How effective are social safety net programs in reducing poverty risks? A comparative study of Taiwan and Southeast China

ABSTRACT

Using the harmonized Panel Study on Family Dynamics 2006, 2011, and 2013/14 data, this is the first study to examine poverty patterns and investigate the role of social safety nets in alleviating poverty over time in East Asia through a comparative lens, focusing on Taiwan and southeast (SE) China. We compared the extent to which public transfers alleviate poverty utilizing multilevel regression models. Our findings indicate that, firstly, the extent of poverty increased at a faster rate in rural areas and among migrants in SE China compared to residents in Taiwan and urban SE China, suggesting widening regional inequality. Secondly, social safety nets reduced poverty in all locations, and the reductions increased over time to a greater extent in rural areas and among migrants in SE China. However, such reductions remained insufficient to narrow the disparity in regional poverty trends. Thirdly, social safety nets in SE China lagged behind Taiwan’s on participation rates in elder, disability and education assistance and led in low-income assistance. Finally, although welfare policy design is more generous in Taiwan than SE China, our findings based on welfare outcomes indicate that Taiwan’s safety net did not alleviate poverty to a greater extent compared to SE China’s.

KEYWORDS:

• Poverty
• social safety net
• public transfers
• impact evaluation
• Taiwan
• China

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. Data from the Luxembourg Income Study (LIS) reveal the relative poverty rate as 20% in mainland China, 15% in South Korea, 12% in Japan, and 11% in Taiwan in 2012/13. Between the 2000s and the 2010s, child poverty rates decreased in Japan, elderly poverty rates increased in China and South Korea, and elderly poverty rates decreased in Taiwan (Luxembourg Income Study, 2020).

2. Coverage of South Korea’s public assistance program was extended to able-bodied adults in 2000 and its family support obligation regulation was further relaxed in 2015 (Nam & Park, 2020).

3. Taiwan extended public assistance benefits to middle-to-low income citizens (in 2011) and Hong Kong initiated a working family allowance program (in 2016) and a work incentive transport subsidy (in 2011).

4. Please see Table A1 for comparisons on detailed characteristics.

5. 4164 individuals from Taiwan 1417 from Zhejiang, 1495 from Fujian, and 548 from Shanghai.

6. Although measuring poverty beyond income (e.g., consumption, deprivation, and multidimension) is ideal (Alkire & Foster, 2011; Alkire & Shen, 2017; Deaton & Paxson, 1997; Laderchi et al., 2003), PSFD either did not collect such information or, in the case of consumption, did not consistently collect such information across years and locations to allow comparisons.

7. {[Poverty line – household income per capita]/poverty line} for household income per capita below the poverty line, and zero for household income per capita equal to or above the poverty line.

8. In PSFD, the business income information captures the net profit/loss from operating a business (including farming, fishing, owning an orchard, owning a shop, or being a street vendor.)

9. PSFD data only collected income information regarding the household head and their spouse. This limits the possibility of generalizing income information to the whole household. We dealt with this limitation using three ways of defining household size for calculating per capita household income, thus more precisely approximating the income resources available for each person in the household: 1. household head, the spouse, and all other likely dependents (described in the following footnote); 2. household head, the spouse, and their children aged below 18 (more likely to overestimate the per capita income because there can be additional dependents); 3. the total number of individuals residing in the household (more likely to underestimate the per capita income because there can be additional working adults). Since the results did not qualitatively differ by definitions, we used the first definition throughout.

10. Household members were defined as the individuals living in the same household: the household head, the household head’s spouse, and their likely dependents, including their parents or in-laws aged over 65, their grandparents and in-laws, their children below age 18, adult children aged between 18 and 25 and were not working, grandchildren if their adult child is not working, and siblings who were below age 18. We opted not to using equivalised household size because neither Taiwan nor mainland China use equivalised household size in their national statistics.

11. Pension income is excluded from our inquiry because it depends on occupation and contributions prior to retirement in Taiwan and SE China.

12. We first changed the data structure to a long format, with two observations each year for each individual, one capturing poverty condition based on pre-transfer income and one capturing poverty condition based on post-transfer income. Each individual responded in at least two and at most three of the survey years. Therefore, between four and six observations were available for each respondent. This way, the dichotomous indicator can capture the differences in poverty outcomes before and after transfer income, indicating the the poverty alleviation effects of social safety nets.

13. We also used poverty rates as the outcome and present these results in Table A3.

14. Definitions of welfare programs are presented in Table 1.

15. Namely, sharing the same composition of characteristics refers to urban residents in Taiwan sharing the same compositions in age groups, marital status, household characteristics, education, work, health, and home ownership as urban residents in China.

16. In other words, having the same coefficients of characteristics refers to, for example, the extent to which gaining one additional level of education related to poverty gap index is the same between urban residents in Taiwan and in SE China.

17. We also conducted the analysis using the poverty rate as the outcome. The results are presented in Table A3.

18. We plotted the results from Models 1 and 2 as shown in Figure A1(a,b).

19. To construct Figure 2(a,b), we relaxed the assumption in Models 3 and 4 that trends were linear and allowed trends to differ across locations in 2006–2011 and 2011–2013/14. We used the -marginsplot- command in Stata to construct these figures.

20. We conducted sensitivity analysis to control for county fixed effects, accounting for time-invariant characteristics in each location, and the findings were similar to the main findings without controlling for county fixed effects. For ease of interpreting location coefficients, we present the model without the county fixed effects.

21. For brevity, we only present coefficients from key variables of interest. Coefficients of all covariates are presented in Table A4.

Additional information

Funding

This work was supported by the The University of Hong Kong [201702159010]; Hong Kong Research Grants Council Early Career Scheme [27611517].

Notes on contributors

Julia Shu-Huah Wang

Julia Shu-Huah Wang is an Assistant Professor at the Department of Social Work and Social Administration, the University of Hong Kong. Her research focuses on social welfare policies, poverty alleviation interventions, immigration policies, and the well-being of families. She is currently working on several research projects, including impacts of social policies on families; impacts of migration policies; poverty dynamics and social safety nets in East Asia; and the policy advocacy in Hong Kong.

Yixia Cai

Yixia Cai is a doctoral candidate at Sandra Rosenbaum School of Social Work, University of Wisconsin–Madison. She is also an economist at the Center for Economic and Policy Research (CEPR). Her research centers on the intersection of economic instability, poverty, social policy, and family well-being. Current research projects examine how employment instability impacts individuals and families. She also investigates the anti-poverty effects of safety net programs and their impact on the life circumstances of low- and middle-income families in the United States.

Qin Gao

Qin Gao is a Professor of Social Policy and Social Work at Columbia University, and she is the founding director of Columbia University’s China Center for Social Policy. Dr. Gao’s research examines the changing nature of the Chinese welfare system and its impact on poverty and inequality; effectiveness of Dibao, China’s primary social assistance program; social protection for rural-to-urban migrants in China and Asian American immigrants; and cross-national comparative social policies and programs.