Good morning,
I am using reghdfe in Stata 16.1 for Windows.
I have a panel dataset with about 10,000 observations. My dependent variable (IIC_110_hosp_new) and independent variable of interest (iv) are both indicator variables. I find a significant relation in levels but not in differences. I have seen several posts with this same question but no suggestions as to why the relation would not hold in differences as expected.
Any guidance is much appreciated. Thanks,
Ginger
I am using reghdfe in Stata 16.1 for Windows.
I have a panel dataset with about 10,000 observations. My dependent variable (IIC_110_hosp_new) and independent variable of interest (iv) are both indicator variables. I find a significant relation in levels but not in differences. I have seen several posts with this same question but no suggestions as to why the relation would not hold in differences as expected.
Any guidance is much appreciated. Thanks,
Ginger
Code:
reghdfe IIC_110_hosp_new iv lrev cmi growth_bedsize bedsize num acqd age_all_apps age_all_apps2 pct_medicaid pct_medicare county_HHI_all purchasegrou
> p backfill_academic backfill_profit religious specialty_hosp rural, absorb (cbsa data_year sysid) cluster(sysid)
(MWFE estimator converged in 56 iterations)
HDFE Linear regression Number of obs = 10,775
Absorbing 3 HDFE groups F( 18, 347) = 7.43
Statistics robust to heteroskedasticity Prob > F = 0.0000
R-squared = 0.6484
Adj R-squared = 0.6184
Within R-sq. = 0.0515
Number of clusters (sysid) = 348 Root MSE = 0.3055
(Std. Err. adjusted for 348 clusters in sysid)
-----------------------------------------------------------------------------------
| Robust
IIC_110_hosp_new | Coef. Std. Err. t P>|t| [95% Conf. Interval]
------------------+----------------------------------------------------------------
iv | .0909766 .0170995 5.32 0.000 .0573449 .1246082
lrev | .0181682 .0145263 1.25 0.212 -.0104025 .0467389
cmi | -.0213183 .0423344 -0.50 0.615 -.1045827 .0619461
growth_bedsize | -.008643 .0234295 -0.37 0.712 -.0547247 .0374386
bedsize | .0001105 .0000595 1.86 0.064 -6.55e-06 .0002276
num | .0017873 .001612 1.11 0.268 -.0013832 .0049578
acqd | -.2150124 .0452321 -4.75 0.000 -.303976 -.1260489
age_all_apps | .0173815 .0167258 1.04 0.299 -.0155151 .0502782
age_all_apps2 | -.0015197 .0009383 -1.62 0.106 -.0033652 .0003257
pct_medicaid | .0003359 .0011223 0.30 0.765 -.0018714 .0025433
pct_medicare | .0005903 .0008958 0.66 0.510 -.0011715 .0023521
county_HHI_all | -5.01e-06 3.99e-06 -1.25 0.211 -.0000129 2.85e-06
purchasegroup | .0118717 .0181146 0.66 0.513 -.0237565 .0474999
backfill_academic | -.0073197 .0166686 -0.44 0.661 -.0401039 .0254646
backfill_profit | .0179319 .0445519 0.40 0.688 -.0696938 .1055576
religious | .0331071 .0200178 1.65 0.099 -.0062644 .0724786
specialty_hosp | -.0140059 .0606577 -0.23 0.818 -.1333089 .105297
rural | .0170688 .034588 0.49 0.622 -.0509598 .0850973
_cons | .2103599 .2029035 1.04 0.301 -.1887156 .6094354
-----------------------------------------------------------------------------------
Absorbed degrees of freedom:
-----------------------------------------------------+
Absorbed FE | Categories - Redundant = Num. Coefs |
-------------+---------------------------------------|
cbsa | 475 0 475 |
data_year | 9 1 8 |
sysid | 348 348 0 *|
-----------------------------------------------------+
* = FE nested within cluster; treated as redundant for DoF computation
.
. reghdfe d.IIC_110_hosp_new d.(iv lrev cmi growth_bedsize bedsize num acqd age_all_apps age_all_apps2 pct_medicaid pct_medicare county_HHI_all purchas
> egroup backfill_academic backfill_profit religious specialty_hosp rural), absorb (cbsa data_year sysid) cluster(sysid)
(dropped 39 singleton observations)
(MWFE estimator converged in 46 iterations)
note: D.backfill_academic omitted because of collinearity
note: D.backfill_profit omitted because of collinearity
note: D.religious omitted because of collinearity
note: D.specialty_hosp omitted because of collinearity
note: D.rural omitted because of collinearity
HDFE Linear regression Number of obs = 9,069
Absorbing 3 HDFE groups F( 13, 318) = 1.75
Statistics robust to heteroskedasticity Prob > F = 0.0508
R-squared = 0.0990
Adj R-squared = 0.0131
Within R-sq. = 0.0226
Number of clusters (sysid) = 319 Root MSE = 0.2860
(Std. Err. adjusted for 319 clusters in sysid)
-----------------------------------------------------------------------------------
D. | Robust
IIC_110_hosp_new | Coef. Std. Err. t P>|t| [95% Conf. Interval]
------------------+----------------------------------------------------------------
iv |
D1. | -.0050465 .0112212 -0.45 0.653 -.0271237 .0170307
|
lrev |
D1. | .0094508 .0119898 0.79 0.431 -.0141384 .0330401
|
cmi |
D1. | .0615533 .0611378 1.01 0.315 -.0587324 .181839
|
growth_bedsize |
D1. | .0034839 .0137963 0.25 0.801 -.0236598 .0306275
|
bedsize |
D1. | .0000669 .0001486 0.45 0.653 -.0002255 .0003593
|
num |
D1. | -.0004232 .0007383 -0.57 0.567 -.0018758 .0010294
|
acqd |
D1. | -.1594614 .0378615 -4.21 0.000 -.233952 -.0849708
|
age_all_apps |
D1. | -.0424908 .0204681 -2.08 0.039 -.0827608 -.0022209
|
age_all_apps2 |
D1. | .0019362 .0011889 1.63 0.104 -.0004028 .0042752
|
pct_medicaid |
D1. | .0004412 .001004 0.44 0.661 -.0015342 .0024166
|
pct_medicare |
D1. | .0018019 .0011286 1.60 0.111 -.0004186 .0040223
|
county_HHI_all |
D1. | 1.77e-06 5.29e-06 0.33 0.738 -8.63e-06 .0000122
|
purchasegroup |
D1. | .0070085 .0115273 0.61 0.544 -.0156709 .029688
|
backfill_academic |
D1. | 0 (omitted)
|
backfill_profit |
D1. | 0 (omitted)
|
religious |
D1. | 0 (omitted)
|
specialty_hosp |
D1. | 0 (omitted)
|
rural |
D1. | 0 (omitted)
|
_cons | .0119001 .0032821 3.63 0.000 .0054426 .0183575
-----------------------------------------------------------------------------------
Absorbed degrees of freedom:
-----------------------------------------------------+
Absorbed FE | Categories - Redundant = Num. Coefs |
-------------+---------------------------------------|
cbsa | 451 0 451 |
data_year | 8 1 7 |
sysid | 319 319 0 *|
-----------------------------------------------------+
* = FE nested within cluster; treated as redundant for DoF computation
Comment