We propose a new tradeoff for sovereign borrowing. Borrowing in local currency insulates sovereigns from default but increases the severity of fire sales in bad times, as local-currency bonds are endogenously held by intermediaries more vulnerable to fire-selling. We document two facts. Intermediaries with open-ended funding hold most external local-currency debt, while intermediaries with stable funding disproportionately hold hard-currency bonds. When open-ended intermediaries face redemptions, local-currency bond prices fall differentially and issuers reallocate to hard currency. We rationalize these facts in a model where foreign households demand money-like claims in their own currency from bond-market intermediaries. Deeper local-currency markets, meant to promote stability, attract intermediaries prone to fire sales.

Regulators in many emerging markets face a tradeoff: Restricting banks from holding currency mismatches on balance sheet may result in more widespread currency mismatch among bank-dependent domestic firms, including small and medium-sized enterprises (SME). This paper studies the macroeconomic costs of widespread dollar-denominated bank lending in a small open economy. Using micro data from Peru, we document that liability dollarization is substantial across the firm size distribution, including non-tradable sectors and for small, bank-dependent firms. We develop a model to quantify the costs of currency devaluations given heterogeneous firm liabilities. Aggregate outcomes following a depreciation depend on the joint distribution of firms’ foreign currency exposures and their marginal propensities to invest out of liquidity, which are particularly high for small, bank-dependent firms. Estimating the model on the universe of Peruvian firm-level borrowing, we find that depreciations are significantly more contractionary than predicted by representative-firm models. We conclude that small firms’ exposure to foreign currency debt is a key macroeconomic channel of devaluation costs.

We introduce a theory where dollarization arises as a risk transfer between domestic firms and households, who hedge against inflation and income risks from local currency depreciation. Dollarization increases when these risks are higher, resulting in large UIP deviations. Country-level evidence shows that 80% of dollarization is driven by income hedging. Using household-level data from Uruguay to account for individual pass-through effects, we find that households save in dollars to protect against exchange rate pass-through into prices and, to a lesser degree, wage reductions. Surprisingly, capital controls increase domestic dollarization, exacerbating firms’ currency mismatches and undermining their original intent.