Multidimensional inequality, earnings mobility and the tunnel effect: Evidence from Indonesia

This thesis analyses multidimensional inequality, earnings mobility and the tunnel effect with empirical data from Indonesia. Chapter 3 applies the hedonic method for the analysis of multidimensional inequality with an expenditure group weighting scheme. The results from the regressions part in this chapter are utilised to get proportional weights of expenditure, health, and education to be used for Maasoumi's multidimensional inequality and Decancq and Lugo's multidimensional Gini indices. It is found that, while equal and all-samples weighting schemes show almost similar results, multidimensional inequality measures with an expenditure group weighting scheme shows higher results, indicating a variation in each expenditure group in valuing dimensional weights. Chapter 4 analyses the segmented labour market between formal and informal labour and earnings mobility of the two types of labour within the context of a monetary crisis. In addition to proving a segmented labour market between formal and informal labour, the results confirm the concept of safety net function of informal labour and the countercyclical nature of the informal sector in the crisis period. In terms of methodology, GMM estimators are also proven to yield more efficient results compared to pooled OLS and fixed effect estimators. Chapter 5 investigates the tunnel effect in terms of the association between expected future economic level and the reference group's income. The ordered probit model is applied to different types of reference groups based on the geographical location of the community, sub-district, district and province. The model also incorporates income inequality and variables related to social capital. The results show the indication of a tunnel effect with the reference group of people who live in the same province, which also indicates that the tunnel effect is not sourced only from altruism. Inequality makes the tunnel effect stronger, while social capital variables do not make significant differences on the effect. When applied to religious groups, the tunnel effect is only found in the Muslim group. It is not proven that the tunnel effect is affected by adaptive expectations based on the regression results of income quintile groups.