Association for Public Policy Analysis & Management

Policy enacted to shift from the status quo to a desirable future state is inherently forward looking; thus, ex ante predictive modeling is central to the design of effective policy. For policies that support residential solar photovoltaic (PV) diffusion, this means predicting how changes in the individual-level adoption decision-making context impact the growth of installed capacity. However, since residential solar PV adopters are often wealthier than non-adopters, a common criticism of policies that support residential solar PV diffusion is that they regressively redistribute public funds. When the opportunity presents to make a mid-course correction, policy makers and program designers face a dilemma – how to increase installed capacity while managing an equity imbalance.

Solar PV adoption decision-making is complex. Both economic aspects (e.g. prices & rebates) and informational aspects (e.g. information exchanges among individual decision-makers) each play a large role in the structure and evolution of diffusion. Forming ex ante expectations of the impact of policy changes requires predictive models that account for both economic and informational aspects of the decision-making context. We compare two approaches that incorporate these aspects in predictive modeling of solar PV diffusion, each with embedded tradeoffs: a top-down, equation-based approach that deals with information flows and individual decision-making in the aggregate and a bottom-up, generative approach that models them explicitly. We exploit the variation across these approaches to investigate how tradeoffs in predictive modeling methods impact the prediction of future solar PV diffusion when the adoption decision-making context experiences a large shift. By comparing the two approaches on a common footing, we contribute to an understanding of the role model choice plays in determining policy recommendations.