North-South lending and endogenous domestic capital market inefficiencies
📄 Summarized from the full manuscript · Human-reviewed for faithfulness before publication
In brief
Why do richer developing countries borrow more, not less, from private creditors? Gertler and Rogoff build a model where lenders cannot observe how a borrower actually invests, so financing terms depend on the borrower's own wealth -- poorer borrowers face bigger spreads and invest less than the first-best. Extended to two countries, this produces a "siphoning" effect: richer-country wealth gains bid up the world interest rate and crowd out investment in poorer countries, and a wealth transfer from poor to rich can cost the poor country more than its face value.
What this paper finds — and why it matters
This 1990 Journal of Monetary Economics paper by Mark Gertler and Kenneth Rogoff builds a model of North-South lending in which information asymmetries between individual borrowers and world capital markets, not sovereign risk, explain why marginal products of capital and borrowing spreads are systematically higher in poorer countries. The source text used here is the authors’ earlier NBER working paper draft, “Developing Country Borrowing and Domestic Wealth” (NBER WP 2887, March 1989) – NBER’s own record lists this working paper as published, under the JME title used here, so the model and results below are the same ones that appeared in the journal. The paper first documents that across seventy developing countries in 1980, a one-percentage-point increase in per capita GNP is associated with more than a one-percent increase in per capita external debt to private creditors, a relationship that holds within Africa, Asia, and Latin America separately. It then builds a small-country model in which a borrower’s choice of investment is unobservable to lenders (moral hazard), so that the optimal contract leaves investment below the first-best level by an amount that shrinks as the borrower’s own wealth W rises – producing the observed positive wealth-debt relationship as a by-product of the financing friction itself, not of any explicit sovereign-risk channel. In a two-country general-equilibrium extension, the model yields a “siphoning” effect: an increase in the wealth of rich-country entrepreneurs widens their own financing terms, raises their demand for investment funds, drives up the world interest rate, and thereby worsens the financing terms facing – and crowds out investment by – entrepreneurs in the poor country, even though the poor country’s own endowments are unchanged. A pure wealth transfer from poor-country citizens to rich-country lenders can, for the same reason, cost the poor country more than the face value of the transfer, a version of the classical Keynes-Ohlin transfer-problem debate arising here from the distribution of wealth across asymmetrically-informed borrowers rather than from terms-of-trade effects. The authors stress that government intervention is not automatically welfare-improving: it would have to be run by an agency facing the same informational constraints as private lenders, but more efficiently, to raise welfare.
Summary of a classic paper, AI-assisted and human-reviewed. See the linked original for the authoritative claims and full conditions.
Questions & answers
Q1. What empirical pattern motivates the paper, and how robust is it?
Across seventy developing countries using 1980 data, a one-percentage-point increase in per capita GNP is associated with more than a proportionate (greater than one-percent) increase in per capita external debt to private creditors (Sec. II, p. 2). Regressing log external debt per capita on log GNP per capita gives a coefficient of 1.05 (s.e. 0.08) for total external debt across all seventy countries, and a larger coefficient of 1.60 (s.e. 0.12) when the sample is restricted to debt owed to private lenders only (Table 2) – the authors note that official/multilateral debt is effectively more senior in practice despite being junior in principle, which biases the all-debt coefficient down. The relationship holds separately within Sub-Saharan Africa, Latin America and the Caribbean, and Asia, and is robust to excluding Nepal (whose zero private debt per capita would otherwise bias the logged regression) and to using 1986 data in the appendix.
Q2. What is the information friction at the center of the small-country model, and what does it imply?
Each individual has access to a risky investment project yielding θ units of output with probability π(k) (increasing, concave in investment k) and zero otherwise; lenders can observe the realized output but cannot observe or verify how much the borrower actually invested, so contracts can only be conditioned on output, not on k itself (Sec. III, pp. 5-6). Under full information the borrower would invest at the first-best level k* where the expected marginal product equals the world interest rate, π′(k*)θ = r (eq. 6). Under asymmetric information, the incentive-compatible contract instead sets k via the “IC curve” π′(k)[θ − Z^g] = r together with the lenders’-zero-profit “MR curve” Z^g = r(k − W)/π(k) (eqs. 9-10); the resulting investment level is strictly below the first-best, k < k*, whenever wealth W is below k* (Proposition 1/Corollary, pp. 8-9, 19-20), because the borrower’s payoff in the bad state is pinned to zero (the non-negativity constraint Z^b ≥ 0 binds), which raises her effective cost of investing beyond its true social cost.
Q3. How does the model generate the observed wealth-debt relationship?
A rise in a borrower’s own wealth W shifts the MR curve down (for given investment, a wealthier borrower needs to borrow less and so can be promised less-punitive contract terms), which raises equilibrium investment k and lowers the good-state repayment Z^g (Sec. III, p. 9, eq. 11 and surrounding discussion). Because external borrowing x = k − W, the paper shows dx/dW is strictly between 0 and 1: external borrowing rises less than one-for-one with wealth, but it does rise, and the model’s comparative statics are “completely consistent” with the paper’s cross-country regressions linking debt and GNP per capita (Sec. III, pp. 10-11). A corresponding result is that the spread between the marginal product of capital and the riskless world rate is predicted to be larger in poorer (lower-W) countries, even though the riskless rate itself is equalized by a perfectly integrated world capital market.
Q4. How is the model extended to two countries, and what is the “siphoning” effect?
In the two-country extension, a rich country R and a poor country P each have a population of entrepreneurs and lenders, with poor-country entrepreneurs endowed with strictly less wealth than rich-country entrepreneurs (W^P < W^R); the world interest rate r is now endogenous, determined jointly with each country’s incentive-constrained investment level k^P, k^R by five equations (eqs. 15-19) represented graphically by a downward-sloping “WW” world-capital-supply curve and an upward-sloping “ρρ” curve (Sec. IV, pp. 12-13, Figure 2). Because of the information friction, k^P < k^R even though technologies are identical. The “siphoning” effect arises when a shock raises rich-country entrepreneurs’ wealth without changing total world wealth: financing terms improve for rich-country borrowers, their demand for investment funds rises, this bids up the world interest rate, and the rise in r worsens poor-country entrepreneurs’ financing terms, so that investment is “siphoned” away from the poor country toward the rich one even though the poor country’s own fundamentals are unchanged (Sec. IV, pp. 14-15).
Q5. What does the model imply about a pure wealth transfer between countries, and how does this relate to the classical transfer problem?
A transfer of wealth from poor-country citizens to rich-country lenders raises k^R and lowers k^P (an upward rotation of the ρρ curve); the direct cost to the poor country is the face value of the transfer, but because the fall in k^P also reduces the project rents earned by poor-country entrepreneurs, the paper shows the indirect costs “always magnify the costs of the transfer” when the poor country is a sufficiently small player in world markets (so the interest-rate effect is small) (Sec. IV, pp. 15-16). The authors explicitly connect this to the Keynes-Ohlin debate over whether the cost of paying a foreign debt can exceed its face value: here the mechanism is intertemporal (via information-constrained investment and the distribution of wealth), not the classical mechanism of adverse terms-of-trade movements from contemporaneous trade (Sec. IV, p. 15).
Q6. Does the paper find a role for government intervention to correct the financing friction?
No – the paper is explicit that although the equilibrium is distorted relative to the first-best, “government intervention cannot be Pareto improving” unless the government agency is more efficient than private lenders at overcoming the same informational problem it would face (Sec. V, p. 18). The friction is modeled as a genuine information asymmetry, not a policy-created distortion, so there is no presumption that a government lender or regulator can simply legislate its way around it.
Q7. What alternative explanation does the paper offer for the Feldstein-Horioka savings-investment correlation puzzle?
The paper notes in its conclusion that in a world of perfect information, a small country’s increase in savings (from a higher endowment) would simply flow abroad; but when borrowing is subject to the paper’s informational friction, a large part of an increase in a country’s savings is instead invested domestically – offering an alternative, non-market-segmentation explanation for the empirically high correlation between national savings and investment rates that Feldstein and Horioka (1980) had used to argue for imperfectly integrated world capital markets (Sec. V, p. 18).
Q8. How does a world productivity shock affect the distribution of investment between the two countries?
A shock to world productivity θ raises per-capita world output and the world interest rate but has no effect on the relative distribution of investment capital between the rich and poor country, because θ enters both countries’ equations symmetrically and “factors out of both sides” of the equilibrium condition (eq. 20) (Sec. IV, p. 17). This contrasts with the wealth-redistribution shocks analyzed above, which do have first-order distributional effects even though they leave total world wealth unchanged.
Key terms in this paper
Definitions below follow the paper's own usage.
- Siphoning effect
- The paper's central two-country mechanism (Sec. IV): because investment is constrained by individual borrowers' observable wealth under asymmetric information, a wealth increase among rich-country entrepreneurs improves their own financing terms, raises their demand for funds, bids up the world interest rate, and thereby worsens financing terms for -- and draws investment capital away from -- poor-country entrepreneurs, even though the poor country's own endowments have not changed.
- Incentive-constrained investment (IC and MR curves)
- The paper's formal characterization (Sec. III, eqs. 9-10) of the optimal lending contract under moral hazard: the IC curve, π′(k)[θ − Z^g] = r, states the borrower's incentive-compatibility condition for her choice of investment k given the good-state repayment Z^g; the MR curve, Z^g = r(k − W)/π(k), is lenders' zero-profit condition. Their intersection pins down equilibrium investment k, which lies strictly below the full-information optimum k* whenever wealth W is less than k*.
- Wealth-transfer magnification (the information-based transfer problem)
- The paper's result (Sec. IV) that a pure wealth transfer from a poor country to a rich country costs the poor country more than the transfer's face value, because the resulting fall in poor-country entrepreneurs' wealth also tightens their financing constraint and reduces the project rents they earn -- an intertemporal analog of the classical Keynes-Ohlin transfer problem, operating through the distribution of wealth across asymmetrically-informed borrowers rather than through contemporaneous terms-of-trade movements.