Impact of socio-economic factors and social affiliation on living standards: a quantile regression approach

Abstract

The influence of socio-economic factors and social affiliation on living standards is shown to be contingent on the living standard status of the household. This new result casts doubt on studies that use conditional mean regression analysis and the Oaxaca–Blinder decomposition analysis to study the impact of ‘characteristics’ and ‘structural’ components in areas such as poverty, health and labour market outcomes.

Keywords:

• quantile regression
• poverty
• caste

JEL Classification:

• C31
• I32
• J15

Notes

¹ Some studies are less ad hoc and use lowest 10%, lowest 25%, middle 50%, highest 25% among other variations for categories. But the problem of outliers and different error distributions within each category may still persist, thereby leading to inaccurate estimates.

² For details, see Blinder (1973) and Oaxaca (1973).

³ Rural India is home to at least 75% of the people in disadvantaged castes.

⁴ The bootstrap procedure is modified to account for clustering as the Indian data set is a two-sample design in which clusters are drawn first followed by a selection of households from each cluster.

⁵ Quantile plots for the other two caste types also exhibit a similar nonlinear pattern and are available upon request.

⁶ To conserve space, we do not report the quantile estimates of the other explanatory variables but by and large, the results indicate that there are variations on the impact of these factors on per capita expenditure across different caste types. Evidence of this can be seen in Fig. 1 for the SC/ST.

⁷ The coefficient effect relies on the possibility that the effect of household characteristics may vary among groups while the characteristics effect reflects how differences in attributes among different groups affect the dependent variable.