Southern African Regional Poverty Network (SARPN) SARPN thematic photo
Events Last update: 2020-11-27  

 Related documents


The impact of conditional cash transfers on human development outcomes

A review of evidence from PROGRESA in Mexico and some implications for policy debates in South and Southern Africa

Paul Gertler, Chief Economist: Human Development

World Bank

27 January 2005

[Printer friendly version - 57Kb < 1min (6 pages)]     [ Share with a friend  ]

One of the greatest tragedies of poverty is its intergenerational transmission. Children who grow up in poor households tend to remain poor. Because they are malnourished, have poor quality education and lower educational qualifications they lack basic capabilities as adults to take advantage of opportunities. To try and overcome these barriers the Mexican government established PROGRESA to overcome both immediate and long term problems of poverty. It provides households with cash to overcome barriers to meeting basic needs on condition that parents invest in improving the nutrition, health and education of their children. While the focus is on children, other family members also have to undergo medical check ups and attend hygiene and health education sessions. By setting conditions for cash transfers PROGRESA attempts to overcome the shortcomings of demand (cash transfers) and supply (provision of services) driven programmes.

PROGRESA is a big programme, unlike the small village programmes that most learning on development is based on. Between 1997 and 2000 it enrolled 2.6 million households from 50 000 rural villages and added about 2 million urban families between 2001-3. Currently it spends about US$2.5 billion a year.

Why is PROGRESA likely to have a bigger impact than traditional cash programmes?

There are two typical approaches we adopt in attempting to improve health outcomes. The one is to provide cash on the assumption that people cannot afford food and basic services. The other approach is to improve the supply of services.

The problem with cash transfers is that families may have other priorities apart from improving health outcomes and may use the money for other purposes. We have very little evidence of the effects of cash transfers on health. A US study shows no effects. A study done in South Africa looks at the impact of the big expansion in pensions after 1995. It found that if the money went to a grandmother the health outcomes of granddaughters improved, but not of grandsons. If the money was given to grandfathers there was no effect on health outcomes. This indicates that cash transfers might work but it depends on whom the money goes to.

Programmes aimed at increasing the supply of services experience problems with take up. In most cases it is those who are already accessing the health care system that benefit most, while those who are most in need are least likely to use the facilities.

PROGRESA overcomes this by providing cash to relax the income constraint but also providing an incentive to use it for the purposes of the programme.

The other thing that PROGRESA overcame was the political problem. Most political leaders in democracies are on a two to five year election cycle and are more inclined to support programmes that give a short term return. The cash transfer provides this and opened the way for politicians to engage in a more technical discussion on securing the longer term health benefits.

Compared to a traditional cash transfer programme the additional costs involved in running the programme were quite small. The cost of monitoring whether people complied with the conditionality was also quite small – about 1% of the cost of the programme.

There was a strong commitment to evaluating the programme both to avoid spending money on things that did not work, and to produce evidence that would ensure that the next generation of policy makers would not abandon the programme. To ensure a rigorous evaluation, external evaluators were brought in to conduct a controlled randomised trial. The fact that Mexico lacked the resources to implement the programme in all eligible communities from the start provided the opportunity to use the communities that did not get the benefits in the first round as a control group.

The decision on which communities would go first was based on a public lottery once the qualifying communities had been identified. This ensured equity and protected the government from charges of corruption. In practice there was some intervention to ensure geographical spread. After the first three years the control group was phased into the programme.

Eligibility was determined on the basis of the national census using a poverty index based on education, family size, assets and income. Communities that were 50-70% poor went first with small rural communities prioritised. Once communities were identified, households that qualified were identified using a proxy means test. Most households (97%) identified agreed to participate and were enrolled in the programme for three years. Those who did not enrol had to wait for the next enrolment in three years time. This was to prevent people migrating to enrol in the programme.

Cash transfers amounting to about a third of household income were given to the female head of the household. About 70% of this money was spent on better food. In order to get the cash, children had to attend school and family members had to attend health clinics and hygiene education programmes. There were prenatal clinics, well baby clinics and clinics for preschool children. Some upgrading of services was required in villages to ensure that services were available.

Impacts on health

One problem with measuring the incidence of illness is that people have different definitions of illness. Random assignment cancels this out because the trial and control groups are likely to have the same range of definitions on average. In fact the treatment group is more likely to report illness than the control group.

The evaluation showed that the programme had a big effect on child health. The number of reported illnesses decreased by about 25% with the reduction increasing the longer the programme went on.

The other two measures used to establish health impacts were anaemia and height. Anaemia decreased by about 13% after 12-18 months and there was also an increase in height – a good measure of cumulative health – for children under three. There was no significant impact on stunting.

In 2003 the evaluation looked at what impact the programme had on school readiness. One of the hopes was that improved nutrition would lead to better cognitive development and hence improved school readiness. As the control group had been incorporated in the programme in 2000, communities that were just too wealthy to qualify for the programme were then used as a control group. The results showed that although there were continuing improvements in health, physical and motor skill development for children in the programme, there was almost no improvement in cognitive development. This was seen to be linked to the limited intellectual stimulation in the children’s communities suggesting that early childhood development activities might be needed to improve cognitive abilities. There was an improvement in school enrolment and in transition from primary to secondary school.

A major question is whether people would have done as well if they were just given cash. The suspicion is no, because of the differential take up.

The other really important thing about PROGRESA was that rigorous evaluation was feasible and not expensive in terms of the overall size of the programme. The new government took ownership of the programme and added important innovations of their own. The programme has had a big influence on policy making with a number of other countries adopting similar programmes. In this respect it was important not only to document the results but also how the programme was implemented.

Dr Gertler’s powerpoint presentation can be accessed at:

Comments by Dr Ingrid Woolard

The presentation underlines the importance of programmes that try to break the intergenerational transmission of poverty and also the importance of rigorous evaluation of the implementation of policies.

Some of the key aspects of PROGRESA are that:
  • It raises the question of whether it is best to give people cash or provide benefits, something that still needs to be resolved in South Africa;

  • It tries to address key elements of health, nutrition and education;

  • It puts women at the centre of the programme;

  • A decentralised and non-partisan agency runs the programme;

  • It has high level political support, and

  • From the start there was agreement on the need for ongoing programme evaluation.
At present 12% of non-interest spending in South Africa goes on social assistance (about 3% of GDP) with 8.9 million people receiving cash transfers. About 5.2 million of these are child support grants (CSGs), which are directly relevant to this discussion. The intention of the CSG is to contribute to the cost of raising children in very poor households. It is a means tested, unconditional cash transfer. The original recommendation was to make the grant conditional on registration of birth and meeting certain health related activities. This was not done and the grant was implemented without any conditions. The CSG is paid to the primary caregiver, in 99% of cases a woman. When it was introduced it was R100 and was limited to children under 6, it is now R170 and is being extended to children under 14. There are however two puzzling aspects about the CSG: the means test for the CSG has remained the same since its inception while the test differs in urban and rural areas. However, there is no official definition of an urban or rural area. In practice it is easier to get the grant in informal settlements and rural areas.

The CSG is intended for the child but for 18% of the households receiving the grant it is the only form of cash income and in poor households it will inevitably be used for other purposes. There has been very little evaluation of the grant to look at ways of improving it. A case study showed that it had improved the ability to take care of children but the study was primed to get a positive response. Data from Hlabisa shows that the grant is going to most of those who should be getting it and that it is working on an operational level. However, it was too soon to tell whether it had had an impact on nutrition and other health outcomes.

In 2004 the Department of Social Welfare undertook a study on the child support grant in KwaZulu-Natal as part of a large-scale ongoing panel study. This has provided anthropometric data and household characteristics and work is being done on constructing a control group. The question is who to use, as children who are not enrolled do not qualify. In the meantime the study is using children who have applied but are not yet receiving the grant as a control group.

Preliminary work on height suggests that the child support grant is not having an impact except for those children who started to receive it when very young. There seems to be a window of intervention between 0-20 months, which yields the biggest impact. This suggests that the CSG matters most for very young children but most families do no rush to enrol very young children for reasons that are not clear.

Introductory comments by the Ambassador of Mexico to South Africa

The presentations have provided good descriptions of the assessment of PROGRESA and of what is happening with the South African programme. The background to PROGRESA is important. In the mid 1990s Mexico had a programme called Solidaridad. It was intended to provide support for families to build schools, hospitals, roads and other infrastructure in their communities. It was an important programme but it showed that such programmes have high costs and without clear definition many of those who benefited were not the intended beneficiaries.

It was also clear at election times that the resources for the programme were often being used to get political support. An evaluation of the programme indicated that it should be abolished and replaced with a radically different programme with far greater emphasis on who should benefit. This led to PROGRESA. It is important that the focus is not only on health but also on education, especially on getting girls into school and on empowering women.

An important element has been the readiness of new government to take on the programme and to add new elements to it. Recently this has included extending the number of years of schooling that get support, scholarships and loans for university study and also support for entrepreneurship to encourage micro business in the villages.

Continuity has been fundamental. Other important elements have included keeping the cost of administration as low as possible, good management and external evaluation. We need to consider whether South Africa and other African countries can take something from this experience.

Questions and comments: responses by Paul Gertler

  • Did recipients use the extra cash to purchase food?
There was a detailed analysis of how the money was spent that looked at what households were spending before the cash transfers in both the treatment and control areas and then compared the change in households that received the cash transfer to the change in households that did not receive it. About 60 to 70 % of the money was spent on food. It was spent not only on more calories but also on more proteins.

A small portion of the money – about 15% and this increased over time – was invested. On farms they bought chickens and pigs, cows, tools and fertiliser. These were long term investments that would yield long term benefits, so if the cash transfers were taken away people would be better off than before. In urban areas people are investing in micro enterprises like small cafeterias and stores. This was an interesting side effect of lifting the previous constraint on investing that was unrelated to any conditionality.

  • The cash given in the programme was about a third of household income. How do you calculate the correct amount? Is there a critical mass that makes cash transfer programmes effective?
The Mexicans decided on the value of the grant based on costing what persons needed to spend to improve their nutritional status. On the school side they looked at the costs of schooling such as uniforms and transport and also at possible loss of income that the children could have earned for the household if they were not in school. They came up with a sliding scale that increased with age and also made it slightly higher for girls on the basis that there was discrimination against girls, which turned out not to be true.

Was that the right way to do it? If the cash transfer was just to get children to school this could probably have been achieved with a lot less money. But the money was also used to buy food and invest in income generating activities. In fact those who invested most were those with the most children in school because they got more in transfers.

On effectiveness: the problem with this kind of evaluation in long term programmes is that, for moral and ethical reasons, you cannot vary the benefits given to different families and study the effects. This is the only way to rigorously determine whether you could use more or less money is to do this. We have to accept this trade off.

  • Have there been any comparisons between this kind of programme and food-based programmes?
On the food versus cash question Mexico had a small pilot to look at this and found that food did not have any benefits over cash and was much more expensive to distribute. Cash can be distributed through banks.

  • What was the extent of coordination with government departments to provide supporting services for the conditionalities? One of the difficulties in South Africa is providing the support services that go with grants from the Department of Social Development.
The coordination of government departments worked in Mexico because President Zedilo, who came to power following the assassination of the president, was an academic who wanted to do something different. He appointed a confidante to run PROGRESA but the head also reported directly to him. He also appointed another confidante to the oversight committee. This person was the number two in the treasury and took on the task of driving the budget for PROGRESA. As a result they had a lot of power over funding and to coordinate things. Despite this it was still very difficult to make sure that schools were there and to get the nutrition supplements out. It took a year or more to scale up. It really took a huge effort to get it going.

  • Was corruption an issue and how was it dealt with? It is a big issue in South Africa in connection with grants.
Monetary compliance turned out to be quite easy in Mexico. To bypass corruption the cash was distributed from the central agency, which paid much higher salaries than other government agencies so there was less incentive to pocket money. The money went directly to banks or to PROGRESA offices in the field with very little leakage of funds. They also put up public listings of how much money people should get so people were able to demand what they were entitled to.

Monitoring was done directly by PROGRESA officials who only denied benefits to about 1% of recipients.

  • What are the fiscal implications of scaling up such a programme? Incentive based programmes have not been popular in Africa because finance ministers argue that we do not have the resources to sustain these kinds of programmes beyond a pilot programme.
In a country like Mexico you are looking at targeting benefits at the bottom 10-12% of the population so the fiscal implications are not too bad. In a place like Honduras where you are looking at 80% of the population benefiting, providing schools and health care facilities may be a higher priority than cash transfers. Honduras does have a cash transfer programme, which gives four percent of household income rather than a third. Nicaragua gave 12%.

  • To what extent is strong fiscal decentralisation important for scaling up a programme like this? This is related to the issue of decentralisation as a whole in an African context where public accountability systems are often weak.

  • Beyond political sustainability, which is an issue in Africa where many countries have a rapid turnover of governments, there is an issue of institutionalising these programmes where many countries lack capacity.
On decentralisation, Gertler noted that this was a “big topic” that would need another session to examine.

In her closing comments Sue Mbaya, SARPN’s Director, noted that in Africa supply side issues were very significant to the debate about social protection. She noted that SARPN would continue to engage with the topic of cash transfer issues.

Octoplus Information Solutions Top of page | Home | Contact SARPN | Disclaimer