In this paper, we investigate the relationship between the transparency of banks and the fragility of the banking system. We show that information-based bank runs may be inefficient because the deposit contract designed to provide liquidity induces depositors to have excessive incentives to withdraw. An improvement in the transparency of a bank may reduce depositor welfare by increasing the chance of an inefficient contagious run on other banks. A deposit insurance system in which some depositors are fully insured and the others are partially insured can ameliorate this inefficiency. Under such a system, bank runs can serve as an efficient mechanism for disciplining banks. We also consider bank managers' control over the timing of information disclosure, and find that bank managers may use their influence to eliminate both inefficient and efficient bank runs.

By devising a real effective exchange rate (REER) index where bilateral exchange rates are weighted for relative trade shares, we find that the REER volatility (differently from the bilateral exchange rate volatility with the dollar) has significant impact on growth of per capita income after controlling for other variables traditionally considered in conditional convergence estimates. We also find that this (cost of volatility) effect can be reconciled with the concurring negative and significant effect on growth of the adoption of a fixed exchange rate regime (advantage of flexibility effect), where the latter may be also interpreted as the cost of choosing pegged regimes without harmonization of rules and macroeconomic policies with main trading partners. The adoption of an REER volatility measure, instead of a bilateral exchange rate with the dollar, has the advantage of making it possible a joint test for these two effects. This is because, while fixed exchange rate regimes are strongly negatively correlated, and almost collinear, with bilateral exchange rate volatility with the dollar, the correlation is much weaker when considering our REER volatility measure.