I am currently trying to find if there is a causal relationship which exists between cuts to government spending and county court judgements.
I have panel data covering multiple regions across 10 years. This data covers: Local authority spending, benefits spending, uneployment, local gross value added (similar to GDP), education, number of county court judements etc... for each region for each year.
I am planning on using a fixed effects regression but I am unsure about whether I should control for both time and entity effects, time effects or just enetity effects.
I have run the regression for entity fixed effects only and for both entity and time fixed effects and the results are pretty different so I am unsure which to use.
Entity Fixed Effects Only
Oneway (individual) effect Within Model
Call:
plm(formula = lg_ccj_p1000 ~ lg_GVA_p_1000 + lg__16._Unemployed_p1000 +
lg_low_income_housing_benefits_p1000 + lg_Total_Expenditure_no_Education_p_1000 +
lg_Population + lg_Non_Disabled_p_1000 + Dependency.Ratio +
lg_Disability_Related.Benefits_p1000, data = df, model = "within",
index = c("LACcd", "Year"))
Balanced Panel: n = 149, T = 10, N = 1490
Residuals:
Min. 1st Qu. Median 3rd Qu. Max.
-0.5712078 -0.1056597 -0.0054156 0.0953477 0.5314735
Coefficients:
Estimate Std. Error t-value
lg_GVA_p_1000 -0.025601 0.084799 -0.3019
lg__16._Unemployed_p1000 -0.197998 0.023132 -8.5594
lg_low_income_housing_benefits_p1000 -1.445102 0.055498 -26.0387
lg_Total_Expenditure_no_Education_p_1000 0.295909 0.062235 4.7547
lg_Population -1.074991 0.297190 -3.6172
lg_Non_Disabled_p_1000 0.674084 0.201692 3.3422
Dependency.Ratio -0.820800 0.388098 -2.1149
lg_Disability_Related.Benefits_p1000 -0.344142 0.079086 -4.3515
Pr(>|t|)
lg_GVA_p_1000 0.7627697
lg__16._Unemployed_p1000 < 2.2e-16 ***
lg_low_income_housing_benefits_p1000 < 2.2e-16 ***
lg_Total_Expenditure_no_Education_p_1000 2.203e-06 ***
lg_Population 0.0003090 ***
lg_Non_Disabled_p_1000 0.0008544 ***
Dependency.Ratio 0.0346212 *
lg_Disability_Related.Benefits_p1000 1.455e-05 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Total Sum of Squares: 90.753
Residual Sum of Squares: 34.017
R-Squared: 0.62517
Adj. R-Squared: 0.5813
F-statistic: 277.904 on 8 and 1333 DF, p-value: < 2.22e-16
Both Entity and Time Fixed Effects
Twoways effects Within Model
Call:
plm(formula = lg_ccj_p1000 ~ lg_GVA_p_1000 + lg__16._Unemployed_p1000 +
lg_low_income_housing_benefits_p1000 + lg_Total_Expenditure_no_Education_p_1000 +
lg_Population + lg_Non_Disabled_p_1000 + Dependency.Ratio +
lg_Disability_Related.Benefits_p1000, data = df, effect = "twoways",
model = "within", index = c("LACcd", "Year"))
Balanced Panel: n = 149, T = 10, N = 1490
Residuals:
Min. 1st Qu. Median 3rd Qu. Max.
-0.4704793 -0.0668939 -0.0046179 0.0670402 0.4971985
Coefficients:
Estimate Std. Error t-value Pr(>|t|)
lg_GVA_p_1000 0.0099026 0.0594994 0.1664 0.867842
lg__16._Unemployed_p1000 -0.0079207 0.0177763 -0.4456 0.655976
lg_low_income_housing_benefits_p1000 -0.0435300 0.0680741 -0.6394 0.522641
lg_Total_Expenditure_no_Education_p_1000 -0.0876279 0.0476867 -1.8376 0.066349
lg_Population -1.2512911 0.2311960 -5.4123 7.38e-08
lg_Non_Disabled_p_1000 -0.0206642 0.1505369 -0.1373 0.890838
Dependency.Ratio -0.9941714 0.3577568 -2.7789 0.005531
lg_Disability_Related.Benefits_p1000 0.2599478 0.0906636 2.8672 0.004207
lg_GVA_p_1000
lg__16._Unemployed_p1000
lg_low_income_housing_benefits_p1000
lg_Total_Expenditure_no_Education_p_1000 .
lg_Population ***
lg_Non_Disabled_p_1000
Dependency.Ratio **
lg_Disability_Related.Benefits_p1000 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Total Sum of Squares: 16.771
Residual Sum of Squares: 16.292
R-Squared: 0.028568
Adj. R-Squared: -0.092495
F-statistic: 4.86698 on 8 and 1324 DF, p-value: 6.1864e-06
Any help would be appreciated!
