Accounting for consumption inequality in Myanmar: 2004/05 and 2009/10

Abstract

This thesis investigates consumption inequality in Myanmar, utilising comprehensive household expenditure data sets from 2004/05 and 2009/10 called the Integrated Household Living Conditions Assessment surveys. The distributions of revised comprehensive total household expenditures per adult equivalent indicate a decline in different measures of consumption inequality over time. These data suggest that both ‘relative inequality’ and ‘absolute inequality’ have fallen over this five year period. Poorer population groups have gained rapid expenditure growth than richer ones over the whole national consumption distribution. The nationwide Gini coefficient for expenditure per adult equivalent decreased from 0.256 to 0.220 over time. Nationally, the declines in the Gini coefficient, Theil index, Mean Log Deviation, and Atkinson indices were each statistically significant. Disparities in socio-economic conditions between rural and urban areas, as well as states and regions have persistently been claimed, especially by people in rural areas and minority states who believe that they do not receive equal redistributions of their country’s resources. Yangon and Taninthayi had the highest inequality in expenditures and Kayin state was the lowest in the ranking of inequality over time. The static inequality decomposition analyses show that the contribution of within-group inequality of rural and urban areas to total inequality in both levels and changes is higher than that of between-group inequality. Over the study years, both the between-group and within-group inequalities of rural and urban areas have decreased significantly. However, the contribution of between-group inequality of rural and urban areas to total inequality in Myanmar decreased over time, while that of within-group inequality to total inequality correspondingly increased. A similar trend is found for the level of, and changes in, the contributions of states and regions to total inequality. Therefore, the results confirm that a substantial part of expenditure inequality in Myanmar is not spatial. Cyclone Nargis also contributed to the decline in inequality that occurred in the Nargis-affected area, as well as to the observed decline in total national inequality. The Fields (2003) regression-based inequality decomposition reveals that locational and regional effects, occupation, and levels of education of household members are key to explaining both the level of, and changes in, consumption inequality. Firstly, regional specific variables are the main contributors to the narrowing of expenditure inequality and these explain about 35% and 43% for all households and panel households, respectively. However, these factors have complex origins. Ideally, other variables that are beyond the available data can be correlated with the region-specific variables considered in this study, and thus, while their impact cannot be captured directly, it is reflected in the regional variables. The second largest contributor is the share of household members with different types of occupation, accounting for 22% (all households) and 16% (panel households). The third major influencing factor is the level of education of working-age adults (aged 15-64) constituting about 14% and 18% for all households and panel households, respectively. This research also finds that the results produced using the Yun (2006) approach are inconsistent, and provide a seemingly arbitrary choice for researchers.