Catch Me If You Can: A Few Thoughts On The Analysis Of Poverty Convergence In LAC
In 1951, Alexander Gerschenkron coined the concept of "advantages of backwardness" to illustrate the fact that less developed countries could borrow from the experience (and technologies of production) of developed ones to "catch up" with them. Such notion can be associated to what formal economic models have called"conditional convergence", the fact that economies with different initial levels of income would converge over time, controlling for heterogeneity in economic policies and institutions.
While convergence in income is derived from classic assumptions regarding the decreasing pattern in terms of marginal returns to capital as the stock increases, if income growth and poverty reduction are closely associated, an analogous phenomenon could be expected to take place when we look at the rates of poverty reduction across countries. Thus, building on the fact that economic growth and poverty reduction are indeed correlated, it would not be strange to observe countries with higher rates of poverty reducing them at a faster pace than those with lower rates—in the same way that poorer countries grow at faster rates than richer ones. It would be an implication of the growth convergence itself, though it involves specific assumptions with respect to the pattern of growth incidence (distribution).
However, when testing this idea seven years ago on data for 90 developing countries, World Bank economist Martin Ravallion did not find the existence of “poverty convergence.” He argued that high levels of poverty were themselves the reason behind this as (i) countries with a higher initial incidence of poverty tended to experience a lower rate of growth and (ii) high poverty rates made it harder to achieve any given proportionate impact on poverty through growth in the mean.
How about convergence of regions within countries? Two colleagues and I tested this hypothesis at the sub-national level with data for Mexico’s municipalities for a long span of time and found that (i) mean per capita income in the poorest municipalities grew consistently faster than in richer counterparts, (ii)the process of income convergence effectively translated into an unambiguous process of poverty convergence; (iii) the growth of income among the poorest in a context of disappointing overall economic growth, promoted sizeable reductions in food poverty rates, which relates to the fact that active redistribution took place during that period (thus: redistribution accelerates convergence).
This #GraphForThoughtlooks at what the data says regarding the idea of poverty convergence through the lens of multidimensional poverty (instead of income poverty), a phenomenon for which there is no actual theoretical foundation. The analysis does so byusing Haiti’s sub-nationaldata from the 2019 global MPI update recently released by UNDP and OPHI. (Haiti is the country in the region where MPI recent trends over time where made available in this version of the report so it’s not obvious to extrapolate the conclusions).
In the figure, regions within Haiti are ordered along the horizontal axis according to their MPI in 2012—from lowest to highest. That is, the richest region (with the lowest MPI) is at the far left and the poorest region (with the highest MPI) is at the far right. The vertical axis shows the absolute reduction in MPI from 2012-2017.As the blue bars tend to grow longer toward the right side of the figure, we can see that many poorer regions are reducing multidimensional poverty faster than richer regions. Although not perfectly, this means that we do observe a pattern in which poorer regions are “catching up” with less poor regions.
Why is this important? As the 2019 global MPI Report “Illuminating Inequalities” stresses, there is considerable heterogeneity among regions within countries (as is the case of Haiti, with a headcount of MPI ranging between 26% and 58%) and it is no longer pertinent to talk about rich versus poor countries. Analysis for Haiti is far from being representative of the bigger question, while being interesting given the low-income context. Understanding how poverty reduction rates vary across regions of the same country is important to keep in mind when designing effective policies and MPIs emerge as a policy coordination toolto focus efforts depending on places or even groups of the population based on MPIs. For instance, a relatively robust pattern in the MPI figures is that ages bracket [0-17] and [65,+] show a higher incidence of poverty than the middle age bracket [18, 64]. Hence,authorities should take note of it and promote policies accordingly.
As mentioned above, there is no theory behind the notion of multidimensional poverty convergence. The mechanisms by which poorer regions reduce poverty faster deserve further scrutiny. We argue here that a model should be founded in a political economy analysis and the dynamics of power asymmetries in the definition of national versus cross-regional allocation of resources. It can be conjectured that these patterns respond to discontinuities of the State across territory in investments in areas linked to the deprivations MPI measures (Education, Health, Living Standards) are part of the explanation. This #GraphForThought is an invitation to think whether and how households in the poorest areas can catch up with those in the richest ones, and what are the implications for the global discussion on SDG “localization”.