Democratic Institutions and Exchange Rates 79 will have more stable expectations,and,as a consequence,cross-border trade and investment will increase.27 Floating exchange rates,according to these arguments, lead to higher exchange-rate variability and,hence,to greater uncertainty and risk.In countries that are less dependent on trade,stabilizing the exchange rate will not be a priority.Instead,governments will want to use macroeconomic policy for domestic policy objectives. We measure a country's dependence on trade with an openness variable composed of imports plus exports as a percentage of GDP.Countries with higher levels of openness-that is,higher dependence on trade-will be more likely to fix their ex- change rates.In alternative specifications we disaggregated the openness variable into its component parts:imports as a percentage of gross domestic product(GDP), exports as a percentage of GDP,and the trade deficit as a percentage of GDP.The alternative measures did not substantially alter the results.28 Vulnerabiliry to Shocks Recent research in economics argues that a country's vulnerability to macroeco- nomic shocks conditions the optimal exchange-rate arrangement.The theoretical and empirical literatures conclude that a country faced with adverse shocks emanating from the real (that is,tradable)sector will be better able to insulate the domestic economy by adopting a floating exchange arrangement.29 A fixed exchange arrange- ment,on the other hand,is more desirable if the country experiences domestic(nomi- nal)disturbances. Following Michael Melvin and Andreas Savvides,we operationalize a country's vulnerability to shocks in several ways.30 The first variable,designed to capture do- mestic shocks,measures the variability in the growth rate of domestic credit over the course of a year,again based on quarterly data.Greater variability in the monetary sector makes a fixed exchange-rate arrangement more likely.To measure real shocks we include a measure of trade openness (discussed earlier)and a measure of the yearly rate of economic growth(discussed later).31 27.Frankel and Rose argue that,even where exchange-rate variability has been high.its effect on trade has been low.Frankel and Rose 1996.They suggest,however,that the exchange-rate variability argument "still carries some weight.It looms large in the minds of European policy-makers and business-people. Promoting trade and investment in Europe was certainly a prime motivation for the European Monetary System and for the planned E.M.U." 28.Including either exports or imports led to very similar empirical results-with the exception of the sign change.Including both indicators at the same time caused both variables to be statistically insignifi- cant because the correlation between imports and exports in our sample is 0.96. 29.For example,Fischer 1977;Melvin 1985:and Savvides 1990. 30.See Melvin 1985:and Savvides 1990. 31.In alterative specifications,we also included a measure of a country's economic size.based on its GDP in constant dollars,as a proxy of a country's vulnerability to shocks.Presumably,larger countries will be less vulnerable to exogenous shocks and,consequently,more likely to choose a floating exchange- rate arrangement.The economic size variable,however,was collinear with measures of openness and interational capital mobility.Democratic Institutions and Exchange Rates 79 will have more stable expectations, and, as a consequence, cross-border trade and investment will increase.27 Floating exchange rates, according to these arguments, lead to higher exchange-rate variability and, hence, to greater uncertainty and risk. In countries that are less dependent on trade, stabilizing the exchange rate will not be a priority. Instead, governments will want to use macroeconomic policy for domestic policy objectives. We measure a country's dependence on trade with an openness variable composed of imposts plus exposts as a percentage of GDP. Countries with higher levels of openness-that is, higher dependence on trade-will be more likely to fix their ex￾change rates. In alternative specifications we disaggregated the openness variable into its component parts: imposts as a percentage of gross domestic product (GDP), exports as a percentage of GDP, and the trade deficit as a percentage of GDP. The alternative measures did not substantially alter the results.28 Vulnerability to Shocks Recent research in economics argues that a country's vulnerability to macroeco￾nomic shocks conditions the optimal exchange-rate arrangement. The theoretical and empirical literatures conclude that a country faced with adverse shocks emanating from the real (that is, tradable) sector will be better able to insulate the domestic economy by adopting a floating exchange a~angement.~~ A fixed exchange arrange￾ment, on the other hand, is more desirable if the country experiences domestic (nomi￾nal) disturbances. Following Michael Melvin and Andreas Savvides, we operationalize a country's vulnerability to shocks in several ways.30 The first variable, designed to capture do￾mestic shocks, measures the variability in the growth rate of domestic credit over the course of a year, again based on quarterly data. Greater variability in the monetary sector makes a fixed exchange-rate arrangement more likely. To measure real shocks we include a measure of trade openness (discussed earlier) and a measure of the yearly rate of economic growth (discussed late^).^' 27. Frankel and Rose argue that, even where exchange-rate variability has been high. its effect on trade has been low. Frankel and Rose 1996. They suggest, however. that the exchange-rate variability argument "still canies some weight. It looms large in the minds of European policy-makers and business-people. Promoting trade and investment in Europe was certainly a prime motivation for the European Monetary System and for the planned E.M.U." 28. Including either exports or imports led to very similar empirical results-with the exception of the sign change. Including both indicators at the same time caused both variables to be statistically insignifi￾cant because the correlation between imports and exports in our sample is 0.96. 29. For example. Fischer 1977; Melvin 1985: and Savvides 1990. 30. See Melvin 1985; and Savvides 1990. 31. In alternative specifications. we also included a measure of a country's economic size. based on its GDP in constant dollars, as a proxy of a country's vulnerability to shocks. Presumably, larger countries will be less vulnerable to exogenous shocks and, consequently, more likely to choose a floating exchange￾rate arrangement. The economic size variable. however. was collinear with measures of openness and international capital mobility