In recent years, the formulation and effectiveness of subnational policy systems in late-industrializing countries such as Brazil and Spain were substantially affected by the transformation of states involved in "dual transitions" - simultaneous conversions to market-oriented reforms and democratization during the 1980's and 1990's. These changes decentralized political authorities and fiscal resources to subnational governments, shifting previously centralized models of industrial policy to regional administrators. Although decades of geographically uneven patterns of industrialization gave subnational government in Brazil and Spain sufficient reason to intervene in regional economies, these administrators had little experience with market-oriented industrial policies. How could subnational governments in these unevenly developed economies build and maintain regional industrial policies in the wake of dual transitions?

My argument is that subnational industrial policy systems are built due to the identification of the political interests of subnational leaders with technocratic solutions to regional economic problems. However, once these institutions are created, their maintenance and performance depend upon the development of an array of horizontal ties among public policy agencies and organizations in political society (unions, parties, and business associations) and vertical associations between development agencies and private firms. The diversity of horizontal ties allows subnational development agencies to accumulate political and technical resources needed to maintain the policy system even when the political support of regional leaders wanes. Through vertical ties public agencies develop an understanding of the economic problems of private firms, thus facilitating the focusing of policy resources more efficiently on areas of production, external economies and quality control that will most enhance the primary and secondary effects of investment. Vertical ties also enhance accountability of both firms and public agencies by creating monitoring functions that prevent mutual shirking. These horizontal and vertical associations (collectively termed embeddedness) increase the ability of industrial policy to address market failures (e.g., scarcity of capital, information asymmetries, and unexploited externalities) which inhibit improvements to the level and quality of regional industrial development. In subnational cases where horizontal ties are weak and political interests subjugate development concerns, vertical ties will malfunction and policy outcomes will generate negative or negligible results.

These arguments are tested in six subnational comparative case studies - three cases in Brazil (Minas Gerais, São Paulo, and Rio de Janeiro) and three cases in Spain (Asturias, Madrid, and Andalucía). Spain and Brazil were selected for their similarities and their differences. The similarities - the regional unevenness of state-led models, the exhaustion of those models and the turn to market-oriented industrial policies during the 1980's, and political decentralization linked with democratization - place both countries in the universe of cases of late industrializers with shifting states in uneven markets. Differences in democratization, relative fiscal constraints, and their position in global markets should make Spain a candidate as a "strong state", a more stable democracy and a relatively more developed "success story" than the case of Brazil. Despite these differences, the subnational comparisons show that the factors in my argument explain common outcomes in regional development systems. These similarities suggest that structural factors affect the nature and performance of subnational industrial policy, but do not determine the outcome. Specifically, the most successful subnational policy systems among the six subnational cases in both countries were those with the highest levels of policy embeddedness.