At the spring 2013 meetings, World Bank President Jim Young Kim set 2030 as the target date for eradicating extreme poverty, defined as subsistence on less than $1.25 per day, across the globe. In line with this goal, the United Nations created a New Global Partnership to lift the 1.2 billion poorest people out of penury in the same time frame. The New Global Partnership or Post-2015 Development Agenda replaces the eight Millennium Development Goals declared in 2000 and calls for a “data revolution” that demands development goals be based on internationally compatible measures.
An important tool for this “data revolution” is poverty mapping — the visual depiction of income and consumption data as indicators of lagging development in particular locations. Such maps are produced using aggregate statistics that rank villages and communes by these indicators and enable governments, international agencies and NGOs to channel aid “more precisely” to the poorest of the poor.
The maps, in the words of World Bank proponents, attempt to “summarize poverty estimates for hundreds or even thousands of towns, villages or urban neighborhoods on a single page and in a visual format that is readily understandable by a wide audience.” [1] One benefit of poverty maps is to shift poverty data away from sole emphasis on the nation-state toward provincial and local variances. Poverty mapping tools have been used to guide a wide range of interventions including the location of basic infrastructure development; the creation of national, provincial and municipal development plans; the allocation of grant monies; and the piloting of conditional cash transfers to replace subsidies.
History
Debates about how to define and measure poverty have been around for hundreds of years.
One of the first poverty maps was Charles Booth’s 1889 depiction of poverty in London. Based on detailed observations, Booth used seven and then eight color-coded classifications along a spectrum to describe the income and social class of individual streets in the British capital. That kind of attention to regional and local variations did not become a field of inquiry until the 1940s, when regional scientists attempted to explain the variables that affect the distribution of wealth across a city or municipality. In fact, regional science and the fascination with statistics led to the quantitative revolution of the 1950s and 1960s in which the discipline of geography became oriented toward the scientific method of inquiry and methodologically dominated by the use of statistics and mathematical modeling.
Poverty maps come out of a larger trend within development that seeks “evidence-based policymaking.” Historically, discussions of poverty have been dominated by debates about how to define and measure the “national poverty head count ratio” — the proportion of a population living below the poverty line. In the early 1990s, economists began to ponder the idea that there might be a spatial dimension to poverty — that is, that a relationship between poverty rates and specific locations might exist. They identified “spatial poverty traps” where poverty tends to be concentrated and began to ponder the relationship between this phenomenon and the environments in which these subjects reside.
Economists’ discovery of geography spurred a “new economic geography” in which econometric and spatial analyses were merged. At the time, geographical information science was at its infancy and subnational poverty measures were only beginning to be collected. This juncture also marked the beginning of investment in poverty mapping, with World Bank leadership, as resources were put toward assembling data and developing statistical and econometrical techniques that could construct maps of where poverty is most concentrated. The 2009 World Development Report, entitled Reshaping Economic Geography, solidified the development industries’ commitment to geographical considerations. Poverty mapping therefore ties into broader debates about the definition and measurement of poverty, including basic needs indexes and multidimensional poverty measurements. But the impetus for targeted programs was a neoliberal framework that left many countries in the global south, particularly in sub-Saharan Africa, with rising poverty rates and dwindling public resources. The mandate to “do something with less” drove innovations in mapping methodologies, as policymakers sought to ensure that the benefits were not leaking upward from the bottom of the pyramid.
Data Matters
There are multiple methodologies used to produce the maps; the most common is a technique called small-area estimation. It involves multiple regression analyses and simulation methods that estimate household consumption and compare it to poverty lines. Some estimates are based solely on income or expenditure and others on composite indexes. Regardless, these methods are indirect and calculated based on an amount of statistical error that is assumed to be “suitably precise.” [2] There are huge variances in poverty maps based on how the census and survey data are combined, when the data was collected and what equations are used to develop the poverty estimates at the commune or village level.
Poverty maps only highlight the variances that are captured in the data; therefore, “poverty maps are built on the assumption that the census and survey data sets represent the same underlying population.” [3] Aside from the fact that the census is usually a decennial exercise, meaning that by the time the maps are produced the data is several years old, there are other problems with data accuracy. As those familiar with census and “big” data collection know, the hardest-to-reach places are also the ones with the worst data. This problem is compounded by the fact that hardest-to-reach populations are frequently the poorest. For example, the number of residents in slums and informal settlements is inevitably underestimated, yet these communities are precisely the ones that are most frequently targeted by these tools. In some cases, slums and frontier provinces are entirely excluded from the poverty mapping exercise because of bad data.
The underlying assumption is that poverty maps are themselves a better form of data than was previously available. But if the maps are based on flawed inputs, the output will be faulty, too, as will the conclusions drawn by policymakers from the maps.
A further problem is that poverty maps rely most heavily on consumption or income as measures of poverty, ignoring education, health care and other indicators that multi-dimensional approaches are supposed to take seriously, albeit through composite indexes that suffer from many of the same data and calculation problems. Poverty maps represent the development industry’s over-reliance on numerical representations and rankings.
Inequalities
There is some discussion of how to improve the poverty mapping tool so as to capture the phenomenon of poverty in all its complexity. Yet the tool itself is rarely subject to scrutiny. The use of poverty maps speaks to persistent inequalities in knowledge production.
Since World War II, the Bretton Woods institutions, in cooperation with national aid agencies, have set international development priorities. With the emergence of the “Washington consensus,” the one-size-fits-all prescriptions of neoliberal reforms for developing economies, a plethora of NGOs and private-sector development consultants took on the task of administering these harsh medicines on the ground. The upshot is that poverty experts concentrated in the global north work with counterparts in the capitals of the global south to produce policies based almost entirely on reductive quantitative data. The colonial legacy remains as poverty knowledge is disseminated from the metropole to the periphery.
In the 1990s, critical development scholars highlighted the unequal power relations embedded in this system and attacked the multi-billion dollar aid industry for perpetuating rather than breaking the cycle of poverty. Some of the most important critiques center around the construction of poverty knowledge and the assumptions underlying development measures. How to measure poverty became a subject of great debate, particularly the question of whether income and consumption are adequate as indicators. Knowledge production about poverty remains rooted in approaches that see poverty as discrete, calculable and technical.
Global poverty experts remove the politics from development by claiming that indicators are standardized; the policies that derive from these tools are assumed to be evidence-based and thus scientific. But statistics themselves must be produced and interpreted — and it matters who has the prerogative to produce and interpret. If statistics are a key apparatus of government, as Foucault has argued, then poverty maps represent an important tool of governance. Despite the immense power embedded in these tools, these “technologies of rule” are often taken for granted. The technical expertise produced by multilateral agencies reinforces poverty knowledge as teleological and fixed in place. Poverty maps have far-reaching implications for where aid goes and under what conditions, yet surprisingly little is known about whether the maps have indeed ameliorated the plight of the poorest of the poor.
An alternative framework for analyzing poverty is relational. A relational analysis puts the processes that produce inequality at the center and examines the interactions between individuals, institutions and neighborhoods. It puts politics back into the equation. Rather than viewing poverty as discrete and static, it sees “persistent poverty as the consequence of historically developed economic and political relations.” [4]
The irony is that poverty mapping itself is best understood as a relation of power and not a snapshot of reality.
Endnotes
[1] Tara Bedi, Aline Coudouel and Kenneth Simler, More Than a Pretty Picture: Using Poverty Maps to Design Better Policies and Interventions (Washington, DC: World Bank, 2007), p. 3.
[2] Ibid., p. 4.
[3] Ibid., p. 18.
[4] David Mosse, “A Relational Approach to Durable Poverty, Inequality and Power,” Journal of Development Studies 46/7 (2010). p 1157.
