Problem with interpretation of results given by reghdfe (binary dependent variable)

[Firangiz Aghayeva]

Dear Forum users,

I am estimating the investor's decision to invest in private equity funds (1 for yes, 0 for no). The reghdfe model includes more than one fixed effect (year, investor ID, manager ID) and double clustering of standard errors. My results look like this:
y=-0.0015x...(other estmation results). How can I interpret the results of my main independent variable? Is it probability or just coefficient?

Kind regards,
Firangiz

[Luca Calianno]

Hi, if x is linear (not scaled) then the interpretation is that of a probability. The coefficient can be interpreted as the change in the probability of investing for a one-unit change in the independent variable.

[Firangiz Aghayeva]

Hi, Luca! I see. What is it is scaled?

[Luca Calianno]

Not something like log(x)

[Carlo Lazzaro]

Firangiz:
it is not that easy to reply positively without reading what you typed and Stata gave you back (as per FAQ). Thanks.

[Firangiz Aghayeva]

Dear Carlo,

This is what I typed and what I got from STATA 17. Hope it helps to explain my issue. I have non-scaled and dummy independent variables. I am not quite sure how to interpret their coefficients.

 reghdfe invest_dummy SixD_hofV_1 Prior_relationship_dummy LP_GDP_currentmnLN LP_EFInvestmentFreedomLN Fund_Size_USDLN Fund_seq_numLN LP_WGILN Manager_WGILN Legal_origin_bias Fund_coinvestment_dummy, absorb(Fund_Vintage Manager_ID) vce(cluster Fund_Vintage Manager_country) noconstant
(MWFE estimator converged in 10 iterations)
Warning: VCV matrix was non-positive semi-definite; adjustment from Cameron, Gelbach & Miller ap
> plied.
warning: missing F statistic; dropped variables due to collinearity or too few clusters

HDFE Linear regression                            Number of obs   = 12,210,492
Absorbing 2 HDFE groups                           F(  10,     22) =          .
Statistics robust to heteroskedasticity           Prob > F        =          .
                                                  R-squared       =     0.0082
                                                  Adj R-squared   =     0.0079
Number of clusters (Fund_Vintage) =         23    Within R-sq.    =     0.0045
Number of clusters (Manager_country) =         67 Root MSE        =     0.0503

                      (Std. err. adjusted for 23 clusters in Fund_Vintage Manager_country)
------------------------------------------------------------------------------------------
                         |               Robust
            invest_dummy | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
-------------------------+----------------------------------------------------------------
             SixD_hofV_1 |  -.0020704   .0004806    -4.31   0.000    -.0030671   -.0010738
Prior_relationship_dummy |    .003192   .0010566     3.02   0.006     .0010007    .0053832
      LP_GDP_currentmnLN |  -.0004935   .0003121    -1.58   0.128    -.0011407    .0001537
LP_EFInvestmentFreedomLN |   -.001253   .0008132    -1.54   0.138    -.0029395    .0004336
         Fund_Size_USDLN |    .001084   .0003121     3.47   0.002     .0004367    .0017314
          Fund_seq_numLN |   .0007423   .0002476     3.00   0.007     .0002287    .0012559
                LP_WGILN |   .0001221   .0015866     0.08   0.939    -.0031683    .0034125
           Manager_WGILN |  -.0023804   .0015964    -1.49   0.150    -.0056912    .0009303
       Legal_origin_bias |  -.0008595   .0006005    -1.43   0.166    -.0021049     .000386
 Fund_coinvestment_dummy |   .9932533   .0007418  1338.91   0.000     .9917149    .9947918
------------------------------------------------------------------------------------------

Absorbed degrees of freedom:
------------------------------------------------------+
  Absorbed FE | Categories  - Redundant  = Num. Coefs |
--------------+---------------------------------------|
 Fund_Vintage |        23          23           0    *|
   Manager_ID |      3415        3415           0    *|
------------------------------------------------------+
* = FE nested within cluster; treated as redundant for DoF computation

.

[Carlo Lazzaro]

Firangiz:
see Interpretation of coefficients in LPM - Statalist

[Firangiz Aghayeva]

Thank you very much for your reply!