We estimate the equation using four different specifications: with only the year dummy as in column
(I); with both year and industry dummy as in column (II); as a fractional logit model as in column
(III); and with year and affiliate dummy, namely, a two-way component fixed effect model (FE) as in
column (IV). Except for (IV), we also include Age as an independent variable. A fractional logit estimation
technique (Frac), proposed by Papke and Wooldridge (1996), ensures that the predicted values of
the dependent variable are in the unit interval, i.e., [0, 1].8 The results are the same except for the coefficients
on employment and age.