The Economics of Immigration: The Best Books, in Order
Immigration is one of the few topics where serious economists working from overlapping evidence reach genuinely opposed conclusions, and this path is built to carry that disagreement rather than to resolve it. It starts with a neutral textbook and the long-run American record, then reads George Borjas at full technical strength, then the economists who read the same natural experiments — the Mariel boatlift above all — and conclude the opposite, then the political and philosophical arguments that economics cannot settle on its own.
The framework and the record
BeginnerLearn the standard apparatus — labour supply and demand, complementarity versus substitution, the fiscal accounting of a migrant over a lifetime — and get the long-run American evidence on assimilation and mobility before meeting any of the arguments.
▸ Study plan for this stage
Pace: Two to two and a half months. Bansak's The Economics of Immigration is 356 pages and is the only genuine textbook on the path — Bansak, Simpson and Zavodny walking through the theory, the measurement problems and the main empirical findings without taking a side. Give it four to five weeks at 25 to
- The basic labour-market model: an increase in labour supply lowers the wage of workers who are substitutes and raises the productivity of those who are complements, so the whole empirical question is who is which.
- Complementarity versus substitution as the pivot of the entire subject. Whether a new arrival competes with or complements a given native worker depends on skill, language, occupation and time horizon, and it is measured rather than assumed.
- The short run and the long run. Capital adjusts, firms enter, and the occupational structure shifts, so a wage effect measured over two years and one measured over twenty are answering different questions.
- Fiscal accounting over a lifetime rather than a year: a migrant arriving as a working adult has had their education paid for elsewhere, and the sign of the fiscal effect depends heavily on the accounting frame chosen.
- Measurement problems Bansak sets out — how immigrant status is recorded, how undocumented populations are estimated, and how skill is proxied — all of which propagate into every headline finding.
- Borjas's redistribution claim: even if the aggregate gain from immigration is positive, it is small, and the distributional transfer from competing workers to employers and immigrants is large relative to it.
- The immigration surplus as a quantity that can be estimated, and the arithmetic showing why a small aggregate number can sit on top of very large transfers.
- Abramitzky and Boustan's intergenerational finding, and its most important feature: it uses linked individual records across generations, which is a different and stronger design than comparing group averages at one moment.
- Under what conditions does an increase in labour supply lower wages, and under what conditions does it not? State the model rather than the conclusion.
- Distinguish complementarity from substitution and give a concrete pair of occupations for each.
- How is a migrant's lifetime fiscal contribution calculated, and which accounting choices most affect the sign of the answer?
- State Borjas's redistribution argument as precisely as you can, including the relative sizes of the surplus and the transfer.
- What does the linked-records design in Streets of Gold allow Abramitzky and Boustan to see that cross-sectional comparisons cannot?
- Which of the disagreements you can already anticipate are about models and which are about data? Make the list now; you will check it at the end.
- Draw the standard labour-market diagram for a supply increase and then redraw it with capital adjusting and with complementary labour. The three pictures are the whole theoretical dispute.
- Take a published estimate of the fiscal impact of immigration and identify its accounting frame — annual or lifetime, individual or household, with or without descendants. Then find an estimate with a different frame and reconcile the two.
- Work through Borjas's surplus arithmetic in We wanted workers with your own numbers, varying the elasticity. The sensitivity of the answer is his point and doing it once makes it stick.
- Write down, before stage two, what evidence would convince you that immigration lowers native wages substantially. Date it. You will test it against both sides.
- Using Abramitzky and Boustan, state one claim about historical immigration you previously believed that the linked records contradict.
Next up: Borjas's popular book gestures at a technical argument about how the comparison group is defined, and the next stage is that argument at full strength.
The one genuine textbook on this list: Bansak, Simpson and Zavodny walk through the theory, the measurement problems and the main empirical findings without taking a side. Start here so you can tell later which disagreements are about models and which are about data.
Borjas's accessible statement of the sceptical position — that immigration mainly redistributes income from competing native workers to employers and to immigrants themselves. Read it early, because he is the scholar every book in stage three is arguing with, and he writes his own case better than his critics summarise it.
Abramitzky and Boustan link millions of historical census records to track immigrant families across generations, and find that children of poor immigrants out-climb children of comparable natives — then and now. The strongest long-run evidence on the path, and a direct empirical answer to the claim that today's immigration is different.
The sceptical case at full strength
IntermediateRead Borjas as an economist rather than as a polemicist. His claim is specific: the wage effect on competing low-skilled workers is real and has been obscured by how other researchers define the comparison group. You should be able to state that argument precisely enough to explain what evidence would refute it.
▸ Study plan for this stage
Pace: Two months for 583 pages, and this is the most demanding stage on the path relative to its length. Immigration Economics is 295 pages of technical monograph — the national skill-cell approach, the surplus calculation, and the critique of local-labour-market research designs — and it assumes comfort
- The national skill-cell approach: define labour markets by education and experience across the whole country rather than by city, on the argument that native workers move in response to immigration and thereby diffuse the effect.
- Why that choice is the crux. If natives relocate away from immigrant-receiving cities, a city-level comparison understates the wage effect by construction, and Borjas's design is built to close that channel.
- The critique of local-labour-market designs, including the specific objection to how comparison cities are chosen and to what an unaffected control city would even mean.
- Borjas's Mariel reanalysis, restricting attention to the group he identifies as most directly competing, and the argument that aggregating over a wider population averages the effect away.
- The immigration surplus calculation in its formal version, and why the aggregate gain is small under most plausible elasticities while the transfers are large.
- Selection in Heaven's Door: which people migrate from which countries depends on the return to skill at home relative to the destination, which generates testable predictions about cohort quality.
- The declining relative skills argument for successive American cohorts, and how much of it turns on the choice of comparison group and time window.
- The points-system case as the policy consequence of selection theory, and what it assumes about a government's ability to identify productive migrants in advance.
- Explain the national skill-cell approach and say precisely what identifying assumption it relies on.
- What is Borjas's objection to city-level research designs, and what evidence does he give that natives relocate in response to immigration?
- Set out his Mariel reanalysis: which subgroup does he examine, why that one, and what happens to the estimate when the group is widened?
- Walk through the surplus calculation and say what drives the size of the aggregate gain relative to the transfers.
- What determines who migrates in Heaven's Door's selection model, and what does the model predict about cohorts from high-inequality origin countries?
- State exactly what evidence would refute Borjas's central wage claim. If you cannot, you have not yet read him closely enough.
- Write out the skill-cell specification and the city-level specification side by side as equations, and mark the term that differs. That single line is what the field has been arguing about for thirty years.
- Reconstruct the Mariel comparison with progressively wider definitions of the affected group and note how the estimate moves. Even done roughly, this shows why subgroup choice is the whole dispute.
- Recompute the immigration surplus under three different elasticity assumptions and present the range rather than a point estimate.
- Take the selection model in Heaven's Door and predict which origin countries should send positively selected migrants. Then check the prediction against published cohort data.
- Summarise Borjas's argument in one page as if you agreed with it entirely, using no hedging language. Doing this before stage three is the only way to give the other side a fair hearing.
Next up: The next stage reads the same natural experiments Borjas has just reanalysed and concludes the opposite, which is why it should follow immediately.
The technical monograph — the national skill-cell approach, the surplus calculation, the critique of local-labour-market designs. Demanding and worth it, because the disagreement with David Card and his successors is a disagreement about identification strategy, and you cannot judge it from the popular books.
His earlier and more policy-facing book, on selection, the declining relative skills of successive American immigrant cohorts, and the case for a points system. Read it after the monograph as the applied version, and note how much of its 1999 data has since been revisited.
The other side of the same evidence
IntermediateNow read the economists who look at the Mariel boatlift and the other natural experiments and find little or no wage damage, and who emphasise complementarity, entrepreneurship and the enormous productivity gain from moving a worker across a border. Hold both positions at once — that is the honest state of the field, not a failure to decide.
▸ Study plan for this stage
Pace: Two and a half months for 909 pages, and it should begin as soon as you close Borjas rather than after a break. Banerjee and Duflo's Good Economics for Hard Times is 416 pages and opens with immigration; three to four weeks, and read the opening chapters twice, because they set out the Card-side rea
- The Mariel boatlift as the field's canonical natural experiment: a sudden large labour supply shock to one city, and the finding on the Card-side reading that measurable wage damage to natives was small.
- Why the same episode supports opposite conclusions. The disagreement is about which workers to compare, over what window, and against which control, not about what happened in Miami.
- Banerjee and Duflo's broader claim that labour markets are sticky in ways the textbook model does not capture — people do not move readily, and firms adjust technology and output mix rather than only wages.
- Firm and technology adjustment as the mechanism absorbing a supply shock, which is why a wage effect can be small without the model being wrong.
- Kerr's evidence on high-skilled migration: immigrant contributions to patenting and firm founding in the United States, and the international competition among states to attract that group.
- The distinct economics of skilled migration. Complementarity is stronger, the fiscal contribution is unambiguous, and the origin-country cost is a real question in a way it is not for low-skilled movement.
- The place premium: the same worker with the same skills earns several times more in a rich country, which implies enormous global efficiency gains from movement and is the empirical basis of Caplan's case.
- Where Caplan's argument stops being economics. The place premium is a measured quantity; the claim that policy should follow it depends on premises about states and obligations that stage four takes up directly.
- Describe the Mariel boatlift and set out both readings of it. Which specific analytical choices separate them?
- What do Banerjee and Duflo mean by saying labour does not behave like a commodity, and what evidence do they offer?
- Through what channels other than wages can a labour market absorb a supply shock? Name three and say how each would be detected.
- What does Kerr's evidence show about high-skilled migration, and why does it not settle anything about low-skilled migration?
- What is the place premium, how is it measured, and what does it imply about global efficiency?
- Having now read both sides, which parts of the disagreement are empirical and which are about identification strategy? Answer with the list you made in stage one.
- Build a table of the Mariel estimates: for each analysis, the comparison group, the control, the window and the result. The table explains the dispute better than any summary of it.
- Take the Borjas one-page summary you wrote in stage two and write the Banerjee and Duflo answer to it, point by point. Where you cannot answer a point, mark it — those are the field's genuinely open questions.
- Separate every claim you have met on this path into a skilled column and an unskilled column using Kerr's distinction. Most public arguments collapse when you do this.
- Check one of Caplan and Weinersmith's footnoted claims against its cited source. The book invites this and the exercise is a good test of how footnoted rigour and rhetorical framing coexist.
- Revisit the dated note you wrote in stage one about what would convince you. Has either side supplied it?
Next up: The wage question is now as settled as it is going to get, which leaves the questions economics was never going to answer on its own.
Banerjee and Duflo open with immigration and set out the Card-side reading of the boatlift and related episodes, arguing that labour does not behave like a simple commodity in these markets. The clearest short statement of the case against Borjas, written by two Nobel laureates who name the disagreement rather than papering over it.
Kerr shifts the question from wage competition to the global race for high-skilled workers, with hard evidence on immigrant patenting and firm founding in the United States. Read it because the skilled and unskilled cases have different economics, and most public argument runs them together.
Caplan and Weinersmith's illustrated case for radically freer movement, built on the place-premium research showing that the same worker earns several times more simply by relocating. Deliberately the most extreme position on the path, drawn as a comic and rigorously footnoted — read it as the argument at its limit.
What economics does not settle
IntermediateWage effects are not the whole question. Read the arguments about what open movement does to the countries people leave, to institutions in the countries they enter, and to what a state may legitimately do at its border — and be able to say which of those claims are empirical and which are not.
▸ Study plan for this stage
Pace: Two and a half months for 825 pages, and this is where the reading changes register from estimation to argument. Collier's Exodus is 319 pages, three weeks; he argues that migration's effects on origin countries and on host-country social trust are systematically under-weighted, and that moderate mi
- The origin-country question. Emigration removes workers, remits income and changes incentives to acquire skills, and Collier argues the net effect for the poorest origin countries is under-examined.
- Brain drain and its offsets — remittances, return migration and the incentive to educate — which is a live empirical dispute rather than a settled loss.
- Collier's social-trust argument, and the fact that it rests on a specific and contested body of research; naming that dependency is part of reading him accurately.
- The moderate-migration conclusion and its structure: Collier's case is that the relationship is non-linear, so a defensible position sits between open and closed rather than at either end.
- Why refugees are a different case. The legal regime, the obligations involved and the alternatives available all differ from those governing labour migration, and Refuge is built on that difference.
- The Betts and Collier proposal: economic zones near origin countries where displaced people may legally work, replacing indefinite camp encampment with employment.
- Miller's qualified right to exclude: states may control admission because political communities have legitimate interests in their own composition and public culture, subject to duties toward the desperate.
- The empirical and normative boundary. Whether immigration lowers wages is empirical; whether a state may exclude someone whose life would improve by entering is not, and no dataset resolves it.
- What are the mechanisms by which emigration affects an origin country, and which direction does each push?
- State Collier's social-trust claim and identify the research it rests on. What would refute it?
- Why is Collier's conclusion moderate rather than restrictionist? Reconstruct the shape of the relationship he posits.
- How does refugee movement differ from labour migration in law, in ethics and in available alternatives?
- Set out the Betts and Collier proposal and the strongest objection to it.
- What is Miller's argument for a qualified right to exclude, and where does he say that right runs out?
- Take one origin country with high emigration and assemble the actual figures for remittances, return rates and skilled emigration. Then decide whether Collier's concern is borne out there.
- Split twenty claims from this stage into empirical and normative columns. The exercise is harder than it sounds and it is the single most useful thing on the path.
- Write Miller's argument and Caplan's argument as two premise-and-conclusion chains and find the premise on which they actually differ. It will not be an empirical one.
- Read the Betts and Collier proposal and write the objection a refugee-rights organisation would make. Then write their answer to it.
- Take one policy you support and identify which of its justifications are empirical and which normative. If you cannot separate them, neither can the argument you are having about it.
Next up: These arguments do not happen in seminars, so the final stage puts them back into the political contest where they are actually decided.
Collier argues that migration's effects on origin countries and on host-country social trust are systematically under-weighted, and that moderate migration is better than either extreme. Controversial among migration economists, several of whom dispute his social-trust evidence directly — which is why it belongs on the path with that stated.
Betts and Collier on refugees specifically, and their case that camps should be replaced by economic zones where displaced people can legally work. Read it after 'Exodus' because refugee movement is governed by different law and different ethics from labour migration, and conflating them is the most common error in this debate.
A political philosopher's defence of a state's qualified right to control immigration, taking the cosmopolitan arguments seriously before rejecting them. Included as the honest philosophical opposite of Caplan, so the normative disagreement is visible rather than smuggled into an empirical one.
Politics, numbers and the long view
IntermediateFinish with three books that put the economics into the political argument it actually lives in — assimilation policy, population growth as a national strategy, and a reminder that the current debate is a rerun of an older one with the same shape.
▸ Study plan for this stage
Pace: Two and a half months for 954 pages, and the first two are best read back to back. Salam's Melting pot or civil war? is 213 pages and takes two weeks; it is a restrictionist argument written by the son of Bangladeshi immigrants and focused on the descendants of low-wage migrants rather than on migra
- Salam's shift of the subject from migrants to their descendants, which changes the relevant evidence from wage effects to intergenerational mobility, education and residential segregation.
- Assimilation as the load-bearing variable in the political argument, and the reason Salam's case survives even if the wage evidence goes entirely against him.
- Yglesias's population-strategy framing: immigration as an input to national scale, with consequences for defence, innovation and the tax base rather than for the labour market alone.
- The shared premise both authors hold, namely that successful integration is what makes a given level of immigration workable, which is why their disagreement is about capacity rather than about principle.
- Housing, transport and public services as the binding constraints on any high-immigration strategy, which Yglesias treats as the real policy work and which connects directly to the urban-economics literature.
- Simon's demonstration that the fiscal and wage questions were being asked and answered with much the same methods decades ago, and that the answers he reached were closer to the modern consensus than his contemporaries expected.
- How little the question set has changed. The persistence of the same debates across thirty-five years is itself evidence about what kind of question this is.
- Reading old empirical work honestly: Simon's data are a period record, his methods are of their time, and both are worth knowing precisely because the current literature is still answering him.
- What is Salam's argument, and why does he focus on the second generation rather than on immigrants themselves?
- Which of Salam's claims would be unaffected if the wage evidence went entirely against him? That set is his real case.
- What is Yglesias's strategic argument for a much larger population, and what does it assume about housing and public services?
- Where do Salam and Yglesias actually agree? Name three shared premises.
- Which of Simon's 1989 findings have held up and which have been overturned by later work?
- Looking at the whole path, which questions have moved in thirty-five years and which have not?
- Write the strongest one-page case for lower immigration and the strongest one-page case for much higher immigration, using only evidence from this path. Then mark every place where the two rely on the same finding.
- Take Yglesias's proposal and cost out its housing implication for one metropolitan area. The number is the reason this is an urban-policy argument as much as an immigration one.
- Pick three of Simon's empirical claims and find the current estimate of each. The comparison is a compressed history of the field.
- Return to your dated note from stage one about what would convince you, and answer it in writing now. Keeping both versions is the honest record of the path.
- Build a one-page map of every book here showing who agrees with whom and on what grounds, marking each disagreement as about data, about identification, or about values. That map is what you actually keep.
Next up: The path ends without a verdict because the field has not reached one, and being able to state the disagreement precisely is the achievable outcome.
A restrictionist argument written by the son of Bangladeshi immigrants, focused on the descendants of low-wage migrants rather than on migrants themselves. Read it as the most serious version of the political case for lower numbers, and notice how much of it rests on assimilation rather than on wages.
The opposite proposition — that the United States should triple its population, largely through immigration, as a matter of national strategy. Read it directly after Salam; the two share more premises about integration than either would admit.
Simon's 1989 study assembled much of the fiscal and wage evidence that the modern debate still runs on, and reached broadly pro-immigration conclusions decades before the current literature. Read it last as the historical anchor: the questions have barely changed, which is worth knowing before assuming this one will be settled soon.
Discussion
Keep reading
Paths that share books, cover the same subject, or open a related topic.
