It is frequently claimed that the most innovative feature of social protection, in contrast to safety nets, is that it has the potential to reduce the vulnerability of poor people to the extent that they can manage moderate risk without external support. This has led to an expansion of large-scale ‘productive safety net’ programmes. The potential to reduce vulnerability so that people can move off social protection provision is popularly termed ‘graduation’. However, the vision for graduation rests on the assumption of the existence of a large population of low-productivity, risk-prone and often poor households. Under this scenario, if risk can be underwritten through appropriate social protection then significant numbers of poor people have the potential to move out of vulnerability and extreme poverty into more productive and resilient livelihoods.
These claims for the livelihood ‘promoting’ functions of social protection and the graduation agenda, lead an enquiring mind to ask, what is the theory of change behind this graduation vision and under what conditions will this theory of change lead to improved outcomes for the poor? At the centre of this vision there typically exists a focus on resource accumulation and house-hold level assets. Surprisingly, we see numerous programmes built on this concept, yet little analysis or understanding of the requirements and constraints to building sustainable pathways to graduation.
The ambition of this paper is to map out the theory of change underpinning the notion of graduation and to set out, conceptually and empirically, the range of enabling and constraining factors that facilitate or undermine this change process. We distinguish ‘threshold’ graduation (an administrative benchmark that signals the point at which a beneficiary is no longer eligible for the programme) from ‘sustainable’ graduation (a state in which livelihoods have been fundamentally transformed through social protection interventions). Furthermore, in keeping with recent work by the authors and others (Dorward et al, 2006; Sabates-Wheeler, Devereux and Guenther, 2008), we emphasise the existence of multiple factors working at different levels, beyond the household, such as market conditions, community investment and scale effects that work to constrain or complement each other, and asset usage.
The notion of development coordination (Dorward and Kydd, 2004) requires that these different levels and initiatives are not analysed in isolation from each other. Consider an asset transfer targeted at poor farmers. This package may bring them above a specified asset threshold, but local markets may be so thin and imperfect that any productivity gains are not translated into higher incomes because of adverse scale effects. In other words, without development coordination (or some way of enabling sustainable graduation) the tendency is towards a stable low-productivity equilibrium only. A related point is the scale of the programme – the size of the livelihood package and of the target group. Even if local markets function well and are able to absorb increases in production, if the livelihood package does not bring enough households above a critical threshold there will be negligible multiplier effects and farmers may be unable to take advantage of potential economies of scale (see Dercon, 2004). These considerations are central to this paper and to a full understanding of the process of graduation.
This paper is structured as follows. In the following section we reflect on the genesis of graduation, in particular the theories of asset accumulation and asset thresholds which provide the basis of support for graduation. After providing some concrete and current examples of large-scale programmes built on this concept, we derive a ‘theory of change’ that is common to these programmes. In the latter part of the paper we critique this theory of change by drawing on a variety of literatures and programme experience that enables us to lay out a range of conditions under which graduation is more or less likely to occur. We conclude by highlighting some basic considerations that need to be taken into account in future social protection programming.
