An Impact Evaluation of China s Urban Resident Basic Medical Insurance on Health Care Utilization and Expenditure Hong Liu China Economics and Management Academy Central University of Finance and Economics Zhong Zhao School of Labor and Human Resources Renmin University of China NATSEM, University of Canberra April 2012
Outline 2 Introduction and Research Questions Background Literatures Data Econometric Methods Results Main Results Test for Estimation Assumptions Robustness Check Heterogeneous Effects Conclusion
Introduction 3 Chinese government has been trying to build up a universal public health insurance system in recent health care reform. Urban Employee Basic Medical Insurance (UEBMI) for the urban employed initiated in 1998 New Cooperative Medical Scheme (NCMS) for the rural residents established in 2003 Urban Resident Basic Medical Insurance (URBMI) covering urban residents without formal employment started in 2007 Three pillars of Chinese ambitious universal health care system
Introduction (cont.) 4 The URBMI is regarded as a landmark reform toward universal coverage in China. government-run voluntary insurance program; operated at the city level; target at 420 million urban residents not covered by UEBMI. first implemented in 79 pilot cities in 2007; expanded to another 229 cities in 2008; aims at covering all cities by the end of 2010. Few studies have been done to evaluate the effectiveness of this public insurance program.
Research Questions 5 This paper aims at estimating the impact of the URBMI on health care utilization and medical expenditure. We also investigate the heterogeneous effects of the program for different age groups, income groups, gender, and regions.
Background about the 6 URBMI Enrollment: children, the elderly, and other unemployed urban residents. on a voluntary basis at individual level or household level. Financing: Financed by individual contributions and government subsidies shared between central and local governments. Levels vary across cities, and on average, 236 RMB for adults, 97 RMB for children. Government subsidies account for about 36 percent of the financing cost for adults, and 56 percent for children. Benefits: Mainly cover inpatient services, and outpatient services for chronic or serious diseases. typically do not cover general outpatient services. The average reimbursement level is 45 percent for inpatient costs.
Literatures 7 A number of studies have examined the effect of the expansion of public health insurance program. primarily focusing on developed countries. Currie and Gruber (1996a;1996b; 1997; 2001):the expansion of Medicaid. Card et al. (2008): the expansion of Medicare. Studies on the effect of Taiwan s national health insurance. Cheng (1997) Chen et al.(2006)
Literatures (cont.) 8 Recent studies on China s NCMS. Yip and Hsiao (2009) Wagstaff et al. (2009) You and Kobayashi (2009) Lei and Lin (2009) Shi et al. (2010) Sun et al. (2009, 2010) Chen and Jin (2012) Lin et al. (2009) study the effectiveness of NRBMI. Selection bias is not addressed with cross sectional data; Outcome variables are limited.
9 Data: China Health and Nutrition Survey Waves: 1989, 1991, 1993, 1997, 2000, 2004, 2006 and 2009
Data (cont.) 10 Data selection Waves 2006 and 2009 Sample with urban hukou and living in urban area. Target population, not covered by free insurance or UEBMI. children under 18, or current student over 18. men over 60 and women over 55: retired, or without job information. adult over 18: no job and not retired, temporary workers.
Data (cont.) 11 The final study sample consists of 3,013 observations, including 1,576 in 2006, and 1,427 in 2009. 682 were sick in 2009 690 enrolled in UEBMI in 2009
Variable 12 Dependent variable: Any formal medical care Any formal medical care for the sick Any inpatient visit Inpatient hospital days Total health expense Out-of-pocket health expense Key independent variable: Enrollment in URBMI at t (0/1) Treatment status: Treated: those who enrolled in URBMI in CHNS 2009. Control: those who are not enrolled in URBMI in CHNS 2009. Post: indicator of CHNS 2009. Indicator of experimental cities
Table 1. Summary Statistics Wave 2006 Wave 2009 Treated Control Treated Control Variable Mean Mean Dependent Variables (in last four weeks) Any formal medical care 0.15 0.13 0.16 0.10 *** Any formal medical care for the sick 0.63 0.58 0.58 0.55 Any inpatient visit 0.01 0.01 0.02 0.01 Inpatient hospital days 0.05 0.07 0.19 0.08 Total health expense 49.67 148.85 152.62 218.19 Out-of-pocket health expense 26.79 38.35 45.18 115.54 13
Table 1. Summary Statistics (cont.) Wave 2006 Wave 2009 Treated Control Treated Control Variable Mean Mean Explanatory Variables Individual charateristics Enrolled in URBMI 0 0 1 0 Education: primary school 0.17 0.13 * 0.19 0.11 *** Education: junior high school 0.23 0.28 * 0.32 0.22 *** Education: senior high school 0.16 0.22 *** 0.21 0.14 *** Education: college 0.01 0.04 *** 0.03 0.03 Total household income (k) 23.97 30.65 *** 43.21 44.49 Age 44.32 39.73 *** 47.32 37.84 *** Female 0.56 0.56 0.56 0.52 * Married 0.56 0.49 ** 0.64 0.35 *** Household size 3.57 3.43 * 3.46 3.45 Student 0.14 0.21 *** 0.16 0.16 Community characteristics Any health facility 0.57 0.51 ** 0.83 0.77 *** Average treatment fee for a cold (k) 0.05 0.07 *** 0.07 0.07 14 Community urbanicity index 81.32 82.42 * 85.92 85.02 *
Econometric Methods 15 Difference-in-Differences Simple DID 1 ( Y treatment treatment ) ( control control post Ypre Ypost Ypre ) Difference out time-invariant unobservables Or time-variant unobservables but with same time trend in treated and control groups Cannot control for time-variant unobservables with different time trend Full model Y Post Treat Post Treat X it 0 1 t 2 it 3 t * it 4 it 5 jt 6 k it
Table 2. Effect of URBMI on Health Care Use and Expenditure OLS OLS wave 2009 full Sample Linear FE DID (1) (2) (3) (4) 1.Any formal health care in last four weeks (logit) Effect of URBMI 0.04** 0.04** 0.08* 0.05** (0.02) (0.02) (0.04) (0.03) 2.Any formal health care for the sick in last four weeks (logit) Effect of URBMI 0.07 0.06 0.36* 0.08 (0.07) (0.06) (0.20) (0.09) 3. Inpatient days in last four weeks Effect of URBMI 0.08 0.08 0.06 0.13 (0.08) (0.08) (0.13) (0.09) 4. Hospital admission in last four weeks (logit) Effect of URBMI 0.003 0.003 0.002 0.002 (0.004) (0.004) (0.01) (0.005) 5. ln(out-of-pocket +1) Effect of URBMI -0.03-0.04 0.10-0.05 (0.07) (0.07) (0.14) (0.10) 6. ln(total health expense +1) Effect of URBMI 0.14 0.14 0.35* 0.15 16 (0.10) (0.10) (0.19) (0.13)
Econometric Methods Placebo Tests 17 Placebo tests for DID assumptions Placebo test I - the analogous estimates for the period 2004-2006, before the implementation of URBMI in 2007. Placebo test II - the analogous estimates for uncovered preventative care.
Table 3. Placebo Test I -- Estimates Using 2004-2006 Data OLS OLS Wave 2006 Full Sample Linear FE DID (1) (2) (3) (4) 1.Any formal health care Effect of URBMI -0.01-0.01 0.03-0.00 (0.02) (0.02) (0.03) (0.02) 2.Any formal health care for the sick Effect of URBMI -0.04 0.03 0.01-0.12 (0.07) (0.06) (0.14) (0.09) 3. Inpatient days Effect of URBMI -0.06-0.07 0.04 0.07 (0.06) (0.04) (0.10) (0.08) 4. Hospital admission Effect of URBMI -0.000-0.001 0.009 0.005 (0.001) (0.003) (0.009) (0.008) Observations 881 3729 3729 3726 5. ln(out-of-pocket +1) Effect of URBMI -0.01-0.05-0.08-0.06 (0.08) (0.07) (0.13) (0.12) 6. ln(total expense +1) Effect of URBMI -0.02-0.01 0.11 0.00 18 (0.10) (0.09) (0.14) (0.13)
Table 4. Placebo Test II -- Estimates for Uncovered Preventive Care OLS OLS Wave 2009 Full Sample FE DID (1) (2) (3) (4) 1. Preventive care use in last four weeks Effect of URBMI 0.01 0.01 0.02 0.02 (0.01) (0.01) (0.02) (0.02) 2. General physical examination in last four weeks Effect of URBMI -0.01-0.01 0.02 0.00 (0.01) (0.01) (0.02) (0.01) 3. Other preventive care use in last four weeks Effect of URBMI 0.02 0.01 0.00 0.02 (0.01) (0.01) (0.02) (0.01) 19 Preventive care: blood test, blood pressure screening, child health examination gynecological examination
20 Econometric Methods Robustness Check Instrumental Variable Estimation EnrollURB MI CityURBM I X u ijt 0 1 jt 4 it 5 jt it Y EnrollURBMI X it 0 1 it 2 it 3 jt it Using project city as an instrumental variable
Table 5. Robustness Check Estimates after Dealing with Endogeneity 2SLS Wave 2009 FE+IV Full Sample (1) (2) Weak instrument test F= 40.57 F= 326.01 1.Any formal health care in last four weeks Effect of URBMI 0.19* 0.08# (0.11) (0.05) 2.Any formal health care for the sick in last four weeks Effect of URBMI 1.07** 0.01 (0.53) (0.26) 3. Inpatient days in last four weeks Effect of URBMI 0.37 0.41** (0.52) (0.18) 4. Hospital admission in last four weeks Effect of URBMI 0.03 0.02 (0.04) (0.02) 5. ln(out-of-pocket +1) Effect of URBMI 0.51 0.35* (0.42) (0.20) 21 6. ln(total health expense +1) Effect of URBMI 1.19** 0.68** (0.60) (0.26)
Econometric Methods Robustness Check 22 DID Estimates based on Different Treatment/Control Groups. Treated: the enrollees living in URBMI cities. Control 1: people living in URBMI cities who chose not to enroll; Control 2: people living in non-nrbmi cities Control 3: people living in non-nrbmi cities who are not enrolled in URBMI.
Table 6. Robustness Check DID based on Different Treatment/Control Groups 23 Control Treated Main Robustness check DID DID DID DID Unenrolled in URMI & Non- URMI cites N=1929 Enrolled N=1038 Unenrolled in URMI cities N=1047 Enrolled In URMI cities N=717 Non-URMI cities N=830 Enrolled In URMI cities N=717 Unenrolled in Non-URMI cities N=628 Enrolled In URMI cities N=717 (1) (4) (5) (6) 1.Any formal health care Treat*Wave2009 0.05** 0.06* 0.13** 0.11* (0.03) (0.03) (0.06) (0.06) Observations 2967 2237 1206 1136 2.Any formal health care for the sick Treat*Wave2009 0.08 0.06 0.46*** 0.35* (0.09) (0.09) (0.14) (0.18) Observations 676 513 274 255 3. Inpatient days Treat*Wave2009 0.13 0.14 0.16 0.16 (0.11) (0.11) (0.10) (0.10) Observations 2967 2237 1206 1136 4. Hospital admission Treat*Wave2009 0.00-0.01-0.00-0.00 (0.01) (0.01) (0.01) (0.01) Observations 2723 2237 1206 1136 5. ln(out-of-pocket +1) Treat*Wave2009-0.05 0.06 0.44*** 0.36** (0.10) (0.11) (0.16) (0.16) Observations 2967 2237 1206 1136 6. ln(total expense +1) Treat*Wave2009 0.15 0.21 0.61*** 0.50** (0.13) (0.15) (0.21) (0.22) Observations 2967 2237 1206 1136
Econometric Methods Heterogeneous Effect 24 We also study heterogeneous effects of URBMI for different age groups: 0-17, 18-59, 60+ income groups: low, medium and high HH income gender: male, female. regions: eastern, central and western
25 Table 8. Effect of URBMI by Population Groups (DID) Sample 0-17 18-59 60 and above Low HH Medium HH High HH income income income (1) (2) (3) (4) (5) (6) 1.Any formal health care Treat*Wave2009 0.08-0.00 0.19*** 0.10* 0.08** -0.05 (0.07) (0.03) (0.07) (0.05) (0.04) (0.05) 2.Any formal health care for the sick Treat*Wave2009 0.27-0.03 0.19# 0.09 0.20# -0.20 (0.22) (0.13) (0.12) (0.12) (0.13) (0.18) 3. Inpatient days Treat*Wave2009-0.01 0.22** 0.05 0.27* 0.16-0.15 (0.08) (0.11) (0.25) (0.15) (0.14) (0.23) 4. Hospital admission Treat*Wave2009 0.00 0.01-0.02 0.01-0.01-0.01 (0.01) (0.01) (0.03) (0.02) (0.01) (0.02) 5. ln(out-of-pocket +1) Treat*Wave2009 0.19-0.15 0.09-0.14 0.15-0.19 (0.17) (0.13) (0.23) (0.18) (0.13) (0.21) 6. ln(total expense +1) Treat*Wave2009 0.32-0.02 0.53# 0.06 0.43** -0.12 (0.23) (0.17) (0.33) (0.26) (0.17) (0.29)
Table 8. Effect of URBMI by Population Groups (DID) Sample Male Female Eastern Central Western China China China (7) (8) (9) (10) (11) 1.Any formal health care Treat*Wave2009 0.08* 0.04 0.08 0.04 0.07* (0.04) (0.04) (0.06) (0.04) 0.04 2.Any formal health care for the sick Treat*Wave2009 0.17# 0.05-0.03 0.05 0.29** (0.11) (0.10) (0.17) (0.12) (0.14) 3. Inpatient days Treat*Wave2009 0.10 0.16 0.09 0.15 0.06 (0.13) (0.12) (0.21) (0.14) (0.12) 4. Hospital admission Treat*Wave2009-0.00 0.00 0.00-0.01 0.01 (0.01) (0.01) (0.02) (0.02) (0.02) 5. ln(out-of-pocket +1) Treat*Wave2009-0.07-0.01 0.29* -0.17-0.10 (0.13) (0.14) (0.16) (0.17) (0.16) 6. ln(total expense +1) Treat*Wave2009 0.19 0.13 0.19-0.03 0.42** 26 (0.18) (0.18) (0.26) (0.22) (0.20)
Conclusion 27 URBMI has significantly increased the utilization of formal medical services, but has no significant effect on the use of inpatient services. There is no evidence that it has reduced out-of-pocket expenditure. The main results are robust to various specifications and multiple estimation strategies. URBMI has improved medical care utilization for the elderly and the low and middle income families, and the relatively poor western regions have benefitted more from the program.
28 THANK YOU! Hong Liu irisliu2000@gmail.com Zhong Zhao mr.zhong.zhao@gmail.com