If I have a variable graphed as a function of time, that is not normally distributed, does it make sense to use its (time-varying) dispersion index, $\mathrm{Var}(x)/E(x)$ to compare its volatility for different members of a panel?
For example, daily temperature for different cities, or weekly exports for different countries. The reason I thought of using the index instead of straight variance or standard deviation is that the values are quite large for some members and low for others, and I would like to transform the series to make the trends comparable. Thanks! I hope I'm being clear.