No Country for Old Men: An Investment Motive for Downward Inter-generational Transfers in Rural China

Transcription

1 No Country for Old Men: An Investment Motive for Downward Inter-generational Transfers in Rural China Chiyu Niu Mary Arends-Kuenning Department of Agricultural and Consumer Economics University of Illinois Urbana Champaign May 21, 2016 Selected Paper prepared for presentation at the 2016 Agricultural & Applied Economics Association Annual Meeting, Boston, Massachusetts, July 31-August 2 Copyright 2016 by Chiyu Niu. All rights reserved. Readers may make verbatim copies of this document for non-commercial purposes by any means, provided that this copyright notice appears on all such copies

2 No Country for Old Men: An Investment Motive for Downward Inter-generational Transfers in Rural China Chiyu Niu Mary Arends-Kuenning Abstract This article looks at the impact of a newly introduced rural pension program on the transfer behaviors between the elders and their children in China. We employ a regression discontinuity design with the pension eligible age as the cutoff to identify the causal impact. We found that the pension program largely crowd out the existing private transfers from children to parents, whereas the transfers from parents to children increase significantly. We further analyze the incentives behind this increase in the downward inter-generational transfers. By utilizing the longitudinal feature of our data, we found that parents are more likely to transfer to children that are more educated and reside in more populous cities. These findings are consistent with an investment motive of the rural elderly parents, who seek for higher returns to investment in the urban areas through their children instead of investing in agricultural production in the rural areas. The policy implication of this research is that the large public transfer efforts in the form of rural pension are not likely to boost the capital accumulation that is needed for the rural development in China. Introduction Aging has been posing social and economic challenges, especially for the developing countries that are growing old rapidly without being rich enough to provide for the old. Tens of millions of older Chinese have been struggling with poverty and loneliness as their children flee villages to cities. Sharp demographic changes such as rapid aging and and increasing dependency ratio due to the one-child policy, as well as the recent trend of rural-to-urban migration as a result of urbanization 1, have frayed the ties that once bound the nation s families together. The left-behind elders have to live off their labor and remittances from their migrant children. In fighting against the increasing old-age poverty in the rural areas, China launched the New Rural Pension Program (NRPP) in 2009, which covered more than 300 million Chinese by the end of Unlike the pension programs in the developed areas, the NRPP could be considered as a conditional cash transfer program, where the conditions are minimal: being registered as rural residents 2, and age 1 The proportion of urban residents in China increased from 17.4% in 1978 to 41.8% in 2005 (NBS, 2006) 2 The Chinese Hukou system, which is the population registration system, abolished its differentiation between agricultural and non-agricultural households, but it could still indicate the type of area that a household live in. 2

3 60 or above. In assessing the welfare impact of a public transfer program as such, it is crucial to understand whether the public transfers substitute for the private transfers (Cox and Fafchamps 2008). If they are substitutes, the private transfers that the old have been relying on will be crowded out (Becker, 1974; 1981), which will dampen the efficacy of the pension program, because the beneficiaries would be the prime-age children instead of their elder parents. It is also critical to see whether the spending patterns of the pension recipients could help boost capital accumulation in the rural areas. Does the program enable the elders to spend more on nutrition and thus become healthier? Are the rural elders spending the pension benefits strategically such as investing in productive capital, which is required for rural development? The policy implications of these questions are paramount not only for the Chinese government, who has listed urbanization and reducing poverty as the major goals in China s 13th five-year plan starting in 2015, but also for the other developing countries that are experiencing urbanization and industrialization without a mature social safety net program for the elders. This article focuses on answering the following questions: Does the public cash transfer program (NRPP) crowd out the private transfers that the rural elders have been receiving? How does the NRPP affect the spending patterns of the rural elders with respect to transfers sent to others, consumption, investment in productive assets and financial assets? What could be the motivation behind the behavioral responses to such a cash transfer program? Using a regression discontinuity design based upon eligibility for the program and a rich rural survey dataset, this research finds that the NRPP decreases both the probability and the amount of private transfers received by the rural elders, which indicates a strong crowding-out effect. In addition, the NRPP has no significant impact on the rural elders consumption, investment in assets, loans and debts. However, the NRPP significantly increases the amount of transfers sent to children from the elders, and at the same time, the amount of transfers sent to the elders siblings decreased. The results of household and individual fixed-effect analyses reveal that the elders tend to transfer more to the more educated children, and also to those who migrated to a more distant region with a bigger population. These findings could be reconciled with an investment motive of the Chinese rural elders, who are treating their migrant children as productive assets that have higher returns than the productive capitals in the rural areas where the financial inclusion level is 3

4 low. The contribution of this study is twofold. First, this study is one of the first empirical analyses that assesses the causal impact of large cash transfer programs on inter-vivo transfer behaviors and expenditure patterns of the Chinese rural elderly. Second, this study complements the literature on the impact of remittances and cash transfer programs on agricultural investment behaviors. The focus on comparing simultaneously the differential impacts of cash transfer programs on transfer behaviors and investment behaviors is particularly novel, which helps to answer the key policy question of whether expanding cash transfer programs for the rural elderly is likely to boost capital accumulation required for rural development in the long run. The article is organized as follows: Section II describes the New Rural Pension Program in China and briefly reviews the literature; Section III presents the empirical model and major econometric issues; Section IV introduces the data; Section V shows empirical findings; Section VI discusses the policy implications, and Section VII concludes. The New Rural Pension Program (NRPP) in China Before the implementation of the New Rural Pension Program, the old Rural Pension Program, in place from 1987 to 2008, was an unsubsidized individual contribution account system. However, the participation rate had been extremely low, which lead to insufficient pension funds that could barely insure the promised benefits for the rural elderly. In 1999, the rural population that participated in the old pension program started to shrink. In September 2009, China launched the New Rural Pension Program, which is now covering over 300 million Chinese. The NRPP aims at a universal coverage of formal and informal workers in rural China. By the end of 2012, the central government and the local governments have invested more than 232 billion RMB ( about 40 billion USD) to the pension fund, and all 2853 counties in China have implemented the NRPP. The pension fund consists of three parts: an individual contribution account, local government subsidy, and central government subsidy. According to the guidance released by the State Council of 4

5 China, the individuals are free to choose from five basic categories of individual premium: 100, 200, 300, 400 and 500 RMB per year per person 1. A local government may introduce more categories to adjust for increasing cost of living, and the more the individuals contribute to their accounts, the more subsidies local government pays. Pensioners receive payment from both the individual account and the basic pension benefit, where the latter is subsidized by the central government. The average benefit level is about 700 RMB per year, which is roughly 10% of the median annual individual income in the rural area. All rural residents aged 16 or above who are not enrolled in the urban basic pension program can enroll in NRPP voluntarily. Participants below age 45 are required to contribute at least 15 years to become eligible for pension payment upon reaching age 60. However, according to the State Council, there are no required minimum years of contribution for those aged between 45 and 60, who are only encouraged to contribute more to meet the shortfall in contributions to their account over their remaining years before age 60. Most people aged 55 to 60 wait until age 60 to receive the free basic pension offered by the central government without any contribution to the individual account (Chen and Wang 2014). Pension payment is non-means tested, and non-contributory to anyone aged 60 or above by the local rollout time of NRPP. Elders who are above age 60 are not required to make any contribution to the pension fund. The literature that studies the impact of old-age pension on private transfers has focused on upward intergenerational inter-vivo transfers, with the concern that the public transfer programs could merely crowd out the existing private old-age support system if the transfers are based on an altruistic motive for the elders. Such crowding out effects are found in many developing countries, such as south Africa (Maitra and Ray, 2003; Jensen, 2004), Korea (Jung et al., 2015) and Mexico (Juarez, 2009). However, the existing studies about private transfers between Chinese adult children and their elder parents have suggested that it could be very unlikely that the upward intergenerational transfers will be crowded out by a public pension system, because the upward inter-generational transfers are largely based on an exchange motive, and the traditional Confucius filial piety is deeply rooted as well as being more of 1 The 100 RMB option is the most popular choice (Lei et al. 2011) 5

6 a social norm than altruism (Secondi, 1997; Cai et al., 2012). Compared to the literature of private transfers received by the elders, very few studies have looked at transfer decisions made by the elders themselves, especially in the developing countries where the rural elders are considered to be much more vulnerable to poverty than the younger generation. However, the fruitful literature on the expansion of south African old-age pension can elucidate the impact of the cash transfer programs on the recipients extended family. Studies have shown that old-age pension helps improve the nutrition status of grandchildren (Duflo, 2000; 2003), relax the credit constraints for the household to send migrant for better employment opportunities (Ardington et al., 2009), decrease the labor supply of the prime-age children who are residing with the pension receivers (Bertrand et al., 2003), and also encourage education investment for grandchildren (Edmonds, 2005). Even though none of these studies have used data on the actual transfers made to the other family members, it could be learned that private transfers constitute an important part of the pension receivers expenditure. Other than private transfers made to the other family members, consumption, productive investment and financial investment generally constitute the rest of the household expenditures. Jung et al. (2015) found that Korean old-age pensions did not increase the elders total expenditures or food expenditures because the expansion of the public pensions largely crowded out the private transfers from adult children and/or own siblings. De Brauw and Roselle (2008) and Zhu et al. (2009) find no impact of remittances on savings and productive investment. This could be explained by the argument that migrant network could substitute for precautionary saving (Giles and Yoo, 2006). To the extent that remittances and old-age pension are substitutes, one might expect little impact from old-age pension on savings and productive investment. However, Zheng and Zhong (2015) shows that NRPP increases both consumption and productive investment in rural China, although their findings do not focus on the elders behavior, and the results are aggregated at the county level instead of the household level. Empirical Model and Econometric Issues To estimate the causal effect of NRPP on the private transfers and other expenditure components of the Chinese rural elders, I use a fuzzy regression discontinuity design (RDD), made possible by the 6

7 arbitrary age eligibility threshold of the NRPP. According to the state council, rural residents who are age 60 and above are eligible for receiving the monthly payment from NRPP. Figure 1 shows the distribution of the running variable. Figure 1: Distribution of Running Variable However, in reality, not all the rural elders who are eligible have received the pension benefits, which implies incomplete compliance. Anecdotal evidence has shown that a few counties refuse to pay pension benefits to the eligible elders if their children are not enrolled in the pension program. One reason for this deviation from the official policies set by the state council is that the central government requires the local governments to subsidize at least 30 Yuan per person per year, which could be a burden if there is inadequate participation from the elders prime-age children. Figure 2 shows the incomplete compliance rate, which motivates the fuzzy regression discontinuity design. Figure 2: Incomplete Compliance at Age Threshold 1 Probability of Receiving Benefits from NRSP age kernel = epanechnikov, degree = 0, bandwidth =.61 7

8 My goal is to estimate the discontinuous changes in transfer behaviors, consumption, saving and investment across the eligibility threshold. Assuming that the unobserved variables are a smooth function of the running variable, this change reflects the causal effect of the New Rural Pension Program. Since the running variable is discrete, it is impossible to observe behaviors that are arbitrarily close to the threshold on both sides. This means that I have to extrapolate the trend of the outcome variables away from the threshold. Potentially, this extrapolation makes my results sensitive to the functional form of the running variable. To show that my results are robust to functional form choices, I follow Lee and Lemieux (2010) and use two different functional forms. The first specification is a nonparametric local linear regression, which estimates the following model using triangular weights to give higher weight to observations that are closer to the threshold: Y ict = β 0 + β 1 R i + β 2 X i + β 3 R i X i + α c + γ t + ɛ ict (1) In equation (1), Y ict denotes the outcome variable for household i, who resides in county c at year t. R i is the running variable, which is the age of the household head. X i is a binary indicator of whether the household head has already received pension benefits from NRPP. α c is a community fixed effect, γ t is a year fixed effects, and ɛ ict is the idiosyncratic error. Since the pension-receiving status X i might be endogenous due to household-specific characteristics, I construct a binary variable which indicates whether the household head is age 60 and above at the time of the survey and use it as an instrument for the variable X i. Because the policy eligibility rule is arbitrarily chosen by the central government, it is exogenous to the pension-receiving status. However, we need to rule out the possibility that the households can manipulate their registration ages, which can affect their pension receiving status. To check this, I use McCrary test (McCrary 2008) to see if there is manipulation of the running variable. The null hypothesis is that the density of the age distribution is smooth around the age threshold, which is 60. Figure 3 shows that the test statistic is with standard error so that we cannot reject the null hypothesis, which indicates that there is no significant manipulation of the running variable. 8

9 Figure 3 The second specification is a regression that flexibly controls for quadratic trends on both sides of the threshold: Y ict = β 0 + β 1 R i + β 2 Ri 2 + β 3 X i + β 4 R i X i + β 5 Ri 2 X i + α c + γ t + ɛ ict (2) In both specifications, the coefficient of X i, which is the pension-receiving status, gives the RDD estimate. My baseline specification uses a triangular kernel with a bandwidth of 6. Because there is no consensus on the selection of the optimal bandwidth, I present results for a wide range of bandwidth between 4 and 8 years in the appendix. The unit of observation is the household-year. Standard errors are clustered at the county level to take care of the potential serial correlation within a county. Another concern about the RD design is that whether the identifying assumptions hold, which include the smoothness of the observable characteristics of the household across the eligibility threshold. If this assumption holds, we should not observe any discontinuities for the household characteristics around the cutoff age. These robustness check results are shown in the appendix. Also, several falsification tests are performed at other pseudo-thresholds where the NRPP eligibility did not change. 9

10 Data Data of this study comes from the China Health and Retirement Longitudinal Study (CHARLS), which aims to collect a high quality nationally representative sample of Chinese residents ages 45 and older. The national sample includes a baseline wave in 2011 and a follow-up wave in About 10,000 households and 17,500 individuals in 150 counties/districts and 450 villages are included in the survey. The CHARLS data is based on the Health and Retirement Study (HRS) in the US and is comparable with the surveys such as the English Longitudinal Study of Aging, as well as the Survey of Health, Aging and Retirement in Europe. The CHARLS questionnaire includes information on demographics, family structure and transfers, health status, work history, pension, consumption and assets acquisition. The sample used for estimation includes only the rural families that have at least one adult child. Table 1 provides the summary statistics. The table shows that the education levels for the rural elders are very low with only about four years on average. As high as 80% of these individuals are still working, which reflects the depiction of endless toil in the labor supply literature on Chinese rural elders. Private transfers are defined to be the transfers sent or received by the elder couple in the household. This is because the question in the survey asks the respondents Have you or your spouse received transfers from your children who are not living with you in the last year? If yes, what was the amount? The private transfers include both monetary and in-kind supports. Table 1 shows that transfer behaviors are very prevalent among rural Chinese, and the amount of transfers is large relative to the elders annual income. About 80% of the children have sent transfers to their parents, and about 30% of the elder parents have sent transfers back to their children. In CHARLS sample, only about 16% of the rural elders are living with their adult children, and the proportion of households that have migrant children is about 85%, which largely reflects the trend of rural-to-urban migration and the phenomenon of left-behind elders. Results A. Transfers Received by the Rural Elders 10

11 Table 1: Summary statistics Variable Mean Std. Dev. Characteristics of the household head Age Male Married Years of Education Annual Income (yuan) Whether received NRPP benefits Whether work Private Transfers From children (binary) From children (yuan per year) To children (binary) To children (yuan per year) From siblings (binary) From siblings (yuan per year) To siblings (binary) To siblings (yuan per year) From friends (binary) From friends (yuan per year) To friends (binary) To friends (yuan per year) Living Arrangements Number of adult children in household Proportion of elders living with adult children Number of migrant children Proportion of households that have migrant children Number of grandchildren in household Proportion of elders living with grandchildren N

12 Figure Probability of Receving Transfers from Children age Transfer Amount Received from Children age Mean Quadratic Fit Nonparametric Fit Mean Quadratic Fit Nonparametric Fit Figure 4 plots the probability of receiving transfers from children and the actual amount of transfers received against the elder parent s age. The scatter dots are the means of the outcome variable by each age group; the solid curves represent the quadratic fits on both sides of the age eligibility threshold; the dashed curves give the nonparametric local linear fits. The graphs show that both the probability of receiving from children, and the actual amount of transfer received decrease sharply across age 60, which is the age threshold for receiving pension benefits from NRPP. According to the corresponding RD results in Table 2, for the Chinese rural elders who are close to age 60, receiving the pension benefits from NRPP reduces the probability of receiving transfers from children by 23.9%. Also, for those who have been receiving transfers from their children, the amount transferred decreased significantly, too. These results indicate that the introduction of the NRPP has significantly crowded out the private transfers that the rural elders have been receiving from their adult children. These findings contradict the previous results shown in the literature on the transfer behaviors in rural China (Secondi 1997; Cai et al. 2012). Column 3 and 4 in Table 2 show the transfer results between the elders and their siblings. Even though the signs of the estimates are negative, they are not statistically different from zero. The corresponding RD graphs for the elders siblings could be found in the appendix. 12

13 Table 2: The Impact of NRPP on Transfers Received From Children From Siblings Probability Amount Probability Amount Specifications (1) (2) (3) (4) Non-parametric linear, no control (0.114) ( ) (0.091) ( ) Non-parametric linear, with control (0.113) ( ) (0.091) ( ) Quadratic, no control (0.110) ( ) (0.194) ( ) Quadratic, with control (0.110) ( ) (0.194) ( ) Control Mean Observations 3,040 3,040 3,040 3,040 Community FE yes yes yes yes Nonparametric local linear regressions use triangular weights with a bandwidth of 6. Control variables include household size, education and income of both the senders and receivers of the transfers, and sex ratio of the transfer senders. Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 B. Transfers Sent from the Rural Elders Figure 5 Transfer Amount Sent to Children Transfer Amount Sent to Siblings age age Mean Quadratic Fit Nonparametric Fit Mean Quadratic Fit Nonparametric Fit Figure 5 shows the plots of the actual amounts of transfers sent from the rural elders to their children and their siblings. The graphs indicate that the amount of transfers sent to children increases 13

14 sharply when the elders start to receive benefits from the NRPP. Whereas for the transfers to siblings, the amount transferred decreases. The corresponding RD results are shown in the first two columns in Table 3. The opposite propensities of transferring to children and own siblings could be puzzling for the following two reasons. First, unlike the pension program in south Africa, where the benefit level is twice as high as the median income, NRPP offers an average benefit that is only about 10% of the median income. Additionally, in the previous section, the results on the transfers received have shown that the private transfers from children are largely crowded out by the NRPP, so the increment in transfers to children might not simply be a result of empowerment from the NRPP. Second, if the increment in transfers to children is indeed granted by the empowerment from NRPP, and the transfers are made based on an altruistic motive, then the transfers sent to the siblings should also increase instead of decreasing. To find out the motivation behind these transfer behaviors, I further analyze the other major components of household expenditure in section C, and also the relationship between the characteristics of the elders adult children and the amount of transfer the children received in section D. C. Impact of NRPP on Consumption, Asset Acquisition, Loans and Debts Table 4 shows the impact of NRPP on the other major components of the household expenditures. Household consumption is further categorized into food, clothes, household goods and service, medical expenses, durable goods; productive assets 1 including the number of productive assets owned and the current value of the productive assets; financial assets include savings, loans and debts. I use the same sample and same econometric method as for the analyses on transfers. The results indicate that there is no significant impact of the NRPP on the expenditure patterns of the rural elders. Similar insignificant findings are also found for the Korean old-age pension program (Jung et al. 2015), where the authors conclude that because the public pension largely crowds out the private transfers, the impact of the pension program on consumption is insignificant. Also, the insignificant results on capital investments are in line with the previous studies on rural Chinese households (De Brauw and Roselle, 2008; Zhu et 1 The productive assets include tractor, thresher, tractor tools, water pump, processing equipment and other fixed capital assets used in production or self-employed activities. 14

15 al., 2009). Table 3: The Impact of NRSP on Transfers Sent Transfers Sent Children Siblings Grandchildren Friends Specifications (1) (2) (3) (4) Linear, no control ( ) ( ) ( ) ( ) Linear with control ( ) ( ) ( ) ( ) Quadratic no control ( ) ( ) ( ) ( ) Quadratic with control ( ) ( ) ( ) ( ) Control Mean Observations 3,040 3,040 3,040 3,040 Community FE yes yes yes yes Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 However, as Alderman and Ymtsov (2014) have reviewed the literature on rural household investment behaviors, cash transfer programs have significant impacts on productive investment in many developing countries around the world. Why are rural Chinese households being so reluctant to invest in productive investments, but at the same time being prone to transferring to children? In order to see whether transferring to children can be understood as a substitute for investment in rural China, I further analyze in the next section the determinants of the amount of transfers adult children received from their elder parents. D. Determinants of Private Transfers to Adult Children Table 5 shows the analyses of the determinants of the amount of transfers sent to the elders children. Two samples of the rural elders children are used for estimation in Table 5: the specification (1) and 15

16 Table 4: The Impact of NRSP on Transfers Sent, Consumption, Assets, Loan and Debts Consumption Productive Assets Financial Assets Food Clothes Household Health Durables Other Number Value Saving Loans Debts Specifications (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Linear, no control (2432.1) (503.6) (3233.0) (4412.8) (2467.1) (5438.9) (0.2) (1792.1) (5375.2) (5048.9) (9581.7) Linear with control (2409.4) (495.1) (3319.5) (4368.6) (2463.8) (5520.5) (0.2) (1768.6) (5444.9) (4965.2) (9735.9) Quadratic no control (3008.4) (1090.4) (6517.8) (9708.5) (5941.9) ( ) (0.4) (2253.3) ( ) (8430.0) ( ) Quadratic with control (2999.6) (1091.1) (6562.4) (9681.3) (5943.3) ( ) (0.4) (2226.6) ( ) (8439.1) ( ) Control Mean Observations 3,040 3,040 3,040 3,040 3,040 3,040 3,040 3,040 3,040 3,040 3,040 Community FE yes yes yes yes yes yes yes yes yes yes yes Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 16

17 (2) use a sample of adult children whose parents have more than one adult child 1 ; the sample that is used for specifications (3) and (4) includes all adult children available in the CHARLS dataset. CHARLS questionnaire asks the amount of transfers that the elders sent to each of their children, and also the demographic and economic information about the children. Several potential determinants are considered in this part of the analysis. Child s years of education can reflect the earnings potential in the labor market; it can also represent the earlier education investment that the parents devoted to this child. Number of contact days is defined to be the number of days during which the child either visited or called the elder parents in the past year. The literature on the motives of private transfers has been using the number of contact days as a proxy for child services that are not easily substitutable in the market. According to the theory of exchange motives for private transfers, the more services that the child provides to the parents, the more transfers the parents would send to the child (Cox, 1987). Child distance is a categorical variable with values from 1 through 7, with 1 being living in the parents household and economically dependent; 2 being living in the parents household but economically independent; 3 being living in adjacent dwellings; 4 being living in another household but in the same village as the parents do; 5 being living in another village but within the same county area; 6 being living in another province; and 7 being living abroad. As the value of the distance variable goes up, the physical distance between the parents and the adult children becomes bigger. Children s annual income and age are also included in the analyses. Specification (1) in Table 5 shows the OLS regression of the amount of transfers sent to each child on the child s characteristics. Both education and income variables are statistically significant and positive, which indicates positive conditional correlations between education, income and the amount transferred to the child. However, one might suspect that some household specific characteristics could be endogenously related to both the child s education and income, and the amount of transfers to children. For example, parents in one family might care more about their children than the parents in another family. To overcome this endogeneity issue, I estimate a within-family regression where the family fixed-effects are used. This means that only the deviations from the family mean are used in 1 In other words, the adult children in this sample have at least one adult sibling. 17

18 the regression, and the unmeasured family-specific characteristics are controlled. Specification (2) in Table 5 shows the within-sibling family fixed effects regression. After controlling for the family specific effects, the coefficient of the child s education remains significantly positive, whereas the coefficient of the child s income switches signs and loses its significance. The within-family analysis shows that the parents would like to transfer more to the child who has more earnings potential regardless of their current income status. However, even after controlling for family-specific effects, the unobserved child-specific characteristics could still bias the point estimates of the determinants of the transfers. For example, parents might adore a child who is more responsible or has a better appearance than his/her siblings by transferring more to this child. These favorable characteristics of this child might help him/her to gain more education, income or even the chance to move away from the rural area. As a result, it is the individual specific traits that lead to receiving more transfers from the parents, instead of the aforementioned determinants such as education, income, and distance from parents. To control for this potential bias, I utilize the longitudinal nature of the CHARLS data, and estimate an individual fixed-effect model. One drawback of the longitudinal fixed-effect model is that the time-invariant variables will be dropped during the estimation. Since the adult children in the sample have all finished their education, the individual fixedeffect model cannot identify the parameter of child education. Specification (4) in Table 5 presents the results for the individual fixed-effect analysis. The coefficient of the child distance becomes positive and significant after controlling for time-invariant individual specific attributes, and this result shows that the elder parents tend to transfer more to the children who are further away from their hometown. Discussion and Policy Implication In section A, it is found that the NRPP significantly crowds out the private transfers that the rural elders used to receive from their children. This finding implies that even though the NRPP is targeting the Chinese rural elders who have been suffering from poverty and loneliness, part of the pension benefits are reaped by the elders children, about 85% of whom are migrants in the urban areas. Even though this crowding-out effect could have dampened the efficacy of the NRPP, the relief from the burden of 18

19 Table 5: Determinants of Transfers Sent to Children Transfers to Children Households with more than one child Households with at least one child OLS Within-Family OLS Individual FE Specifications (1) (2) (3) (4) Child Education (604.7) (791.9) (244.4) Number of Contact Days (9.9) (8.4) (2.5) (9.1) Child Distance (1065.2) (1725.9) (342.3) (684.3) Child Income (728.7) (1349.7) (406.3) (624.1) Child Age (143.3) (269.7) (23.2) (16.4) Constant (9125.1) (21.1) (3122.7) (4950.9) Family FE no yes no yes Individual FE no no no yes Control Mean Observations Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 19

20 providing for the old might help the younger rural-to-urban migrants to assimilate into the urban areas and facilitate the urbanization process in China. In addition to the crowding-out effect, the results in section B shows that after receiving the pension benefits from the NRPP, the elders tend to transfer more to their children. However, there are not enough pieces of evidence to show that transferring more to children is a result of empowerment by the NRPP. Because the elders also tend to reduce the transfers to their siblings, and at the same time, there is no evidence that the elders are spending more in consumption, productive assets, and financial assets as shown in section C. It is not surprising that there is no significant impact of the NRPP on consumption due to the crowding-out effect, and the rural elders being reluctant to investment behaviors are in line with the previous studies (De Brauw and Roselle, 2008; Zhu et al. 2009). By reconciling all these findings, I hypothesize that responding to the NRPP by transferring to children is at least partially based on an investment motive. This is because in rural China where financial inclusion level is still low, elder farmers can hardly find access to financial products to store their wealth. On the other hand, they can invest in their migrant children to help them assimilate into the urban areas, and expect higher remittances later on or even expect the children to move them out of the rural areas. The returns to these kinship assets could be higher than the returns to productive investment in the rural area. Also, the risk of investing in migrant children is mitigated by the kinship. The analyses in section D provides further shreds of evidence for this investment motive hypothesis. The analyses of transfer determinants show that more educated children tend to receive more transfers from their elder parents. Because education reflects a child s earning potential, the parents might treat the more educated children as more productive assets, which deserve more investment. Furthermore, the result that the more distant children tend to receive more transfers could also be explained by the investment motive. Because more distant children tend to be in an area with higher administrative level, such as a city that has better income opportunities than the areas that are close to the children s rural hometown. The availability of jobs and the promising income opportunities could potentially make the children who are more distant from home more productive assets in their parents minds. The findings here can be used to answer the key policy question that whether the expansion of 20

21 cash transfer programs such as NRPP could boost the capital accumulation that is required for rural development in China. The answer is negative for two reasons. First, part of the public transfers will crowd out private transfers, which means the wealth distribution over the rural and urban area could remain intact. Second, as long as the returns to capital are lower than those in the urban area, the rural elders would tend to divert their wealth gained from public transfers back to the urban areas through their migrant children. Thus, expanding public cash transfer programs might assist the urbanization process by indirectly providing the young rural-to-urban migrants to assimilate into the urban area. However, it is unlikely that these public transfer programs could contribute to accumulating productive capitals that are necessary for rural development. Conclusion Public pension program is an important tool for alleviating old-age poverty, especially in rural areas where financial inclusion is usually low. However, it is unclear that whether such programs could improve the standard of living and capital accumulation for those elders who have been relying on the old-age support from their children. This study analyzed the impact of the recently introduced New Rural Pension Program in China on the rural elders transfer behaviors and expenditure patterns. Using a fuzzy regression discontinuity design, it is found that NRPP largely crowds out the private transfers from children, and the elders tend to transfer more to their children upon receiving the benefits from NRPP, even though there is no significant impact of NRPP on the other expenditure patterns such as consumption and investment. The analysis of the determinants of the amount of transfers sent to children reveals that rural elder parents tend to transfer more to their more educated and more distant children. These results are consistent with the hypothesis that the Chinese rural elders transfer to their migrant children based on an investment motive. The potential unequal returns to capital in the urban and rural areas might lead the rural elders to invest in their migrant children instead of productive assets in the rural area. Therefore, further expansion of public cash transfers program in rural China might consider incentivizing the elders to invest in both productive and financial assets to boost capital 21

22 accumulation for rural development. 22

23 References Alderman, Harold, and Ruslan Yemtsov. "How Can Safety Nets Contribute to Economic Growth?." TheWorld Bank Economic Review (2013): lht011. Ardington, Cally, Anne Case, and Victoria Hosegood. Labor supply responses to large social transfers: Longitudinal evidence from South Africa. No. w National Bureau of Economic Research, Becker, Gary. "A theory of social interactions." (1974). Becker, Gary S. "Altruism in the Family and Selfishness in the Market Place."Economica (1981): Bertrand, Marianne, Sendhil Mullainathan, and Douglas Miller. "Public policy and extended families: Evidence from pensions in South Africa." the world bank economic review 17.1 (2003): Cai, Fang, et al. The elderly and old age support in rural China. World Bank Publications, Chen, X., & Wang, T. (2014). Old-age pension, subjective well-being and mental health. In Paper presented in the NBER Summer Institute Cox, Donald. "Motives for private income transfers." Journal of political economy 95.3 (1987): Cox, Donald, and Marcel Fafchamps. "Extended family and kinship networks: economic insights and evolutionary directions." Handbook of development economics 4 (2007): De Brauw, Alan, and Scott Rozelle. "Migration and household investment in rural China." China Economic Review 19.2 (2008): Duflo, Esther. "Child health and household resources in South Africa: Evidence from the Old Age Pension program." The American Economic Review 90.2 (2000): Duflo, Esther. "Grandmothers and Granddaughters: Old Age Pensions and Intrahousehold Allocation in South Africa." The World Bank Economic Review 17.1 (2003): Edmonds, Eric V. "Child labor and schooling responses to anticipated income in South Africa." Journal of development Economics 81.2 (2006): Giles, John, and Kyeongwon Yoo. "Precautionary behavior, migrant networks, and household consumption decisions: An empirical analysis using household panel data from rural China." The Review of Economics and Statistics 89.3 (2007): Juarez, Laura. "Crowding out of private support to the elderly: Evidence from a demogrant in Mexico." Journal of Public Economics 93.3 (2009): JUNG, HAEIL, MAUREEN PIROG, and SANG KYOO LEE. "Do public pensions crowd out private transfers to the elderly?: evidence from South Korea." Journal of Pension Economics and Finance: 1-23.

24 Jensen, Robert T. "Do private transfers displace the benefits of public transfers? Evidence from South Africa." Journal of Public Economics 88.1 (2004): Lee, David S., and Thomas Lemieux Regression Discontinuity Designs in Economics. Journal of Economic Literature 48 (2): Lei, Xiaoyan, Chuanchuan Zhang, and Yaohui Zhao. "Incentive problems in China s new rural pension program." Research in Labor Economics 37 (2013): Maitra, Pushkar, and Ranjan Ray. "The effect of transfers on household expenditure patterns and poverty in South Africa." Journal of development Economics 71.1 (2003): McCrary, Justin. "Manipulation of the running variable in the regression discontinuity design: A density test." Journal of Econometrics (2008): Secondi, Giorgio. "Private monetary transfers in rural China: are families altruistic?." The Journal of Development Studies 33.4 (1997): Zheng, Huamao, and Teng Zhong. "The impacts of social pension on rural household expenditure: evidence from China." Journal of Economic Policy Reform (2015): Zhu, Yu, et al. "Where did all the remittances go? Understanding the impact of remittances on consumption patterns in rural China." Applied Economics46.12 (2014):