2010年-世界发展银行全球_Poverty_Dynamics_in_Vietnam_2002-2006_35页_713kb

Summary of Poverty Dynamics in Vietnam, 2002-2006

Core Content

This paper presents a detailed analysis of poverty dynamics in Vietnam using panel data from the Vietnam Household Living Standards Surveys (VHLSS) of 2002, 2004, and 2006. It explores how households transition in and out of poverty and identifies the factors contributing to chronic poverty and poverty transitions in rural areas.

Main Findings

• Poverty Reduction Success: Vietnam has significantly reduced poverty between 1993 and 2006, with the poverty headcount falling by between two-thirds and three-quarters. This success is attributed to economic growth and policy reforms.

• Poverty Dynamics: Despite progress, many households did not move far above the poverty line. Transition matrices and contour plots show that while some households moved out of poverty, a tenth of rural households remained chronically poor.

• Chronic Poverty: The chronically poor (PPP) are defined as households that were poor in all three survey years. These households have significantly lower mean and median expenditures compared to others, indicating they are the poorest group.

• Poverty Traps: Initial conditions such as household size, ethnic minority status, and residence in northern Vietnam are important in trapping households in poverty.

• Drivers of Poverty Transitions: Education, particularly secondary schooling and non-farm employment, plays a significant role in reducing the risk of falling into poverty and increasing the chances of escaping poverty. Infrastructure such as permanent roads also helps in reducing poverty risk.

• Geographic Vulnerability: Households in the Northern Uplands and North Central Coast are more likely to be chronically poor compared to other regions.

• Modeling Approach: The paper uses multinomial logit (MNL), sequential logit, and nested logit models to analyze poverty dynamics. It also employs quantile regression to examine the impact of variables across different expenditure levels.

Key Information

• Panel Data: The VHLSS provides a rotating panel of around 4,000 households for rural areas. The attrition rate is moderate, at 14.0% between 2002 and 2004, and 9.5% between 2004 and 2006.

• Transition Matrices: These matrices show that the percentage of households in poverty decreased from 14.2% in 2002-04 to 10.8% in 2004-06. However, the percentage of households moving out of poverty also declined, from 12% to 8.5%, while the percentage moving into poverty fell from 4.7% to 4.1%.

• Contour Plots: These plots provide a continuous representation of poverty dynamics and highlight that a large number of households remain vulnerable to falling back into poverty, especially those near the poverty line.

• Multinomial Logit Model: The model has a pseudo R2 of 0.275 and a Wald chi2 statistic of 32189.830, indicating a good fit. It shows that households with ethnic minority heads are more likely to be poor and less likely to be non-poor. Education levels, particularly secondary and tertiary, are strongly associated with reduced poverty risk. Infrastructure like permanent roads and access to electricity and clean water also play a role in reducing poverty risk.

• Predictive Ability: The MNL model correctly predicts 70% of poverty dynamics outcomes. It performs better at predicting households that remain poor or non-poor than those that move in or out of poverty.

Conclusion

The paper emphasizes the importance of understanding poverty dynamics beyond just headcount reductions. It highlights that while poverty has decreased significantly, certain sub-groups remain particularly vulnerable. The use of advanced models like quantile regression and alternative categorical models provides a more nuanced understanding of the factors influencing poverty transitions and chronic poverty.