Suchergebnisse

Policy analysis, the scientific evaluation of policy impact, must include both the technical transformation and political decision process. This analysis is plagued by limited data that leads to model uncertainty. Not only the derived models are uncertain, but also political decision-makers have to deal with this. They form simple mental models, policy beliefs. Therefore, a political economy equilibrium framework, the CGPE model, is developed. The CGPE models the political and economic system together and allows the disentanglement of political performance gaps into knowledge and incentive gaps. Structural model uncertainty is handled by a large simulation sample, while for parameter uncertainty, a MCMC sample is derived. A distributed simulation tool has been developed. A metamodeling approach is applied to model the transformation of economic growth into outcomes. Sector-specific policy impact functions are estimated using observational and expert data in a Bayesian estimation framework. We applied this framework empirically to the case of the CAADP in Ghana, Senegal, and Uganda. Indicators for key sectors, key policies, and optimal policies are derived, and the impact of model uncertainty on them is assessed. A theoretical framework for measurement and evaluation of participatory network structures is developed. The network generating process is estimated using exponential random graph models, and separate measures for lobbying and informational influence are derived. By combining this, individual policy beliefs are estimated, and their political performance gaps are measured and disentangled into knowledge and incentive gaps. The central results of these applications are: In the technical evaluation, model uncertainty is important, as derived messages can change dramatically. Designing efficient participation structures is hard. Beyond biased incentive gaps, biased beliefs are important for the observed political performance gaps. Using different constitutional and participation scenarios, we show that such a design does not solve the performance gaps. A way out are transdisciplinary research approaches, as they connect the science world with the society, and in doing so, new knowledge is generated.