Monetary Policy and Redistribution in Open Economies

Beyond the standard uninsurable idiosyncratic labor-income risk, households in this model also differ in “real integration” – whether their income comes from the home tradable sector or the nontradable sector – and “financial integration” – whether they can save and borrow in internationally traded securities at the foreign interest rate, or are restricted to domestically traded securities priced off the domestic policy rate (eq. 2-4, Sec. 2.1). Both integration statuses evolve as independent, persistent two-state Markov processes, and are calibrated using Canadian data: the share of households working in tradable sectors is set to Canada’s actual tradable-employment share (35%), and the share of financially integrated households is set to 15%, identified empirically as households holding external assets or U.S.-dollar-denominated savings in the Canadian Financial Monitor Survey. The paper’s stated goal is to build “a laboratory economy that has at play the main mechanisms of monetary transmission of open-economy models, combined with realistic distributions of wealth and marginal propensities to consume … and uneven exposures to external shocks.”

At the aggregate level, an expansionary monetary policy shock behaves much as in the canonical representative-agent open-economy New Keynesian model: lower nominal and real rates raise consumption, the currency depreciates, export demand rises, and firms respond by raising output and prices in both the tradable and nontradable sectors (Sec. 3.2, Fig. 1). The paper explicitly notes this aggregate similarity “is consistent with the findings of Auclert et al. (2020) [the companion paper on the real income channel], who provide general conditions under which household heterogeneity does not lead to an aggregate response significantly different from that of the representative-agent open-economy New Keynesian model” – signaling that this paper’s calibration and shock lie in the region where those neutrality-like conditions approximately hold, so its main contribution lies in the distributional, not the aggregate, response.

Domestic monetary policy shocks generate similar consumption responses for tradable- and nontradable-sector workers, but substantially larger responses for financially non-integrated households (who are directly exposed to the domestic policy rate) than for financially integrated households, and for low-asset (high marginal-propensity-to-consume) households than for high-asset households (Sec. 3.2, Fig. 2). This shows that, for the specific case of a domestic monetary shock, financial (not real) integration and wealth are the dimensions of heterogeneity that matter most for the distribution of the policy’s effects, consistent with the mechanism operating through the domestic interest rate that only unintegrated households are directly exposed to.

An external demand shock (higher foreign demand for the home tradable good) hits real-integration status hardest, since tradable-sector workers directly benefit from the resulting output and wage gains, whereas a foreign monetary policy shock hits financial-integration status hardest, since only financially integrated households directly benefit from lower foreign interest rates (Sec. 4.1, Fig. 3-4). For the foreign monetary shock specifically, the paper finds that “the main source of these heterogeneous responses comes from the differential response of households integrated to international capital markets and those not,” with real integration and wealth producing “more modest differences” by comparison – the clearest evidence in the paper that different types of external shocks are transmitted unevenly through different, shock-specific channels of international integration.

A standard Taylor rule does not mitigate, and can actively exacerbate, uneven exposure to external shocks: for instance, a contractionary external demand shock induces currency depreciation, which the central bank offsets by raising interest rates, and that rate increase falls more heavily on financially non-integrated households (who cannot smooth the shock by accessing foreign asset markets) than on integrated households (Sec. 4.1, Introduction). This is the paper’s first piece of evidence for its broader claim that traditional macro-stabilization policy, designed with aggregate objectives in mind, is not neutral with respect to the distributional consequences of globalization.

Fixed exchange rates amplify the aggregate consumption response to external shocks relative to a Taylor rule – a standard result in representative-agent open-economy New Keynesian models – but they reduce the cross-household dispersion of that response, so that achieving less inequality in consumption responses requires accepting more aggregate (and inflation) instability (Sec. 4.2, Fig. 5). The mechanism is that, following an expansionary external shock (higher foreign demand or lower foreign rates) that would otherwise appreciate the currency, a central bank defending a fixed exchange rate must cut domestic interest rates more sharply than a Taylor-rule central bank would, and this larger domestic rate cut disproportionately benefits households not directly exposed to the shock in the first place – financially non-integrated households and nontradable-sector workers – narrowing the gap between them and the directly-exposed households. The paper frames this explicitly: “monetary authorities might face a trade-off between maintaining aggregate stability and reducing income and consumption inequalities.”

Under dollar-currency pricing – where exporting firms face costs of adjusting prices in foreign, rather than domestic, currency – monetary policy’s ability to stimulate exports through the expenditure-switching channel is weaker, reducing its aggregate stimulus to consumption, and this same weakening of the export channel makes monetary policy’s effects on consumption more unevenly distributed across households than under producer-currency pricing (Sec. 4.3, Fig. 6). The mechanism is that with muted export effects, a monetary expansion boosts nontradable-sector income and consumption relatively more than tradable-sector income and consumption (since exporters gain less), widening the tradable/nontradable consumption gap. The paper frames this as a “corollary” of the standard aggregate result (Devereux and Engel 2003; Mukhin 2018) that dollar pricing blunts monetary stimulus – showing that the international price system shapes not just how effective monetary policy is, but how evenly its effects fall across households.

Higher real or financial integration reduces the aggregate power of domestic monetary policy but increases the aggregate exposure to external shocks, while simultaneously reducing the unevenness of that external exposure across households (Sec. 5, Fig. 7-8). On the domestic side, monetary policy loses its grip on nontradable-sector activity when few households remain in the nontradable sector (high real integration), and loses its direct interest-rate channel when few households remain financially unintegrated – in an economy where almost all agents borrow and save in foreign securities, the paper finds monetary policy’s effect on consumption is “three times smaller” than in the baseline calibration. On the external side, greater real integration raises the aggregate consumption response to external demand shocks, and greater financial integration raises the aggregate response to foreign monetary shocks, but in both cases the dispersion of the response across households shrinks as integration rises, because a larger, more representative share of the population is now directly exposed to the shock rather than a narrow minority absorbing most of its effect.

The paper concludes that the discontents of globalization documented in the literature “might arise, perhaps paradoxically, from international integration’s not being sufficiently generalized,” rather than from deep integration itself (Sec. 6, Conclusion). This follows directly from the result in Q8: an economy with only partial (low) real or financial integration concentrates the consumption impact of external shocks on the relatively small, exposed subgroup of households (tradable-sector workers, or the financially integrated), generating the sharp, visible inequality that motivates policy concern, whereas broader integration – while raising the economy’s aggregate exposure to the global cycle – actually spreads that exposure more evenly across the population.

The paper positions itself alongside de Ferra, Mitman and Romei (2020) and Auclert, Rognlie, Souchier and Straub (2020) as part of a growing literature introducing household heterogeneity into open-economy New Keynesian models, but with a distinct focus: rather than a single dimension of heterogeneity (foreign-currency debt exposure in de Ferra et al.; aggregate market incompleteness in Auclert et al.), this paper’s household heterogeneity is explicitly two-dimensional – real and financial integration – chosen specifically to speak to the classic Mundell-Fleming policy questions from a distributional angle (Introduction, “Related literature”). Unlike Auclert, Rognlie, Souchier and Straub (2021b), whose central finding is a strong real-income channel that can make aggregate depreciations contractionary, this paper’s baseline calibration produces aggregate responses to monetary shocks that closely track the representative-agent benchmark, consistent with the parameter region in which Auclert et al.’s own neutrality conditions approximately hold; the paper’s novel contribution is therefore concentrated in the cross-sectional, rather than aggregate, response to shocks.