Linear combinations of parameters
Linear combinations of parameters for FixedEffectModels.jl
and EventStudyInteracts.jl
like Stata package lincom
. As Stata package lincom
help document explains:
lincom
computes point estimates, standard errors, t or z statistics, p-values, and confidence intervals for linear combinations of coefficients after any estimation command, including survey estimation. Results can optionally be displayed as odds ratios, hazard ratios, incidence-rate ratios, or relative-risk ratios.
lincom
can be used to aggregating event study estimates and estimate the average effect when use the Stata package eventstudyinteract
provided by Sun and Abraham (2021).
I wrote the EventStudyInteracts.jl
package, which is a Julia replication of the Stata package eventstudyinteract
provided by Sun and Abraham (2021). However, there is currently no package in Julia similar to lincom, so I wrote this package.
This package can also be used for t-tests of linear combinations of results from FixedEffectModels.jl.
The package is registered in the General
registry and so can be installed at the REPL with ] add LinComs
.
After estimating the results using EventStudyInteracts.jl
, you can refer to the following code to estimate the ATE.
rel_varlist1 = [:g_3,:g_2 ,:g0 ,:g1 ,:g2 ,:g3 ,:g4]
m1 = eventreg(df, formula1, rel_varlist1, control_cohort1, cohort1, vcov1)
expr = :((g0+g1+g2+g3+g4)/5)
lincom(m1,expr)
# Which will return a result like this.
lincom
=============================================================================
ln_wage | Estimate Std.Error t value Pr(>|t|) Lower 95% Upper 95%
-----------------------------------------------------------------------------
Linear Combination | 0.0561422 0.012538 4.47776 0.000 0.0315618 0.0807226
=============================================================================
Thanks to newbing.