Use Log(Stock Prices) or (Stock Returns)

[Mina Sami]

I am running a regression model to find the impact of Federal Funds Rate(FFR) on Stock market.
I have monthly data over the period 2000-2014 and cross countries (Europe and America).

Why papers are always using stock returns and not log stock prices in Panel Data? is any econometric issue to believe that stock returns are more robust that log prices?
If I run:
Log(stock Prices) as Dependent not (Returns) is this an econometric issue ?

[Attaullah Shah]

Stock prices represent level data, and carry cumulative effect from the day of inception of the company, while stock returns represent difference data, which is easy to associate with changes in the independent variables. Further, difference data does not usually suffer from the non-stationarity issue.

[Mina Sami]

Dear Attaullah,
But If I am dealing with Panel data and not time series, We care that much about stationarity?

[Attaullah Shah]

In panel data, still, the logic of fixed effects applies, which by default takes the first difference of each variable. Further, share prices in levels can pose a problem of hetro due to significant variations in the prices across different companies. Researchers, therefore, use either stock returns or some sort of standardized variables such as Tobin's Q.