布鲁金斯学会-报告：超过10亿人生活在贫困地区（英文）-2021.3-36页_5mb

Summary of "Poverty Hotspots and the Correlates of Subnational Development"

Core Content

This paper investigates the spatial distribution of economic growth and poverty at the subnational level across 169 countries, analyzing 2,894 subnational administrative units. The study emphasizes the importance of spatial autocorrelation in understanding subnational development patterns, suggesting that conventional regression models often fail to account for the influence of neighboring regions on growth and development. It also explores the role of geographic, climatic, and institutional factors in shaping subnational economic outcomes.

Main Viewpoints

• Spatial Inequality and Growth: Economic growth is not evenly distributed across space. Even as national incomes converge, subnational areas show increasing disparities. Spatial clustering of growth and development is a key factor in explaining these disparities.

• Spatial Autocorrelation: Spatial autocorrelation is a critical determinant of subnational growth. The paper confirms that Ordinary Least Squares (OLS) estimates of subnational growth are potentially inconsistent and biased due to ignoring spatial dependencies.

• First-Nature Characteristics: Physical geographic features such as soil suitability for agriculture, elevation, and malaria ecologies significantly influence subnational development. These characteristics are independent of national economic policy.

• Conflict Legacy: The presence of conflict has a consistent, negative impact on subnational growth. Conflict-related deaths and violence are associated with long-term underdevelopment and poverty.

• Conditional Convergence: Lower-income subnational areas tend to grow faster than higher-income ones, indicating conditional convergence. However, absolute convergence is not consistently observed.

• Poverty Hotspots: There are 538 poverty hotspots—subnational units that remained classified as low-income in both 2000 and 2015—with a population of 1.12 billion in 2015. These hotspots are concentrated in certain regions, such as the Amazon basin, sub-Saharan Africa, and Central-South Asia.

• Spatial Spillovers: Growth in one region can spill over to neighboring regions, even across national borders. Spatial dependence is a significant factor in explaining growth patterns.

• Public Policy Implications: The findings suggest that spatial policies are necessary to address persistent poverty and inequality. Investments in infrastructure, human capital, and conflict mitigation can have positive spillover effects on neighboring regions.

Key Findings

• Spatial Autocorrelation: The paper finds strong spatial clustering in subnational growth, with significant Moran's I values indicating non-random distribution.

• Geographic Factors:

  • Favorable soil conditions and lower elevation are associated with higher growth.
  • Malaria temperature suitability negatively affects growth, with a one-unit increase reducing growth by 1%.

• Human Capital: Expected years of schooling have a strong positive effect on subnational development.

• Infrastructure and Connectivity:

  • Average distance to the nearest port and travel time to major cities have no direct effect on growth, but travel time within countries significantly hampers growth.

• Conflict: The presence of deadly armed conflicts has a negative impact on growth, which becomes more pronounced when controlling for endogeneity.

• State Exposure: Historical exposure to state capitals has a small but significant effect on growth, indicating the influence of political institutions on development.

• Conditional Convergence: Subnational areas with lower initial income levels tend to grow faster than those with higher income levels, consistent with previous findings in the U.S.

Methodology

• Data: The study uses a geo-referenced, cross-sectional dataset of subnational units based on the Database of Global Administrative Areas (GADM 2018).

• Empirical Strategy:

  • The paper employs spatial autoregressive (SAR) models to account for spatial autocorrelation.
  • It uses the Moran's I statistic to assess global and local spatial autocorrelation.
  • Local spatial autocorrelation analysis identifies clusters of poverty and high-income areas.

• Instrumental Variables: To address endogeneity, the study instruments conflict with drought indicators (standardized evapotranspiration index, SPEI), showing that the negative impact of conflict on growth is more significant when endogeneity is controlled for.

• Spatial Weighting Matrices: The study tests different spatial weighting matrices (queen contiguity, rook contiguity, and inverse distance) and finds that the results are robust across these specifications.

Policy Implications

• Spatial Targeting: Identifying and targeting poverty hotspots requires spatially informed policies that consider the influence of neighboring areas.

• Infrastructure Development: Improving access to cities and ports can have indirect positive effects on growth in surrounding areas.

• Conflict Mitigation: Reducing conflict and its impact is essential for long-term development, especially in low-income regions.

• Human Capital Investment: Enhancing education and skills in subnational areas can drive growth and reduce spatial inequality.

• Need for Comprehensive Analysis: The study underscores the importance of considering both direct and indirect effects of spatial variables in policy design.

Conclusion

The paper highlights the importance of spatial factors in subnational development and poverty. It argues that public policy must account for spatial spillovers and dependencies to effectively address inequality and promote inclusive growth. The findings support the need for targeted interventions in poverty hotspots and emphasize the role of geography, climate, and governance in shaping economic outcomes.