Daniele Malerba, PhD Researcher, Global Development Institute 
The 2016 DSA conference, themed around politics in development, involved an incredible number of interesting panels (and more than 600 delegates!). I was very pleased to present in the one focusing on “Poverty dynamics: shame, blame and responsibility”, which also had a strong focus on multidimensional poverty. The panel, running almost all day on Tuesday (chaired brilliantly by Keetie Roelen from IDS), was divided into three sessions, each covering a different sub-topic: discourse, measurement and social protection.
I was very pleased to present on this panel for many reasons. One of them is that the conference was held in Oxford, the hometown of the Multidimensional Poverty Index (MPI, developed by OPHI), giving a romantic element to the event – yes, researchers also have feelings… But more important was the relevance of the topics discussed. Rising inequality, especially in high-income countries, is increasing vulnerability to poverty. As the abstracts of the papers are available in the conference program I will not go into the details of the presentations but want to reflect on the main points discussed and what needs to be done next.
Current issues: definitions of poverty, long-term objectives and responsibility
Throughout all the presentations (and in other panels I attended) three extremely linked themes caught my attention as fundamental to the theme of the panel. The first is the definition and measurement of poverty. Do we conceptualize poverty as monetary or as multidimensional? If we consider the latter, what are the dimensions to include? This is not a new issue. But it is of increasing importance now with the widespread use of direct antipoverty policies in developing countries. Some of them (such as conditional cash transfers, CCT, or integrated poverty reduction programmes like the Chile Solidario) focus on the accumulation of human and physical capital. These programs align closer to the multidimensional definition of poverty and have a longer term vision. On the other hand, unconditional cash transfers address poverty more as a lack of income. It is clear that the conceptualisation of poverty drives policies.
This distinction is a preamble for the second point of interest. Decreasing long term (vulnerability to and) poverty does not imply just temporarily filling temporary income gaps alone. There has to also be a focus on transformational and productive aspects which can create benefits and payoffs also in the long run. Examples include increased education, which brings higher labour market outcomes and earnings, or investment in agricultural machinery which leads to higher output – both typically seen as components of CCT or integrated programmes. This does not mean that unconditional cash transfers cannot achieve these goals as well: for example, higher lump sum transfers are associated with more physical capital accumulation. Any poverty alleviating policy should take these long term goals into account. It is worth noting that some of these transformative dimensions (such as learning quality) are difficult to measure or the data is not available. This is the main reason for their exclusion from indicators such as the MPI.
Thirdly, the responsibility cannot be given to a single person or a single actor. If transformative effects of antipoverty policies need to be achieved, all agents need to play a role. If a CCT programme has the effect of increasing school enrolment, but the school infrastructure is poor and the teachers are absent or of poor quality, how can children improve their education? The objective of CCTs and other human development antipoverty policies is human capital accumulation so final outcomes (learning, health status) should be the ones that matter the most, compared to intermediate goals (school enrolment or visits to health clinics). Responsibilities should be then shared among the agents responsible for the supply-side services (such as infrastructure) as well as the recipients.
What comes next: better data, social and geographical poverty traps, and the need for complementary interventions in antipoverty policies
Given these core issues, what can and should be done? Firstly, improve data quality and its collection frequency, to improve poverty estimations and to include additional relevant dimensions (such as test scores or individual learning outcomes).
Secondly, the goals of antipoverty programs need to take into account the geographical and social poverty traps where many poor people are located. This is important not just because the normalisation of poverty is a big danger for many social groups and geographical areas, but also because the concentration of the poorest in disadvantaged clusters entails more structural policies to address vulnerability and poverty. This is important for development as a whole, as we see that the majority of the global poor now live in specific pockets in middle income countries – and not in low income countries. And despite growth and available resources at the national level, they still remain poor and are excluded from the development process.
Finally, antipoverty programs, especially the ones looking to have a transformational effect through, for example, investments in human capital, need to be complemented by supply side programs through coherent policy architecture, such as ensuring that school enrolment is complemented by good quality teaching. A positive example from Brazil comes from “Brasil Sem Miseria”, a programme launched in 2011 which takes a holistic approach to poverty eradication. This programme complements pre-existing direct antipoverty transfers (such as the well-known CCT Bolsa Família) with other policies related to the supply of good infrastructure or labour force policies.
Reflecting on these last two points in particular, I was wondering if we need to push for a second wave of impact evaluations for many social policies. Most of these policies are implemented at a national level and what we usually know is the overall impact, when information and analysis of the local mechanisms might be more insightful.
In fact, an average positive effect of a program could also be the result of high positive impacts in some areas and low, or even negative, ones in other locations. And it is of fundamental policy relevance to understand what drives these different impacts. Following the last Nobel price winner in economics Angus Deaton, should we not move beyond average effects of these policies and look at deeper and important causal mechanisms?