Essays on income inequality, poverty and mobility

Abstract

The current cumulative doctoral dissertation consists of four essays. Common aim of all essays is to provide empirically backed insights on different aspects of the distribution of income. The two initial essays address key social problems of the German welfare state. The first essay examines the economic situation of the elderly with a particular focus on pensioners, the second essay investigates poverty. While these two contributions take a cross sectional perspective using cross sectional data, the third essay complements their findings by adding the longitudinal dimension of earnings dynamics in terms of mobility and volatility. In sum, these three essays provide a comprehensive picture on the long run trends of the German income distribution. The fourth essay investigates the conceptual understanding of income distributions, scrutinizing the methodology usually applied when equivalent incomes are investigated as in the first two essays of this thesis. The first essay entitled Incomes and Inequality in the Long Run: The Case of German Elderly is a joined work with Carsten Schröder and Katharina Schulte who each contributed one third to the overall project. Furthermore, it is accepted for publication in the German Economic Review. In this essay German Sample Survey income data is used to examine the income distribution for elderly individuals during the period from 1978 to 2003, an era particularly interesting for the development of the statutory German pay-as-you-go (PAYG) pension system as it was subject to several fundamental reforms. Indeed, in the late 1970s, the German PAYG system was expanded to one of the world’s most generous ones, in terms of both replacement rates and early retirement provisions. Population aging, German reunification and high unemployment rates, however, caused a raising fiscal imbalance and, in consequence, the eligibility age has been raised, replacement rates have been lowered and subsidies have been introduced to stimulate private old-age provisions. The reforms undertaken and in preparation have direct implications for the financial situation of Germany’s actual and future elderly. In order to investigate the implications of these institutional changes, the elderly population, defined as people of age 55 and older, is decomposed by people resident in the Old and New Federal States. Further, we distinguish between persons receiving old-age pensions and persons who do not. Inequality estimates are decomposed by income components, and the bootstrap method is used to test for statistical significance of results. In sum, taking stock of the changes in the income distribution of the elderly in the last decades provides a useful yardstick for taxing the costs and benefits of the ongoing reform process. In common with the first essay’s period under investigation and data source, the second contribution entitled Poverty in Germany – Statistical Inference and Decomposition is aimed at completing the overall picture on long-term trends by looking at the bottom of the German income distribution. The essay is a joined work co-authored by Carsten Schroeder (his contribution is fifty percent) and is accepted for publication in the Journal of Economics and Statistics (Jahrbücher für Nationalökonomie und Statistik). Poverty poses a key social problem, both on the individual level as well as for the society as a whole. Therefore, its measurement, explaining its causes and its consequences is on top of the research agenda of scholars from various disciplines. On the individual level, a slim budget not only restrains the actual possibility to consume. Growing up poor is likely to have negative effects on children’s learning and social capabilities, and on their future life chances. Medical studies indicate that poverty during infancy and childhood is an important predictor of mortality risk. In addition, the loss of autonomy and social participation can work as a psychological stress for deteriorating health, the so-called status syndrome. As mentioned above, poverty is not only an individual dilemma. High poverty rates are likely to create social costs and lower income growth. Credit constraints may prevent people with low income from undertaking efficient human capital investments. Substantial income and wealth disparities may discourage and frustrate people. In turn, deprived people might withdraw from social life, stop looking for work, or turn their backs on the democratic system. Individuals who feel powerless in view of large economic disparities may see no other chance to improve their economic situation but to infringe social and ethical rules and norms. To investigate poverty in Germany, the second essay provides insights of inter-temporal changes in poverty for Germany from year 1978 to 2003. Again, we employ the bootstrap method to test for the statistical significance of results. All estimates are decomposed by household type and region. Across household types, we find poverty estimates are particularly high for single parents. The regional decomposition reveals that poverty is particularly high in the New states. In addition, a nonlinear Oaxaca-Blinder decomposition is conducted to quantify the separate contribution of regional differences in households’ characteristics to the probability of being poor. Whereas the first and second essays investigate poverty and inequality in (repeated) cross sections, the single-authored third essay entitled Cohort Earnings Inequality and Mobility: Evidence from German Social Security Records focuses on the dynamics of earnings. Thus, the findings of the first to essays are complemented by adding the dimension of income mobility and volatility. Again, long run trends in Germany are investigated but different data is deployed. Here, rich longitudinal data on individual earnings biographies obtained from social security administration records is analyzed to research the long-term evolution of earnings inequality and mobility in Germany for the period 1967 to 2007. Categorized into four age cohorts, West German males’ annual earnings are investigated. Each age cohort encompasses ten years. Annual earnings inequality is U-shaped in age and increases steadily for all age cohorts over the period under investigation. Short as well as long-term earnings mobility, in the opposite, has remained stable. The variance in annual log earnings is increasing over the full period. This trend can almost exclusively be explained by an increase in the permanent variance of earnings. In sum, essays one, two and three provide a comprehensive picture on the long term evolution of inequality, poverty and mobility in Germany. The last essay contributes to the conceptual understanding of income distributions and their implications for the distribution of living standards. The paper is co-authored by Carsten Schröder (his contribution is fifty percent) and is entitled Country Inequality Rankings and Conversion Schemes. The essay aims at deepening the insights on the distribution of living standards in a society comprised of heterogeneous households, a topic of interest not only to researchers but to the general public. Thereby, living standard of a household’s members is determined by the material comfort derived from available goods and services. Economists consider the income distribution as a close proxy for the distribution of living standard. When heterogeneous household types are involved two complications emerge. First, different household types have different needs. Members of differently sized/structured households with the same household income may attain different living standards. To obtain a measure that reflects differences in living standards across household types, household incomes must be adjusted for differences in needs. Second, household size heterogeneity also raises the issue of an adequate household weighting when the distribution of living standards is derived. Two conversion schemes are usually employed for assessing personal-income inequality from household equivalent incomes: to weight household units by size or by needs. Using data from the Luxembourg Income Study, we show the sensitivity of country inequality rankings to conversion schemes and explain the finding by means of inequality decomposition. A bootstrap approach is implemented to test for statistical significance of our results.

Die vorliegende kumulative Dissertation besteht aus vier Beiträgen. Der erste Beitrag Incomes and Inequality in the Long Run: The Case of German Elderly ist eine Gemeinschaftsarbeit mit Carsten Schröder und Katharina Schulte, wobei jeder Autor einen eigenen Beitrag von 33% geleistet hat. Darüber hinaus wurde das Papier zur Veröffentlichung im German Economic Review akzeptiert. In dem Beitrag werden sechs Wellen der Einkommens- und Verbrauchsstichprobe (EVS) benutzt um die Einkommensverteilung der älteren Bevölkerung über den Zeitraum von 1978 bis 2003 zu untersuchen. Die ältere Bevölkerung, definiert als Individuen, die mindestens das 55te Lebensjahr erreicht haben, wird nach neuen und alten Bundesländern zerlegt. Weiterhin wird zwischen Personen mit Renten- und/ oder Pensionsbezug und solchen ohne diese Bezüge unterschieden. Ungleichheitsmaße werden darüber hinaus nach Einkommenskomponenten zerlegt und die Bootstrapmethode wird angewandt, um die statistische Signifikanz der Ergebnisse zu überprüfen. Der zweite Beitrag Poverty in Germany – Statistical Inference and Decomposition ist wiederum ein gemeinsames Werk mit Carsten Schröder. Sein Anteil liegt in diesem Fall bei 50% und der Artikel ist angenommen zur Publikation im Journal of Economics and Statistics (Jahrbücher für Nationalökonomie und Statistik). Inhaltlich setzt sich der Beitrag mit der Armutsentwicklung in Deutschland über den Zeitraum von 1978 bis 2003 auseinander. Wie schon dem ersten Beitrag liegen diesem Artikel sechs Wellen der EVS zugrunde und die Signifikanz von Resultaten wird mit der Bootstrapmethode überprüft. Weiterhin werden Armutsrisiko und Armutsintensität nach Haushaltstypen und Regionen zerlegt. Ein Vergleich der Haushaltstypen zeigt, dass insbesondere Alleinerziehende ein hohes Armutsrisiko tragen, und eine Zerlegung nach Regionen offenbart, dass das Armutsrisiko besonders in den neuen Bundesländern sehr ausgeprägt ist. Welche Rolle die unterschiedliche Verteilung der Haushaltscharakteristika zwischen den beiden Regionen auf das unterschiedliche Armutsrisiko hat, wird mittels einer nichtlinearen Oaxaca- Blinder-Zerlegung quantifiziert. Der dritte Beitrag Cohort Earnings Inequality and Mobility: Evidence from German Social Security Records ist ein alleiniges Projekt und nutzt Längsschnittdaten der Sozialversicherung um auf Basis individueller Erwerbsbiographien die langfristige Entwicklung von Lohnungleichheit und Lohnmobilität in Deutschland zu untersuchen. Der Untersuchungszeitraum erstreckt sich über 40 Jahre von 1967 bis 2007. Kategorisiert nach vier Alterskohorten werden die Jahreseinkommen aus sozialversicherungspflichtiger Beschäftigung westdeutscher Männer untersucht. Jede Alterskohorte umfasst 10 Jahre. Es wird gezeigt, dass die Lohnungleichheit in der jüngsten und ältesten Alterskohorte höher als für die mittleren Kohorten ist und insgesamt über den untersuchten Zeitraum für alle Alterskohorten zunimmt. Demgegenüber ist sowohl die kurzfristige als auch langfristige Lohnmobilität unverändert geblieben. Eine Zerlegung der Ungleichheit in eine permanente und eine transitorische Komponente hat zum Ergebnis, dass die Zunahme der Ungleichheit fast ausschließlich auf einen Anstieg in der permanenten Ungleichheit zurückzuführen ist. Der vierte Teil Country Inequality Rankings and Conversion Schemes leistet einen Beitrag zum konzeptionellen Verständnis von Einkommensverteilungen und ist wiederum ein gemeinsames Projekt mit Carsten Schröder. Sein Anteil beträgt auch hier 50%. In dem Papier wird untersucht, welchen Einfluss zwei unterschiedliche Gewichtungsstrategien von äquivalenten Haushaltseinkommen auf die gemessene Ungleichheit von individuellen Lebensstandards haben. Entweder wird das äquivalente Haushaltseinkommen mit der Anzahl der Mitglieder des Ursprungshaushalts oder ihrer Bedürfnisse gewichtet. Wir zeigen mit Querschnittdaten der Luxembourg Income Study (LIS) wie sensitiv auf Ungleichheitsmaßen basierende Länderrankings auf eine Änderung der Gewichtungsmethode reagieren und erklären die Ergebnisse mit einer Dekompositionsanalyse. Um die statistische Signifikanz der Ergebnisse zu überprüfen wird wiederum das Bootstrapverfahren verwendet.