REGHDFE | Quick Start Guide
Basic Syntax
-
The basic syntax of reghdfe is the same as areg. Thus,
. areg depvar indepvars, absorb(absvar)
Becomes:
. reghdfe depvar indepvars, absorb(absvar1 absvar2 …)
-
With IV/GMM regressions, use the ivregress and ivreg2 syntax:
. reghdfe depvar indepvars (endogvars=iv_vars), absorb(absvars)
-
Similarly, for robust standard errors:
. reghdfe depvar indepvars , absorb(absvars) vce(robust)
. reghdfe depvar indepvars , absorb(absvars) vce(cluster clustervars)
Examples
Estimate an OLS Regression with two levels of fixed effects
. sysuse auto
(1978 Automobile Data)
. reghdfe price weight length, absorb(turn trunk)
(dropped 9 singleton observations)
(converged in 12 iterations)
HDFE Linear regression Number of obs = 65
Absorbing 2 HDFE groups F( 2, 38) = 12.85
Prob > F = 0.0001
R-squared = 0.6403
Adj R-squared = 0.3942
Within R-sq. = 0.4035
Root MSE = 2254.6288
------------------------------------------------------------------------------
price | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
weight | 5.688144 1.302818 4.37 0.000 3.050727 8.325561
length | -44.21191 57.60797 -0.77 0.448 -160.8332 72.40933
-------------+----------------------------------------------------------------
Absorbed | F(24, 38) = 1.548 0.110 (Joint test)
------------------------------------------------------------------------------
Absorbed degrees of freedom:
---------------------------------------------------------------+
Absorbed FE | Num. Coefs. = Categories - Redundant |
-------------+-------------------------------------------------|
turn | 13 13 0 |
trunk | 12 13 1 |
---------------------------------------------------------------+
Estimate an OLS Regression with two-way clustering
. reghdfe price weight length, absorb(turn trunk) vce(cluster turn trunk)
(dropped 9 singleton observations)
(converged in 12 iterations)
HDFE Linear regression Number of obs = 65
Absorbing 2 HDFE groups F( 2, 12) = 16.59
Statistics robust to heteroskedasticity Prob > F = 0.0004
R-squared = 0.6403
Adj R-squared = 0.3778
Number of clusters (turn) = 13 Within R-sq. = 0.4035
Number of clusters (trunk) = 13 Root MSE = 2284.8937
(Std. Err. adjusted for 13 clusters in turn trunk)
------------------------------------------------------------------------------
| Robust
price | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
weight | 5.688144 2.837967 2.00 0.068 -.4952552 11.87154
length | -44.21191 87.68177 -0.50 0.623 -235.2541 146.8302
------------------------------------------------------------------------------
Absorbed degrees of freedom:
---------------------------------------------------------------+
Absorbed FE | Num. Coefs. = Categories - Redundant |
-------------+-------------------------------------------------|
turn | 0 13 13 * |
trunk | 0 13 13 * |
---------------------------------------------------------------+
* = fixed effect nested within cluster; treated as redundant for DoF computation
Estimate an IV regression
. reghdfe price weight (length = gear), absorb(turn trunk)
(dropped 9 singleton observations)
(converged in 12 iterations)
HDFE IV (2SLS) estimation
-------------------------
Estimates efficient for homoskedasticity only
Statistics consistent for homoskedasticity only
Number of obs = 65
F( 2, 38) = 12.22
Prob > F = 0.0001
Total (centered) SS = 323844734.5 Centered R2 = 0.6293
Total (uncentered) SS = 323844734.5 Uncentered R2 = .
Residual SS = 199090229 Root MSE = 2289
------------------------------------------------------------------------------
price | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
length | -106.3953 407.0322 -0.26 0.795 -930.3889 717.5984
weight | 6.578179 5.915196 1.11 0.273 -5.396509 18.55287
------------------------------------------------------------------------------
Underidentification test (Anderson canon. corr. LM statistic): 1.342
Chi-sq(1) P-val = 0.2467
------------------------------------------------------------------------------
Weak identification test (Cragg-Donald Wald F statistic): 0.801
Stock-Yogo weak ID test critical values: 10% maximal IV size 16.38
15% maximal IV size 8.96
20% maximal IV size 6.66
25% maximal IV size 5.53
Source: Stock-Yogo (2005). Reproduced by permission.
------------------------------------------------------------------------------
Sargan statistic (overidentification test of all instruments): 0.000
(equation exactly identified)
------------------------------------------------------------------------------
Instrumented: length
Included instruments: weight
Excluded instruments: gear_ratio
------------------------------------------------------------------------------
Absorbed degrees of freedom:
---------------------------------------------------------------+
Absorbed FE | Num. Coefs. = Categories - Redundant |
-------------+-------------------------------------------------|
turn | 13 13 0 |
trunk | 12 13 1 |
---------------------------------------------------------------+
If you also want the first stage or the OLS version of this regression, check out the stages(...) option (which also supports the reduced form and the “acid” version).