From Oliver Twist to Ethiopia’s PSNP: How did workfare become so productive?

JustKidding184x94The Just KIDDing blog is Dr. Stephen Kidd’s take on key issues in social policy in international development.

“Productive” safety nets seem to be all the rage, with Ethiopia’s Productive Safety Net (PSNP) spawning a crop of “wannabes” – such as the PSNPs of Mozambique and Tanzania – and significant enthusiasm among donor agencies. In fact, so popular are “productive” safety nets that low-income countries have been persuaded to take on significant loans to finance them. After all, they’re productive, so they must be a good investment.

Yet, on closer examination “productive” safety nets seem to bear a remarkable resemblance to traditional workfare: to receive cash benefits, people have to undertake hours of hard labour, often under a hot and unforgiving sun. As the World Bank has pointed out – see Subbarao et al (2013) – workfare commenced with the workhouse of Victorian England and has continued as a favoured social protection instrument of neoliberals and those who fear that unconditional transfers create dependency. Workfare ensures that the poor don’t receive “handouts:” instead, they’re made to work for their bread, just like Oliver Twist

Yet, the term productive safety net suggests a very different type of instrument to traditional workfare, which is usually regarded as a rather demeaning form of social security. So, is this contemporary form of workfare really that different and is it as productive as its advocates would lead us to believe? Just to be boring, why don’t we look at the evidence? 

There have been two recent evaluations undertaken of Ethiopia’s PSNP, one by Berhane et al (2011) and another by the Young Lives Project (Tahere and Woldehanna 2012). Unfortunately, the evidence from these evaluations does not support the notion of the PSNP as productive: in fact, it’s rather more Oliver Twist than budding entrepreneur.

One of the most striking findings from Berhane et al (2011) is that, among participants on the PSNP, “the proportion of primary food sources contributed from own production declines.” So, participants on the PSNP actually become less productive and, indeed, more dependent on the PSNP transfer. In fact, the programme seems to be achieving the opposite of its aim of “graduating” participants out of poverty.

So, does the Young Lives study tell a different story?

Unfortunately, Tahere and Woldehanna (2012) appear to confirm the findings of Berhane et al (2011). Indeed, the picture they paint is even more alarming. They found that participation in the PSNP’s public works had a surprisingly negative impact on both food and non-food consumption: “On average, participation of households in the [public works] programme reduces per capita consumption expenditure per month by 33 birr (US$1.86), per capita food consumption per month by 21 birr (US$1.18) and per capita non-food expenditure per month by 12 birr (US$0.67).” They also found that the PSNP is not building the assets of participants, one of the key features of a “productive” social security scheme.

Tahere and Woldehanna (2012) argue that their “results seem to support the general public opinion that the PSNP is making some people dependent on aid and is not lifting households permanently out of poverty.” They note that: “People who were not included in the PSNP worked hard to increase the amount of payment they obtained from off-farm employment (wage labour and non-farm business), while many PSNP beneficiaries waited for low-paying public work, which they saw as less risky.”

However, Tahere and Woldehanna went further: they found that, not only are households consuming less, children are suffering. Participation by households in PSNP’s public works has led to children spending more time on household chores and on paid and unpaid labour outside the home. In effect, when their parents are engaged in public works, they substitute for them. Furthermore, even though it is against the programme’s regulations, many children also miss school to work on the PSNP. Others combine working on the PSNP and going to school but they end up tired, which must affect their performance.

Similar concerns with “productive safety nets” – or workfare – have been found elsewhere. Manley et al (2012) undertook a systematic review of the impacts of cash transfers on child nutrition. While unconditional cash transfers have a positive impact, workfare appears to increase child undernutrition. Another study by Helen Keller International (undated) examined Bangladesh’s Chars Livelihoods Programme and found that the participation of women in workfare led to children having poorer nutrition than when men were involved, presumably because the children were denied essential care when their mothers were away. Furthermore, the women themselves were more likely to be thin when compared to women in households with only men engaged in the public works.

Of course, one of the problems with workfare is that it is a very inefficient form of social protection. Participants in workfare face significant opportunity costs since they have to give up other income earning opportunities. In Bangladesh, for example, Ahmed et al (2007) found consumption increasing by only 30 cents for every dollar earned. Generally, opportunity costs range between 30% and 70%. This contrasts dramatically with unconditional transfers which are often associated with a multiplier effect: families use their income to invest in their own income generating activities, rather than being pulled away to work for others.


So, are we seeing the creation of a new myth of the “productive safety net,” similar to those we’ve discussed in our recent papers on BOLSA UnFAMILIAr and the Seven Deadly Myths of Social Protection. Or, perhaps we are witnessing a clever piece of marketing, transforming workfare into a “must have” social protection instrument for developing countries, which leads to the next “must have,” a soft loan.

Yet, it is evident that workfare is a much less productive instrument than well-designed unconditional transfers. Instead of obliging people to invest their energy in public works – from which they often derive no benefits – recipients of unconditional transfers can invest in their own income generating activities, generating higher – not lower – consumption. And, importantly – and in significant contrast to workfare – well-designed unconditional transfers bring positive benefits for children, improving nutrition and school attendance.

So, let’s have less Oliver Twist and more investment in social security schemes that really do enable families to be more productive.

(By the way, if anyone wants a very quick overview of workfare schemes, I’d suggest reading McCord 2005).


Berhane, G., J. Hoddinott, N. Kumar, A. Taffesse, M. Diiressie, Y. Yohannes et al. (2011)Evaluation of Ethiopia’s Food Security Program: Documing Progress in the Implementation of the Productive Safety Nets Programme and the Household Asset Building Programme.  International Food Policy Research Institute, Institute of Development Studies, University of Sussex.

McCord, A. (2005) Win-win or Lose? An Examination of the Use of Public Works as a Social Protection Instrument in Situations of Chronic Poverty. Paper presented at the conference on Social Protection for Chronic Poverty, University of Manchester, 23-24 February 2005.

Subbarao, K, C. Del Ninno, C. Andrews and C. Rodriguez- Ales (2013) Public Works as a Safety Net: Design, Evidence and Implementation. Washington DC, World Bank.

Tafere, Y., and E. Woldehanna (2012) Beyond Food Security: Transforming the Productive Safety Net Programme in Ethiopia for the Well-being of Children.  Working Paper No. 83. Young Lives, Oxford Department of International Development, University of Oxford: UK, Oxford.

12 Responses to “From Oliver Twist to Ethiopia’s PSNP: How did workfare become so productive?”

  1. William Wiseman Reply

    Hi there. Interesting piece. Can you clarfiy if the 2011 Berhane evaluation is of the PSNP or of the Social Cash Transfers Pilot Programme, Tigray Region. As the citation would suggest. This is a different program from the PSNP and aims to \” It aims to support the most vulnerable, specifically targeting labour-constrained and ultra-poor households. The programme seeks to improve the quality of life of orphans and vulnerable children (OVC), elderly people and people with disabilities (PWD) by providing a regular cash transfers and linking those with the provision of social welfare services.\”…. I think it is important to clarify. Although it is possible that the evaluation of the Social Cash Transfer Program in Tigray looked at the PSNP, it seems strange not to use data from the large-scale multi-year panel survey of the PSNP itself, also undertaken by IFPRI. Perhaps that survey doesn\’t support the argument as cleanly or maybe there was another reason.

  2. naomi hossain Reply

    Very interesting, very Victorian – the deserving poor never really disappear, do they? I suppose we should expect more emphasis on workfare social protection in development, in these days of European austerity and workfare.

  3. Stephen Kidd Reply

    Dear Will: thanks very much for pointing out the error. I’ve corrected the citation. As you surmise, I was referring to the IFPRI survey on PSNP.



  4. John Hoddinott Reply


    I am one of the co-author\’s of one of the studies cited in this blog (Berhane et al 2011). I am interested in writing a blog post in response to this. Is that possible?

    John Hoddinott

  5. Stephen Kidd Reply

    Dear John: we\’d be delighted to receive a blog. One of our aims is to promote debate. Best, Stephen

  6. Malcolm Marks Reply


    Some big issues with the HKI report on CLP cash for work you quote. For reference the survey work was carried out in late 2006 and the initial report – that you appear to be quoting – made statements that the data did not support; especially since a proper control group was not used. This led to the CLP contracting HKI to repeat the study during monga 2007 but with an appropriate control group, the results were reversed and the benefit of the CFW programme clearly shown by HKI for women in the workforce and their children.

    I would be happy to send you the later report …

  7. Malcolm Marks Reply


    Some big issues with the HKI report on CLP cash for work you quote. For reference the survey work was carried out in late 2006 and the initial report – that you appear to be quoting – made statements that the data did not support; especially since a proper control group was not used. This led to the CLP contracting HKI to repeat the study during monga 2007 but with an appropriate control group, the results were reversed and the benefit of the CFW programme clearly shown by HKI for women in the workforce and their children.

    I would be happy to send you the later report …



    I find your latest investigation of the PSNP interesting and relevant to Uganda. I am sure you have been following the development of the Draft National Social Protection Policy Framework. Public Works is one of the strategies identified for reducing poverty among the rural population. What recommendations would you give to ensure that the envisaged promotion of public works as a social protection net does not turn out to be the\’ Oliver Twist model\’ ?

  9. Stephen Kidd Reply

    Dear Malcolm,

    Thanks for this. I’ve put aside my natural scepticism about development partners repeating analysis when the first results don’t give the desired result – we’ve published on that elsewhere – and found a paper by Mascie-Taylor et al (2010) which, presumably, has the results from the second survey. Correct me if I’m wrong, but it seems that the study was undertaken just at the end of the public works programme when the difference between the intervention and control groups would have been at their maximum. However, it would probably have been better to have undertaken the study perhaps 6 or 12 months later to see if there were any permanent gains. For example, we don’t know what the control group was doing during the short period of the public works and any investments they may have made may have paid off in the longer term, meaning that they could have been doing better than the intervention group some months later. The latter may have had to give up investment opportunities while doing the public works, which may have meant that later on, their consumption may have fallen compared to the control group. We just don’t know. But, this seems to have been the problem in the PSNP: households were made to work on public works when they should have been working on their own agricultural plots, which reduced their subsequent returns from their own plots.

    Of course, each situation is different. In reality, I suspect that the CLP public works was an infrastructure programme rather than a workfare programme, as the main aim was to build plinths. Furthermore, if the objective had been social protection rather than infrastructure, we have no evidence from the study on the comparison between an unconditional transfer and public works (although an unconditional transfer would almost certainly had a more favourable outcome as recipients would have been able to invest in other activities, without the opportunity cost associated with a workfare scheme).

    Finally, it seems that – in the second study – there was no difference found in the children between households in which women worked and those in which they didn’t (but the men did). This contrasts significantly with the first study and the difference has nothing to do with the absence of a control group. Why were there such different results?



  10. Stephen Kidd Reply

    Dear Harriet,

    It’s great that Uganda is developing a National SP Strategy and, in this context, the country needs to make some important decisions on priorities. DFID is funding a pilot Old Age Pension in Uganda which appears to be having some very positive results, not only for older people but also for children. So, the question for Ugandan policy makers is why would they promote workfare – which may well have negative impacts and is not a cost-effective social protection instrument – when the Old Age Pension has not yet been scaled up nationally? Indeed, a much better option than workfare for working age families in Uganda would be to implement a child grant (please see the paper on Graduation that I’ve just published).

    You do need to look at the drivers behind workfare. If there is a push for Uganda to take a soft loan for workfare – as has happened in Mozambique and Tanzania – you need to be concerned about the motives of those promoting workfare (and, indeed, their understanding of social protection).

    However, we do need to differentiate between public works that are infrastructure programmes and workfare. The former may be important for building and maintaining key national infrastructure – such as roads or terracing – and, rather than using machines, it is preferable to employ people, especially when there is high youth unemployment or underemployment (we call such programmes “labour intensive public works.” But, such programmes need to be part of a national infrastructure strategy rather than a national SP strategy. If it is workfare that is being promoted – where social protection is the main aim – then this is unlikely to be in the best interests of low income families in Uganda and is not a cost-effective means of providing social protection. An unconditional transfer will almost certainly have much larger impacts and is more likely to enable families to engage in the labour market.



  11. Malcolm Marks Reply

    Hi Stephen,

    Lots of interesting points raised in your response and many of the points need more than a sentence or two to respond. The original HKI study (that you quote) was meant to have taken before and after anthropometric (and many additional dietary and socioeconomic) measurements from a treatment (working) and a control group. However, they simply reported on end measurements of a group that worked and a group that did not – and tried to make the negative impact hypothesis work in that manner. The comparison is highly dubious and failed to deliver the complete study as requested. Hence the insistence that the work be repeated in the following year. The reality was more complex than I report here, and I would be happy to give you the full details (perhaps we could Skype?)

    Yes, this was an infrastructure programme (plinth raising) and timed to inject cash into extreme poor communities at times when little or no work (termed monga) is available. Also, any calculation of opportunity costs is clouded, rather negated, by the fact that many workers got their household plinths raised for “free”.

    There are a couple of other related papers on the CLP website that you might find interesting: the impact of providing a cash grant to families without a person physically able to work on plinth raising (written by me), the impact of CFW on reducing the severity of coping strategies utilised during monga (Kate Conroy), the value of plinths during floods (me), and a CBA of plinths (Philip White, to be finalised).

    Skype: drmkmarks

  12. Anna McCord Reply

    Thanks for this fascinating blog and discussion.

    I think the reality of PWP impact, and their potential role within a social protection system is highly situation specific – depending on i) the nature of poverty and ii) the sensitivity of programme design to the context.

    There certainly is evidence of i) kids taking on domestic work in place of parents working in PWPs, ii) parents neglecting child care in favour of PWP participation, iii) malnourished workers struggling to meet conventional work norms adopted under PWP, such as cubic meters of earth moved per day, and iv) of the opportunity cost of participation (both in terms of forgone revenue generating and subsistence activities) reducing the net income gain to perhaps half of the gross PWP wage transfer (see discussion and examples in McCord, 2013).

    And most evaluations focus heavily on numbers of people employed, days of employment provided, and income paid out, and from this assuming a set of social protection benefits at household level, Similarly an appraisal of the ‘productive impact’ is often similarly overlooked, and based on assumed benefits.

    The upshot is, you can have a programme which is technically a success, (based on evaluation against process rather than outcome indicators, which is typical of many IFI designed and supported PWP (see IEG Social Safety Nets Evaluation, 2011)) but may confer little or no social protection benefit when viewed from a household perspective, and have potentially limited ‘productivity’ impacts. It really depends on wage level, terms of employment, duration of employment relative to needs, the broader economic climate (which will influence the likelihood of graduation) etc, and critically also the implementation of complementary interventions, eg agricultural extension, micro-credit etc. These would be the things to consider in relation to the proposed Uganda programme.

    The excellent study by Berhane, Hoddinott, Kumar and Taffesse, ‘The impact of Ethiopia’s Productive Safety Nets and Household Asset Building Programme: 2006-2010’ (IFPRI, 2011) really teases out how the combination of i) repeated support and ii) complementary agricultural extension support has been effective in promoting productivity and household asset accumulation and offers significant food for thought in terms of revision to conventional PWP design.

    The key consideration is that if the PWP is intended to provide social protection for the working age poor as one component of a wider social protection system, then provision needs to be consistent with needs in terms of providing access to support a significant proportion of those in need on an ongoing basis– most programmes, particularly in SSA offer only one off or short term employment to a small section of the working age poor – this is not consistent with meaningful social protection provision, but is often confused with social protection provision by programme designers and funders. Short term labour intensive construction programmes do not offer social protection provision for the working age poor, in contexts of chronic impoverishment.

    The VUP in Rwanda, and the PSNP in Ethiopia are particularly interesting as they attempt to provide some form of ongoing provision on a large scale, and offer complementary inputs to stimulate productivity – although as Stephen mentions, the costs of providing social protection in this way are significantly higher than through cash transfers, so the value of the output (the quality of assets created) becomes particularly relevant here.

Leave a Reply

Your email address will not be published. Required fields are marked *