This paper investigates the effects of skill bias technical change at the frontier on the evolution of output and human capital in the adopting countries. The framework introduces a novel feature by connecting the direction of technology adoption to a sequential process of skill accumulation, where the returns of advanced human capital depend on the quality of basic education. I find that moderate skill bias produces convergence in output per capita, while strong skill bias leads to convergence clubs. The predictions of the model are broadly consistent with the evolution of world income distribution and educational attainments during 1960-2000.

We investigate the relationship between inequality and public education funding in a model of probabilistic voting where the private option is available and voting participation differs across income groups. A change in inequality can have opposite effects at different income levels: higher inequality decreases public spending per student and increases enrollment in public schools in poor economies, while the opposite holds in the rich ones. A change in the tax base can also have non-monotonic effects. These theoretical predictions, with support in U.S. school district-level data, reconcile previous contradictory results in the political economy literature on redistribution and inequality.

Regional income disparities have increased in many European countries during the last three decades, even as national and supra-national policy instruments were created to correct them. To explain these evolutions, we develop a two-region, two-sector model with migration and public investment in infrastructure and education. Accumulation and creation of new ideas and technologies are at the core of differential regional growth. Together with regional migration, these forces are also responsible for diverging industrial structures, with the lagging region also falling behind in innovation. In this framework, we assess the effectiveness of structural funds, modelled on the EU policy. In a numerical example calibrated to Portugal, we find that, to diminish the initial gap in income per capita, the backward region needs to receive around 9% of its own GDP in structural funds, while the actual disbursements were around 4%. We also find that maximizing innovation in the backward region conflicts in the short run with the goal of maximizing its income per capita. Moreover, the rich region has an incentive to bias the allocation of structural funds towards human capital formation.

We study the optimal degree of fiscal decentralization in a dynamic federal economy where governments decide on budget size and its allocation between public education and infrastructure spending. We find that full centralization of tax and expenditure policies is optimal when infrastructure productivity is similar across regions. When differences are not too large, partial centralization is optimal. With strong differences, full decentralization becomes optimal. National steady-state output tends to be highest under full decentralization. We provide a justification for the mixed evidence regarding the Oates conjecture by showing that full dominates partial decentralization, despite being inferior to complete decentralization. On étudie le degré optimal de décentralisation fiscale dans une économie fédérale dynamique où les gouvernements décident de la taille du budget et de son allocation entre les dépenses en éducation publique et en infrastructures. On découvre qu'une pleine centralisation des taxes et des dépenses est optimale quand la productivité des infrastructures est la même dans toutes les régions. Quand les différences ne sont pas trop importantes, une centralisation partielle est optimale. En présence de fortes différences, la pleine décentralisation est optimale. La production nationale en régime permanent tend àêtre la plus élevée quand il y a pleine décentralisation. On fournit une justification pour les résultats mixtes obtenus quant à la conjoncture de Oates, en montrant que la pleine décentralisation est préférable à la décentralisation partielle même si elle est inférieure à la décentralisation complète.

We study the optimal degree of fiscal decentralization in a dynamic federal economy where governments decide on budget size and its allocation between public education and infrastructure spending. We find that full centralization of tax and expenditure policies is optimal when infrastructure productivity is similar across regions. When differences are not too large, partial centralization is optimal. With strong differences, full decentralization becomes optimal. National steady-state output tends to be highest under full decentralization. We provide a justification for the mixed evidence regarding the Oates conjecture by showing that full dominates partial decentralization, despite being inferior to complete decentralization. On étudie le degré optimal de décentralisation fiscale dans une économie fédérale dynamique où les gouvernements décident de la taille du budget et de son allocation entre les dépenses en éducation publique et en infrastructures. On découvre qu'une pleine centralisation des taxes et des dépenses est optimale quand la productivité des infrastructures est la même dans toutes les régions. Quand les différences ne sont pas trop importantes, une centralisation partielle est optimale. En présence de fortes différences, la pleine décentralisation est optimale. La production nationale en régime permanent tend àêtre la plus élevée quand il y a pleine décentralisation. On fournit une justification pour les résultats mixtes obtenus quant à la conjoncture de Oates, en montrant que la pleine décentralisation est préférable à la décentralisation partielle même si elle est inférieure à la décentralisation complète.