Fernando Letelier

How does the composition of tradable income shape financial fragility during sudden stops? Using the Bianchi and Mendoza (2020) systemic sudden stop chronology, I show that economies entering an episode with greater commodity dependence undergo larger absolute adjustment: moving from the 25th to the 75th percentile of pre-crisis exposure is associated with a 2.93 percentage point larger current account reversal one year after onset, alongside larger consumption and equity price losses. I then extend the Fisherian collateral model of Bianchi (2011) by allowing commodity exposure to change the sensitivity of tradable income to a common world price. Holding the inherited balance sheet and shock fixed, moving across the empirically mapped exposure interquartile range raises the current account reversal by 4.88 percentage points. When exposure is anticipated, households borrow less ex ante: on a fixed set of crisis dates, the severity differential peaks at intermediate exposure and falls by 46 percent by α = 0.40. A borrowing rule counterfactual shows that this endogenous adjustment substantially attenuates the exposure gradient. Greater exposure also raises the marginal external cost of leverage at a common balance sheet. Financial fragility therefore depends not only on leverage, but also on the composition of the income backing it and on endogenous balance sheet adjustment.

What inflation expectations are most informative for central banks in open economies? Using a quarterly panel of twelve open economies from 2000 to 2025, I compare one year ahead inflation expectations from non-expert agents and expert forecasters. Unlike recent U.S. evidence, experts are more accurate in both demand and supply driven inflation episodes. The paper's main result is that this expert advantage declines systematically with commodity terms of trade exposure during global supply episodes. The narrowing reflects lower non-expert forecast errors rather than deteriorating expert performance. Fair–Shiller encompassing regressions further show that the non-expert/expert forecast gap becomes increasingly informative about subsequent inflation as exposure rises, with the incremental information most clearly reflected in future services inflation. Contemporaneous exchange rate movements do not account for the exposure gradient. A parsimonious small open economy New Keynesian model shows that this informational channel, rather than commodity transmission alone, narrows the difference between optimal policy responses across demand and supply regimes.

This paper introduces iceberg transportation costs that depend on firm productivity into a model of international trade with monopolistic competition, firm-level heterogeneity, and both constant and variable markups, following the framework of Arkolakis et al. (2019). Using shipment-level customs data from Chile, I show that larger firms face systematically lower trade costs: a 1% increase in a firm's total imports is associated with a 0.4–0.6 percentage point decline in its iceberg transport cost, measured as freight over CIF. This evidence challenges the standard assumption that trade costs are uniform across firms within a given origin-destination pair. I incorporate a productivity-dependent iceberg cost into the model and show that it strengthens the selection effect, raising the productivity cutoff for exporting and improving the fit to observed patterns of firm participation and the distribution of export sales. In a counterfactual exercise with a 25% increase in U.S. tariffs, the model with heterogeneous trade costs predicts smaller declines in aggregate exports and in the fraction of exporters than a benchmark with constant iceberg costs, but larger welfare losses, as trade becomes more concentrated in a small set of high-markup firms. These results suggest that ignoring firm-level heterogeneity in transportation costs can bias quantitative assessments of trade policy.

This paper shows that the accuracy and behavior of U.S. inflation expectations depend critically on whether inflation is driven by demand or supply shocks. Combining one year ahead expectations from the SPF, Michigan Survey, and Cleveland Fed with Shapiro's decomposition, we find a reversal in forecast rankings: consumers forecast CPI inflation more accurately than experts in demand driven episodes, while professional and market based expectations dominate in supply driven episodes. Forecast inefficiencies and error persistence are also regime specific. A simple New Keynesian noisy information framework with divine coincidence in demand regimes and its breakdown in supply regimes rationalizes these patterns and their policy implications. A state dependent Taylor rule which conditions on the prevailing demand/supply mix can reduce welfare losses by around 20 percent, highlighting the monetary policy gains from treating expectations as regime contingent rather than uniform.