3.1 Gender Gap in Health Insurance Coverage: Based on Survey Data of 1% of the Population in 2005 (Sample A)
In the first stage of the analysis, we used Sample A. Figure 1 shows that for all age groups, the proportion of women covered by health insurance was lower than that of men. This was due to the higher labor participation rate of men. Meanwhile, by age group, the proportions of men and women with insurance coverage revealed different distribution trends. For men, the proportion of those covered by health insurance increased continually with age; while for women aged below 50 years, the proportion increased continually with age as well. However, this proportion kept declining continuously for the older women (aged above 50 years) and more obviously for those aged above 60 years. Considering the educational level (Figure 2), the UEBMI coverage rates for men and women aged below 50 years were almost equivalent. However, for the groups aged over 50, for women, the proportion of those with insurance coverage was significantly lower than that for men only in the less educated group (primary school graduates and below), and this gap tended to widen with age. In general, this suggests that education level might have influenced the low rate of insurance coverage for the older people, while this effect of education did not exist for the younger people.
Next, in the regression estimation, we classified the population by age, with [50, 55)being the control group. Table 1 reports the result of the regression, in which columns 1 and 2 correspond to regression models with and without control variables (such as education), respectively. As shown in Table 1, when variables such as the educational level were not controlled for, the proportion of women being covered by UEBMI in all age groups was always lower than that of men (the coefficients of both female and female*age are significantly positive). This is consistent with the previous results as shown in Figure 2. Moreover, when controlling for variables such as the educational level, the coefficient of women became significantly positive, indicating a higher proportion of women being covered by health insurance in the age group of [50, 55)as compared with men. However, since the other interaction term coefficients of gender and age are both significantly negative and the absolute values are greater than the coefficient of female, this indicates that generally, the women’s health insurance coverage rate was still significantly lower than that of men. By further observing the coefficients of the interaction term, we can clearly find their obvious changes below and above the age bracket of [50, 55). Below this age bracket, there is no clear trend in these coefficients of the interaction term, while above this age bracket they continue to increase in absolute value. This means that the health insurance coverage gap between women and men has been continuously widening for those above the age of 55.
Taking a step further, we examined the gap of health insurance coverage between men and women considering different education levels (see Table 2). The results of the regression show that it was only in the groups of less educated (primary school graduates and below) people that this phenomenon of continuous widening of the gap existed. In groups of senior middle school graduates and below, women’s health insurance coverage rates were all lower than that of men’s but did not show an expanding trend. Among higher educated people, women’s health insurance coverage rates were not necessarily lower than that of men’s, signifying that the gender gap in health insurance was not obvious among the old, higher educated people.
Figure 1 The Urban Employee Basic Medical Insurance (UEBMI) Coverage Rate: Based on Survey Data of 1% of the Population in 2005 and 2015
Figure 2 The Urban Employee Basic Medical Insurance (UEBMI) Coverage Rate in 2005 by Educational Levels
Table 1: Gender Gap in Health Insurance Coverage: Based on Population Census in 2005
Urban Employee Basic Medical Insurance (UEBMI)
(Dependent Variable: Covered by UEBMI or Not)
|
|
(1)
|
(2)
|
(3)
|
|
Female
|
-0.0937***
(0.00977)
|
-0.125***
(0.0106)
|
0.0548***
(0.0110)
|
|
Female *age
[15,25)
|
0.0629***
(0.0135)
|
0.0690***
(0.0149)
|
-0.148***
(0.0161)
|
|
Female *age
[25,35)
|
-0.0395***
(0.0119)
|
-0.0300**
(0.0131)
|
-0.199***
(0.0138)
|
|
Female *age
[35,45)
|
-0.0498***
(0.0115)
|
-0.0484***
(0.0127)
|
-0.119***
(0.0132)
|
|
Female *age
[45,50)
|
-0.00956
(0.0139)
|
-0.00845
(0.0151)
|
-0.0191
(0.0158)
|
|
Female *age
[55,60)
|
-0.0810***
(0.0151)
|
-0.130***
(0.0163)
|
-0.121***
(0.0169)
|
|
Female *age
[60,65)
|
-0.196***
(0.0166)
|
-0.269***
(0.0179)
|
-0.250***
(0.0186)
|
|
Female *age
[65,70)
|
-0.239***
(0.0175)
|
-0.320***
(0.0188)
|
-0.249***
(0.0197)
|
|
Female *age
[70,75)
|
-0.288***
(0.0189)
|
-0.384***
(0.0203)
|
-0.277***
(0.0211)
|
|
Female *age
[75,80)
|
-0.358***
(0.0227)
|
-0.472***
(0.0241)
|
-0.344***
(0.0253)
|
|
Female *age
[80+,100)
|
-0.426***
(0.0253)
|
-0.531***
(0.0270)
|
-0.426***
(0.0284)
|
|
X
|
No
|
No
|
YES
|
|
City
|
No
|
YES
|
YES
|
|
Observations
|
1,904,763
|
1,904,763
|
1,903,627
|
|
|
|
|
|
|
|
Notes: 1) The control age group is [50, 55).
2) The control variable X includes registered permanent residence, marital status, educational level, nationality, and so on. City represents fixed effects after controlling for cities.
3) ***, **, * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. The numbers in parentheses are standard errors of robust SE.
Table 2: Gender Gap in Health Insurance Coverage in 2005 by Educational Levels
Dependent Variable: Covered by the Urban Employee Basic Medical Insurance (UEBMI) or Not
|
|
(1) Primary School Graduates and below
|
(2) Junior Middle School Graduates
|
(3) Senior Middle School Graduates
|
(4) University Graduates and above
|
Female
|
-0.0241
(0.0170)
|
0.195***
(0.0191)
|
0.229***
(0.0340)
|
0.211***
(0.0758)
|
Female *age
[15,25)
|
-0.0506
(0.0337)
|
-0.282***
(0.0239)
|
-0.375***
(0.0425)
|
-0.219***
(0.0842)
|
Female *age
[25,35)
|
-0.0156
(0.0263)
|
-0.343***
(0.0224)
|
-0.548***
(0.0384)
|
-0.323***
(0.0792)
|
Female *age
[35,45)
|
0.0592***
(0.0230)
|
-0.304***
(0.0218)
|
-0.373***
(0.0376)
|
-0.183**
(0.0814)
|
Female *age
[45,50)
|
0.0753***
(0.0271)
|
-0.223***
(0.0267)
|
-0.168***
(0.0410)
|
-0.169*
(0.0992)
|
Female *age
[55,60)
|
-0.0753***
(0.0246)
|
-0.130***
(0.0325)
|
0.0693
(0.0612)
|
-0.134
(0.123)
|
Female *age
[60,65)
|
-0.152***
(0.0262)
|
-0.418***
(0.0391)
|
-0.202***
(0.0670)
|
-0.167
(0.133)
|
Female *age
[65,70)
|
-0.176***
(0.0263)
|
-0.295***
(0.0515)
|
-0.134
(0.0826)
|
-0.197
(0.133)
|
Female *age
[70,75)
|
-0.182***
(0.0272)
|
-0.283***
(0.0708)
|
-0.186
(0.125)
|
-0.255
(0.174)
|
Female *age
[75,80)
|
-0.285***
(0.0312)
|
-0.180
(0.111)
|
-0.573***
(0.170)
|
0.395
(0.295)
|
Female *age
[80+,100)
|
-0.380***
(0.0339)
|
-0.847***
(0.136)
|
-0.675***
(0.210)
|
-0.271
(0.335)
|
X
|
YES
|
YES
|
YES
|
YES
|
County
|
YES
|
YES
|
YES
|
YES
|
Observations
|
771,963
|
755,007
|
248,055
|
128,602
|
Notes: 1) The control age group is [50, 55).
2) The control variable X and fixed effects of cities have been added to all analyses.
3) ***, **, * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. The numbers in parentheses are standard errors of robust SE.
3.2 Gender Gap in Health Insurance Coverage: Based on Survey Data of 1% of the Population in 2015 (Sample B)
In the following analysis, we further used Sample B.
Figure 3 The Urban Employee Basic Medical Insurance (UEBMI) Coverage Rate in 2015 by Educational Levels
Figure 1 indicates that the results are similar to those generated from using Sample A, except that the turning point here has shifted to the age of 60. Below the age of 60, men and women’s UEBMI coverage rates were almost equivalent, and the coverage rate for young women was even higher than that for men. Meanwhile, the proportion of women aged above 60 covered under UEBMI was much lower than that of men. Considering the level of education, the coverage rates for men and women below 60 were almost equivalent and not influenced by the level of education. Meanwhile, for those aged over 60, only for the less educated groups (primary school graduates and below) had a significantly lower coverage of women than men, and this gap continued to grow with age.
We applied model (1) on Sample B to perform the same regression and report the corresponding results in Tables 3 and 4. As seen from the tables, for those below the age bracket of [60, 65), there is no clear trend in these interaction term coefficients, but the coefficients continuously increase (with a negative sign) for those above this age bracket. This indicates that the gap in UEBMI coverage rates between women and men has been getting wider and wider for those above the age of 60. Moreover, considering the level of education, the regression analysis shows that this phenomenon of continuous increase in insurance coverage gap for the [60, 65)age group and above only exists among the less educated population (primary school graduates and below).
Combining the results above, the study using Sample B yielded almost the same results compared with the analysis based on Sample A. The only difference is the age at which the change in the gender gap in health insurance coverage occurred, which increased from age 50 to age 60. This suggests that the situation (i.e., failing to get insurance coverage) has not changed 10 years later for part of the women aged above 50.
Table 3: Gender Gap in Health Insurance Coverage: Based on Population Census in 2015
Urban Employee Basic Medical Insurance (UEBMI)
(Dependent Variable: Covered by UEBMI or Not)
|
|
(1)
|
(2)
|
(3)
|
Female
|
-0.177***
(0.0226)
|
-0.258***
(0.0241)
|
0.240***
(0.0270)
|
Female *age
[15,25)
|
0.526***
(0.0344)
|
0.558***
(0.0366)
|
-0.0568
(0.0401)
|
Female *age
[25,35)
|
0.238***
(0.0269)
|
0.255***
(0.0289)
|
-0.232***
(0.0328)
|
Female *age
[35,45)
|
0.0418
(0.0264)
|
0.0716**
(0.0282)
|
-0.244***
(0.0319)
|
Female *age
[45,50)
|
-0.0382
(0.0294)
|
0.00757
(0.0314)
|
-0.229***
(0.0350)
|
Female *age
[50,55)
|
0.0610**
(0.0306)
|
0.104***
(0.0327)
|
-0.0182
(0.0365)
|
Female *age
[55,60)
|
0.179***
(0.0335)
|
0.202***
(0.0358)
|
0.165***
(0.0405)
|
Female *age
[65,70)
|
-0.190***
(0.0349)
|
-0.211***
(0.0372)
|
-0.221***
(0.0416)
|
Female *age
[70,75)
|
-0.371***
(0.0389)
|
-0.386***
(0.0412)
|
-0.386***
(0.0472)
|
Female *age
[75,80)
|
-0.373***
(0.0436)
|
-0.411***
(0.0460)
|
-0.294***
(0.0543)
|
Female *age
[80+,100)
|
-0.707***
(0.0464)
|
-0.693***
(0.0491)
|
-0.454***
(0.0572)
|
X
|
No
|
YES
|
YES
|
County
|
No
|
No
|
YES
|
Observations
|
838,544
|
838,544
|
838,544
|
Notes: 1) The control age group is [60, 65).
2) The control variable X includes registered permanent residence, marital status, educational level, nationality, and so on. County represents fixed effects after controlling for counties.
3) ***, **, * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. The numbers in parentheses are standard errors of robust SE.
Table 4: Gender Gap in Health Insurance Coverage in 2015 by Educational Levels
Dependent Variable: Covered by Urban Employee Basic Medical Insurance (UEBMI) or Not
|
|
(1) Primary School Graduates and below
|
(2) Junior Middle School Graduates
|
(3) Senior Middle School and Technical Secondary School Graduates
|
(4) University Graduates and above
|
Female
|
0.0157
(0.0427)
|
0.385***
(0.0442)
|
0.188**
(0.0938)
|
-0.141
(0.180)
|
Female *age[15,25)
|
-0.200
(0.341)
|
-0.435***
(0.0868)
|
-0.0729
(0.105)
|
0.517***
(0.184)
|
Female *age[25,35)
|
-0.181
(0.135)
|
-0.468***
(0.0592)
|
-0.246**
(0.0992)
|
0.249
(0.182)
|
Female *age[35,45)
|
-0.214***
(0.0730)
|
-0.422***
(0.0512)
|
-0.0973
(0.0990)
|
0.193
(0.185)
|
Female *age[45,50)
|
-0.266***
(0.0766)
|
-0.377***
(0.0549)
|
0.0547
(0.106)
|
0.0980
(0.195)
|
Female *age[50,55)
|
-0.158**
(0.0804)
|
-0.136**
(0.0569)
|
0.243**
(0.107)
|
0.194
(0.203)
|
Female *age[55,60)
|
0.182**
(0.0731)
|
0.170***
(0.0638)
|
0.150
(0.116)
|
0.0617
(0.242)
|
Female *age[65,70)
|
-0.327***
(0.0603)
|
-0.114
(0.0782)
|
-0.109
(0.166)
|
0.0317
(0.296)
|
Female *age[70,75)
|
-0.583***
(0.0643)
|
-0.333***
(0.0970)
|
-0.0613
(0.179)
|
0.135
(0.327)
|
Female *age[75,80)
|
-0.533***
(0.0692)
|
-0.466***
(0.133)
|
0.0874
(0.225)
|
0.0465
(0.356)
|
Female *age[80+,100)
|
-0.734***
(0.0693)
|
-0.386**
(0.173)
|
-0.568**
(0.251)
|
-0.515
(0.398)
|
X
|
YES
|
YES
|
YES
|
YES
|
County
|
YES
|
YES
|
YES
|
YES
|
Observations
|
306,558
|
324,610
|
126,747
|
80,629
|
Notes: 1) The control age group is [60, 65).
2) The control variable X and fixed effects of counties have been added to all analyses.
3) ***, **, * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. The numbers in parentheses are standard errors of robust SE.
3.3 Gender insurance gap before and after the implementation of the UEBMI reform
Before the UEBMI reform in 1998, labor insurance, in effect, had covered family dependents. Thus, women, even though without a job, were able to obtain medical insurance benefits through the coverage of their working spouse. However, after the reform, the UEBMI's feature of being tied to work and not including other family members, made it harder than before for these unemployed women to obtain health insurance, which might have led to a widening of the gender gap in health insurance coverage. To verify this conjecture, we used Sample C, covering data before and after the 1998 reform, which enabled us to compare changes in health insurance between men and women.
First, the changes in the UEBMI coverage rate from 1991 to 2011 reappear in Figure 4 (what should be noted here is that we combined labor insurance and various forms of employee medical insurance modes before 1998 into UEBMI for a comparison before and after that year). Figure 4 shows that there was a significant decline in the health insurance coverage rates for all age groups before 2000 but these began to increase rapidly starting from the implementation of the UEBMI. In 2011, the UEBMI had already covered 25% of the population.
Meanwhile, Figure 5 shows no obvious difference in health insurance coverage rates between men and women among the investigated population in 1991 and 1993. However, the increasingly apparent gender gap was observed in the survey samples of 1997 and later, in 2000, 2004, 2006 and 2009, this gap mainly existed among the older population.
Figure 4 The Urban Employee Basic Medical Insurance (UEBMI) Coverage Rate: 1991–2011
Note: Suffix m is for male; f is for female.
Figure 5 Gender Gap of The Urban Employee Basic Medical Insurance (UEBMI) Coverage Rate: 1991-2009