Growth and Convergence in Income Per Capita and Income Inequality in the Regions of the EU

Abstract

The neoclassical growth model predicts that there will be both a reduction of income per capita disparities over time and long-term convergence in income inequality levels. This paper examines whether this holds true for the EU using data from the European Community Household Panel for 102 regions over the period 1995–2000. The analysis is conducted using cross-sectional and panel data growth models with spatial interaction effects. The results reveal the presence of a conditional convergence in income per capita after controlling for educational attainment, unemployment, sectoral composition, spatially lagged growth of income per capita, and regional fixed effects, and that of an unconditional convergence in income inequality.

Expansion et convergence dans les revenus par habitant et inégalité des revenus dans les régions de l'UE

RÉSUMÉ Le modèle de croissance néoclassique prévoit que l'on assistera, à terme, à une réduction des disparités entre les revenus par habitant, ainsi qu’à une convergence à long terme des disparités dans l'inégalité des revenus. La présente communication examine l'applicabilité éventuelle de ce modèle à l'UE, en utilisant des données de l'European Community Household Panel pour 102 régions, au cours de la période 1995–2000. On effectue cette analyse en utilisant des modèles de croissance transversaux et de commissions, avec des effets d'interaction spatiale: ses résultats révèlent l'existence d'une convergence des revenus par habitant, après avoir contrôlé les réalisations éducationnelles, la composition sectorielle, la croissance du retard spatial des revenus par habitant, ainsi que des effets régionaux fixes, ainsi que la présence d'une convergence inconditionnelle dans l'inégalité des revenus.

Crecimiento y convergencia en ingresos per capita y desigualdad de ingresos en las regiones de la UE

RÉSUMÉN El modelo neoclásico de crecimiento predice que con el tiempo se producirá una reducción de las disparidades en los ingresos per capita, así como una convergencia a largo plazo de los niveles de desigualdad de ingresos. Este trabajo examina la veracidad de este caso en la UE utilizando datos procedentes del European Community Household Panel aplicables a 102 regiones durante el período 1995–2000. El análisis se conduce utilizando modelos de crecimiento de datos transeccionales y de panel con efectos de interacción espacial. Los resultados revelan la presencia de una convergencia condicional en los ingresos per capita después de controlar el rendimiento educacional, desempleo, composición sectoral, crecimiento limitado espacialmente de los ingresos per capita y efectos regionales fijos, así como la presencia de una convergencia incondicional en la desigualdad de ingresos.

Keywords:

• Income inequality
• convergence
• spatial econometrics
• Europe

JEL CLASSIFICATION:

• C21
• C23
• O12
• O52

Acknowledgements

The author wishes to thank Andrés Rodríguez-Pose (LSE) and Steve Gibbons (LSE) for their useful comments, and Eurostat for granting access to the European Community Household Panel (ECHP). Eurostat bears no responsibility for the results and conclusions of this paper.

Notes

1. Sicily, Sardinia, Nisia Aigaiou and Crete, the Canaries, Azores, Madeira and Ireland.

2. The random-effects estimators, which assume that the unobserved heterogeneity is stochastic across regions but time invariant within regions, are not reported because, apart from measurement error—which is not expected to be the case in convergence models—the pooled OLS and random-effects estimates should be similar when most of the variation is cross-sectional (Griliches & Mairesse, 1984; Barro, 2000).

3. More specifically, information on personal income is collected by the ‘Total net personal income (detailed, NC, total year prior to the survey)’ variable. Data on income are collected: (1) for each individual in the household, so as to measure income of any given individual, and (2) for each normally working (15+ hours/week) individual in the household—using the variable ‘Main activity status—Self defined (regrouped)’—in order to measure the income of normally working people.

4. Income per capita for the whole of the population increased from [euro]9,760 in 1995 to [euro]12,810 in 2000, while income inequality decreased from 0.42 to 0.36 (Theil index). Income per capita for normally working people increased from [euro]13,190 in 1995 to [euro]16,620 in 2000, while income inequality decreased from 0.24 to 0.21 (Theil index).

5. Moran's I and Geary's C tests for spatial autocorrelation are calculated using the normal approximation, the randomization assumption, and the pseudo-significance levels based on an empirical distribution derived from 9,999 permutations from 1995 to 2000. The results will be provided upon request.

6. The standardized values of the Moran's I and Geary's C indices will be provided upon request.

7. Educational attainment and unemployment are extracted from the ECHP data survey. More specifically, individuals are classified into three educational categories which are mutually exclusive and allow for international comparisons as they are defined by the International Standard Classification of Education. These categories are: recognized third-level education completed, second stage of secondary-level education completed, and less than second stage of secondary-level education completed. I describe the educational attainment within a region in terms of the percentage of the population who have successfully achieved the above three levels of formal education (Rodríguez-Pose & Tselios, 2009b). The percentage of people with less than second stage of secondary-level education completed is taken as the base category. Unemployment is the percentage of unemployed respondents. Finally, the sectoral composition variables are extracted from the Eurostat database. ‘Lagged agriculture’ is the lagged share of added value of agriculture, hunting, forestry and fishing in total added value; ‘lagged industry’ is the lagged share of added value of mining and quarrying, manufacturing, electricity, gas and water supply, and construction in total added value; and ‘lagged services’ is the lagged share of added value of services (excluding extra-territorial organizations and bodies) in total added value (Tselios, 2008). ‘Lagged agriculture’ is taken as the base category.

8. The distributions of income per capita and income inequality and their growth rates are relatively compact. More specifically, there are no outliers in the distributions of income inequality (Gini coefficient), income per capita for the whole population, and of the growth rates of income per capita and income inequality. Outliers are present in the distribution of income per capita among normally working people. In 1998, for example, Luxembourg and Île de France were outliers at the upper end of the distribution, while the Portuguese regions of Centro, Algarve, Madeira, and Alentejo were outliers at the lower end. These outliers do not alter the sign and significance of the regressions results.