Financial Structure and International Trade

The paper compares a “primitive” financial structure, in which a single real bond can be traded only after each period’s productivity shocks are already known, with a “sophisticated” complete-markets structure in which Arrow securities – one for every possible realization of the four country/project productivity shocks – can be traded before the shocks are known. The ex post bond structure “can only fulfill” the function of separating the timing of consumption from the timing of production, since “trading in these bonds occurs only after the resolution of uncertainty, and the payoffs are not state-contingent”; the Arrow-securities structure additionally “allows risk-sharing.”

Each of two countries produces a single nonstorable consumption good from two projects – one yielding output immediately, the other only in the next period – using labor as the only input, and project output is subject to i.i.d. country- and project-specific productivity shocks. Under the ex post bond structure, “the international bond market has served to equalize the ratio of marginal products of labor between countries” (equation 16); the world (intertemporal) interest rate is the sole channel through which a productivity shock in one country can affect the other country’s agents.

With a complete Arrow-securities market, domestic and foreign agents’ marginal utilities of consumption are equalized state by state, so “for any given state of the world, the consumptions of the foreign and domestic agents are identical” – full insurance against each other’s idiosyncratic productivity shocks. Labor effort, by contrast, is not equalized across countries even though the ratio of marginal products of labor is, because “one’s productivity shock is a nontraded input, and laborers are immobile across countries” – the agent with the higher productivity parameter works harder but does not consume more than the agent with the lower parameter.

Under complete Arrow-securities markets, an agent would prefer, ex post, that the other agent’s productivity parameters be higher than his own – the opposite of what holds under the primitive ex post-bond structure, where an agent prefers his own productivity to be higher. This “ex post preference reversal first noted by Stockman and Delias (1986)” arises because, once the agents have pooled risk via Arrow securities, a high realization of the other agent’s productivity raises the net transfer the first agent receives under their implicit insurance contract.

Moving to the more complete financial structure lowers the variance of domestic consumption, because wealth becomes less tied to domestic output and more tied to foreign output via risk-pooling, but its effect on the variance of labor effort and the trade balance is ambiguous. Labor-effort variance can rise “if, as the evidence seems to indicate, labor effort is procyclical,” because the substitution effect of a productivity-driven real-wage change is no longer damped by the negative income effect that risk-pooling has removed; the trade balance’s variance is pulled in opposite directions by the resulting rise in output variance (which raises it) and the fall in consumption variance (which lowers it).

Under the primitive (ex post bond) structure the trade balance equals saving, so its present value is always zero and – absent initial debt – it coincides with the current account, which must also be zero in present-value terms; under the sophisticated (Arrow-securities) structure this equivalence breaks down, because the trade balance no longer nets out the ex post state-contingent transfers the agents have contracted into. As a result, “it is quite possible here for a country to have a trade deficit (or surplus) in both periods if its productivity shocks are low (or high) relative to those abroad,” so “the more sophisticated financial structure would seem to allow for more persistent deficits in the trade balance.”

Cole and Obstfeld (1991) argue that for developed countries with small, fairly correlated output fluctuations, the welfare gains or losses from greater international risk-sharing “are likely to be small”; Baxter and Crucini (1992), using a quantitative two-country real-business-cycle model, qualitatively confirm the paper’s predictions but find that the size of the effects “depends crucially on the fine details of the stochastic process governing the fluctuations in the productivity parameters” – the structure of financial markets matters much more when productivity shocks are treated as difference-stationary (permanent) than when they are trend-stationary (persistent but transitory). Baxter and Crucini’s quantitative model also reproduces a sign switch implied by the paper’s own model: cross-country consumption correlation rises with market sophistication while cross-country output and labor-effort correlations fall, with labor effort’s correlation becoming negative in every case they consider – a result the paper flags as “substantially at variance with the generally positive cross-country correlations of labor effort that we see in the data.”

This document is explicitly not the original 1988 International Economic Review article: its own cover note states it is “a revised version of a paper published in the International Economic Review (May 1988, vol. 29, no. 2, pp. 237-59): ‘Financial Structure and International Trade’ by Harold Cole,” republished in 1993 as a Federal Reserve Bank of Minneapolis Quarterly Review piece “with the permission of the University of Pennsylvania.” The restyled text itself cites the original “Cole 1988” as a separate source for some derivations (footnote 5’s claim about the direction of labor-effort changes is “verified” “in Cole 1988” rather than here) and discusses post-1988 quantitative work (Cole and Obstfeld 1991; Baxter and Crucini 1992) that could not have appeared in the 1988 original – confirming this is a restyled, updated exposition of the same model and results rather than a verbatim reprint. The byline on this document lists the author’s affiliation as “Economist, Research Department, Federal Reserve Bank of Minneapolis,” which is this 1993 republication’s affiliation; the original 1988 IER paper’s affiliation for the author is not stated in this text and is not independently confirmed in this record.

The paper closes by flagging two unresolved questions: how sensitive the results are to the assumed domestic financial structure (whether there are greater gains from international risk-sharing when domestic financial markets are themselves more primitive), and what role international financial markets play “in risk-sharing and in ensuring an efficient allocation of capital across countries.” It notes international financial markets may matter more when countries are very dissimilar (developed versus developing, fast- versus slow-growing) than when they are similar.