Support UnspecifiedParallel (#9)

Author: junyuan-chen

src/did.jl

@@ -10,14 +10,14 @@ const RegressionBasedDID = DiffinDiffsEstimator{:RegressionBasedDID,
 const Reg = RegressionBasedDID
 
 function valid_didargs(d::Type{Reg}, ::DynamicTreatment{SharpDesign},
-        ::TrendParallel{Unconditional, Exact}, args::Dict{Symbol,Any})
+        ::TrendOrUnspecifiedPR{Unconditional,Exact}, args::Dict{Symbol,Any})
     name = get(args, :name, "")::String
     treatintterms = haskey(args, :treatintterms) ? args[:treatintterms] : TermSet()
     xterms = haskey(args, :xterms) ? args[:xterms] : TermSet()
     solvelsweights = haskey(args, :lswtnames) || get(args, :solvelsweights, false)::Bool
     ntargs = (data=args[:data],
         tr=args[:tr]::DynamicTreatment{SharpDesign},
-        pr=args[:pr]::TrendParallel{Unconditional, Exact},
+        pr=args[:pr]::TrendOrUnspecifiedPR{Unconditional,Exact},
         yterm=args[:yterm]::AbstractTerm,
         treatname=args[:treatname]::Symbol,
         subset=get(args, :subset, nothing)::Union{BitVector,Nothing},

src/procedures.jl

@@ -186,8 +186,7 @@ See also [`MakeTreatCols`](@ref).
 function maketreatcols(data, treatname::Symbol, treatintterms::TermSet,
         feM::Union{AbstractFixedEffectSolver, Nothing},
         weights::AbstractWeights, esample::BitVector,
-        cohortinteracted::Bool, fetol::Real, femaxiter::Int,
-        ::Type{DynamicTreatment{SharpDesign}}, time::Symbol,
+        cohortinteracted::Bool, fetol::Real, femaxiter::Int, time::Symbol,
         exc::Dict{Int,Int}, notreat::IdDict{ValidTimeType,Int})
 
     nobs = sum(esample)
@@ -286,18 +285,19 @@ const MakeTreatCols = StatsStep{:MakeTreatCols, typeof(maketreatcols), true}
 
 required(::MakeTreatCols) = (:data, :treatname, :treatintterms, :feM, :weights, :esample)
 default(::MakeTreatCols) = (cohortinteracted=true, fetol=1e-8, femaxiter=10000)
-transformed(::MakeTreatCols, @nospecialize(nt::NamedTuple)) = (typeof(nt.tr), nt.tr.time)
+# No need to consider typeof(tr) and typeof(pr) given the restrictions by valid_didargs
+transformed(::MakeTreatCols, @nospecialize(nt::NamedTuple)) = (nt.tr.time,)
 
-combinedargs(step::MakeTreatCols, allntargs) =
-    combinedargs(step, allntargs, typeof(allntargs[1].tr))
-
-# Obtain the relative time periods excluded by all tr in allntargs
-function combinedargs(::MakeTreatCols, allntargs, ::Type{DynamicTreatment{SharpDesign}})
+# Obtain the relative time periods excluded by all tr
+# and the treatment groups excluded by all pr in allntargs
+function combinedargs(::MakeTreatCols, allntargs)
     exc = Dict{Int,Int}()
     notreat = IdDict{ValidTimeType,Int}()
-    @inbounds for nt in allntargs
+    for nt in allntargs
         foreach(x->_count!(exc, x), nt.tr.exc)
-        foreach(x->_count!(notreat, x), nt.pr.e)
+        if nt.pr isa TrendParallel
+            foreach(x->_count!(notreat, x), nt.pr.e)
+        end
     end
     nnt = length(allntargs)
     for (k, v) in exc
@@ -315,14 +315,18 @@ end
 Solve the least squares problem for regression coefficients and residuals.
 See also [`SolveLeastSquares`](@ref).
 """
-function solveleastsquares!(tr::DynamicTreatment{SharpDesign}, pr::TrendParallel,
+function solveleastsquares!(tr::DynamicTreatment{SharpDesign}, pr::TrendOrUnspecifiedPR,
         yterm::AbstractTerm, xterms::TermSet, yxterms::Dict, yxcols::Dict,
         treatcells::VecColumnTable, treatcols::Vector,
         cohortinteracted::Bool, has_fe_intercept::Bool)
 
     y = yxcols[yxterms[yterm]]
     if cohortinteracted
-        tinds = .!((treatcells[2] .∈ (tr.exc,)).| (treatcells[1] .∈ (pr.e,)))
+        if pr isa TrendParallel
+            tinds = .!((treatcells[2] .∈ (tr.exc,)) .| (treatcells[1] .∈ (pr.e,)))
+        else
+            tinds = .!(treatcells[2] .∈ (tr.exc,))
+        end
     else
         tinds = .!(treatcells[1] .∈ (tr.exc,))
     end

test/did.jl

@@ -94,6 +94,35 @@
         Fixed effects: none
         ──────────────────────────────────────────────────────────────────────"""
 
+    r0 = @did(Reg, data=hrs, dynamic(:wave, -1), unspecifiedpr(),
+        vce=Vcov.cluster(:hhidpn), yterm=term(:oop_spend), treatname=:wave_hosp,
+        treatintterms=(), xterms=(fe(:wave),),
+        cohortinteracted=false, solvelsweights=true)
+    # Compare estimates with Stata
+    # gen rel = wave - wave_hosp
+    # gen irel?? = rel==??
+    # reghdfe oop_spend irel*, a(wave) cluster(hhidpn)
+    @test coef(r0) ≈ [-1029.0482, 245.20926, 188.59266, 3063.2707,
+        1060.5317, 1152.3315, 1986.7811] atol=1e-4
+    @test diag(vcov(r0)) ≈ [764612.86, 626165.68, 236556.13, 163459.83,
+        130471.28, 294368.49, 677821.36] atol=1e0
+    pv = VERSION < v"1.6.0" ? " <1e-8" : "<1e-08"
+    @test sprint(show, MIME("text/plain"), r0) == """
+        ──────────────────────────────────────────────────────────────────────
+        Summary of results: Regression-based DID
+        ──────────────────────────────────────────────────────────────────────
+        Number of obs:               3280    Degrees of freedom:            12
+        F-statistic:                 9.12    p-value:                   $pv
+        ──────────────────────────────────────────────────────────────────────
+        Sharp dynamic specification
+        ──────────────────────────────────────────────────────────────────────
+        Relative time periods:          7    Excluded periods:              -1
+        ──────────────────────────────────────────────────────────────────────
+        Fixed effects: fe_wave
+        ──────────────────────────────────────────────────────────────────────
+        Converged:                   true    Singletons dropped:             0
+        ──────────────────────────────────────────────────────────────────────"""
+
     sr = view(r, 1:3)
     @test coef(sr)[1] == r.coef[1]
     @test vcov(sr)[1] == r.vcov[1]

test/procedures.jl

@@ -152,8 +152,7 @@ end
     pr = nevertreated(11)
     nt = (data=hrs, treatname=:wave_hosp, treatintterms=TermSet(), feM=nothing,
         weights=uweights(N), esample=trues(N), default(MakeTreatCols())...)
-    ret = maketreatcols(nt..., typeof(tr), tr.time, Dict(-1=>1),
-        IdDict{ValidTimeType,Int}(11=>1))
+    ret = maketreatcols(nt..., tr.time, Dict(-1=>1), IdDict{ValidTimeType,Int}(11=>1))
     @test size(ret.cells) == (20, 2)
     @test length(ret.rows) == 20
     @test size(ret.treatcells) == (12, 2)
@@ -175,8 +174,7 @@ end
     @test all(w[ret.treatcells.wave_hosp.==10].==163)
 
     nt = merge(nt, (treatintterms=TermSet(term(:male)),))
-    ret1 = maketreatcols(nt..., typeof(tr), tr.time, Dict(-1=>1),
-        IdDict{ValidTimeType,Int}(11=>1))
+    ret1 = maketreatcols(nt..., tr.time, Dict(-1=>1), IdDict{ValidTimeType,Int}(11=>1))
     @test size(ret1.cells) == (40, 3)
     @test length(ret1.rows) == 40
     @test size(ret1.treatcells) == (24, 3)
@@ -195,8 +193,7 @@ end
     @test ret1.cellweights == ret1.cellcounts
 
     nt = merge(nt, (cohortinteracted=false, treatintterms=TermSet()))
-    ret2 = maketreatcols(nt..., typeof(tr), tr.time, Dict(-1=>1),
-        IdDict{ValidTimeType,Int}(11=>1))
+    ret2 = maketreatcols(nt..., tr.time, Dict(-1=>1), IdDict{ValidTimeType,Int}(11=>1))
     @test ret2.cells[1] == ret.cells[1]
     @test ret2.cells[2] == ret.cells[2]
     @test ret2.rows == ret.rows
@@ -210,8 +207,7 @@ end
     @test ret2.cellweights == ret2.cellcounts
 
     nt = merge(nt, (treatintterms=TermSet(term(:male)),))
-    ret3 = maketreatcols(nt..., typeof(tr), tr.time, Dict(-1=>1),
-        IdDict{ValidTimeType,Int}(11=>1))
+    ret3 = maketreatcols(nt..., tr.time, Dict(-1=>1), IdDict{ValidTimeType,Int}(11=>1))
     @test ret3.cells[1] == ret1.cells[1]
     @test ret3.cells[2] == ret1.cells[2]
     @test ret3.cells[3] == ret1.cells[3]
@@ -232,8 +228,7 @@ end
     feM = AbstractFixedEffectSolver{Float64}(fes, wt, Val{:cpu}, Threads.nthreads())
     nt = merge(nt, (data=df, feM=feM, weights=wt, esample=esample,
         treatintterms=TermSet(), cohortinteracted=true))
-    ret = maketreatcols(nt..., typeof(tr), tr.time, Dict(-1=>1),
-        IdDict{ValidTimeType,Int}(11=>1))
+    ret = maketreatcols(nt..., tr.time, Dict(-1=>1), IdDict{ValidTimeType,Int}(11=>1))
     col = reshape(col[esample], N, 1)
     defaults = (default(MakeTreatCols())...,)
     _feresiduals!(col, feM, defaults[2:3]...)
@@ -254,9 +249,13 @@ end
     @test combinedargs(MakeTreatCols(), allntargs) ==
         (Dict{Int,Int}(), IdDict{ValidTimeType,Int}())
 
+    allntargs = NamedTuple[(tr=tr, pr=pr), (tr=tr, pr=unspecifiedpr())]
+    @test combinedargs(MakeTreatCols(), allntargs) ==
+        (Dict(-1=>2), IdDict{ValidTimeType,Int}())
+
     df.wave = settime(Date.(hrs.wave), Year(1))
     df.wave_hosp = settime(Date.(hrs.wave_hosp), Year(1), start=Date(7))
-    ret1 = maketreatcols(nt..., typeof(tr), tr.time,
+    ret1 = maketreatcols(nt..., tr.time,
         Dict(-1=>1), IdDict{ValidTimeType,Int}(Date(11)=>1))
     @test ret1.cells[1] == Date.(ret.cells[1])
     @test ret1.cells[2] == Date.(ret.cells[2])
@@ -272,7 +271,7 @@ end
     df.wave = RotatingTimeArray(rot, hrs.wave)
     df.wave_hosp = RotatingTimeArray(rot, hrs.wave_hosp)
     e = rotatingtime((1,1,2), (10,11,11))
-    ret2 = maketreatcols(nt..., typeof(tr), tr.time,
+    ret2 = maketreatcols(nt..., tr.time,
         Dict(-1=>1), IdDict{ValidTimeType,Int}(c=>1 for c in e))
     @test ret2.cells[1] == sort!(append!((rotatingtime(r, ret.cells[1]) for r in (1,2))...))
     rt = append!((rotatingtime(r, 7:11) for r in (1,2))...)
@@ -284,7 +283,7 @@ end
     df.wave = settime(Date.(hrs.wave), Year(1), rotation=rot)
     df.wave_hosp = settime(Date.(hrs.wave_hosp), Year(1), start=Date(7), rotation=rot)
     e = rotatingtime((1,1,2), Date.((10,11,11)))
-    ret3 = maketreatcols(nt..., typeof(tr), tr.time,
+    ret3 = maketreatcols(nt..., tr.time,
         Dict(-1=>1), IdDict{ValidTimeType,Int}(c=>1 for c in e))
     @test ret3.cells[1].time == Date.(ret2.cells[1].time)
     @test ret3.cells[2].time == Date.(ret2.cells[2].time)