The ‘Do Public Works Programmes Work? Design and implementation features for programme success’ webinar took place on 17 January 2019. The event discussed new research on the effectiveness of public works programmes (PWPs), and on the design and implementation features that lead to programme success or failure.
Watch the video recording and access the slide presentation here, or read on for a summary.
The webinar presented the findings of two recent studies: “Do Public Works Programmes Work? A systematic review of the evidence from programmes in low and lower-middle income countries in Africa and the MENA region” and “Design and Implementation Features of Public Works Programmes: What do we know about the drivers of success and failure to achieve programme objectives?”, both of which are the result of a collaboration between Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH on behalf of the German Federal Ministry for Economic Cooperation and Development (BMZ) and the University of Passau, published on behalf of the German Ministry of Economic Cooperation and Development (BMZ).
The webinar, which was also organised by GIZ and the University of Passau, was moderated by Ralf Radermacher (Head of Social Protection Sector Initiative, GIZ) and speakers, Michael Grimm (Professor of Development Economics and Dean, University of Passau), Stefan Beierl (Research Assistant at the Chair of Development Economics, University of Passau), and discussant Rajeev Ahal (Project Director of Natural Resource Management in India, GIZ).
A review of PWPs in low and lower-middle income countries in Africa and the MENA region
PWPs are social protection mechanisms based upon the dual objective of providing temporary employment to vulnerable populations, while also conducting labour-intensive infrastructural projects, such as rural roads or the maintenance of public spaces. These programmes aim to address unemployment, providing income support to poor people, and demand for infrastructure development simultaneously.
PWPs can be divided into two main types, which can be tailored to different circumstances by involving additional measures:
The main impacts of PWPs can be divided into three areas:
1. Wage vector: Participants earn a wage, which increases their income resources and financial stability;
2. Asset vector: The infrastructure developed by the programme;
3. Skills vector: Beneficiaries may acquire new skills that can facilitate their post-programme integration into the labour market.
Although PWPs differ greatly, they are all defined by the same underlying programme theory, related to outcomes tied to longer-term outcomes, such as poverty reduction and sustainable development, as demonstrated by the table below:
Both publications presented focused on the systematic reviews of the experiences of PWP’s in Ethiopia, Côte d’Ivoire, Ghana, Malawi, Sierra Leone, Rwanda and Yemen. Most of the evidence of the outcome areas came from Ethiopia, due to its more established programmes. The research presents a review of studies that evaluated PWPs in these country cases according to their impact in relation to specific outcomes:
The Figure below provides an overview of the number of studies for each outcome area and the number of measured significant positive and negative impacts. Given the dominance of studies that look into the Ethiopian case, the findings are separately presented for Ethiopia and the group of all other countries.
· Income, consumption & expenditure
Michael Grimm summarised to what extent PWPs have affected outcomes related to income, consumption and expenditures. Although overall the picture is inconclusive, i.e. in some case PWPs have positively affected these outcomes and in others not, there is some evidence for substantially increased earnings, higher spending on food (Type 1 and 2) and in the case of Côte d’Ivoire (Type 1) also for increased productivity among self-employed workers.
· Food consumption and food security:
Michael Grimm demonstrated how PWPs can be efficient in improving food security, especially in the case of Ethiopia’s Type 2 Plus Productive Safety Net Programme (PSNP). Cases outside of Ethiopia mainly consist of Type 1 programmes, and the evidence surrounding the cases of Ghana, Rwanda and Sierra Leone are inconclusive, while there were no observed impacts in the cases of Yemen and Malawi.
· Agricultural technology:
Results were inconclusive, however, when evaluating the outcomes of expenditure on farm equipment, the use of technology and the quantity of fertiliser used, it was found that Type 2 Plus programmes outperform those of a simple Type 2 tier. The additional measures of training in fertiliser usage and giving assistance in stone terracing and fencing proved to increase the impact of PWPs.
The findings in most outcome areas are quite heterogeneous. Hence, the assumed benefits of PWPs can by no means be taken for granted even with respect to the limited objective of enabling consumption smoothing. In fact, in all outcome areas expected to be positively influenced by PWPs, there are several studies that do not support this expectation. At the same time, in all these outcome areas there are also studies that detect positive trends. This strongly suggests that PWPs can work, depending on how they are designed, how they are implemented and on the specific context.
The drivers of success and failure to achieve PWP objectives
Stefan Beierl explained how there are many crucial design decisions to be made when establishing a PWP related to the wage vector (benefits, periodicity of work, benefits delivery, etc.). The ultimate choices will vary by context, but it is always advisable to decide based on an analysis conducted in terms of the real total transfer value that addresses the main trade-offs, which are often left implicit when taking these decisions. The real transfer value refers to the sum of the following aspects involved in establishing a PWP:
1. Duration and timing
Duration and predictability play a key role in determining the scope of impacts that can be achieved through a PWP. Duration in this regard concerns the following aspects:
· Number of working days beneficiaries can work in a PWP each year
· Number of months/years individual beneficiaries remain on the programme
Longer seems better in both respects but the existing evaluations of PWP in Africa do not allow for disentangling the income effect (i.e., the longer time spent participating in the programme means that people earn more) from the insurance effect (i.e., predictability regarding access to extra income for a considerable period of time).
However, evidence relating to MGNREGA, the large-scale employment guarantee in India, suggests that the income insurance that the programme provides (by offering access on demand for up to 100 days a year) encourages entrepreneurial risk taking. While there are no employment guarantees of that kind in Africa, Type 2 PWP may have a similar effect.
Regarding the timing of the work activities in PWPs, the evidence suggests that the bulk of activities should be scheduled when other employment opportunities are rare, such as outside the planting and harvest season. However, there may still be room for some time-critical PW activities during peak labour demand periods, e.g., for environmental activities.
Traditionally, PWPs rely on self-targeting. By deliberately setting a very low wage, this is meant to ensure that only the poorest of the poor are willing to work. However, this approach is not suitable for poverty-reduction due to the low real total transfer value, serving rather as a temporary poverty alleviation mechanism. If the poorest are to be reached and tangibly benefit from participation, a more sophisticated targeting mechanism should be used.
Approaching the asset vector of PWPs, Beierl presented different approaches in defining the choice of public work activities:
There are several factors determining asset quality and relevance, namely:
· Usage of quality materials;
· A labour intensity that does not undermine asset quality;
· Availability of adequate technical expertise, management capacity and oversight;
· Involvement of communities in a well-facilitated selection process;
· Maintenance arrangements that are explicitly agreed upon and then financed and implemented using the subsidiarity principle;
· Embedding PWP activities not just in a PWP-specific work plan but in a general local development plan.
There is skepticism in the research concerning the skills acquired by programme beneficiaries, mainly regarding the inclusion of complementary skill transfer components as part of PWPs. Even if an increase in knowledge is observed amongst participants, there is little evidence that it translates into improved employability or an increase in income. On the other hand, some positive examples can be observed:
· Knowledge gains through on-the-job training closely linked to the PW activities can also be applied in beneficiaries’ daily lives;
· There may be some knowledge spillover to non-PW beneficiaries.
Therefore, on-the-job training that promotes skills that are closely linked to the work activities of the programme can be an effective extra component of PWPs. However, caution is warranted regarding ambitious efforts to transfer less related skills.
The webinar closed with a return to the definition of the types of PWPs, and their efficiency considering the evaluations conducted:
· Type 1: Only suitable in contexts of acute poverty and to achieve basic objectives (e.g., consumption smoothing), but even this is not guaranteed, especially if transfer values are inadequate;
· Type 2: Wider literature suggests that even in contexts of wide-spread chronic poverty, Type 2 may facilitate the accumulation of enough savings to withstand moderate shocks and marginally boost post-PWP income. This outperforms Type 1 in improving food security but there is no strong evidence that it also outperforms it in facilitating asset accumulation and, thus, in putting households on an upward trajectory.
· Type 2 Plus: This is the most promising in terms of impacts but also the most expensive. It does well for asset accumulation and agricultural technology adoption but displays no strong indications of an increase in income or agricultural output in the medium-term.