Question

What is the advantage of exponential smoothing over moving average

Answer 1

Unless there is a good reason to expect smoothing over a specific number of observations, exponential smoothing makes more sense because you don’t get jumps when an observation falls out of the moving average window.

If you have good information, you should model the process. If you don’t have good information, exponential smoothing is a better general technique because a small difference in the decay parameter makes less difference than the effect of making the moving average window one observation bigger or smaller. Also most lagged effects that do not have a specific time lag will likely have effects that look more exponential than zero/one.