Hi All,
I am dealing with company level unbalanced panel data (attached). There are broadly two types of control variables in the study - a) firm level control variables, FIV1 (market cap, return on assets, cash holdings, carbon risk etc.) which varies for a firm cross-sectionally as well as in time and (b) macro or market variables, MIV1 (risk free rate, excess market return etc.) which do not vary across firm level but vary only across the time.

When I run the regression, I get an estimated coefficient of ‘0’ for macro-economic variables if I take the time and industry fixed effect. Stata provides the warning message that such variables are perfectly correlated with the fixed effects, hence cannot be estimated and provides ‘0’ coefficients for such variables. This serial correlation is understood because the values of these control variables will remain same for the time effect dummy across all the firms. Hence, when I remove the time fixed effects, then as expected this issue does not occur and I get the estimated coefficients.

Similar macro-economic variables are used in the other studies where the results always mention that the time fixed effects and cross-sectional fixed effects are being considered. I am not able to understand why these studies did not face the same issue that I am facing? I need to have time fixed effect in my analysis but the collinearity issue is giving '0' value for the market level coefficients. Please guide me.

Code:
reghdfe DV FIV1 MIV1, absorb(IndustryId Date) vce(cluster CompanyId Date)

Warning from Stata - MIV1 is probably collinear with the fixed effects
MIV1 omitted because of collinearity