reghdfe with wild bootstrap

[Clotilde Mahe]

Dear all,

I am currently running estimations in a difference-in-differences setting along the lines of

reghdfe y i.a##i.b c d e, absorb(municipality) cluster(municipality)

The coefficient estimate I am interested in is

1.a#1.b

(DID estimator)

Since the number of cluster (municipality) is around 40, sometimes less, I would like to estimate standard errors via wild bootstrapping.

However, after going through various posts and articles, I am still having difficulties understanding how to properly to do so. Could one of you point me to the right direction?

Thank you.

[Anshuman Tiwari]

Hi Clotilde,

Were you able to figure out this issue? I have a very similar situation that I am trying to figure out in either R or Stata.

Thanks,
Anshuman

[Anshuman Tiwari]

Just a clarification - the original post has one fixed effect. In that situation 'boottest' works well after reghdfe.

But I have multiple fixed effects and boottest cannot handle that. This is the problem I am trying to figure out.

Thanks again

[John Kirk]

I am also trying to figure out how to bootstrap in a reghdfe regression (or similar alternative that allows for absorbing multiple fixed effects).

[Maxence Morlet]

Same here for the wild cluster restricted bootstrap of Roodman et al. (2018)

[Joro Kolev]

If you are to hope to have an resolution of "the issue", you should try to explain what this issue is. For what I see the current version of the user written -boottest- can handle the current version of the -reghdfe-. Here:

. sysuse auto
(1978 automobile data)

. reg price mpg i.rep

      Source |       SS           df       MS      Number of obs   =        69
-------------+----------------------------------   F(5, 63)        =      4.39
       Model |   149020603         5  29804120.7   Prob > F        =    0.0017
    Residual |   427776355        63  6790100.88   R-squared       =    0.2584
-------------+----------------------------------   Adj R-squared   =    0.1995
       Total |   576796959        68  8482308.22   Root MSE        =    2605.8

------------------------------------------------------------------------------
       price | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         mpg |  -280.2615   61.57666    -4.55   0.000    -403.3126   -157.2103
             |
       rep78 |
          2  |   877.6347   2063.285     0.43   0.672     -3245.51     5000.78
          3  |   1425.657   1905.438     0.75   0.457    -2382.057    5233.371
          4  |   1693.841   1942.669     0.87   0.387    -2188.274    5575.956
          5  |   3131.982   2041.049     1.53   0.130    -946.7282    7210.693
             |
       _cons |   10449.99   2251.041     4.64   0.000     5951.646    14948.34
------------------------------------------------------------------------------

. boottest mpg=0

Wild bootstrap-t, null imposed, 999 replications, Wald test, Rademacher weights:
  mpg=0

                           t(63) =    -4.5514
                        Prob>|t| =     0.0000

95% confidence set for null hypothesis expression: [−414.5, −168]

. areg price mpg, absorb(rep)

Linear regression, absorbing indicators          Number of obs     =        69
Absorbed variable: rep78                         No. of categories =         5
                                                 F(1, 63)          =     20.72
                                                 Prob > F          =    0.0000
                                                 R-squared         =    0.2584
                                                 Adj R-squared     =    0.1995
                                                 Root MSE          = 2605.7822

------------------------------------------------------------------------------
       price | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         mpg |  -280.2615   61.57666    -4.55   0.000    -403.3126   -157.2103
       _cons |   12112.77   1347.968     8.99   0.000      9419.07    14806.47
------------------------------------------------------------------------------
F test of absorbed indicators: F(4, 63) = 1.072               Prob > F = 0.378

. boottest mpg=0

Wild bootstrap-t, null imposed, 999 replications, Wald test, Rademacher weights:
  mpg=0

                           t(63) =    -4.5514
                        Prob>|t| =     0.0000

95% confidence set for null hypothesis expression: [−414.5, −162.8]

. reghdfe price mpg, absorb(rep)
(MWFE estimator converged in 1 iterations)

HDFE Linear regression                            Number of obs   =         69
Absorbing 1 HDFE group                            F(   1,     63) =      20.72
                                                  Prob > F        =     0.0000
                                                  R-squared       =     0.2584
                                                  Adj R-squared   =     0.1995
                                                  Within R-sq.    =     0.2475
                                                  Root MSE        =  2605.7822

------------------------------------------------------------------------------
       price | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
         mpg |  -280.2615   61.57666    -4.55   0.000    -403.3126   -157.2103
       _cons |   12112.77   1347.968     8.99   0.000      9419.07    14806.47
------------------------------------------------------------------------------

Absorbed degrees of freedom:
-----------------------------------------------------+
 Absorbed FE | Categories  - Redundant  = Num. Coefs |
-------------+---------------------------------------|
       rep78 |         5           0           5     |
-----------------------------------------------------+

. boottest mpg=0

Wild bootstrap-t, null imposed, 999 replications, Wald test, Rademacher weights:
  mpg=0

                           t(63) =    -4.5514
                        Prob>|t| =     0.0000

95% confidence set for null hypothesis expression: [−407.7, −163.3]

. 

[Mitch Downey]

Hi Joro,

As Anshuman said, the problem only emerges when you have two fixed effects in reghdfe, not when you have only one. "The issue" is that this doesn't estimate, and returns the error "doesn't work."

. sysuse auto
(1978 Automobile Data)

. reghdfe price mpg, absorb(rep foreign)
(MWFE estimator converged in 3 iterations)

HDFE Linear regression                            Number of obs   =         69
Absorbing 2 HDFE groups                           F(   1,     62) =      22.37
                                                  Prob > F        =     0.0000
                                                  R-squared       =     0.2758
                                                  Adj R-squared   =     0.2057
                                                  Within R-sq.    =     0.2651
                                                  Root MSE        =  2595.6201

------------------------------------------------------------------------------
       price |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         mpg |  -299.6068   63.34525    -4.73   0.000    -426.2322   -172.9815
       _cons |   12524.63   1384.339     9.05   0.000     9757.376    15291.88
------------------------------------------------------------------------------

Absorbed degrees of freedom:
-----------------------------------------------------+
 Absorbed FE | Categories  - Redundant  = Num. Coefs |
-------------+---------------------------------------|
       rep78 |         5           0           5     |
     foreign |         2           1           1     |
-----------------------------------------------------+

. boottest mpg
Doesn't work after reghdfe with more than one set of absorbed fixed effects or with absorbed interaction terms.
r(198);