The way countries define poverty is going to matter for their probability of achieving Sustainable Development Goal 1, Target 1.2. It calls for reducing at least by half the proportion of men, women and children of all ages living in poverty in all its dimensions according to national definitions by 2030. This means that national governments can establish the standards against which they will be measuring progress in just over a decade.
For example, if we measure multidimensional poverty in a way that the starting rate is too high, we will struggle to halve it. Define it at too low a level, and further progress may be harder to achieve. Try to game Target 1.2 by fudging your dimensions and you betray the spirit of the Sustainable Development Goals (SDGs). So how can governments define multidimensional poverty in a way that halving the poverty rate will be realistic and amenable to policy intervention, while representing a true improvement in people’s well-being?
In a recent paper we simulate different scenarios for lowering multidimensional child poverty in two small middle-income post-socialist European countries: Armenia and Bosnia and Herzegovina (BiH). Their UNICEF offices carried out child poverty studies using data from household budget surveys collected in 2011-2013. UNICEF chose the dimensions of child poverty in consultation with government and civil society counterparts to reflect national standards and priorities.
Each study used seven dimensions of poverty from this list: clothing, education or educational resources, housing, information access, leisure, nutrition, social participation or social relations, and utilities. UNICEF had initiated these studies before the SDGs have been adopted, so no one worried about halving the resulting poverty rate by 2030 when they were coming up with a definition of poverty. The rate of multidimensional child poverty was twice as high in Armenia as in BiH: four in five (80%) versus two in five (40%) school-age children, respectively, were deprived in two or more out of seven dimensions.
There were differences between the two countries in the intensity of multidimensional child poverty and way various dimensions interacted with each other. In BiH, one dimension influenced child poverty disproportionately (i.e. information access) and was not highly correlated with other dimensions. In contrast, no single dimension dominated in Armenia, and the majority of children deprived in two or more were deprived in four dimensions (i.e. leisure, housing, social relations, and utilities).
To understand the mechanics of reducing multidimensional child poverty, we simulated several scenarios of lowering deprivation in different dimensions at a time. We played around with switching the deprivation status from 1 “deprived” to 0 “non-deprived” for a random selection of children in the dataset for different combinations of deprivations. For example, if all school-age children in BiH had a networked computer at home (i.e. no deprivation in information access), the multidimensional poverty rate would go down by more than one-third (35%). This goes a long way towards halving poverty to reach the Target 1.2.
An alternative, but similarly effective strategy for BiH, would be to eliminate the correlation between the two dimensions that have the highest deprivation count (i.e. information access and leisure), while maintaining the proportion of children deprived in each of them. As long as it is no longer the same children who are deprived in both of these dimensions, but some deprived in one and others in the other, the overall multidimensional poverty headcount would also fall by over one-third (35%).
However, these strategies would not work in Armenia, where the majority of school-age children are deprived in two out of four dimensions at once and no single dimension stands out. Of the four scenarios we considered, the best we could do would be to lower the poverty rate by just over one-quarter (28%) by halving the deprivation rates in three dimensions and reducing it by one-tenth in the other four.
We also simulated the effects of giving different amounts cash to the households where multidimensionally poor children live. We did this by modelling the associations between household consumption and children’s deprivations in different dimensions. Cash transfers to the poor can be a powerful tool for improving children’s outcomes in nutrition, health and education, to name just a few (see https://transfer.cpc.unc.edu/).
Some deprivations (e.g. nutrition and clothing) are more sensitive to household consumption, so cash transfers would be more effective in tackling them. Others (e.g. utilities) depend more on the local services infrastructure. Our simulations suggest that in a country like BiH, giving all consumption-poor households with children enough money to lift them out of monetary poverty would also eliminate multidimensional child poverty. In a country like Armenia, where children tend to be deprived in a greater number of dimensions simultaneously, even such an expensive strategy would not make a sizeable dent in multidimensional child poverty.
We learned from our analysis that a country’s potential to halve multidimensional child poverty by 2030 hinges on the definition of the poverty measure they adopt in the first place. It influences both the “baseline” rate of poverty against which progress will be measured and the policy levers to achieve the goal. An effective strategy is likely to involve a multi-sectoral approach with cash transfers, information provision and investment in public services and infrastructure.