先用了random模型,结果如下:
- > form4<-t_biders ~ lag(t_biders, 1) +lag(log(t_amount+1),1)+ weekday+request_number+overdue_number+repaid_number+nopaid_number+log(loan_amount+1)+log(loan_total+1)+user_month+age+gender+rate+loan_life_month+desc_length+title_length+repay_type+income+firm_scale+edu+occupation+marriage+job_length+grade+log(credit+1)+isestate+isestate_loan+iscar+iscar_loan
- > renren_random <- plm(form4, data = data, model = "random")
- > summary(renren_random)
- Oneway (individual) effect Random Effect Model
- (Swamy-Arora's transformation)
- Call:
- plm(formula = form4, data = data, model = "random")
- Unbalanced Panel: n=342459, T=1-6, N=2039565
- Effects:
- var std.dev share
- idiosyncratic 0.6908 0.8311 0.196
- individual 2.8417 1.6857 0.804
- theta :
- Min. 1st Qu. Median Mean 3rd Qu. Max.
- 0.5578 0.8027 0.8027 0.8023 0.8027 0.8027
- Residuals :
- Min. 1st Qu. Median Mean 3rd Qu. Max.
- -159.000 -0.030 0.003 -0.009 0.030 226.000
- Coefficients :
- Estimate Std. Error t-value Pr(>|t|)
- (Intercept) 1.9205e+01 6.7643e-01 28.3913 < 2.2e-16 ***
- lag(t_biders, 1) 2.5434e-01 6.9954e-04 363.5846 < 2.2e-16 ***
- lag(log(t_amount + 1), 1) 5.5273e-01 3.0442e-03 181.5657 < 2.2e-16 ***
- weekday 3.6981e-03 2.0087e-03 1.8410 0.0656151 .
- request_number -1.7076e-02 2.3580e-03 -7.2415 4.438e-13 ***
- overdue_number -1.5645e-02 1.1122e-02 -1.4067 0.1595182
- repaid_number -2.9952e-01 7.6353e-03 -39.2288 < 2.2e-16 ***
- nopaid_number 5.2061e+00 4.6915e-01 11.0970 < 2.2e-16 ***
- log(loan_amount + 1) 3.2357e-02 3.6205e-03 8.9370 < 2.2e-16 ***
- log(loan_total + 1) -5.7871e-01 4.9410e-03 -117.1241 < 2.2e-16 ***
- user_month -9.8023e-03 7.8359e-04 -12.5095 < 2.2e-16 ***
- age 1.5335e-02 6.7982e-04 22.5572 < 2.2e-16 ***
- gender 1.5732e-03 1.1209e-02 0.1404 0.8883828
- rate -3.9678e-03 1.1880e-03 -3.3400 0.0008379 ***
- loan_life_month -4.3588e-03 4.5372e-04 -9.6068 < 2.2e-16 ***
- desc_length 4.0769e-03 9.7201e-05 41.9433 < 2.2e-16 ***
- title_length -1.0879e-02 1.0969e-03 -9.9178 < 2.2e-16 ***
- repay_type 2.0886e+00 6.6687e-01 3.1319 0.0017369 **
- income 7.3108e-02 2.9737e-03 24.5851 < 2.2e-16 ***
- firm_scale -2.5688e-02 3.8693e-03 -6.6389 3.161e-11 ***
- edu -7.6334e-02 4.7044e-03 -16.2261 < 2.2e-16 ***
- occupation -3.0348e-02 6.2655e-03 -4.8436 1.275e-06 ***
- marriage -8.1764e-03 7.5286e-03 -1.0860 0.2774613
- job_length -4.1467e-02 4.2155e-03 -9.8367 < 2.2e-16 ***
- grade -3.1355e+00 1.1345e-02 -276.3658 < 2.2e-16 ***
- log(credit + 1) 1.8067e-01 2.4535e-03 73.6375 < 2.2e-16 ***
- isestate -5.6168e-02 1.0724e-02 -5.2377 1.626e-07 ***
- isestate_loan -2.3396e-01 1.5060e-02 -15.5347 < 2.2e-16 ***
- iscar 8.3053e-02 1.2520e-02 6.6334 3.280e-11 ***
- iscar_loan -2.8448e-01 2.2205e-02 -12.8120 < 2.2e-16 ***
- ---
- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
- Total Sum of Squares: 3066500
- Residual Sum of Squares: 2430000
- R-Squared : 0.20762
- Adj. R-Squared : 0.20762
- F-statistic: 18421.1 on 29 and 2039535 DF, p-value: < 2.22e-16
然后用within模型,结果只有两个滞后项的系数显示出来了,其他项目的系数都没有,不知道是什么原因?
- > renren_with <- plm(form4, data = data, model = "within")
- > summary(renren_with)
- Oneway (individual) effect Within Model
- Call:
- plm(formula = form4, data = data, model = "within")
- Unbalanced Panel: n=342459, T=1-6, N=2039565
- Residuals :
- Min. 1st Qu. Median 3rd Qu. Max.
- -176 0 0 0 186
- Coefficients :
- Estimate Std. Error t-value Pr(>|t|)
- lag(t_biders, 1) -0.14406663 0.00087849 -163.99 < 2.2e-16 ***
- lag(log(t_amount + 1), 1) 0.33537329 0.00252595 132.77 < 2.2e-16 ***
- ---
- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
- Total Sum of Squares: 1196900
- Residual Sum of Squares: 1172300
- R-Squared : 0.020528
- Adj. R-Squared : 0.017082
- F-statistic: 17784.5 on 2 and 1697104 DF, p-value: < 2.22e-16