Table 3. Estimation results of the probit and recursive bivariate probit models for predicting poor self-rated healtha (N = 1,352,788).
| Regression model | Probit | Bivariate probitb | ||||
|
| ||||||
| Dependent variable | Poor self-rated health | Poor self-rated health | Work | |||
|
| ||||||
| dy/dxc | 95% CI | dy/dx | 95% CI | dy/dx | 95% CI | |
| Work | −0.064 | (−0.066, −0.063) | −0.067 | (−0.072, −0.062) | ||
| Exogenous variables | ||||||
| Mandatory retirement | −0.115 | (−0.118, −0.112) | ||||
| Eligibility of basic pension benefit | −0.092 | (−0.095, −0.089) | ||||
| Eligibility of wage-proportional pension benefit | −0.064 | (−0.067, −0.061) | ||||
| Prefecture-level variables | ||||||
| Job offers ratiod | 0.004 | (−0.001, 0.009) | ||||
| Primary industryd | 0.279 | (0.157, 0.402) | ||||
| Second industryd | 0.618 | (0.496, 0.739) | ||||
| Individual-level covariates | ||||||
| Female | −0.016 | (−0.017, −0.015) | −0.017 | (−0.019, −0.015) | −0.297 | (−0.299, −0.296) |
| Aged | −0.022 | (−0.026, −0.017) | 0.019 | (0.016, 0.021) | −0.107 | (−0.111, −0.104) |
| Age-squared/100d | 0.019 | (0.015, 0.023) | 0.006 | (0.003, 0.008) | −0.034 | (−0.037, −0.031) |
| Marital status | ||||||
| Unmarried | 0.027 | (0.024, 0.031) | 0.027 | (0.023, 0.031) | −0.166 | (−0.171, −0.162) |
| Divorced/separated | 0.005 | (0.003, 0.008) | 0.007 | (0.004, 0.009) | −0.045 | (−0.048, −0.042) |
| Household structure | ||||||
| Alone | 0.010 | (0.007, 0.013) | 0.009 | (0.006, 0.013) | 0.109 | (0.105, 0.113) |
| Wife and husband and an unmarried child(ren) | −0.003 | (−0.007, −0.013) | −0.004 | (−0.006, −0.002) | 0.053 | (0.050, 0.055) |
| Single parent and an unmarried child(ren) | 0.014 | (0.010, 0.018) | 0.012 | (0.008, 0.017) | 0.128 | (0.123, 0.134) |
| Three-generation | 0.002 | (0.000, 0.004) | 0.002 | (0.000, 0.004) | 0.077 | (0.075, 0.080) |
| Other | 0.026 | (0.024, 0.028) | 0.025 | (0.023, 0.027) | 0.076 | (0.073, 0.078) |
| Household expenditure and housing | ||||||
| Expenditure: 1st quartile (lowest) | 0.014 | (0.024, 0.028) | 0.014 | (0.012, 0.016) | 0.004 | (0.001, 0.006) |
| Expenditure: 2nd quartile | 0.000 | (0.024, 0.028) | 0.000 | (−0.001, 0.002) | −0.011 | (−0.013, −0.009) |
| Expenditure: 3rd quartile | −0.001 | (−0.003, 0.028) | −0.001 | (−0.003, 0.001) | −0.018 | (−0.020, −0.016) |
| Expenditure: unknown/unanswered | 0.004 | (0.024, 0.028) | 0.004 | (0.002, 0.007) | −0.027 | (−0.030, −0.023) |
| Having own house | −0.039 | (−0.041, −0.037) | −0.038 | (−0.040, 0.036) | 0.008 | (0.005, 0.010) |
CI, confidence interval.
aControlled for exogenous variables, individual-level covariates, and prefecture- and survey year-level fixed effects.
bThe Wald test for the null hypothesis of no correlation between the errors of the two probit models for health and work was rejected (P < 0.001).
cMarginal effect. For a binary explanatory variable, this indicates an expected change in the probability of poor self-rated health in response to a change from 0 to 1 of the explanatory variable.
dContinuous variables (others were binary variables).