Enhancing pooling levels strengthens the risk resilience of healthcare insurance: a case study of basic medical insurance fund operations data in Gansu, China

Table 7 Comparison of Tobit panel data regression results on the operating efficiency of basic medical insurance funds (N = 42)^a

Variable	Urban and rural resident		Urban employee
Variable	RE	FE	RE	FE
Per capita hospitalization cost	-0.0000574***	-0.0000515*	0.00000629	-0.00000574
Per capita hospitalization cost	(-3.35)^b	(-2.13)	(0.28)	(-0.13)
Per capita outpatient cost	0.00139**	0.00114	-0.00157	0.0000742
Per capita outpatient cost	(2.62)	(1.63)	(-1.60)	(0.06)
Per capita fund balance	-0.000132***	-0.000116***	-0.000106***	-0.0000787***
Per capita fund balance	(-3.76)	(-3.49)	(-4.15)	(-3.91)
Per capita cumulative fund balance	0.0000326*	0.0000363*	0.0000204**	0.0000014
Per capita cumulative fund balance	(2.45)	(2.05)	(2.64)	(0.09)
Ln per capita gdp	0.0568	0.0152	0.244***	0.225
Ln per capita gdp	(1.54)	(0.13)	(3.98)	(1.15)
Tertiary hospital proportion	0.038	0.00674	0.00116	0.000247
Tertiary hospital proportion	(0.69)	(1.20)	(0.04)	(0.03)
Actual reimbursement ratio	0.0042	0.00596*	0.0101	-0.0015
Actual reimbursement ratio	(1.22)	(2.14)	(1.94)	(-0.42)
Medical institution density	0.000512	0.000217	-0.00179	-0.00314***
Medical institution density	(0.70)	(0.48)	(-1.73)	(-3.48)
Hausman test^c	χ² =-2.32		χ² =-10.02

Note *P < 0.05, **P < 0.01, ***P < 0.001.
RE represents the random effects panel data Tobit model, while FE represents the fixed effects panel data Tobit model.
^a Given the limited sample size and the collinearity among variables the study did not employ a mixed-effects model.
^bNumbers in parentheses are test statistic Z-values.
^cThe Hausman Test examines whether to choose the FE model when P < 0.05 or the RE model when P > 0.05. The negative value of the Hausman test statistic indicates that the null hypothesis is untenable, and therefore, the FE model should be employed [31].

ISSN: 1471-2458