participants - 12th Edition Artificial Economics Rome 20th-21st September 2016

12th Artificial Economics Conference
co-located with Social Simulation Conference


Paper submission: April 18th 2016

Acceptance notification: June 6th 2016

Early registration: July 15th 2016


20th-21st September 2016

Fiscal Transfers and Regional Economic Growth

Herbert Dawid, Philipp Harting and Michael Neugart


Recent developments in the European Union have highlighted the need to gain a better understanding of the effect of inter-regional fiscal transfers in an economic union which consists of regions that have independent fiscal policies and are characterized by heterogeneities with respect to productivity, skill endowments and growth rates. Taking into account the feedback between technological evolution in a region and (global) competitiveness of local producers as well as the resulting (fiscal) revenues, effects of fiscal measures are hard to predict in such a setting. 
In this paper we employ a two-region agent-based macroeconomic model (Eurace@Unibi) to study these feedbacks and their implication for fiscal policy design. The model captures heterogeneity of firms with respect to the quality of their physical capital and of workers with respect to their specific skills. Specific skills of a worker increase over time if her employer uses technology that is above the worker’s current specific skills. Productivity of a firm is determined by complementarity between the (evolving) quality of its physical capital stock and the specific skills of its workforce. Physical capital is available in different vintages and the firms’ vintage choice when investing depends on a comparison of the firms expected returns from different vintages. Due to this, diffusion of new vintages in a technology is driven by the level of investment in physical capital undertaken by firms, which depends on the demand dynamics, and the firms’ vintage choice, which is strongly influenced by the specific skills in the firms’ workforce. The model has been shown to reproduce a large set of empirical stylized facts on different levels of aggregation. 
A setting of the model is considered where the economy consists of a large region and a small region, where the average quality of the physical capital endowment and the specific skills are initially substantially larger in the large region. Firms from different regions globally compete on product markets but labor markets are assumed to be local. In the default setting fiscal policy is regional and policy makers in each region finance unemployment payments through taxes on household income and firm profit in that region. 
Due to the endogeneity of firms’ investment and technology choice the model dynamics in this default setting exhibit a widening of the initial gap between the regions leaving firms in the small region at a persistent competitive disadvantage. The resulting large unemployment in the small region implies that the government in that region runs persistent deficits and eventually has no longer access to credit. In the absence of fiscal transfers between regions this results in severe tax increases or cuts in public spending. 
Starting from this base scenario it is studied how fiscal transfers from the large technologically more advanced region to the small region affects technological change and growth in both regions. In particular, also the net effect of the policy for the large region which is the source of the transfers is analyzed. Different variants of the policy are compared and it is also studied how such a policy can be combined with technology policies which foster through targeted investment subsidies the choice of high technologies by firms in the laggard region. This policy analysis is based on systematic simulation studies based on large sets of batch runs under the different policy scenarios. Dynamic policy effects are estimated using dynamic statistical models based on penalized spline methods. Furthermore, the economic mechanisms driving the policy effects are carefully analyzed employing time series of micro- and meso-level variables in the simulation data.