Performing Department
Resource Economics
Non Technical Summary
Residential solar power is an important technological innovation that holds promise for a cleaner energy future. Out of 2.5 million households in the state of Massachusetts, those who installed solar photovoltaic(PV) systems grew from a mere 14 households to 60,465 households between 2010-2017. Between 2015-2017, the residential installations are growing at an even higher rate of 50% (Data source: Massachusetts Department of Energy Resources). It is crucial to understand what factors are determining the household decisions in the process of adopting the solar PV system.Each potential adopter faces a crucial decision--to lease or to own the panels. Leasing the panels from a solar company incurs little to no upfront cost. A fixed lease payment, however, reduces the net saving from the electricity generated by the panels. On the other hand, owning solar panels includes a high immediate cost with a higher return in future periods resulting from the absence of lease payments. Moreover, solar owners can receive additional revenue from selling Solar Renewable Energy Certificates (SREC). The price of SRECs is determined in the marketplace for such certificates and is subject to significant price fluctuation. The tradeoff is obvious. Owning is risky and involves high upfront cost, but can potentially yield higher future returns from SREC revenues. Leasing is risk-free with no front-end payment, but lessees forgo SREC revenues. The choice between these two options naturally depends upon the decision maker's tendency to tradeoff payment over time (known in behavioral economics as "time preference") and tolerance towards risk (namely "risk preference").In this study, we propose a conceptual model that considers the household's costs and benefits in both owning and leasing options. More importantly, the model takes into accountdecision makers' preference for delayed return and risky outcomes. We then empirically examine the connection between an individual's behavioral preference and their own-vs-lease choice. To implement this project, we plan to recruit solar homeowners from Massachusetts and measure their time and risk preferences using state-of-the-art decision tasks recently developed in the field of behavioral economics. Participants of the study will receive real monetary payoff based on their own decisions. This "revealed" preference approach may provide a more accurate measurement of individual preference than the stated preference approach. We then link homeowners' responses in these decision tasks with information from a detailed survey on their solar system specifications and history of energy usage.The choice of owning versus leasing has significant policy relevance. Currently, among the households that have solar PV systems in Massachusetts, 68% are lessors and 32% own their systems. State governments, such as that of Massachusetts, provide many incentives on top of the federal tax credit in hopes that more residents would adopt solar energy. The return to such public investment varies depending on the ownership of the panels. The state government prefers solar ownership since its residents fully retain the benefit of the subsidy. Leasing panels from a national solar installer, on the other hand, channels part of the subsidy benefit outside of the state. By understanding how individual behavioral traits determine the decision between owning and leasing the residential solar system, we will be able to suggest ways to increase the public "bang-of-the-buck" for solar incentives at the state level.In addition, the findings in this study have broader relevance for the adoption of other green technologies such as battery storage and electric vehicles that can either be bought or leased.
Animal Health Component
0%
Research Effort Categories
Basic
100%
Applied
0%
Developmental
0%
Goals / Objectives
The primary goal of this study is to link behavioral traits, specifically risk and time preferences, to own-vs-lease decisions in the context of solar panel adoption. The insights obtained from this study will help us better tailor incentives and marketing strategy towards solar ownership at the state and local level.A secondary goal of the project is to provide support and training for two early-career Ph.D. students (Ming Ge and Emma Grazier) and allow them to learn the literature and techniques of experimental and environmental economics through providing research assistance to the two principal investigators.
Project Methods
This project uses three primary methods from the discipline of economics: theoretical modeling, data collection using field experiments and econometric analysis.A formal model will be developed adopting the general framework used in the literature of lease-vs-own decision modeling. We will modify the general model to incorporate unique costs and benefits faced by decision-makers in the context of solar adoption (e.g. risky SREC market revenue, etc). More importantly, to investigate individual's choice under risk and time delay, we expand beyond the classical expected present value approach and allow the choice to be governed by maximizing the utility function with flexible behavioral parameters (e.g. Constant Relative Risk Aversion and Prelec one-parameter discount factor). By doing so, our model will be a generalization of the expected-present-value types of models.Experimental data has been a well-established source for testing theories and for the design and analysis of public policies (Smith 1982). Field experiments collecting data from the non-student population is especially valuable as it provides stronger confidence in the external validity of the drawn conclusions. We filed a public record request to the Massachusetts Department of Energy Resources to gain access to a list of all solar homeowners residing in the state of Massachusetts. Additional key variables in the same dataset include (1) system capacity, (2) pre-tax-credit installation cost, (3) timing of the installation and (4) whether the system is owned or leased by the resident. The last variable is particularly important since it is the choice variable we are interested in. Since we intend to conduct the experiment at a campus computer lab, we choose to focus our recruiting on the sample of 2,646 residential addresses in Hampshire county, all of which are in short driving distance to UMass Amherst. For sufficient statistical power, we intend to recruit 200 households for this study.During each experimental session, we will measure each participating household's behavioral traits using two sets of incentivized choices--one for risk preference and one for time preference. Risk preference elicitation asks an individual to choose between paired lottery options with outcomes at different risk levels. The preferred lottery reveals the level of risk tolerance of the decision maker. Time preference elicitation offers two payment amounts with different time delays. The patient individual would prefer the higher later payment than lower sooner payment. We design these incentivized tasks so it is optimal for individuals to reveal their true preference since non-optimal choice bears economic costs. These so-called "incentive-compatible" tasks are considered to produce a more accurate measure of behavioral traits than un-incentivized survey questions (Charness et. al 2013). We believe eliciting preferences with incentivized tasks is a formal, replicable, and relatively inexpensive means of measuring individual behavioral traits. To link such measures with field behavior is especially valuable in understanding environmental decisions such as the solar own-vs-lease choice.At the completion of preference elicitation tasks, each participant will be asked to fill in a detailed survey providing information on their solar system's specifications, the forecasted return from the solar installer, the history of actual energy generation and usage, as well as their household's social-demographic characteristics (age, income, etc). This information will be used as control variables in the regression analysis.All of the experiments will be conducted in the Cleve E. Willis Experimental Economics Laboratory in the Department of Resource Economics at the University of Massachusetts Amherst campus. The experiment interface uses Z-tree, a software developed specifically for economic experiments and employed widely in the field of experimental economics (Fischbacher 2007). The preference elicitation tasks have already been designed and coded by Ming Ge, a graduate student in the Department of Resource Economics. We expect the prototype to be approved soon by the Institutional Review Board of the University of Massachusetts Amherst.Individual choice data collected from the preference elicitation tasks yields both parametric and non-parametric measures of risk aversion and patience for each individual. The preference measure can be combined with the household and solar system information and analyzed with appropriate statistical techniques. Specifically, with binary dependent variables (own==1, lease==0), a standard approach is to use a logit or probit regression model. Alternatively, we can design a hazard model that takes into consideration of the timing of installation. Statistical packages such as Stata have pre-programmed routines to estimate these models, and the investigators have substantial experience with these statistical techniques.