12.2.3 Rによる固定効果モデルの推定
import delimited "/Users/kitagawa/Documents/bookwriting/empanar/prefecture.csv", case(preserve) clear
sort pref_id year
list in 1/4
+---------------------------------+
| pref year suicide unemp |
|---------------------------------|
1. | 北海道 1997 19.6 3.7 |
2. | 青森 1997 26.5 3.9 |
3. | 岩手 1997 25.8 2.4 |
4. | 宮城 1997 18.6 3.1 |
+---------------------------------+
encode pref, gen(pref_id)
regress suicide unemp ibn.pref_id, noconstant
Source | SS df MS Number of obs = 1,081
-------------+---------------------------------- F(48, 1033) = 1718.59
Model | 567038.287 48 11813.2976 Prob > F = 0.0000
Residual | 7100.67017 1,033 6.87383366 R-squared = 0.9876
-------------+---------------------------------- Adj R-squared = 0.9871
Total | 574138.957 1,081 531.118369 Root MSE = 2.6218
------------------------------------------------------------------------------
suicide | Coefficient Std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
unemp | 3.066965 .086185 35.59 0.000 2.897847 3.236083
|
pref_id |
三重 | 10.84608 .6028334 17.99 0.000 9.663159 12.02899
京都 | 6.187962 .6653313 9.30 0.000 4.882407 7.493517
佐賀 | 11.35607 .6222227 18.25 0.000 10.13511 12.57704
兵庫 | 7.097472 .6744254 10.52 0.000 5.774072 8.420873
北海道 | 8.711331 .6837609 12.74 0.000 7.369612 10.05305
千葉 | 8.475695 .6345959 13.36 0.000 7.230451 9.720939
和歌山 | 13.67467 .61624 22.19 0.000 12.46544 14.88389
埼玉 | 7.598172 .6548335 11.60 0.000 6.313216 8.883127
大分 | 10.90964 .6295185 17.33 0.000 9.674362 12.14492
大阪 | 5.509389 .7203893 7.65 0.000 4.095796 6.922983
奈良 | 6.245202 .6433229 9.71 0.000 4.982833 7.507571
宮城 | 7.912826 .6759649 11.71 0.000 6.586405 9.239247
宮崎 | 15.64264 .6363151 24.58 0.000 14.39402 16.89126
富山 | 15.37911 .6037843 25.47 0.000 14.19433 16.5639
山口 | 12.98887 .6155497 21.10 0.000 11.781 14.19675
山形 | 15.64775 .6116364 25.58 0.000 14.44755 16.84794
山梨 | 13.30341 .6096331 21.82 0.000 12.10715 14.49967
岐阜 | 12.6032 .6001765 21.00 0.000 11.4255 13.7809
岡山 | 7.968494 .6280375 12.69 0.000 6.736119 9.200869
岩手 | 18.20351 .6363151 28.61 0.000 16.95489 19.45213
島根 | 18.44705 .5911101 31.21 0.000 17.28713 19.60696
広島 | 9.585032 .6258364 15.32 0.000 8.356976 10.81309
徳島 | 7.288137 .6378544 11.43 0.000 6.036499 8.539775
愛媛 | 11.19369 .6313849 17.73 0.000 9.954745 12.43263
愛知 | 8.684236 .615722 14.10 0.000 7.476027 9.892444
新潟 | 17.17808 .624745 27.50 0.000 15.95217 18.404
東京 | 7.037263 .6610885 10.64 0.000 5.740034 8.334493
栃木 | 12.16242 .6263843 19.42 0.000 10.93329 13.39155
沖縄 | 2.537746 .7875862 3.22 0.001 .9922947 4.083197
滋賀 | 9.59004 .6148622 15.60 0.000 8.383518 10.79656
熊本 | 9.569775 .6533933 14.65 0.000 8.287645 10.8519
石川 | 10.60312 .6097991 17.39 0.000 9.406536 11.79971
神奈川 | 6.676491 .6457045 10.34 0.000 5.409449 7.943533
福井 | 12.5827 .5906834 21.30 0.000 11.42363 13.74178
福岡 | 6.669192 .7033529 9.48 0.000 5.289028 8.049355
福島 | 11.93389 .645107 18.50 0.000 10.66802 13.19976
秋田 | 20.34367 .6564885 30.99 0.000 19.05546 21.63187
群馬 | 13.10188 .6207967 21.10 0.000 11.88371 14.32005
茨城 | 10.64122 .6315724 16.85 0.000 9.401912 11.88053
長崎 | 9.680528 .6489149 14.92 0.000 8.407186 10.95387
長野 | 12.76867 .6047422 21.11 0.000 11.58201 13.95533
青森 | 12.89297 .7017055 18.37 0.000 11.51604 14.2699
静岡 | 10.31733 .6091365 16.94 0.000 9.122044 11.51262
香川 | 8.925075 .6199111 14.40 0.000 7.708646 10.1415
高知 | 11.74309 .6569038 17.88 0.000 10.45407 13.0321
鳥取 | 11.97814 .6167596 19.42 0.000 10.7679 13.18839
鹿児島 | 11.66578 .6441142 18.11 0.000 10.40186 12.9297
------------------------------------------------------------------------------
bysort pref_id: egen suicidebar = mean(suicide)
bysort pref_id: egen unempbar = mean(unemp)
generate suicide2 = suicide - suicidebar
generate unemp2 = unemp - unempbar
regress suicide2 unemp2, noconstant
Source | SS df MS Number of obs = 1,081
-------------+---------------------------------- F(1, 1080) = 1323.97
Model | 8704.6819 1 8704.6819 Prob > F = 0.0000
Residual | 7100.67017 1,080 6.5746946 R-squared = 0.5507
-------------+---------------------------------- Adj R-squared = 0.5503
Total | 15805.3521 1,081 14.6210472 Root MSE = 2.5641
------------------------------------------------------------------------------
suicide2 | Coefficient Std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
unemp2 | 3.066965 .0842889 36.39 0.000 2.901577 3.232354
------------------------------------------------------------------------------
reghdfe suicide unemp, absorb(pref_id year) vce(cluster pref)
(MWFE estimator converged in 2 iterations)
HDFE Linear regression Number of obs = 1,081
Absorbing 2 HDFE groups F( 1, 46) = 6.66
Statistics robust to heteroskedasticity Prob > F = 0.0131
R-squared = 0.8710
Adj R-squared = 0.8621
Within R-sq. = 0.0228
Number of clusters (pref) = 47 Root MSE = 1.8218
(Std. err. adjusted for 47 clusters in pref)
------------------------------------------------------------------------------
| Robust
suicide | Coefficient std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
unemp | .7708683 .298724 2.58 0.013 .1695681 1.372168
_cons | 19.59688 1.132139 17.31 0.000 17.318 21.87576
------------------------------------------------------------------------------
Absorbed degrees of freedom:
-----------------------------------------------------+
Absorbed FE | Categories - Redundant = Num. Coefs |
-------------+---------------------------------------|
pref_id | 47 47 0 *|
year | 23 0 23 |
-----------------------------------------------------+
* = FE nested within cluster; treated as redundant for DoF computation
list pref_id __hdfe1__ if mod(_n, 23) == 0
+----------------------+
| pref_id __hdfe1__ |
|----------------------|
23. | 三重 -1.9823115 |
46. | 京都 -3.3060939 |
69. | 佐賀 -.32426722 |
92. | 兵庫 -1.9772961 |
115. | 北海道 .05584997 |
|----------------------|
138. | 千葉 -2.5357812 |
161. | 和歌山 1.6549044 |
184. | 埼玉 -2.3950357 |
207. | 大分 -.37137483 |
230. | 大阪 -1.5887391 |
|----------------------|
253. | 奈良 -4.3170379 |
276. | 宮城 -1.0920616 |
299. | 宮崎 4.7210108 |
322. | 富山 2.6106225 |
345. | 山口 .92918042 |
|----------------------|
368. | 山形 3.3584413 |
391. | 山梨 .89431274 |
414. | 岐阜 -.39489948 |
437. | 岡山 -3.392388 |
460. | 岩手 7.2818804 |
|----------------------|
483. | 島根 4.8399829 |
506. | 広島 -1.8956471 |
529. | 徳島 -3.5536278 |
552. | 愛媛 .01250052 |
575. | 愛知 -3.3654756 |
|----------------------|
598. | 新潟 5.6375061 |
621. | 東京 -2.656453 |
644. | 栃木 .71168929 |
667. | 沖縄 -1.9548115 |
690. | 滋賀 -2.5095867 |
|----------------------|
713. | 熊本 -.49331359 |
736. | 石川 -1.7959955 |
759. | 神奈川 -3.7659529 |
782. | 福井 -1.0543099 |
805. | 福岡 -1.1377319 |
|----------------------|
828. | 福島 1.4614935 |
851. | 秋田 10.430325 |
874. | 群馬 1.3416761 |
897. | 茨城 -.5299815 |
920. | 長崎 -.60218704 |
|----------------------|
943. | 長野 .06007812 |
966. | 青森 5.0161638 |
989. | 静岡 -2.1217193 |
1012. | 香川 -2.8850441 |
1035. | 高知 1.8497092 |
|----------------------|
1058. | 鳥取 -.01167187 |
1081. | 鹿児島 1.1434689 |
+----------------------+
sort pref_id year
list year __hdfe2__ in 1/23
+-------------------+
| year __hdfe2__ |
|-------------------|
1. | 1997 -2.1605963 |
2. | 1998 3.4228034 |
3. | 1999 2.4048345 |
4. | 2000 1.6042981 |
5. | 2001 .79790084 |
|-------------------|
6. | 2002 1.4016126 |
7. | 2003 3.3661109 |
8. | 2004 2.1530947 |
9. | 2005 2.5850204 |
10. | 2006 2.6350294 |
|-------------------|
11. | 2007 3.0396469 |
12. | 2008 2.4702245 |
13. | 2009 2.0611943 |
14. | 2010 .68827919 |
15. | 2011 .32122874 |
|-------------------|
16. | 2012 -1.3296468 |
17. | 2013 -1.2747591 |
18. | 2014 -2.3618466 |
19. | 2015 -3.1123704 |
20. | 2016 -4.336074 |
|-------------------|
21. | 2017 -4.5363731 |
22. | 2018 -4.8443857 |
23. | 2019 -4.9952265 |
+-------------------+
quietly reghdfe suicide unemp, absorb(year)
estimates store POLS
quietly reghdfe suicide unemp, absorb(year pref_id) vce(cluster pref_id)
estimates store TWFE
etable, estimates(POLS TWFE) keep(unemp) showstars showstarsnote column(estimates) title(Dependent Var. = Suicide Rate)
Dependent Var. = Suicide Rate
--------------------------
POLS TWFE
--------------------------
unemp 0.513 ** 0.771 *
(0.130) (0.299)
N 1081 1081
--------------------------
** p<.01, * p<.05