3.1 Government cash transfers
Government cash transfer programs—both conditional and unconditional—are the most ubiquitous (and most evaluated) social protection policies globally—with 52 countries with a conditional (CCT) and 119 countries with an unconditional (UCT) cash transfer program (Gentilini et al. 2014). Government CCTs are typically targeted at poor households with children and involve small, regular cash payments to the household conditional on a certain set of behaviors usually related to children’s education or health. UCTs also typically target the poor, but are not conditioned on any specific behavior. Old-age pension programs and child support grants are the most common forms of UCTs, but some programs also target orphans and vulnerable children. As mentioned above, there is a wealth of evidence on the impact of government CCTs and UCTs on health and education, but a much more limited body of evidence on adult labor market outcomes.
CCTs and UCTs, targeted at poor families with children, can have effects on adult labor market outcomes for two distinct groups of target populations: the children, whose future participation and remuneration in the labor market is affected through increased human capital accumulation, and adults, who may change their labor market participation in response to the increased current income and within-household substitution effects (e.g., in response to decreasing child labor).Footnote 3 Below, we describe the effects of cash transfer programs on labor market outcomes of adult beneficiaries in the shorter run (generally 2- to 5-year follow-ups), followed by effects on future outcomes of adolescents in the longer run (typically 5- to 10-year follow-ups).
The most well-known government CCT program is Mexico’s PROGRESA, which later became Oportunidades. Parker and Todd (2017) provide a review of its effects on labor market outcomes for adult beneficiaries and find no effects on work or leisure (citing Parker and Skoufias 2000; Rubio-Codina 2010). They note some effect on women substituting for children’s time within the household, because child labor (including paid, domestic, and household agricultural) goes down under PROGRESA/Oportunidades. Bianchi and Bobba (2013) find that expected future transfers increase the probability of entrepreneurship (self-employment) in the short run. However, this effect disappears in the medium run, remaining only in areas with low levels of self-employment at baseline, (canceled by a negative effect in areas with high self-employment at baseline). Finally, Gertler et al. (2012) observe a positive short-term impact on non-agricultural microenterprises and total agricultural income among treated households. However, in the medium term, i.e., 5–6 years after the start of the program, they find no effects on outside wages. In summary, the effect of Mexico’s CCT program on the labor market outcomes of its adult beneficiaries is small at best.
Banerjee et al. (2017) re-analyze the results of seven randomized controlled trials of government-run CCT programs from six countries to examine impacts on labor supply.Footnote 4 The authors do not observe a significant effect either individually or pooled on employment or hours of work. Similarly, they find no pooled effect on whether work is self-employed/within family vs. outside the household. They do find a decrease in outside work and an associated increase in within household work in PROGRESA, but the opposite pattern holds for RPS (which has a similar transfer size). Similarly, there are no overall pooled effects on any outcomes when disaggregating by gender.Footnote 5
Elsewhere, examining two similar unconditional cash transfer programs in Malawi and Zambia targeting labor-constrained households (i.e., households with high dependency ratios), (de Hoop J, Groppo V, Handa S: Household micro-entrepreneurial activity and child work: evidence from two African unconditional cash transfer programs, submitted) find that such households substitute away from working as wage laborers and start spending more time in own-agricultural work in both countries. In Zambia, overall economic activity increases, along with participation in non-agricultural household businesses, while these changes are not observed in Malawi.Footnote 6 Evaluating the effects of a 2-year child grant program in Zambia (targeted to households with children under the age of three), Prifti et al. (2017) find similar switches from off-farm paid work to own-farm labor, along with increases in hired agricultural labor among beneficiary households. Using the same RCT, Handa et al. (2018) find increases in the likelihood of operating a non-farm enterprise and revenues from those enterprises for the Child Grant Program (CGP), but not the multiple category targeted program (MCP) for labor-constrained households.Footnote 7
While Child Support Grants target families with children, old age pension programs target the elderly. These schemes affect not only the labor supply of the elderly beneficiaries themselves, but also of the prime-aged adults and children, who are related to pensioners or would-be pensioners. Below, we summarize the findings from South Africa’s old age pension scheme (OAP), which is perhaps as well-known and well-studied as Mexico’s PROGRESA.
Bertrand et al. (2003) find sharp decreases in the labor supply of prime-aged individuals living with elderly (household members, aged 16–50) around the time they become eligible for pensions, especially when the pensioner is female.Footnote 8 The effects are both on employment status (extensive margin) and on hours worked (intensive margin), larger for older adults, and largest for the oldest son. Posel et al. (2006), however, find that when they focus on non-resident household members, they find that African females, but not males, are more likely to be migrant workers in female-pensioner households. Ardington et al. (2009) argue that the OAP increases employment among prime-aged adults, by easing credit constraints for migration and job search and by increasing the availability of elderly to care for small children. Ranchhod (2006) finds that pension eligibility causes large declines in labor force participation among the elderly themselves, as well as an increase in flextime work along with a reduction in hours worked among those who stay in the labor force. Lam et al. (2006) also find that age of pension eligibility is associated with accelerated rates of retirement, faster for men than women, but not as high as the jumps in retirement observed in European countries.Footnote 9 They also point out that many elderly persons live in three- or skipped-generation households, which may in part be a response to the OAP itself (Hamoudi and Thomas 2014) and makes interpretation of studies that treat the OAP as exogenous more difficult. The causal identification of the effects of pension eligibility is therefore likely to be more credible for the labor market responses of the elderly themselves (Lam et al. 2006; Ranchhod 2006) than the working age adult relatives living in their households (Bertrand et al. 2003; Ardington et al. 2009).
We conclude this section by discussing the labor market outcomes of young adults, who were exposed to a CCT or a UCT program as children in beneficiary households. Behrman et al. (2011) observe weak positive effects on working (labor force participation) for females who were 13–15 just prior to the start of PROGRESA 6 years later. For males 13–15 at baseline, they find a weak negative effect on working in agriculture—possibly since some are still in school.Footnote 10 Several recent evaluations of cash transfer programs from Asia, Latin America, and Africa have added to the evidence base on the effects of cash transfers on the future outcomes of child beneficiaries. In Cambodia, Filmer and Schady (2014) examine the effects of a 3-year secondary school scholarship program and find no effects on employment or earnings 5 years after baseline (2 years after the end of transfers/scholarship funds).Footnote 11 In Ecuador, Araujo et al. (2016) find no effect of large cash transfers on employment 10 years after the start of the program among young adults aged 19–25. In Nicaragua, Barham et al. (2017) examine a CCT program, focusing their analysis on a sub-group most at-risk for dropout from school (boys, aged 9–12), so that program impacts on human capital accumulation are substantive. In this sub-group, 10 years after baseline, they find significant and meaningful increases in labor force participation (0.236 SD effect on a LFP index) and earnings (0.192 SD effect on an earnings index). Off-farm employment, migrating for work, salaried non-agricultural employment, and monthly earnings are all substantially higher for this group of young men. In Honduras, 2 years of receiving vouchers for health and education (conditional on mothers attending pre-natal check-ups and children attending school, among other requirements) caused no long-term effect on the labor market outcomes of young adults (Ham and Michelsen 2018). However, when combined with supply-side incentives tied to improvements in the quality of service delivery, there were small effects on young women’s labor force participation, but not on men’s.
In contrast to the findings in Nicaragua, but more in line with those in Cambodia and Ecuador, (Baird S, McIntosh C, Özler B: When the Money Runs Out: Do Cash Transfers Have Sustained Effects?, unpublished) evaluate the medium-term labor market effects of a 2-year cash transfer experiment targeted to initially never-married young women in Zomba, Malawi.Footnote 12 Five years after baseline (and 2 years after the cessation of transfers), the authors find no impact of either the conditional or the unconditional cash transfer intervention on the likelihood of any wage work in the past 3 months, effective daily wage in the past week, hours spent in self-employment or paid work, or labor income in the past five seasons (15 months), nor were there any effects on assessments of basic labor market skills relevant for rural Malawi, such as reading and following instructions to apply fertilizer, making correct change during a hypothetical market transaction, sending a text message and using a calculator on a mobile phone, or calculating profits for a hypothetical business scenario.
3.2 Charitable giving and humanitarian cash transfers
GiveDirectly is a non-profit charity organization that provides cash transfers to poor households via mobile phone in Kenya, Uganda, and Rwanda. Haushofer and Shapiro (2016; Haushofer J, Shapiro J: The long-term effects of unconditional cash transfers: experimental evidence from Kenya, unpublished) evaluated an early version of their efforts in Kenya. In this study, villages in the target region were selected based on population and the share of households that lacked metal roofs, with households with thatch roofs selected for transfers within the villages. There was experimentation with the transfer modality (monthly vs. lump-sum; female vs. male recipients; small vs. large), but the total transfer amount was approximately PPP $1000, made unconditionally to selected households via a mobile payment (M-PESA) account they were required to have. Approximately 9 months after the first transfer, there were no effects on labor supply or occupational choice. Households owned more livestock, owing mainly to those receiving large transfers (PPP $1525), but their earnings or profits from agricultural and business activities were unchanged (revenues were up but so were costs). The findings indicate a relaxation of liquidity constraints, especially in the large transfer arm, but the windfall income was used primarily for consumption, assets, and upgrading to metal roofs, rather than investments causing changes in labor market outcomes.Footnote 13
In 2015, the Overseas Development Institute (ODI) issued a report of a high level panel on humanitarian cash transfers (ODI 2015). While the report calls for increased use of cash transfers in humanitarian assistance settings, it is largely silent on the effects of cash transfers on labor market outcomes. This is natural as increased labor market participation or change in occupational choice is not a primary goal of humanitarian assistance programs, which aim to prevent adverse outcomes for vulnerable households by ensuring adequate coverage of basic needs. Furthermore, relief operations, especially in cases of emergency, do not lend themselves to well-identified impact evaluations. However, it should be noted that refugee status is increasingly more permanent, or at least long-term, rather than temporary and, hence, assistance programs do aim to increase work opportunities and earnings for this population.
As a result, many studies of humanitarian assistance report impacts on food security and nutrition, but not on labor market outcomes. For example, randomized experiments comparing food assistance by the World Food Programme (WFP) with cash transfers (or vouchers) for Colombian refugees in Ecuador and food insecure households in Niger do not report any impacts on labor supply, earnings, or occupational choice (Hoddinott et al. 2018; Hidrobo et al. 2014). Lehmann and Masterson (Lehmann C, Masterson D: Emergency economies: the impact of cash assistance in Lebanon, unpublished) examine the effects of a winter cash assistance program targeted to Syrian refugees in Lebanon. Eligible refugees in need of assistance were selected for the program if they lived at or more than 500 m above sea level, which provides the identification strategy for the study. The transfers, which amounted to about $575 during November 2013 to March 2014, reduced the share of households that reported any of their children working during the past month from 10% in the comparison group to 4% among the beneficiaries. The program also reduced adult labor supply during the past 4 weeks by 13% from 3.1 days in the comparison group to 2.7 days in the treated group. The evidence here is consistent with the standard economic model of labor supply, where beneficiaries may prefer leisure to costly (and perhaps dangerous or risky) work.
3.3 Private transfers of cash in the form of remittances
Remittance flows to developing countries totaled $429 billion in 2016, exceeding the total of all international aid flows (World Bank 2017). Just as with public transfers, there is often expressed fear that these private transfers or remittances will “make people lazy” (e.g., GMA News online 2007). In response to this fear, there is a relatively large literature that attempts to measure the impact of remittances on labor outcomes. However, unlike studies of public transfers that randomly allocate transfers to some households and not others, no study has randomized private transfers to measure their response. Instead, one needs to compare the behavior of households that receive remittances to those that do not, raising concerns about the quality of much of this evidence. We therefore restrict our discussion to those studies we consider the most reliable.
One type of study considers remittances as largely unconditional cash transfers, examining what happens when a household receives a shock to the amount of remittances they receive. The clearest example is Yang (2008), who examines Filipino households with migrants abroad, where exchange rate shocks during the Asian financial crisis meant that households with migrants in some destinations suddenly experienced a large increase in remittances compared to households with migrants in destinations, where the local currency did not appreciate against the Philippine peso. He finds that households receiving more favorable shocks keep children in school longer and reduce child labor hours. This may be a pure income effect, or could also reflect a liquidity constraint effect that prevented households from investing as much as they wanted in schooling. The impact on total adult labor hours is positive, but not statistically significant. Adults increase the number of hours they spend in self-employment, and households are more likely to start new entrepreneurial activities. This leads to a positive, but not statistically significant, change in self-employment income, and negative, but not statistically significant, change in wage income. The impact for adults appears to be through the self-employment liquidity effect—a channel also emphasized by Woodruff and Zenteno (2007) for Mexican migrants.
A second type of study considers remittances as conditional cash transfers, where the condition to receive remittances is that the household has to send a member abroad to migrate. As a result, what can be measured is the overall impact of migration, which captures both the direct effect of higher remittances, and the indirect effects that arise from the absence of a household member. Gibson et al. (2011) use a migration lottery for Tongans wanting to migrate to New Zealand to measure these impacts on household members remaining behind in Tonga. They find no significant changes in adult labor supply or self-employment of these remaining adults. Clemens and Tiongson (2017) use a policy discontinuity that arises from Filipino migrants having to pass a language test to be able to migrate to Korea. Comparing households where this test is just passed (and so migration occurs) to those where it is just failed, they find no significant impact of migration on the hours worked or self-employment of other adults in the household.
Taken together, the more reliable studies in the literature do not support the view that remittances make recipients lazy and work less. Where good opportunities for self-employment are available, the pure remittance effect can instead alleviate credit constraints and cause more work in own businesses. But when households also lose members from migration and are in remote areas with few opportunities for business generation, this self-employment effect may not apply. Finally, the general equilibrium effect of other community members migrating to earn remittances has been to raise wages for those remaining behind (e.g., Mishra 2007; Akram et al. 2017).