|panel data and volatility of price
||[May. 3rd, 2010|03:56 pm]
Is there a way to measure the impact of price volatility on investment with a panel dataset, when prices are assumed to be the same for all firm and we use a vector of time dummies?
The problem is that as price volatility is the same for all firms, there is no between variation (variation over N) but only within variation (variation over T). Therefore, we end-up with perfect collinearity and we cannot identify the effect because it is merged into the time dummy intercept, right?
A friend of mine suggested the use of interaction variable (eg: volP*SizeFirm), but I think it doesn't solve the problem of absence of any cross-section -variation. Any other way?