联合国西亚经济社会委员会-西亚经社会贫困线的划分_验证研究（英）-2025_34页_1mb

Summary of the Document: Demarcation of ESCWA Poverty Lines - A Validation Study

Core Content

This document presents a validation study on the ESCWA quasi-relative poverty lines, which are designed to enable cross-country and cross-time poverty comparisons by reflecting the concave relationship between a country's average household income and the cost of a basic needs basket. The study evaluates the conceptual and empirical validity of these lines against alternative poverty measures, including the World Bank’s Societal Poverty Lines (SPLs) and James Foster’s hybrid poverty lines (HPLs).

The main objective is to develop a consistent and reliable framework for measuring poverty that accounts for varying living standards and basic needs across countries and contexts, while also being suitable for poverty projection in previously uncovered countries and years.

Main Points

1. Conceptual Framework

• Absolute poverty lines are not uniform across countries due to differences in basic needs, product availability, and public goods.
• The ESCWA poverty lines follow a concave relationship with mean household income, mirroring the principle of Engel’s law.
• Engel’s law suggests that as income increases, the proportion of income spent on basic needs decreases.
• The log-log model is used to estimate the relationship between poverty lines and income, capturing this constant-elasticity feature.

2. Data and Methodology

• The study uses data from 151 countries (1,087 observations) spanning 1963–2023, with 90% of data from after 2000.
• Data sources include the World Bank Poverty and Inequality Platform, PovcalNet, and household income and expenditure surveys.
• For countries without absolute NPLs, absolute poverty lines are imputed using income distribution data and reported poverty rates.
• The log-log regression model is weighted to ensure equal representation across all countries.

3. Model Specification

• The model is expressed as:
  $$
  \text{Log} \overline{{\mathcal{N} P L}} = -0.0089 + 0.6905 \log(M) + 0.0781 C
  $$
  where:

  • $ \overline{{\mathcal{N} P L}} $: National Poverty Line
  • $ M $: Mean household income
  • $ C $: Control variable for whether consumption data is available

• The smearing correction is applied to the exponentiated values to account for regression bias.

• A floor of $2.15 per day (2017 PPP) is set for all countries, reflecting the minimum subsistence level.

4. Validation of the Model

• The log-log model outperforms other parametric and non-parametric specifications in terms of fit and accuracy.
• It is shown to align more closely with national poverty lines compared to the World Bank’s SPLs and HPLs.
• SPLs tend to overestimate poverty in high-income countries and underestimate it in middle-income countries, as they are based on relative income thresholds.
• HPLs, which combine absolute and relative elements, are also evaluated. However, ESCWA lines are found to be more accurate and robust, especially in low- and middle-income countries.

5. Empirical Evidence for Concavity

• Engel’s law is extended to non-food basic needs expenditures, showing a consistent declining share of basic needs in total household expenditure as income increases.
• This is supported by microdata from Kuwait and cross-country data from the Luxembourg Income Study, demonstrating that the concave pattern holds for broader bundles of basic needs, including housing, health, education, and transport.
• Non-parametric methods like LOWESS, kernel smoothing, CHAID, and random forest also confirm the concave relationship between poverty lines and income.

6. Conclusion

• The ESCWA poverty lines are well-suited for global poverty monitoring, as they provide a consistent and adaptable measure that reflects changing living standards.
• They are conceptually and computationally robust, and empirically outperform existing alternatives in terms of accuracy and alignment with national poverty lines.
• The model is applied to all countries and years, even where data is limited, making it a practical and scalable tool for poverty analysis and projection.

Key Information

• ESCWA poverty lines are based on a concave log-log relationship between mean income and the cost of basic needs.
• The floor of $2.15 per day (2017 PPP) ensures minimum subsistence is considered.
• Engel’s law is extended to broader basic needs expenditures, supporting the concave assumption.
• The log-log model provides a better fit than other parametric models and is validated by non-parametric methods.
• SPLs and HPLs are compared, with ESCWA lines showing superior alignment with national poverty lines.
• The model is applicable to all countries and years, including those with limited data.

Figures and Tables

• Figure 1: Absolute NPLs vs household mean income per capita
• Figure 2: Mean income vs predicted ESCWA poverty lines
• Figure 3: Expenditure shares for alternative basic needs bundles across 537 surveys, by population deciles
• Figure 4: Expenditures and shares for alternative basic needs bundles, by population quantile: Kuwait 2021
• Figure 5: Poverty rates using SPLs vs ESCWA poverty lines
• Figure 6: HPLs and ESCWA poverty lines vs household income
• Figure 7: HPLs and ESCWA poverty lines vs NPLs
• Figure 8: HPLs (non-linear) and ESCWA poverty lines vs NPLs
• Figure 9: Non-parametric models of national poverty lines
• Table A1: Sample of countries used for estimation
• Table A2: Results of regression analysis (not fully visible in the text)