The geography of poverty hotspots

The pan-aurora borealis is visible over Vancouver, British Columbia and Seattle, Washington at night in this NASA handout photo received by Reuters on February 10, 2012. REUTERS/NASA/Handout  (CANADA - Tags: ENVIRONMENT) FOR EDITORIAL USE ONLY. NOT FOR SALE FOR MARKETING OR ADVERTISING CAMPAIGNS. THIS IMAGE HAS BEEN SUPPLIED BY A THIRD PARTY. IT IS DISTRIBUTED, EXACTLY AS RECEIVED BY REUTERS, AS A SERVICE TO CLIENTS - GM1E82B0OFX01
Editor's note:

This blog shares insights from “Spatial Targeting of Poverty Hotspots,” the authors’ chapter from a new Brookings Press book “Leave No One Behind: Time for Specifics on the Sustainable Development Goals.”

Since at least Adam Smith’s Wealth of Nations in 1776, economists have asked why certain places grow, prosper, and achieve a higher standard of living compared to other places. Ever since growth started to accelerate following the industrial revolution, it has been characterized by, above all, unevenness across places within countries. Appalachia, the Italian “Mezzogiorno,” and the Habsburg Empire east of the Tisa River, are a few well-known examples of laggards on the path towards industrialization.

These patterns of subnational “dualism” are still prevalent across the world, in rich and poor countries alike. The poorest counties in the United States have a per capita income less than that of Mongolia and Peru, while the poorest census-designated places are poorer than Myanmar or Nigeria.

From a policy perspective, it is increasingly important to know where poverty is concentrated at a level that is more granular than a country. In a new book chapter, we have found that a majority of developing countries will still have at least one region where extreme poverty is likely to persist in 2030. These regions have a combination of characteristics that make development difficult: poor agricultural suitability, a high burden of communicable disease, a high risk of natural disasters, water stress, and other geographic factors. Typically, these are areas that are away from urban areas where economic opportunities are concentrated. These poverty hotspots represent a serious challenge to the “leave no one behind” spirit of the Sustainable Development Goals.

Using spatial data on nighttime luminosity from satellite images, we can generate more granular, detailed forecasts for subnational growth and poverty. Figure 1 shows a global map with shaded areas for all poverty hotspots, defined as subnational regions (districts or provinces within a country) that are on track to have a per capita GDP of $4,900 or less in 2011 PPP terms in 2030. We chose the cut-off of $4,900 to represent a proxy for the mean level of development in a region where extreme poverty is likely to have been eradicated. We find 840 poverty hotspots globally, home to 1 billion people, from a universe of 3,609 administrative units one level below the nation-state (districts, states, and provinces). 102 countries have at least one poverty hotspot. Immediately, four principle clusters of hotspots are visible:

  • Tropical Africa: Zones extending from the Sahel to northern Angola, to the southern borders of Zambia, Zimbabwe, and Mozambique, as well as Madagascar.
  • Tropical Latin America: Scattered parts of Central America, the Caribbean coast of Venezuela along with most of its central and southern regions, part of Ecuador and Colombia, Suriname, French Guiana, and northeastern Brazil.
  • Central-South Asia: Areas stretching from Tajikistan and Kyrgyzstan to most of Afghanistan, northwestern Pakistan, Kashmir on both sides of the line of control, much of Nepal, north-eastern Indian states, and parts of Bangladesh and Myanmar.
  • Southeast Asia-Western Oceania: Sections of Cambodia, Vietnam, the Philippines, Indonesia, East Timor, Papua New Guinea, and the Solomon Islands.

Figure 1: Subnational poverty hotspots, 2030

Figure 1: Subnational poverty hotspots, 2030

The map shows that persistent poverty in lagging regions is indifferent to national boundaries. If we were to map poverty at an even finer level of disaggregation, we would find more scattered hotspots across the world.

What can be done to ensure that these areas are not left behind? Solutions such as encouraging emigration or spatial targeting of growth policies are inherently difficult. If migrants leave, and in so doing, take scarce capital with them, they can depress their source areas even more.

The reality is that migration is a complex decision, dependent on many factors. As the table shows, populations in poverty hotspots are expected to grow at more than twice the rate of other regions in developing countries because poorer populations with less access to health and education services have far higher fertility rates. The out-migration that does occur due to a per capita income growth rate that, for the past decade, has been barely one-sixth that of non-hotspots, is not rapid enough to outweigh higher natural population growth. Emigration, while a potential long-run solution, is not the answer for poverty hotspots in the timeframe of Agenda 2030.

Hotspot Other
Average annual growth rate, GDP/capita (2006 – 2015) 0.9% 6.0%
Forecasted population growth rate (2015 – 2030) 2.4% 1.0%
Subnational Human Development Index (HDI) score 0.47 0.66

An alternative is to accelerate the economic growth of poverty hotspots through targeted policies. In most countries, growth is most efficient when it builds on market forces of agglomeration, specialization, and trade—factors that favor urban centers, not rural, low-density regions. Market forces, if left unattended, can exacerbate the persistent inter-regional disparities that have given rise to laggard regions in the first place. In other words, “spatial” targeting will require specialized interventions that aim to alleviate the geographic sources of persistent poverty.

Our findings suggest that three types of spatially-targeted policies can make a difference: (1) those that improve human capital; (2) those that improve physical infrastructure and market connectivity; (3) those that enhance the resilience of regions to shocks such as like droughts, civil conflict, and natural disasters.

Figure 2 summarizes the marginal effects of improvements in various determinants of average annual per capita growth in subnational regions. For example, doubling the road network density within a subnational area will add just over 1 percentage point to growth rates. Mitigating the impact of droughts by half adds another percent. An additional percent increase in growth may be achieved by raising the subnational human development index (HDI) score by 10 percent (at the country level, this would be the equivalent of Rwanda raising its HDI score to that of Angola, Iraq raising its score to that of Thailand, or South Africa raising its score to that of Brazil). Cutting the rate of conflict deaths by half increases growth by an additional 0.5 percentage points. Increasing the accessibility of the subnational population to urban areas adds about 0.4 percentage points to growth.

Figure 2: Marginal effects of changes in subnational factors on annual growth

Marginal effects of changes in subnational factors on annual growth

These five subnational reforms, therefore, could add 3.8 percentage points to a subnational region’s annual per capita growth over the next ten years. This is enough to add over 45 percent to a region’s GDP per capita in a decade. Additional reforms at the national level can complement these reforms—for example, doubling the country “rule of law” score increases per capita growth by another 0.75 percent annually.

Some of these reforms require policy changes. Others require a reallocation of resources, by the public sector as well as by aid donors. Countries such as India and Papua New Guinea have each successfully used geospatial data to target improvements to health and educational outcomes for marginalized groups in lagging regions. In the West Bank, satellite information has been used to focus resources on rebuilding roads to enable multiple access points to larger road networks. Farmers in Mozambique increased their productivity through the use of a network of recreational drones that provided highly granular data on crop yields. In the Democratic Republic of the Congo, a group of university students, using geo-referenced data from household surveys and government records, compiled, mapped, and publicized land ownership claims in a northern region of the country, thereby reducing encroachment and tensions over land use.

These are a few recent examples of how geospatial targeting can prompt improvements in lagging areas. Spatial targeting offers considerable promise in ensuring that geography will not become destiny for people left behind in developing and developed countries alike.