Titolo della tesi: Three Essays on Inequality of Opportunity
The present thesis focuses on various aspects related to inequality of opportunity (IOp).The theory of inequality of opportunity has well-defined philosophical foundations, starting from the assumption that not all differences among individuals should be ironed out (Dworkin, 1981). Consequently, a metric to identify fair and unfair inequalities is needed. The rationale consists in drawing a line between the two of them through the concept of responsibility. Any type of interpersonal difference that can be sourced back to a purely individual choice that anybody could make should be regarded as a fair difference, and thus acceptable. The other way around, all disparities deriving from initial (dis)advantages – such as the family of origin – do not imply any personal responsibility. As such, they should be considered an unfair kind of inequality and thus be tackled (Roemer, 1998). In synthesis, IOp is a particular type of socio-economic disparity among individuals that derives from factors beyond personal control, called circumstances. It regards differences in achievements (e.g., in incomes) due to causes individuals are not responsible for, which thus derive from personal (dis)advantages. An extensive literature has flourished since the seminal work of Roemer (1998), who translated in economics the philosophical foundations laid down by Dworkin (1981). In this field, many different types of contributions have been proposed, focusing on the measurement of inequality of opportunity as well as on its causes and consequences on a range of other economic phenomena. The present thesis contributes to this strand of literature with three chapters, each enclosing an original article.
Chapter 1 encloses original estimates of inequality of opportunity at the regional level in Europe, using the EU-SILC data reported in the ad-hoc modules in the waves of 2005, 2011, 2019. The relevance of such contribution lies in the sub-national socio-economic heterogeneities otherwise overlooked in previous country-level estimations of inequality of opportunity. Indeed, a national index of IOp must start from assuming that circumstances affect final outcomes irrespectively of the local specificities in terms of development and socio-economic structure. This is a strong hypothesis to start from, especially in consideration of the relevant sub-national heterogeneities characterizing several European countries (e.g., Northern vs Southern France, etc.). Furthermore, two additional contributions are provided. First, it is investigated the existence of a positive association between inequality of opportunity and income inequality itself, resembling what already found between the latter and intergenerational mobility (the so-called "Great Gatsby Curve"). Second, inequality of opportunity is related to the average level of per-capita income, verifying whether the pattern mimics the relationship between the latter and income inequality (the so-called "Kuznets Curve").
Chapter 2 focuses on the role playted by inequality of opportunity in shaping economic growth. Such article is placed in the strand of literature sourcing back to the contribution by Okun (1975), holding that equity is alternative to efficiency. An optimal resources’ allocation can only be reached by allowing interpersonal disparities, while a fairer distribution can only be achieved by giving up efficiency. Many attempts have been made to question such perspective, supporting the idea that a fairer outcome distribution can be beneficial to efficiency itself, intended as economic growth. Likewise, many other studies attempted to prove the opposite thesis, reinforcing Okun’s theory. No unanimous consensus about the sign of the relationship between equity and efficiency emerged. In this respect, most of the literature used income equality as a definition of equity. Recently, several authors proposed the adoption of a different definition, namely (in)equality of opportunity. In this respect, the underlying rationale takes IOp as a proxy of the talents’ mis-allocation and dormancy, emerging as crucial factors hindering growth. However, no definitive consensus about its role merged in this sub-strand of literature as well. Moreover, most of previous works focused on the relationship of interest adopting a national dimension. This might lead the estimation to overlook the sub-national differences in the interplay between inequality of opportunity and growth. Such aspect is particularly relevant, since each country might show both a strongly heterogeneous economic structure and different levels of IOp at a sub-national level (take, for instance, the case of the industrial structures in Southern and Northern regions in Italy), such that also their interaction might differ across regions. Additionally, previous works proposed each a different set of controls to be included in the model specification, dampening the comparability of results across the literature. The present article contributes to this literature by providing original estimates of the association between inequality of opportunity and economic growth at the regional level in Europe, using data from the EU-SILC, Eurostat, PWT and Polity V. All controls fragmentarily proposed by previous research are pooled in a unique model specification, regularized through a machine learning algorithm (Post-Double LASSO), to tackle both overidentification and endogeneity issues. Furthermore, an original perspective about the dynamic interaction of IOp with growth is also proposed by using local projections.
Chapter 3 gives emphasis to the measurement issues that may emerge in inequality of opportunity comparisons across different countries. This article’s rationale starts from remarking the main aim of IOp, namely capturing the fairness in the income allocation process. In particular, it should measure of how much of the current inequality is due to the (unfair) return of income to circumstances. Consistently, a high value of IOp should mean that a great part of the observed income inequality depends on the role played by circumstances in driving final incomes. However, from a purely statistical point of view, also other aspects matter in shaping the value of IOp. In particular, the statistical composition of circumstances matters in the decomposition of income inequality. This can be easilyexemplified considering a country where income is extremely affected by parental education but the great majority of individuals have very educated parents. Although the income allocation process is unfair (i.e., personal income depends a lot on the education level of own parents, which is an uncontrollable factor), the observed IOp level will be low since there is small variability in the circumstance of interest. In other words, as everybody will "benefit" from such advantage, the strong unfairness will be hidden by the composition of circumstances. As long as IOp is used as a within-country index for a specific time, this fact is of minor relevance. Indeed, it can still be interpreted as a measure of "intensity" of the consequences due to an unfair allocative mechanism. Taking the example of before, a low IOp would mean that few people experience a disadvantage from low parental education, although their impact on incomes is strong, simply because that disadvantageous
circumstance is rarely observed. However, when making comparisons over time or among countries, this aspect becomes extremely relevant, since it might invalidate the conclusions about their relative position in terms of fairness of the income allocation process. For instance, consider to compare the toy country exemplified before with another one where there is a lot of variability in parental education, although with lower income returns to it. In principle, taking IOp as an index of fairness, the former country should report a higher value for it, since the income returns to parental education are higher. However, it is probable that the latter will show a higher IOp due to compositional causes. Indeed, being parental education uniformly distributed in the first country, its impact on the final income distribution will not emerge, contrarily to the second country, where the mass of people is more sparsely distributed in terms of parental education. This will lead to misleading conclusions about fairness deriving from a biased comparison between the two countries. The present article starts from this remark and proposes a composition-robust comparison of IOp levels across European countries, using data from the EU-SILC. In particular, a benchmark country is chosen adopting a "data-driven" approach, and all other countries are docked to it in compositional terms to iron out all differences in circumstances’ distribution. This benchmark is defined as the "average" country in terms of circumstances, namely the one that on average is the closest to all the others. Then, all other countries’ circumstances’ composition is adjusted to the benchmark through a reweighting technique, enabling to compute inequality of opportunity starting from the same distribution of circumstances. In this process, a Shapley decomposition is implemented to investigate the most relevant circumstances in driving such cross-national compositional heterogeneities.