as we had it before 1914 is not possible today. Why not? Because it is one thing to have free immigration to jobs, it is another thing to have free immigration to welfare” (Friedman 1977).

The question of finding the optimal immigration policy, and how it interacts with domestic redistribution programmes, has become even more important since Friedman's address. In both Europe and the US, immigration policy has become a ‘hot-button issue’, influencing electoral outcomes (Alesina et al. 2018). 

The anti-immigration sentiment that is germinating in many countries is puzzling given the body of research referencing the ‘immigration surplus’ (Borjas 1995). It has been shown that the fall in the labour income of the native population caused by immigration is more than compensated by the rise in the income of non-labour factors, such as land. The net result is a positive immigration surplus for the native population as a whole. Developed countries have technologies that allow their workers to be much more productive than workers in the developing world. So why wouldn’t developed countries welcome immigrants in what seems to be a win–win policy for both the immigrants and the native population?

Existing research on the immigration surplus generally assumes that the native population is homogeneous. There is no role for redistribution policies, so there is no interaction between income distribution and immigration. What is the optimal immigration policy when income is unequally distributed in the domestic population and the government wants to reduce this inequality? We study this question in our latest paper (Guerreiro et al. 2020).

We consider an economy with high- and low-skill workers. The government designs an optimal redistributive welfare system and supplies public goods. No one can be excluded from the consumption of public goods and the provision of these goods is subject to congestion. To simplify, we abstract this setting from other externalities associated with immigration. The government chooses the immigration policy that delivers the highest welfare to the native population. 

We start by considering a ‘first-best’ setting where the government can charge different lump sum taxes or make different lump sum transfers to each of the four different groups: low-skill workers (native and immigrant) and high-skill workers (native and immigrant). In this setting, free immigration is optimal – there is no role for immigration quotas. Immigrants can enter the country freely, as long as they pay a charge for the congestion they create in the provision of public goods. Immigrants do not receive transfers, but they also pay no domestic taxes, other than the public goods congestion charge.          

Next, we consider two ‘second-best’ settings where the government faces information constraints in distinguishing between low- and high-skill workers (in the style of Mirrlees 1971). In the first setting, the government can discriminate between native and immigrant workers. In the second setting, this discrimination is not possible, so the same schedule of taxes and transfers applies to both groups. 

Free immigration is optimal when the ‘Mirrleesian planner’ can discriminate between native and immigrant workers. The reason for this result is that it is preferable to affect immigration flows using immigrant-specific taxes rather than quotas because taxes generate revenue. 

We consider both cases in which skill types are perfect and imperfect substitutes. If skill types are perfect substitutes (as in the traditional Mirrleesian setting), then the optimal immigration policy is the same as in the first best case: immigrants can enter freely as long as they pay congestion taxes that reflect their use of public goods. 

When low- and high-skill workers are imperfect substitutes, immigration affects the skill premium through general-equilibrium effects. Encouraging high-skill immigration and discouraging low-skill immigration reduces the skill premium, improving the government's ability to redistribute income from high-skill natives to low-skill natives. As a result, the optimal immigration policy is to charge more than the congestion tax to low-skill immigrants and less than the congestion tax to high-skill immigrants. 

The optimal immigration policy is quite different when discriminating between immigrants and natives is infeasible. Free immigration is no longer optimal and there is a role for immigration quotas. Since the government wants to redistribute income toward low-skill native workers, and immigrants and natives must be treated alike, the government bans low-skill immigration. The reason for this ban is that low-skill immigrants add to the pool of workers who receive transfers that need to be financed with distortionary taxes on high-skill workers. The optimal immigration policy may feature free immigration for high-skill workers. However, these workers may choose not to immigrate when heavy taxes are levied on all high-skill workers, natives and immigrants alike.

These striking results hold regardless of whether or not immigration has general-equilibrium effects on the skill premium. However, they are reinforced if immigration affects the skill premium. The reason is that low-skill immigration increases the skill premium, making it harder for the planner to redistribute in favour of low-skill native workers.

We calibrate our model to compute the optimal US immigration policy for the 1994-2008 period. Both in the case of unrestricted taxes, and in the case of Mirrleesian income taxes with discrimination between native workers and immigrants, we find that the optimal total immigration flows are close to those observed in the data. However, the composition of the optimal flows differs from that of actual flows, featuring more high-skill immigration and less low-skill immigration.

In our analysis, we assume that governments care only about the welfare of the current citizens. This cold-hearted assumption is, unfortunately, likely to be consistent with the way in which decisions about immigration policy are made in the real world. Politicians represent the interests of people who already live in the country and vote in elections, not the interests of foreigners who could potentially immigrate. 

Some degree of altruism towards immigrants would go a long way towards transforming the lives of many people around the world.  


Alesina, A, A Miano and S Stantcheva (2018), “Immigration and redistribution”, NBER Working Paper 24733. 

Atkinson, A B and J E Stiglitz (1976), “The design of tax structure: direct versus indirect taxation”, Journal of Public Economics 6 (1-2): 55-75. 

Borjas, G J (1995), “The economic benefits from immigration”, Journal of Economic Perspectives 9: 3-22. 

Diamond, P A and J A Mirrlees (1971), “Optimal taxation and public production i: Production efficiency”, American Economic Review 61 (1): 8–27. 

Friedman, M (1977), “What is America?”, lecture at University of Chicago, 03 October 1977. Transcript published in The Economics of Freedom, 1978.

Guerreiro, J, S Rebelo and P Teles (2020), “What Is the Optimal Immigration Policy? Migration, Jobs, and Welfare”, Journal of Monetary Economics, forthcoming.

2,205 Reads