Using a panel data set of 180 countries spanning from 1971 to 2000, I find evidence that exchange rate policy affects economic growth not directly but through its effect on macroeconomic volatility. I consider two measures of economic performance: i) per capita GDP growth and ii) the volatility of per capita GDP growth and investigate the nature of their dependence on two alternative characterizations of exchange rate policy. My first measure of policy is a de facto classification of a country's exchange rate regime as either fixed or floating. My second characterization of exchange rate policy measures whether a country's de facto policy is consistent with its publicly stated exchange rate policy or its de jure exchange rate regime. Employing de facto classification I find the significant statistical relationship between exchange rate policy and growth which is not robust to the inclusion of conventional growth control variables. A coarse division of policy into fixed versus floating exchange rates using the de jure metrics is not significant but de facto classification identifies the dichotomy between countries in terms of GDP growth volatility. Moreover, a more nuanced characterization shows that countries that exhibit 'fear of floating,' which arises when countries de jure float but de facto fix have the most stable GDP growth when compared to other reference groups. These countries as well have lower volatility of GDP growth during the periods when they exhibited 'fear of floating' compared to their own experience when they followed other exchange rate policies. Finally, I demonstrate that the 'fear of floating' type of behavior of the exchange rate is related to the inflation targeting (IT) policy by the Central Bank. I find that in an emerging small open economy environment the combination of the publicly announced flexible foreign exchange rate regime with the pursuit of the inflation targeting lite (ITL) strategy induces the Central Bank to actively employ the domestic interest rate as an instrument to offset the foreign inflationary shocks. This policy results in the ex post stability of the nominal exchange rate which I identify as a 'fear of floating' behavior, a reduction of the inflation volatility, and a significant reduction of the volatility of real GDP growth. This evidence allows me to argue that the 'fear of floating' policy is the most advantageous open economy policy which reduces overall macroeconomic volatility. I identify two possible channels why this result is in place. First, the 'fear floaters' enjoy the lower volatility of the real interest rate which suggests that they are exposed to less real shocks. Second, they have lower real FDI volatility which stabilizes the gross capital formation and the real GDP.