Why Doesn't Capital Flow from Rich to Poor Countries?

Robert E. Lucas, Jr., American Economic Review Papers and Proceedings 1990. Read from the published article (five pages, no tables or figures). No talk recording exists; written from the paper alone.

Some papers earn their immortality with a model, some with a dataset; this one did it with a single division problem. Lucas takes the most standard assumptions in economics — two countries, same Cobb-Douglas constant-returns technology, homogeneous capital and labor — and asks what they imply about returns to capital. Output per worker in the United States is about fifteen times India’s (Summers–Heston numbers). If technology is common, that gap can only come from capital per worker, and diminishing returns then prices the scarcity: with y = Ax^β, the marginal product of capital is

so the return ratio is the income ratio raised to the power (1−β)/β. With β = 0.4, India’s marginal product of capital comes out fifty-eight times the American one. At that differential, no new investment should occur anywhere in the rich world; every marginal dollar of world savings should be pouring into Bombay. It isn’t, it never has, and — Lucas’s real point — “there is nothing at all delicate” about the failed prediction. The assumptions must be “drastically wrong,” and the question of which one is, he says, “a central question for economic development.” He then auditions four suspects, doing the arithmetic for each.

Suspect one: workers aren’t the same. Correct labor input for human capital using Krueger’s estimates — roughly five Indian workers to one American in productive equivalence — and the effective income gap shrinks from 15× to 3×, cutting the predicted return ratio from 58 to 5. A big dent; not an acquittal. And Lucas immediately points out that this fix, taken alone, creates a new problem: if effective-labor arithmetic equalized capital returns, it would equalize wages for equally skilled workers too — “contrary to the evidence provided by millions of Mexicans” crossing a border precisely because identical labor earns more in the United States.

Suspect two: human capital has external effects. Following his own 1988 growth paper, let the average skill level h multiply everyone’s productivity — y = Ax^β h^γ, with the spillover accruing within the country. Calibrating γ from Denison’s 1909–58 U.S. growth accounting gives γ ≈ 0.36, and rerunning the India comparison with Krueger’s numbers produces a return ratio of 1.04. The paradox evaporates — and Lucas is visibly, honestly startled: “I am surprised how well it works,” since a parameter estimated from fifty years of U.S. time series has no business exactly closing a 1959 India–U.S. cross-section gap. But he flags the load-bearing assumption himself: the calculation requires knowledge spillovers to stop at the border, and “without some real evidence on the scope of these external effects, I do not see how to advance this quantitative discussion any further.”

Suspect three: political risk. International loans need enforcing — this is Eaton–Gersovitz and Bulow–Rogoff territory, and Lucas states the repudiation logic in three sentences: the borrower gains by walking out when the repayment phase begins, and the lender, foreseeing it, never lends. His objection is historical: sovereign risk can’t explain the pre-1945 shortfall, when “a European lending to a borrower in India or the Dutch East Indies could expect his contract to be enforced with exactly the same effectiveness… as a contract with a domestic borrower.” Colonial-era capital markets had no commitment problem, yet capital-labor ratios never converged. Whatever limits flows has been limiting them under two entirely different enforcement regimes.

Suspect four: monopoly. The paper’s least-remembered and most Smithian section. Model the imperial power as a monopsonist over the colony’s labor: choosing capital per worker x to maximize repatriated income, it sets f′(x) = r − x f″(x) — deliberately starving the colony of capital to hold wages down, sustaining a colonial return about 2.5 times the European one. The institutional record — exclusive trading companies, the carving-up of the Third World — fits, and Lucas notes drily that monopoly rents did not lose their appeal at independence: heavy “private taxation” of capital inflows in Indonesia, the Philippines, and the Shah’s Iran deserves “a Smithian skepticism” toward its official justifications.

Why the taxonomy matters is the closing argument, and it is a policy syllogism of unusual bluntness. If the human-capital stories are right, public transfers of capital to poor countries are fully offset — private capital wasn’t flowing because returns weren’t actually higher, so aid-financed capital crowds out one-for-one. If the monopoly story is right, transfers are offset too: “giving goods to a monopolist does not reduce his interest in exploiting potential rents.” Only under the political-risk story does fixing the friction unleash equalizing flows. Fifty years of development policy built on shipping capital goods southward had, on three of four readings of the evidence, been pushing on a string.

For this reading list, the paper is the hinge into block four. It poses the level question — why so little capital heads to poor countries — whose allocation refinement comes next: Gourinchas and Jeanne will show that among developing countries, capital flows toward the slow-growing ones, a pattern that wounds even the repaired neoclassical model; and Ohanian, Restrepo-Echavarría and Wright will take Lucas’s suspects, rename them wedges, and measure which one actually governed postwar flows to Asia and Latin America. It is also quietly connected backward: Lucas’s political-risk discussion is the sovereign-debt block compressed to a paragraph, and his colonial-enforcement observation remains one of the best empirical challenges the reputation-versus-sanctions debate ever received.