Presentation on theme: "Peer Effects in the Diffusion of Solar Photovoltaic Panels Marketing Science Science-to-Practice Initiative Bryan Bollinger and Kenneth Gillingham."— Presentation transcript:
Peer Effects in the Diffusion of Solar Photovoltaic Panels Marketing Science Science-to-Practice Initiative Bryan Bollinger and Kenneth Gillingham
Motivation: Social Contagion A causal peer effect is necessary if marketers wish to utilize the social spillover effect, in which marketing activity directed at one person then has a spillover effect to their peers.
Application - The Diffusion of Solar Photovoltaic Panels (PV) Policies to promote adoption of solar photovoltaic (PV) panels have been gaining momentum throughout the world. – In the US: 30% solar energy federal investment tax credit. – In California: $3.3 billion 10 year California Solar Initiative (CSI). The nature of the peer effects will determine how to optimally allocate marketing effort by firms and policymakers.
Accelerating Empirical Adoption Rate for Solar PV in California
Clustering of Installations
The Method Bollinger and Gillingham (2012) estimate causal peer effects by determining the effects of a difference in the local installed base of solar panels on the difference in the adoption rate for different zip codes. By using differences, the authors control for both static and time-varying unobserved factors in the zip code which lead to different adoption rates (see paper for details).
Findings #1 Causal peer effects exist in the diffusion of solar panels. For the average zip code, an additional installation increases the probability of an adoption in the zip code by 0.78 percentage points. There is significant heterogeneity in the peer effects across zip codes which can be partly explained by demographics.
Findings #2 The peer effect is slightly increasing over time. This indicates that firms and policy markers may be effectively leveraging them in their marketing efforts.
Findings #3 Peer effects in the diffusion of solar panels work even more efficiently at a smaller geographic scale (the street-level). Using a Bass model to explain diffusion separately for each zip code does a good job at estimating the peer effect (the coefficient of imitation) but significantly over-estimates the market size due to the significant acceleration of adoption.
Key implications #1 Once weve obtained a better / more precise estimate what can we do with it? – Identify low hanging fruit (regions where peer effects will work hard for you to maximize the impact of initial marketing efforts) – Allocate resources more efficiently – Design product attributes that are conducive to stimulating peer effects (for solar, installers will post signs to increase the visibility of completed installations)
Key implications #2 For what kinds of products should we expect to see peer effects? – Experience goods which have low trialability (see Rogers Five Factors), increasing the value of Word-of-Mouth – Visible products that communicate the consumers values (in this case, environmental stewardship)
Summary Causality of peer effects is important in allocating marketing resources. Estimating causality is a challenge. Causal peer effects can be identified with rich enough data on the timing of adoption (Ideally, experiments could be used to identify and measure causal peer effects.) Peer effects can be effectively utilized in marketing efforts.