In the first paper, I use industry-level data to investigate the impact of exchange rate arrangements on the productive structure of the economy. The identification strategy has similarities with the methodology followed in the literature on heterogeneous effects of financial development. A de facto exchange rate regime classification is used to sort pegs and floats. My findings suggest that industries that have higher working capital needs grow faster under exchange rate stability. A fixed exchange rate regime could lower currency or country risk, leading to greater availability of funds and a reduction in the cost of financing. Since loans are often denominated in foreign currency or indexed to the exchange rate in developing countries, firms with higher working capital needs would prefer exchange rate stability, which may lower interest rates in foreign currency and provide easier access to credit.

The second paper investigates the behavior of output across large devaluations and depreciations. First, I define a currency crisis as an episode in which the nominal exchange rate increases by 15%. Then I proceed to classify them into devaluations and depreciations using Reinhart and Rogoff (2004) exchange rate classification. Once these episodes are sorted out, I analyze the behavior of output across them.

As in previous studies, I find that the majority of the currency crisis episodes have been contractionary for 1970-2007. When I separate currency crisis episodes between devaluations and depreciations, I find that the former have been associated with larger output losses for middle income economies.

These findings are consistent with the fact that middle income countries are often subject to currency mismatches. As a result, they may opt for an exchange rate regime that exhibits relatively more stability. This is well documented in Calvo and Reinhart (2002), a behavior they termed "fear of floating". However, in the case of a currency crisis, the negative impact on output growth is likely to be larger for countries that have adopted a fixed exchange rate regime. This result is also supported by the larger magnitude of the estimated output losses when we use a higher threshold to define currency crisis episodes.