Reghdfe: All unit fixed effects dropped.

[Anupam Ghosh]

I am trying to estmate the effect of natural disasters on a set of outcomes and I have baseline covariates interacted with a linear trend (year FE) and county FE in my model. However, on adding the county FE, the results are becoming both insignificant and all the county FE, coefficients are claimed to be redundant. It would be really helpful if someone can clarify why is this.


clear
input long fips double year float(LProp_p Damage12 Damage35 Damage610 Proportion_Black UR Land_Area) double Coast_Dist float PVR
1001 1980 4.7084475 0 0 0 .2232672 .02815284 594.44 213.8443333333333 .1656579
1001 1981 4.7184052 -.21806997 0 0 .2232672 .02835704 594.44 213.8443333333333 .16685946
1001 1982 4.5825167 -.21806997 0 0 .2232672 .02831013 594.44 213.8443333333333 .16658343
1001 1983 4.7491283 -.21806997 -.21806997 0 .2232672 .02829335 594.44 213.8443333333333 .1664847
1001 1984 4.461182 -.21806997 -.21806997 0 .2232672 .02822907 594.44 213.8443333333333 .16610645
1001 1985 4.263832 -.21806997 -.21806997 0 .2232672 .028125776 594.44 213.8443333333333 .16549863
1001 1986 4.6001096 -.21806997 -.21806997 -.21806997 .2232672 .02757258 594.44 213.8443333333333 .1622435
1001 1987 4.5548244 -.21806997 -.21806997 -.21806997 .2232672 .027265076 594.44 213.8443333333333 .1604341
1001 1988 4.57803 -.21806997 -.21806997 -.21806997 .2232672 .026964355 594.44 213.8443333333333 .15866457
1001 1989 4.491393 -.21806997 -.21806997 -.21806997 .2232672 .02667961 594.44 213.8443333333333 .15698905
1001 1990 4.510293 -.21806997 -.21806997 -.21806997 .2232672 .02640235 594.44 213.8443333333333 .1553576
1001 1991 4.85509 -.21806997 -.21806997 -.21806997 .2232672 .025906883 594.44 213.8443333333333 .15244216
1001 1992 4.780004 -.21806997 -.21806997 -.21806997 .2232672 .02520495 594.44 213.8443333333333 .1483118
1001 1993 4.708352 -.21806997 -.21806997 -.21806997 .2232672 .02454469 594.44 213.8443333333333 .1444267
1001 1994 4.564984 -.21806997 -.21806997 -.21806997 .2232672 .02375216 594.44 213.8443333333333 .13976327
1001 1995 4.637351 -.21806997 -.21806997 -.21806997 .2232672 .023189815 594.44 213.8443333333333 .13645428
1001 1996 4.6606817 3.2465315 -.21806997 -.21806997 .2232672 .02255826 594.44 213.8443333333333 .13273808
1001 1997 4.6021194 3.2465315 -.21806997 -.21806997 .2232672 .02199428 594.44 213.8443333333333 .12941946
1001 1998 4.3919597 -.21806997 3.2465315 -.21806997 .2232672 .02154087 594.44 213.8443333333333 .12675153
1001 1999 4.440213 -.21806997 3.2465315 -.21806997 .2232672 .02111119 594.44 213.8443333333333 .12422317
1001 2000 4.5178523 -.21806997 3.2465315 -.21806997 .2232672 .0206038 594.44 213.8443333333333 .1212376
1001 2001 4.434579 -.21806997 -.21806997 3.2465315 .2232672 .020205395 594.44 213.8443333333333 .11889327
1001 2002 3.916569 -.21806997 -.21806997 3.2465315 .2232672 .019756475 594.44 213.8443333333333 .11625171
1001 2003 4.5243864 -.21806997 -.21806997 3.2465315 .2232672 .01938034 594.44 213.8443333333333 .11403846
1001 2004 4.558999 -.21806997 -.21806997 3.2465315 .2232672 .018752843 594.44 213.8443333333333 .1103461
1001 2005 4.562413 -.2141202 -.21806997 3.2465315 .2232672 .018258315 594.44 213.8443333333333 .10743619
1001 2006 4.451425 -.2141202 -.21806997 -.21806997 .2232672 .017670667 594.44 213.8443333333333 .10397834
1001 2007 4.558672 -.21806997 -.2141202 -.21806997 .2232672 .017307509 594.44 213.8443333333333 .10184143
1001 2008 4.518067 -.21806997 -.2141202 -.21806997 .2232672 .017024232 594.44 213.8443333333333 .10017456
1001 2009 3.8695276 -.21806997 -.2141202 -.21806997 .2232672 .016754411 594.44 213.8443333333333 .09858686
1001 2010 4.0272512 -.21806997 -.21806997 -.2141202 .2232672 .016550189 594.44 213.8443333333333 .09738518
1001 2011 4.4832673 -.21806997 -.21806997 -.2141202 .2232672 .016334102 594.44 213.8443333333333 .09611367
1001 2012 4.601538 -.21806997 -.21806997 -.2141202 .2232672 .016325576 594.44 213.8443333333333 .0960635
1001 2013 4.573265 -.21806997 -.21806997 -.2141202 .2232672 .016312363 594.44 213.8443333333333 .09598576
1001 2014 4.561709 -.21806997 -.21806997 -.2141202 .2232672 .016183713 594.44 213.8443333333333 .09522875
1001 2015 4.184183 -.21806997 -.21806997 -.21806997 .2232672 .016106691 594.44 213.8443333333333 .09477554
1001 2016 4.40002 -.21806997 -.21806997 -.21806997 .2232672 .015932405 594.44 213.8443333333333 .09375
1001 2017 4.337876 -.21806997 -.21806997 -.21806997 .2232672 .015824275 594.44 213.8443333333333 .09311374
1001 2018 4.80581 -.21806997 -.21806997 -.21806997 .2232672 .015706444 594.44 213.8443333333333 .09242038
1001 2019 4.419293 -.21806997 -.21806997 -.21806997 .2232672 .01556654 594.44 213.8443333333333 .09159715
1001 2020 3.519463 -.21806997 -.21806997 -.21806997 .2232672 .015398459 594.44 213.8443333333333 .09060813
1003 1980 4.925783 0 0 0 .1522375 .03107069 1589.78 21.390600000000003 .15979937
1003 1981 4.869672 -.25621065 0 0 .1522375 .03055099 1589.78 21.390600000000003 .1571265
1003 1982 4.7569404 -.25621065 0 0 .1522375 .02979473 1589.78 21.390600000000003 .15323697
1003 1983 4.677062 -.25621065 -.25621065 0 .1522375 .029209336 1589.78 21.390600000000003 .15022625
1003 1984 4.6053376 -.25621065 -.25621065 0 .1522375 .02827568 1589.78 21.390600000000003 .14542437
1003 1985 4.668983 -.25621065 -.25621065 0 .1522375 .02743756 1589.78 21.390600000000003 .14111383
1003 1986 4.7228713 .05620267 -.25621065 -.25621065 .1522375 .026865115 1589.78 21.390600000000003 .1381697
1003 1987 4.654547 .05620267 -.25621065 -.25621065 .1522375 .02631579 1589.78 21.390600000000003 .13534448
1003 1988 4.655588 -.25621065 .05620267 -.25621065 .1522375 .02591599 1589.78 21.390600000000003 .13328825
1003 1989 4.568747 -.25621065 .05620267 -.25621065 .1522375 .025498165 1589.78 21.390600000000003 .13113937
1003 1990 4.477442 -.25621065 .05620267 -.25621065 .1522375 .024789045 1589.78 21.390600000000003 .1274923
1003 1991 4.390391 -.25621065 -.25621065 .05620267 .1522375 .0239504 1589.78 21.390600000000003 .12317906
1003 1992 4.2924056 -.25621065 -.25621065 .05620267 .1522375 .023012336 1589.78 21.390600000000003 .11835452
1003 1993 4.5450416 -.25621065 -.25621065 .05620267 .1522375 .022016587 1589.78 21.390600000000003 .1132333
1003 1994 4.3168235 -.25621065 -.25621065 .05620267 .1522375 .021044053 1589.78 21.390600000000003 .10823146
1003 1995 4.370506 -.25621065 -.25621065 .05620267 .1522375 .020290166 1589.78 21.390600000000003 .10435416
1003 1996 4.301359 .008997701 -.25621065 -.25621065 .1522375 .01955953 1589.78 21.390600000000003 .10059644
1003 1997 4.478709 .008997701 -.25621065 -.25621065 .1522375 .018845456 1589.78 21.390600000000003 .09692388
1003 1998 4.238174 -.1694204 .008997701 -.25621065 .1522375 .018245514 1589.78 21.390600000000003 .09383833
1003 1999 4.4162526 -.16616987 .008997701 -.25621065 .1522375 .017832866 1589.78 21.390600000000003 .09171604
1003 2000 3.751883 -.16616987 .008997701 -.25621065 .1522375 .017355068 1589.78 21.390600000000003 .08925868
1003 2001 3.268167 -.25621065 -.16616987 .008997701 .1522375 .016931837 1589.78 21.390600000000003 .08708197
1003 2002 3.827456 -.25621065 -.16616987 .008997701 .1522375 .016579142 1589.78 21.390600000000003 .08526802
1003 2003 4.068827 -.25614622 -.16616987 .008997701 .1522375 .016190458 1589.78 21.390600000000003 .08326898
1003 2004 3.873265 -.25614622 -.25621065 .008997701 .1522375 .015697593 1589.78 21.390600000000003 .08073413
1003 2005 3.2498794 4.534098 -.25614622 .008997701 .1522375 .015238864 1589.78 21.390600000000003 .07837486
1003 2006 3.5440485 4.534098 -.25614622 -.16616987 .1522375 .014590682 1589.78 21.390600000000003 .0750412
1003 2007 4.187344 .2396466 4.534098 -.16616987 .1522375 .014228208 1589.78 21.390600000000003 .07317696
1003 2008 3.725557 -.25621065 4.534098 -.16616987 .1522375 .013951213 1589.78 21.390600000000003 .07175235
1003 2009 3.824602 -.25621065 4.534098 -.25614622 .1522375 .013672899 1589.78 21.390600000000003 .07032095
1003 2010 3.657632 -.25621065 .2396466 4.534098 .1522375 .013388496 1589.78 21.390600000000003 .068858236
1003 2011 4.0903153 -.25621065 -.25621065 4.534098 .1522375 .013115122 1589.78 21.390600000000003 .06745225
1003 2012 3.9761336 -.25621065 -.25621065 4.534098 .1522375 .012844613 1589.78 21.390600000000003 .066061005
1003 2013 3.884495 -.25621065 -.25621065 4.534098 .1522375 .012512498 1589.78 21.390600000000003 .0643529
1003 2014 3.760057 -.25621065 -.25621065 4.534098 .1522375 .012221552 1589.78 21.390600000000003 .06285655
1003 2015 3.669851 -.25621065 -.25621065 .2396466 .1522375 .01198181 1589.78 21.390600000000003 .06162353
1003 2016 3.650986 -.25621065 -.25621065 -.25621065 .1522375 .01169231 1589.78 21.390600000000003 .06013461
1003 2017 3.686227 -.25621065 -.25621065 -.25621065 .1522375 .0113987 1589.78 21.390600000000003 .05862454
1003 2018 4.0964565 -.25539538 -.25621065 -.25621065 .1522375 .011105527 1589.78 21.390600000000003 .05711673
1003 2019 3.582316 -.18740447 -.25621065 -.25621065 .1522375 .010813218 1589.78 21.390600000000003 .05561335
1003 2020 2.58499 -.18740447 -.25539538 -.25621065 .1522375 .01051801 1589.78 21.390600000000003 .05409508
1005 1980 1.6949587 0 0 0 .4435507 .0310441 884.88 173.12466666666668 .30554995
1005 1981 1.231052 -.16834024 0 0 .4435507 .03103783 884.88 173.12466666666668 .3054882
1005 1982 1.44198 -.16834024 0 0 .4435507 .03099899 884.88 173.12466666666668 .30510595
1005 1983 1.104082 -.16834024 -.16834024 0 .4435507 .03097274 884.88 173.12466666666668 .3048476
1005 1984 . -.16834024 -.16834024 0 .4435507 .03077663 884.88 173.12466666666668 .3029174
1005 1985 . -.16834024 -.16834024 0 .4435507 .03071877 884.88 173.12466666666668 .3023479
1005 1986 . -.16834024 -.16834024 -.16834024 .4435507 .030791435 884.88 173.12466666666668 .3030631
1005 1987 . -.16834024 -.16834024 -.16834024 .4435507 .030591516 884.88 173.12466666666668 .3010954
1005 1988 . -.16834024 -.16834024 -.16834024 .4435507 .030438745 884.88 173.12466666666668 .2995918
1005 1989 1.2143424 -.16834024 -.16834024 -.16834024 .4435507 .0303114 884.88 173.12466666666668 .2983384
1005 1990 .57894766 -.16834024 -.16834024 -.16834024 .4435507 .030111743 884.88 173.12466666666668 .29637325
1005 1991 .3201055 -.16834024 -.16834024 -.16834024 .4435507 .02897457 884.88 173.12466666666668 .28518072
1005 1992 1.1715804 -.16834024 -.16834024 -.16834024 .4435507 .028506737 884.88 173.12466666666668 .28057608
1005 1993 1.8107288 -.16834024 -.16834024 -.16834024 .4435507 .028058894 884.88 173.12466666666668 .2761682
1005 1994 1.799209 -.16834024 -.16834024 -.16834024 .4435507 .02767468 884.88 173.12466666666668 .27238658
1005 1995 1.9605528 -.16834024 -.16834024 -.16834024 .4435507 .02757234 884.88 173.12466666666668 .27137932
1005 1996 1.017731 .3378575 -.16834024 -.16834024 .4435507 .027139727 884.88 173.12466666666668 .26712134
1005 1997 1.242276 .3378575 -.16834024 -.16834024 .4435507 .02702798 884.88 173.12466666666668 .26602146
end
[/CODE]


. reghdfe LProp_p Damage12 Damage35 Damage610, absorb(i.fips i.year#c.Proportion_Black i.year#c.UR i.year#c.Land_Area i.year#c.Coast_Dist i.yea
> r#c.PVR ) vce(cluster fips)
(MWFE estimator converged in 34 iterations)

HDFE Linear regression Number of obs = 44,594
Absorbing 6 HDFE groups F( 3, 1102) = 2.75
Statistics robust to heteroskedasticity Prob > F = 0.0414
R-squared = 0.7296
Adj R-squared = 0.7214
Within R-sq. = 0.0004
Number of clusters (fips) = 1,103 Root MSE = 0.4652

(Std. err. adjusted for 1,103 clusters in fips)

Robust
LProp_p Coefficient std. err. t P>t [95% conf. interval]

Damage12 -.0020696 .0031253 -0.66 0.508 -.0082017 .0040626
Damage35 -.0020033 .002866 -0.70 0.485 -.0076267 .0036202
Damage610 -.0070167 .0026899 -2.61 0.009 -.0122947 -.0017387
_cons 4.117899 .0011508 3578.30 0.000 4.115641 4.120157


Absorbed degrees of freedom:

Absorbed FE Categories - Redundant = Num. Coefs
-
fips 1103 1103 0 *
year#c.Proportion_Black 41 0 41 ?
year#c.UR 41 0 41 ?
year#c.Land_Area 41 0 41 ?
year#c.Coast_Dist 41 0 41 ?
year#c.PVR 41 0 41 ?

[Carlo Lazzaro]

Anupam:
the reasons for omission that spring to my mind are:
1) time-invariant predictors;
2) collinearity with the fixed effect(s), that look too many in your regression specification.

[Anupam Ghosh]

The interaction terms "i.year#c.Proportion_Black i.year#c.UR i.year#c.Land_Area i.year#c.Coast_Dist i.year#c.PVR" are time invariant as they are observed only for the baseline year 1980. Thus, I am trying to gauge the term X#a_t. Should I not include it? Also, can you advise, if I should weight my regression by total population or not, the outcome variable is already in per-capita terms. Weighting improves statistical significance.

[Carlo Lazzaro]

Anupam:
your interatction is very elaborated and difficult to diagnosing.
I would code a more parsimonious on.
I would give weighting a shot.

[Anupam Ghosh]

Carlo,
I am not quite sure what do you mean by coding a more parsimonious specification. I am following the literature in this specification. Also since, the county FEs are being redundant, should I drop them? That vastly changes the estimate.

[Carlo Lazzaro]

Anupam:
I meant am interaction that includes a lower number of predictors, if feasible.
Country fe are clearly time-invariant. I think you can drop them.