Optimizing the medical equipment investment in primary care centres in rural China: evidence from a panel threshold model

Table 2 Test results for threshold effect of urbanization in different township health centres (THCs) subgroups

Subgroup		F-value	p-value	Threshold estimates (%)	95% CI
THCs in the neighboring towns of the county seat		47.8	0.05	69.89	(68.87,70.87)
THCs in the remote towns		78.2	< 0.001	5.99	(4.63, 6.75)

Notes: A test for a threshold effect was conducted to determine whether the coeffecients of urbanization degree are the same in each regime. The null hypothesis and the alternatvie hypothesis (the linear versus the single-threshold model) are H₀: β₁ = β₂; H_a: β₁ ≠ β₂. The F statistic is constructed as F=(S₀-S₁)/ σ², where S₁ and S₀ represent the residual sum of squared errors with and without threshold effects, respectively, and σ² represents the variance of the residual. We used bootstrap on the critical values of the F statistic to test the significance of the threshold effect (Pr = I(F > F1)). The method of bootstrapping with 300 samples were applied to obtain the confidential intervals of the threshold values with robust standard errors. THCs = township health centres

ISSN: 1472-6963