Poverty, income fluctuations and work : Argentina, 1991-2002
This thesis presents results on the economics of poverty and labour markets, using data from Argentina in the 1991-2002 period for empirical applications. The thesis is divided into two Parts. The first Part concentrates on poverty and the effects of income fluctuations on well-being. The 1995-2002 period in Argentina was characterised by recurring economic crises that produced large fluctuations in household income. The empirical applications of this Part rely on a rotating panel dataset from the Greater Buenos Aires region to study the effects of this variability. The first Chapter introduces the data and its characteristics, and describes the economic context of the period. The second Chapter defines a family of indicators of well-being, based on the theory of choice under uncertainty, that account for the negative impact of income fluctuations on household welfare. Chapters 3 and 4 present risk adjusted measures of income and the transient-chronic poverty decomposition, respectively, two methodologies for the study of poverty with panel data which are related to the indicators defined in Chapter 2. In Chapters 3 and 4, the household characteristics associated with income fluctuations and their impact on well-being are identified through regression analysis. Part II deals with fertility and women's labour supply from an empirical perspective, and uses data from the 1991 Argentine Census for its applications. Chapter 5 presents the theoretical and econometric framework employed to deal with the endogeneity of the fertility decision. This identification strategy exploits parental sex preferences as instrumental variables for further child-bearing. Chapter 6 discusses its validity in developing countries, and provides empirical evidence for Argentina. Chapter 7 presents the main results of the estimation, which state that additional children cause a reduction in their mother's labour supply. Finally, Chapter 8 proposes a new test for the generality of instrumental variables results, which is illustrated with the same dataset.