Demand Estimation for Solar Photovoltaics in the United States – An Instrumental Variable Approach

Abstract

Worldwide the demand for solar photovoltaics (PV) has increased significantly over the past decades. This was driven by a price reduction for solar PV systems. A two-stage least squares linear regression yields insights into the price sensitivity for residential customers in the U.S., and California in particular. The specification includes instrumental variables as well as fixed effects to account for the common issues of endogeneity and data heterogeneity in demand estimation problems, respectively. The variation in the sales tax rate on solar PV and the movements of polysilicon spot prices are used to instrumentalise PV price changes. The regression results imply an inelastic demand with a long-term price elasticity of -0.443, accounting for differences over state and time. Investigating price elasticities for various income groups shows that lower-income customers react more strongly to price changes compared to those with relatively high income (-0.521 vs. -0.195). Likewise, regions with lower population density are more sensitive to price changes (-0.473 vs. -0.338). Besides price, installation costs and technological efficiency majorly impact the system size installed. Results of this study can provide data-driven guidance to efficient policy design and pricing strategies.

Keywords: Price elasticity; solar photovoltaic; instrumental variables; demand estimation.