Hello,

i should perform a panelregression with industry and year FE and cluster by firm and year at the end
My data is ranging from 2002 until 2019 and includes all firms from the s&p 500.

i cannot go
Code:
xtset industry_key year
repeated time values within panel
then i tried this
Code:
regress firm_beta_w esg_single_lag1 i.industry_key i.year, vce (cluster company_key year)
However I noticed that with the LSDV approach by using OLS you can only cluster by one variable

Then i tried this
Code:
reghdfe firm_beta esg_single_lag1, absorb(industry_key year) vce (cluster company_key year)
(MWFE estimator converged in 4 iterations)

HDFE Linear regression                            Number of obs   =      6,196
Absorbing 2 HDFE groups                           F(   1,     15) =       0.12
Statistics robust to heteroskedasticity           Prob > F        =     0.7367
                                                  R-squared       =     0.1404
                                                  Adj R-squared   =     0.1367
Number of clusters (company_key) =        494     Within R-sq.    =     0.0001
Number of clusters (year)    =         16         Root MSE        =     0.4120

                         (Std. Err. adjusted for 16 clusters in company_key year)
---------------------------------------------------------------------------------
                |               Robust
      firm_beta |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
esg_single_lag1 |   .0202003   .0589782     0.34   0.737    -.1055087    .1459094
          _cons |   1.103985   .0276022    40.00   0.000     1.045153    1.162818
---------------------------------------------------------------------------------

Absorbed degrees of freedom:
------------------------------------------------------+
  Absorbed FE | Categories  - Redundant  = Num. Coefs |
--------------+---------------------------------------|
 industry_key |        11           0          11     |
         year |        16          16           0    *|
------------------------------------------------------+
* = FE nested within cluster; treated as redundant for DoF computation

(MWFE estimator converged in 4 iterations)

HDFE Linear regression                            Number of obs   =      6,196
Absorbing 2 HDFE groups                           F(   1,     15) =       0.12
Statistics robust to heteroskedasticity           Prob > F        =     0.7367
                                                  R-squared       =     0.1404
                                                  Adj R-squared   =     0.1367
Number of clusters (company_key) =        494     Within R-sq.    =     0.0001
Number of clusters (year)    =         16         Root MSE        =     0.4120

                         (Std. Err. adjusted for 16 clusters in company_key year)
---------------------------------------------------------------------------------
                |               Robust
      firm_beta |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
esg_single_lag1 |   .0202003   .0589782     0.34   0.737    -.1055087    .1459094
          _cons |   1.103985   .0276022    40.00   0.000     1.045153    1.162818
---------------------------------------------------------------------------------

Absorbed degrees of freedom:
------------------------------------------------------+
  Absorbed FE | Categories  - Redundant  = Num. Coefs |
--------------+---------------------------------------|
 industry_key |        11           0          11     |
         year |        16          16           0    *|
------------------------------------------------------+
* = FE nested within cluster; treated as redundant for DoF computation
as you can see, the results do look very bad. Are there problems in the data or did i do anything wrong?
I woul appreciate some help since i am really struggling

Thanks Patrick