I think that the capacitor size indeed affect the response time. In fact it affects the loop gain and phase (therefore reponse time).
So I think the cap shouldn't be too big.
Furthermore, they shouldn't be too small, otherwise the switch parasitics cannot be neglected. Especially the charge injection that would make the output voltage differ from the target value.
As for the ratio, it is said in the book by Johns and Martin that a 1:4 to 1:10 is a good value. I remember once using a 1:5.