bilateral exchange rate volatility definition I am trying to estimate panel gravity model and need exchange rate volatility between country i and j but I don't know how to do it, is it possible to use GARCH model in this case? or Moving average standard deviation? Would you mind helping me please. Thanks in advance 
 A: Essentially, first you want to calculate the first difference in the natural logs of the exchange rates, month to month: $\delta_t = \ln x_t - \ln x_{t-1}$. This will be zero if the exchange rate follows a constant trend, which corresponds to zero uncertainty. If you have daily data, people generally use the last day of the month, though that has never made much sense to me. Then you calculate the standard deviation of these $\delta$s over a one-year period to get short-run volatility in that year. You can use a longer horizon (five years) as well if that makes sense. That is your measure of exchange rate volatility.  
The other big choice is whether to use nominal or real exchange rates for this calculation. The real route demands more data since you need consumer prices, and is preferable on theoretical grounds, especially if you're looking at a longer time span. On the other hand, if you're looking at the short-run, you can assume that the costs of production, export and import prices are all fixed, so the exposure of firms is to the nominal exchange rate.
Take a look around page 30 of this IMF survey for more details and alternative measures.
