Researchers have studied family cash assistance programs across a range of settings and designs, from randomized controlled trials in Ghana to decades-long follow-ups of Americans born in the early 20th century.
The results consistently find that cash benefits to families improve children's well-being, across indicators of physical and mental health, education, and long-term outcomes like earnings and even longevity. That is, child poverty causes problems in childhood and adulthood, and policies that reduce child poverty, such as child allowances, would address these problems to a large degree.
This paper contains research from the United States and Canada, as well as from sub-Saharan Africa (SSA). As these sets of studies differ in context (developed vs. developing economies) and approach (SSA studies are more often randomized controlled trials), we organize them separately, beginning with the U.S. and Canada.
In 1911, the US government established its first welfare program, the Mother's Pension Program, which provided cash transfers representing 12 to 25 percent of family income, generally for about 3 years.
Male children of mothers accepted for the US Mother's Pension Program lived one year longer than male children of mothers who were rejected after initial acceptance {cite}aizer2016long.
This study also reported other medium-term outcomes, which can help estimate the effects of other cash transfer results for which mortality data is not yet reliably available.
The program "resulted in a significant 50 percent decrease in under-nutrition, a 13 percent increase in income, and an increase of 0.4 years of school among young adults."
By comparing these results to other research on mortality and being underweight {cite}flegal2005excess, income {cite}deaton2001mortality, and education {cite}cutler2006education, the authors establish that education and income explain 75 to 95 percent of the longevity increase, while the underweight channel explains a small share.
Nutrition can sometimes have mixed benefits on outcomes like obesity, which most but not all evidence suggests is a consequence of poverty.
A study of Alaska's Permanent Fund Dividend, a UBI ranging from $1,000 to $3,000 per person per year (including children) since its creation in 1976, found that each additional $1,000 in payments reduced child obesity by as much as 4.5 percentage points.
Research on the Food Stamp Program (now Supplemental Nutrition Assistance Program, or SNAP) shows that these benefits persist: access to food stamps as children reduced adult obesity {cite}hoynes2016long.
Since food stamps are treated as cash by most households {cite}hoynes2009consumption, or may even shift consumption toward less healthy groceries like soda {cite}whitmore2002food, these effects probably follow from a general increase in household resources, and would thus also apply to UCTs like child allowances.
However, a study of payments to Eastern Cherokee households found that they increased weight among children from poor families {cite}akee2013young.
When Manitoba, Canada, increased its child benefit in 2001, researchers found that the change improved motor and social development among children ages 0 to 3, and reduced aggression and anxiety among children ages 4-5, especially among girls {cite}milligan2009child.
A meta-analysis of 34 studies isolating income changes in OECD countries found that an income increase of $1,000 improved children's cognitive outcomes by 5 to 27 percent of a standard deviation {cite}cooper2017does.
It also identifies studies showing that income improves maternal health and well-being, which could be a factor in children's development.
A study of Earned Income Tax Credit reforms found that "a $1,000 increase in income raises combined math and reading test scores by 6 percent of a standard deviation in the short-run" {cite}dahl2012impact.
Similarly, research of Canadian child benefit reforms found that C$1,000 in benefits raised math and vocabulary scores by 7.4 and 6.8 percent of a standard deviation, respectively {cite}milligan2011child.
The above-mentioned OECD meta-analysis associated a $1,000 income increase with school performance gains ranging between 5 and 17 percent of a standard deviation.1
By reducing the cost of childbearing, child benefits modestly increase fertility (births per female).
In the US, a $1,000 subsidy increases the probability of having a child in the following two years by 0.4 percent, an effect driven by the Earned Income Tax Credit and personal tax exemptions on low-income, married women in their 30s {cite}mumford2017fertility.
Canada's child benefit had a larger effect of 16.9 percent per C$1,000, possibly due to its greater salience from being a standalone monthly program, while German child benefits only raised fertility among high-income parents considering a second child {cite}riphahn2017fertility.
As of 2016, women in the US report wanting an average of 2.7 children, yet they are having only 1.8 children---the largest gap in 40 years {cite}stone2018.
This suggests that additional fertility from child allowances would be welfare-enhancing.
In developing contexts, there is a large volume of literature on the efficacy of cash transfers in improving child outcomes. In particular, in recent times, countries in sub-Saharan Africa have routinely been the site of rigorously evaluated government and international organization-led interventions, creating a rich comparative literature on the impacts of cash transfers.
This report summarizes evidence from experimental and pseudo-experimental studies on the impacts of cash transfers on metrics of poverty, food security, health, educational attainment, child labour and empowerment-related outcomes. Thereby, it seeks to provide an indication of both the range of outcomes affected by cash transfers and the magnitude of those effects. Of course, the impacts of any intervention vary markedly between contexts, even between the programs and countries considered by this report.
In reporting evaluation impacts, where possible, percentage point changes are reported alongside the baseline treatment group mean in brackets. An asterisk is included after each statistically significant coefficient at the
This report considers 20 programs spanning 11 countries: 13 pure unconditional cash transfers (UCTs), 2 pure conditional cash transfers (CCTs), 4 programs involving independent treatment arms that included both CCTs and UCTs, and one old age pension. Table 1 shows the program title, country, conditionality, and a short-hand name by which we refer to each program throughout the report. While the requirements put upon recipients of individual conditional cash transfers are detailed under their associated outcomes, in general, they tend to include any of: regular attendance at health checkups, enrollment in health insurance and school attendance.
Table 1: Studies considered by this report
| Country | Program Name | Transfer Type | Project Reference | Studies |
|---|---|---|---|---|
| Burkina Faso | Nahouri Cash Transfers Pilot Project | Both | Burkina Faso NCTPP | {cite}BurkinaFaso_final 2 {cite}BurkinaFaso_School 3 |
| Ghana | Livelihood Empowerment Against Poverty (LEAP) | Both | Ghana LEAP | {cite}Ghana_final |
| Kenya | Hunger Safety Net Programme | UCT | Kenya HSNP | {cite}Kenya_HSNP_final |
| Kenya | Orphans and Vulnerable Children Cash Transfer | UCT | Kenya CT-OVC | {cite}CTOVC_food 4 {cite}HANDA201536 5 |
| Kenya | Kenya GiveDirectly Experiment | UCT | Kenya GD | {cite}KenyaGD_domvio |
| Kenya | Kenya GiveDirectly UBI Experiment | UCT | Kenya UBI | {cite}Kenya_UBI |
| Lesotho | Child Grant Programme | UCT | Lesotho CGP | {cite}Lesotho_CGP |
| Malawi | Social Cash Transfer Programme | UCT | Malawi SCTP | {cite}Malawi_SCTP_final |
| Malawi | Zomba Cash Transfer Programme | Both | Malawi ZCTP | {cite}MZCTP |
| Rwanda | Rwanda GiveDirectly Experiment | UCT | Rwanda GD | {cite}RwandaGD |
| South Africa | Child Support Grant and Foster Grant | UCT | South Africa CSG | {cite}SouthAfrica_CSG_final |
| South Africa | South Africa Old-Age Pension | Pension | South Africa OAP | {cite}SA_Duflo |
| Tanzania | Tanzania Social Action Fund | CCT | Tanzania TASAF | {cite}Tanzania_TASAF_final |
| Uganda | Social Assistance Grants for Empowerment: Vulnerable Family Support Grant | UCT | Uganda SAGE | {cite}Uganda_SAGE_final |
| Uganda | WFP Karamoja cash transfer | CCT | Uganda KWFP | {cite}Uganda_WFP_food |
| Uganda | Uganda GiveDirectly Experiment | UCT | Uganda GD | {cite}Uganda_GD_final |
| Zambia | Monze Cash Transfer Pilot | UCT | Zambia MCTG | {cite}Zambia_MCTG_final |
| Zambia | Child Grant Programme | UCT | Zambia CGP | {cite}Zambia_CGP_final |
| Zimbabwe | Community-led Cash Transfer Program | Both | Zimbabwe CCTP | {cite}Zimbabwe_CCTP_Final |
| Zimbabwe | Harmonised Social Cash Transfer | UCT | Zimbabwe HSCT | {cite}Zimbabwe_HSCT_final |
Poverty metrics offer a simple means of estimating a household's capacity to provide for the material needs of its children. Studies on cash transfers in sub-Saharan Africa tend to report 3 such metrics, namely changes in:
- the poverty headcount, a measure the share of the population earning or consuming below a nationally-determined poverty line;
- the poverty gap index, a measure of the intensity of poverty; and
- the squared poverty gap, a measure of the severity of poverty that more heavily weights relatively poor households.
Against these three measures, there is clear evidence from the Ghana LEAP, Kenya HSNP, Malawi SCTP and Zambia CGP that cash transfers are effective tools in reducing the incidence, intensity and severity of poverty. The Uganda SAGE found similar, though less statistically significant impacts by defining a poverty line with respect to consumption expenditure. That said, the Lesotho CGP and Zimbabwe HSCT evaluations found no significant effects on any measure of poverty, though the studies' authors note that this may relate to the irregularity of payments through the Lesotho CGP and a large reduction in net gifts and remittances in the Zimbabwe HSCT.
The findings of these studies are summarized in Table 2, which shows the percentage point (pp) reduction in measures of poverty driven by each of the programs alongside the baseline treatment mean. In its totality, the evidence indicates that cash transfers do reduce household poverty.
Table 2: Evidence from sub-Saharan Africa: Cash transfers reduce poverty
| Program | Headcount | Poverty gap | Squared poverty gap |
|---|---|---|---|
| Ghana LEAP | -2.1** [93] | -2.6** [51] | |
| Kenya HSNP | -3.9 [88] | -7.5** [41] | -6.9** [23] |
| Lesotho CGP | 3.8 [68] | 3.2 [47] | |
| Malawi SCTP | -14.9*** [82] | -10.9*** [60] | -12.2*** [41] |
| Uganda SAGE | -8.3* [44] | -2.2* [10] | -0.83* [3] |
| Zambia CGP | -4.1** [94] | -8.4** [60] | -7.6** [43] |
| Zimbabwe HSCT | -0.5 [92] | -0.0 [55] | -0.2 [37] |
To determine whether increases in household income translate to improvements in the material well-being of children, many programs measure changes in child ownership of basic items, namely, a pair of shoes, a blanket and a spare set of clothing. Most evaluations reported highly significant effects on child deprivation metrics, reflecting households' tendency to spend transfers on meeting the material needs of children, as well as the potential efficacy of cash transfers as an intervention to improve the material wellbeing of children.
All programs that reported on child shoe ownership found significant percentage point increases following the cash transfer and evidence on each of the other indicators was similarly strong. However, findings across the remaining indicators (posession of a blanket, a change of clothing and all 3) are not as clear, likely due to ceiling effects. In the Zimbabwe HSCT for example, 77% of children had blankets at baseline and this increased substantially at mid line, leaving almost no room for statistically signficant increases at endline. These findings are summarized in Table 3, which shows the percentage point impact of cash transfers on measures of child deprivation.
Table 3: Evidence from Sub-Saharan Africa: Cash transfers reduce child material deprivation
| Program | Owns all | Blanket | Shoes | Change of clothes |
|---|---|---|---|---|
| Ghana LEAP | 10*** [22] | 9.5*** [63] | ||
| Malawi SCTP | 30.6*** [13] | 29.2*** [36] | 32.0*** [20] | 10.2 [76] |
| Tanzania (TASAF) | 17.9*** [42] | 6 | ||
| Zambia MCTG | 22.7*** [16] | 16.5*** [61] | 22.6*** [21] | 4.8** [77] |
| Zimbabwe HSCT | 26.1*** [37] | 1.3 [77] | 25.1*** [41] | 2.7 [78] |
There is also some evidence that cash transfers improve nutrition of children and other household members. Households that received a cash transfer tended to spend significantly more money on food, and these households tended to benefit from greater diversity in their diet and more meals per day.
The change in food consumption as a percentage of the total transfer was high across most interventions.
For example, a 2018 World Bank meta-analysis found that the cash transfer elasticity of food consumption was high across Zambia's CGP (110%**), Malawi's SCTP (148%**), and Kenya's CT-OVC (60%**).
On the other hand, some transfers---most often smaller or irregular transfers---had no impact on food consumption.
Across all of the meta-analysis subjects, for each dollar transferred, average household food consumption increased by 36 cents** {cite}World_Bank_meta.
This increase in household food consumption is reflected in more concrete measures of nutritional outcomes for recipient households. The number of meals eaten per day unilaterally increased across the programs that measured it: 0.09*** [2.6] in the Ghana LEAP and 0.29** [1.9] in the Malawi SCTP. Similarly, the proportion of people consuming more than one meal per day increased in both the Malawi SCTP 13 pp*** [79] and the Zambia MCTG 11** pp [71] and the majority of programs reported improvements in indices of food securty.
However, the evidence on other measures of food security is less clear. For example, dietary diversity improved in the Ghana LEAP*** and Malawi SCTP***, while null effects were found in the Kenya CT-OVC. Looking specifically at child outcomes, there is mixed evidence on whether transfers reduced the number of meals that children are forced to skip as a result of economic pressures. The Uganda GiveDirectly program found a 45% reduction in meals skipped by children7 but the Lesotho CGP found no effect on the same metric.
Evidence from the Uganda WFP program indicates that cash transfers may outperform food transfers of similar value across a range of food security measures. The study found that cash transfers outperformed food transfers in increasing both individual and household indices of dietary diversity as well food consumption.
While food security and nutrition are improved markedly by cash transfers, there is weaker evidence of an impact on child health outcomes. Across child anthropomorphic outcomes, healthcare utilization, morbidity and mental health outcomes, there is only limited evidence that child health improves as a result of a cash transfer.
The GiveDirectly Rwanda program found small positive impacts of transfers on two anthropomorphic measures, Height-for-age and Weight-for-age (0.091 standard deviations** [0.15] and 0.067 standard deviations** [0.15], respectively), as did the South African CSG. However, most other evaluations, including the Ghana LEAP and Zambia CGP, found null effects, despite Ghana LEAP's provision of free enrollment in a national health insurance program. The Lesotho CGP evaluation even found a significant reduction in Weight-for-age among program recipients, though there was no corresponding increase in the proportion of children suffering from malnutrition. These small and inconsistent effects may result from the long delays between nutrition and its anthropomorphic benefits. There is also some evidence that the gender of the recipient affects the impact of cash transfers on health anthropomorphic outcomes. The South African Old Age Pension only had significant impacts on Weight-for-height for girls if their grandmother (and not grandfather) was eligible for the pension.
Despite many cash transfers aiming to improve access to healthcare, there is little evidence that they increase household expenditure on health-related services. The Malawi SCTP reports significantly higher utilization of healthcare services (and particularly curative services) on receipt of the transfer. However, the World Bank meta-analysis finds no significant divergence in mean expenditure on health services and both the Tanzania TASAF and Zimbabwe HSCT actually report decreased expenditure. That said, there is some evidence from the Burkina Faso NCTTP that conditioning transfers on quarterly visits to the local health clinic for child growth monitoring for children under 6 (as well as 90% attendance for children 7-15) leads to greater healthcare service usage. It found no effect for the unconditional transfer treatment arm but the conditional transfer increased the number of routine preventative health clinic visits by 0.43** [1.03]
There is similarly weak evidence of improvement across a range medium-term health outcomes. The Kenya UBI found between 3 and 6 pp** [44] reductions in the probability of a household member being sick (depending on time-scale), and the Malawi SCTP found a 6 pp reduction** [30] in the incidence of illness in the past two weeks. However, the majority of studies, including the Ghana LEAP and Uganda SAGE, found no similar effects. The Zimbabwe CCTP and Ghana LEAP also found no evidence of increased vaccination coverage. Looking specifically at child outcomes, there is tentative evidence from the Rwanda GD that large cash transfers reduce child mortality by -1** pp [3.3] and this tentatively positive evidence characterises what we know about the impact of cash transfers on a broad range of health-related outcomes.
There is interesting, though inconclusive, evidence of mental health impacts. Kenya GiveDirectly found a positive*** impact on a village-level psychological well-being index and the Kenya UBI found a significant decrease*** in depression on the CES scale. By contrast, the Zimbabwe HSCT found a weakly significant increase in depression* but a highly significant improvement in subjective well-being***, while the Malawi SCTP, Tanzania TASAF and Zambia MCTG found null effects.
There is some evidence that unconditional cash transfers improve indicators of educational attainment.
For example, the Malawi SCTP found that receipt of the transfer increased enrollment by 9 pp*** [69%] and the Zambia MCTG increased enrollment by 9 pp** [70%] for young males (7-14) and 19 pp** [65%] for older females (15-17).
The Zambia CGP similarly found a 4 pp** [-] increase at 30 months, though this was non-significant at 36 months, largely because of 'catch-up' among the control group.
However, the evidence isn't conclusive. The Kenya CT-OVC and Ghana LEAP only found significant enrollment effects on children age 13 to 17. The Zimbabwe HSCT found no enrollment effects, though the authors indicate that this could be due to recipients losing means tested scholarships that they would have otherwise held. In its totality, the evidence likely indicates that cash transfers cause a small improvement in enrollment outcomes.
On the intensive margin, there is some weak evidence that unconditional cash transfers increase school attendance.
The Malawi SCTP raised regular attendance (school attendance without withdrawl from school for two or more consecutive weeks) by 13 pp*** [59%].
However, most individual studies, as well as a World Bank meta-analysis, arrived at null attendance results {cite}World_Bank_meta.
The programs reported similarly mixed results for the number of grades completed by children (grade attainment).
There is some evidence that conditioning transfers on school attendance may induce more consistent educational outcomes compared to unconditional cash transfers. The Tanzania TASAF, which required recipients aged 0-5 to attend a clinic 6 times per year and aged 7-15 years to be enrolled with 80% attendance, also found a 6 pp* increase in the proportion of children aged 0-18 years who had ever attended school, though there was no effect on current attendance.
Some studies included both CCT and UCT arms, offering the opportunity to compare their impacts. The Burkina Faso NCTPP CCT arm, which required recipients aged 7-15 enroll and retain 90% attendance each quarter, increased the share of school days attended by recipient students by 13.4 pp***, while its UCT arm found a null effect. Similarly, while enrollment increases for transfers conditional on child attendance (17.9 pp***), there is no effect for unconditional transfers. Malawi's ZCTP also found that its CCT (which required 80% attendance) improved the fraction of days that students attended school by around 8 pp** as well as varied student test scores by roughly 0.15*** standard deviations each, while its UCT did not significantly improve these outcomes. Thus, there appears to be some evidence of stronger educational impacts where transfers are conditioned on attendance.
The studies considered found inconsistent evidence on the effects of cash transfers on child labor.
There was limited consistency in the definition used to define child labor, with some studies using the UNICEF thresholds for identification of an excessive number of hours spent working that would lower school participation and many using similar though slightly altered definitions.8
While the Kenya CT-OVC found a modest reduction (-12%*** [42%]) in own-farm labor, the vast majority of studies found no effect of cash transfers on child labor incidence, though this may be a result of substantial ceiling effects.
For example, it was unsurprising that the Kenya CT-OVC found no reduction in wage labor participation, as at baseline less than 2% of children aged 10-15 were engaged in wage labor.
That said, there were some unusual results which highlight the weakness of evidence on the relationship between cash transfers and child labour incidence. The Malawi SCTP found a 9 pp*** increase in child labour incidence, with the majority of the increase in hazardous activities.
The absence of evidence is reflected in the World Bank meta-analysis, which found no significant effect {cite}World_Bank_meta.
In recent years there has been substantial interest in whether cash transfers, particularly if they are given to women, improve metrics of women's empowerment. As yet, there is little evidence that cash transfers improve women's empowerment.
The majority of studies reported the incidence of marriage and pregnancy among female-identifying recipients. For example, the UCT arm of the Malawi ZCTP reported large reductions in the likelihood of marriage and pregnancy, 9 pp*** [17.3] and 8 pp*** [12.1] respectively while there were null effects in the CCT arm, and the Kenya CT-OVC reported a 4.9 pp** [15.0] reduction in the incidence of pregnancy among cash transfer recipients and a null result for marriage. The Ghana LEAP found a reduction in the probability of marriage for women from 12-24 years at baseline but not women 12-17 and no effect on pregnancy. The majority of studies, including the Malawi SCTP, as well as the Tanzania TASAF reported null effects.
Across less commonly reported metrics, such as empowerment indicies, the usage of modern contraceptives and incidence of sex- and gender-based violence, the evidence is similarly weak. While the Kenya GiveDirectly program reported improvements in an index of women's empowerment*** and a reduction in both physical*** and sexual violence** indices for female recipients, there were null results in similar outcome metrics across the Ghana LEAP, Tanzania TASAF and Zambia MCTG.
While the evidence from sub-Saharan Africa is complex and difficult to interpret, it is clear that cash transfers significantly improve the lives of children that live in households that receive them. In particular, these children are substantially less likely to be poor and more likely to have their material needs satisfied, and they also receive some educational and health benefits.
:filter: docname in docnames
Footnotes
-
Based on randomized controlled trials and quasi-experiments listed in Table 7; one study found a 23 percent performance increase per $1,000 among boys. ↩
-
This study covers health-related indices from the Burkina Faso NCTPP. ↩
-
This study covers education-related indices from the Burkina Faso NCTPP. ↩
-
This study covers nutrition and food security-related indices from the Kenya CT-OVC. ↩
-
This study covers youth pregnancy, marriage and a range of empowerment-related indicators from the Kenya CT-OVC. ↩
-
The study reports higher expenditure on child clothing. ↩
-
While the duration over which meals were skipped was not reported, the study noted a -1.93*** [4.22] reduction in meals skipped by children. ↩
-
Consider, for example, the definition of child labor used by the Malawi SCTP program:
-
For economic activities the following age-specific thresholds are applied to identify child labour: * Ages 5-11: 1 any engagement in the week prior to the interview; * Ages 12-14: 14 hours or more in the week prior to the interview. * Ages 15-17: 43 hours or more in the week prior to the interview.
-
For household chores (in combination with economic activities) the following age-specific thresholds are applied to identify child labour: * Age 5-14: 28 hours or more in the week prior to the interview; * Age 15-17: 43 hours or more in the week prior to the interview.
-
Hazardous economic activities carried out in the year prior to the interview are considered to be child labour for all children under the age of 18.
This differs marginally from the standard UNICEF indicator used by the Lesotho CGP. UNICEF’s indicator similarly classes hazardous economic activities but the thresholds for economic and unpaid (household) labour are as follows:
- Ages 5 to 11 years: At least 1 hour of economic work or 21 hours of unpaid household services per week.
- Ages 12 to 14 years: At least 14 hours of economic work or 21 hours of unpaid household services per week.
- Ages 15 to 17 years: At least 43 hours of economic work per week.
-