2007年-世界发展银行全球_Kosovo___Poverty_Assessment_Volume_2_Estimating_Trends_from_Non-Comparable_Data_38页_2mb

Kosovo Poverty Assessment Summary (Volume II)

Core Content

This document provides an in-depth analysis of the challenges in estimating poverty trends in Kosovo using the Household Budget Survey (HBS) data. It highlights the difficulties in comparing poverty estimates across different survey waves due to changes in survey design, sampling methods, and data collection techniques. The report also discusses the implications of these changes on consumption and poverty estimates and offers recommendations for improving future poverty monitoring.

Main Views and Key Information

1. Data Comparability Challenges

• Problem 1: Diary versus Recall

  • The shift from a two-week diary to a weekly recall method led to underreporting of consumption, particularly for frequently purchased items like food.
  • This change is likely to result in lower reported consumption and higher poverty estimates.

• Problem 2: Survey Design - Redefinition of Consumption Items

  • The level of disaggregation of consumption items changed over time, with the number of categories decreasing from over 85 to 12.
  • This change likely resulted in underreporting of consumption and overestimation of poverty.

• Problem 3: Survey Design - LSMS versus HBS

  • The LSMS (2000) and HBS (2002 onwards) differ in several aspects, including recall period, sampling frame, and the way consumption items are reported.
  • These differences make it difficult to compare poverty estimates across surveys.

2. Sampling Weights and Uncertainty

• The use of outdated population frames and limited field supervision introduces significant uncertainty into the HBS data.
• Sample weights amplify this uncertainty, leading to volatility in poverty estimates.
• The estimated population declined by about 25% from 2002 to 2005, which is inconsistent with the absence of conflict or major mortality shocks.
• This volatility affects the accuracy of poverty and inequality measures.

3. Impact on Consumption and Poverty Estimates

• Consumption Trends

  • The mean of total consumption in 2002 was about 10% higher than in 2003 and 15% higher than in 2005.
  • Food consumption dropped by 13% from 2002 to 2003 and by 21% from 2002 to 2005.
  • Own consumption was significantly affected by the changes in recall period and the aggregated list of items, with a 30% drop in own production between 2002 and 2005.

• Poverty Estimates

  • Poverty rates fluctuated widely across survey waves, with estimates ranging from 35% to 45% between 2002 and 2005.
  • The weighted poverty rates were more volatile than the unweighted ones.
  • For example, in urban areas, the weighted poverty rate dropped by 5 percentage points, while the unweighted dropped by only 3.
  • The inclusion of non-food consumption significantly affects poverty estimates, leading to unrealistic swings in welfare indicators.

4. Ethnic and Regional Variations

• Poverty rates by ethnicity and location were highly volatile.
• For the Serbian ethnic group, poverty rates ranged from 35% to 80%, depending on the survey wave.
• The Albanian ethnic group had rates ranging from 32% to 43%.
• These variations are likely due to changes in survey design and data collection methods.

5. Methodological Approaches

• Several methods were used to construct comparable poverty estimates, including post-stratification, reweighting, and using consistent consumption aggregates.
• These methods aim to reduce the impact of survey design changes and sampling errors.
• The report concludes that without consistent survey design and updated population frames, poverty trends cannot be reliably estimated.

Recommendations

• Maintain consistency in the survey questionnaire to ensure comparability of data.
• Conduct a population census to update the sampling frame and improve the accuracy of population estimates.
• Improve survey administration and documentation to reduce errors and biases.
• Use alternative methodologies, such as post-stratification and consistent consumption aggregates, to enhance the reliability of poverty estimates.

Key Tables and Figures

• Table 1.1: Population size by survey wave and year.

• Table 1.2: Summary of survey constraints and their effects on poverty estimates.

• Table 1.3: Poverty headcount by location and ethnic areas using PA05 methodology.

• Table 1.4: Poverty headcount by household head ethnicity.

• Table 2.1: Overview of the results of methodologies for comparable poverty estimates.

• Table 2.2: Summary of poverty estimates from the methodologies used.

• Table 2.3: Poverty rates with current weights and reweighted.

• Table 2.5: Poverty rates using the PA05 and comparable CA methodologies.

• Table 2.6: Robust poverty lines based on consistent food items.

• Table 2.7: Poverty rates using the abbreviated consumption bundle methodology.

• Table A.1: Comparison of previous methodologies.

• Table A.2: Survey comparison.

• Table A.3: Percent changes in main aggregates from survey to survey comparison.

• Table A.4: Alternative consumption aggregate definitions and poverty rates.

• Table A.5: Consistently asked questions over the four surveys.

• Table A.6: Definition of consumption aggregates for the different methodologies.

• Table A.7: Poverty lines in different methodologies.

• Table B.1: Introduction of new questionnaires.

• Table B.2: Poverty statistics using PA05 methodology.

• Table B.3: Poverty rates using PA05 methodology.

• Table B.4: Detailed poverty diagnostics with revised consumption aggregate.

• Table B.5: Poverty rates using alternative consumption and poverty line methodologies.

• Figure 1.1: Total population in millions and household size.

• Figure 1.2: Average monthly household consumption in nominal prices.

• Figure 1.3: Poverty rate estimates and the effect of changes in the questionnaire.

• Figure 2.1: Cumulative and density distribution of consumption for the bottom 50 percentile of the population.

• Figure 1.3 (Box 2.1): Example of sampling without a census.

• Figure 1.3 (Box 2.2): Analysis of changes.

Conclusion

The report concludes that poverty in Kosovo remained high, around 45%, between 2002 and 2006, with no clear evidence of sustained improvement in household welfare. The challenges in data comparability, changes in survey design, and sampling errors make it difficult to draw reliable conclusions about poverty trends. Therefore, the report emphasizes the need for a consistent and updated survey methodology to ensure accurate and reliable poverty monitoring in the future.