Panel Paper: The Effects of Policy on Market Structure in the U.S. Residential Solar PV Installation Industry

Thursday, November 8, 2018
Jackson - Mezz Level (Marriott Wardman Park)

*Names in bold indicate Presenter

Eric O'Shaughnessy, National Renewable Energy Laboratory


Thousands of firms compete to install residential solar photovoltaic (PV) systems in the United States. The installation industry’s structure varies across local markets. Some markets comprise many small-scale local installers, other markets comprise a few large-scale installers, and most markets are some mix of both structures. What accounts for differences in local PV market structure, and what is the role of policy? This study econometrically tests these questions using a rich data set of installed residential PV systems. Study data are from the Lawrence Berkeley National Laboratory’s Tracking The Sun data set, the largest extant source of U.S. PV data consisting of more than one million systems. The effects of policy on market structure are estimated through regressions of policy variables on the distribution of local market shares, known as market concentration. The models control for various non-policy market structure determinants such as latent demand. Preliminary results suggest that more favorable policy environments yield less concentrated markets, meaning markets with more installers and relatively even distributions of market shares. This preliminary result accords with economic theory that favorable policies increase perceived profitability and induce market entry. The preliminary result indicates that local policies may foster local PV installation industries, suggesting that local economic development may be a co-benefit of policies designed to increase PV adoption. However these potential policy effects may not be purely beneficial, given findings that market concentration can, under certain circumstances, yield lower PV prices. The implications of the effects of policy on PV market structure, deployment, and prices are explored.

Full Paper: