Using security-level holdings of globally diversified mutual funds, this paper applies a granular instrumental variable approach to identify currency demand shocks at the bilateral exchange rate level. These shocks cause significant exchange rate movements in both emerging market (EM) and advanced economy (AE) currencies, consistent with the view that FX markets are inelastic—or “shallow.” The estimates show that for flows of the same size, EM currencies respond about nine times more than AE currencies, underscoring their much lower market depth. The results further reveal state dependence along two dimensions. First, during periods of high expected exchange rate volatility, FX markets are highly inelastic, whereas in tranquil periods they are nearly perfectly elastic. Second, in episodes of mutual fund outflows, FX markets are shallower than during inflows. The first property holds for both EM and AE currencies, while the second is specific to EMs. A quantitative small open economy model with segmented FX markets featuring limited risk-bearing capacity and balance sheet constraints rationalizes these empirical findings.

We investigate the optimal time-consistent use of foreign exchange interventions (FXI) in a small open economy model driven by endowment and portfolio flow shocks, with endogenous FX market depth and a lower bound constraint on FX reserves. In a competitive equilibrium, large capital flows increase conditional exchange rate volatility and make FX markets more shallow. Unlike in the unconstrained case, the central bank's optimal interventions are not solely targeted at offsetting inefficient fluctuations in the UIP premium but also incorporate a forward-looking element due to the risk of depleting reserves. We show that this environment leads to optimal time-consistent FXI policy that is highly state-dependent. FX sales are more effective than FX purchases, and the policy may respond less or more than one-for-one to capital outflows, depending on their size and the economy's net foreign assets position. Adopting the policy delivers sizable welfare gains, significantly exceeding those from a simple rule directed at stabilizing current capital outflows, but only if the initial level of FX reserves is sufficiently high.

This paper introduces foreign investment in a small open economy model and studies its implications for foreign exchange intervention (FXI) policy. Due to aggregate demand externalities, overborrowing and unemployment may arise in the economy. In this case, FXI can increase employment by improving the net foreign asset position ex-ante, but comes with the cost of crowding out higher-yielding foreign investment. As a result, the optimal policy problem features two local optima: one with low FXI and positive foreign investment holdings, and one with high FXI and no foreign investment. Whether the former or latter constitutes the global optimum crucially depends on the risk/return profile of foreign investment.