Household Solar Analysis for Policymakers: Evidence from U.S. Data.

• Author: Best, Rohan

1. INTRODUCTION

   1.1 Motivation

   Understanding the drivers of household solar-panel uptake is potentially challenging due to the diversity in prior-study results. For instance, there are mixed results for explanatory variables such as income and age (Alipour et al. 2020). Education is often investigated and only sometimes has a significant impact on solar-panel uptake (Muller and Trutnevyte 2020). There are many studies finding that race or ethnicity have significant effects on solar uptake (Alipour et al. 2020), although these apparent impacts are potentially partly explained by omitted socioeconomic controls. Clarifying the relationships for the many debated influences on solar-panel uptake has major potential to add value to the policy-design process.

   A further factor which is often absent from empirical analysis of actual solar-panel uptake is some measure of wealth. This is especially the case for studies using household-level data. There is likely to be a strong link between key economic variables related to accumulated resources and the uptake of household solar panels. Wealth effects have a long history of analysis in economics (Modigliani 1966).

   There is major potential for emissions reduction benefits if policymakers can further promote household solar-panel uptake. Understanding of the drivers of household solar uptake in the United States is valuable, given its large share of global economic activity and related emissions. While emissions reduction through support for household solar uptake has historically been expensive (Hughes and Podolefsky 2015; Crago and Chernyakhovskiy 2017; Gillingham and Tsvetanov 2019), there is potential for more cost-effective support if policies can be better targeted and as the cost of solar panels falls. For example, cost reductions following high solar uptake in Australia have contributed to relatively cost-effective emissions reduction (Best et al. 2019a).

   Policymakers can also enhance management of another key policy issue, household energy affordability, through improved understanding and targeted policy support for solar-panel uptake. A reduction in energy insecurity related to the affordability of energy, due to solar-panel uptake by constrained households, can be a major co-benefit of environmental policy targeting emissions reduction. While some benefits are offset by rebound effects (Qiu et al. 2019), household solar panels can contribute to substantial reductions in dimensions of household energy insecurity through lower use of grid-supplied electricity and less financial stress related to energy bills (Best and Burke 2019).

   1.2 Contributions

   The combination of analysis of a large household survey and inclusion of housing values as a wealth component is a key contribution of this paper. Many studies of actual solar-panel uptake use data aggregated at regional levels, and most do not control for any wealth-related variables. Recent Australian studies show that wealth is a key variable for actual uptake but not intentions (Best et al. 2019b; Best et al. 2021a), although these studies did not focus on education, race, or respondent age. There is major value in comprehensive household analysis of the United States. Household studies can enhance understanding in conjunction with valuable aggregated studies, which are especially useful for explanatory variables that do not vary substantially within an aggregated region.

   The empirical contributions of the paper include the use of multiple methods with a larger household sample than most prior studies. These multiple methods include logit, probit, and linear probability models. This is useful to ensure robustness of the results. The paper also uses entropy balancing to motivate causal interpretations.

   The key findings of this study differ to many prior studies. One possible explanation for this difference is the more comprehensive analysis of economic variables in this paper with a large household survey. This paper identifies respondent age and wealth, as measured by housing values, as two key drivers of solar-panel uptake. Education, race, and income are shown to be insignificant in most cases when comprehensive control sets are included. Interaction analysis supports the key findings.

   The key findings are potentially relevant for policy management. There are a range of policies that can be used, such as upfront subsidies, feed-in tariff subsidies for ongoing production of solar energy, loans to address the upfront capital constraint, or exemptions from property taxes. To promote efficiency, any of these policies could be targeted to households who may only decide to install solar panels if they receive policy support, as opposed to households who can afford to install solar panels even without policy assistance. This paper helps policymakers in targeting policy assistance. It reveals which variables could be efficiently used for targeting policy and which variables are likely to be less efficient. Our findings suggest that wealth is a key variable such that policy support for households with low wealth would be addressing a key capital constraint.

2. DEBATED INFLUENCES ON HOUSEHOLD SOLAR UPTAKE

   2.1 Education

   Education could conceivably have a positive influence on solar-panel uptake if knowledge of the benefits of solar panels is restricted to those with higher levels of education. Alternatively, education may not be important, if knowledge of the impact of solar panels in reducing electricity bills and pollution emissions is sufficiently broad across society. This theoretical ambiguity motivates empirical analysis.

   Education is one of the most commonly included variables in the solar uptake literature and it is often found to have a positive and significant impact (Alipour et al. 2020). But there are also many instances where education coefficients are found to be insignificant (Alipour et al. 2020; Miller and Trutnevyte 2020). It is possible that apparent effects of education could be explained by other omitted variables, such as variables related to wealth.

   2.2 Age

   Age is another variable where impacts on solar uptake are theoretically ambiguous. Younger households may be more willing to adopt innovative technologies. But solar panels have become more common over the last decade and may no longer be seen as a new technology. Further, younger households may follow more transient housing patterns that reduce incentives for investments in their current residences. As a result, it is possible that households with older occupants may have higher solar uptake. The relationship between age and solar uptake could also be non-linear, as the upper end of the age distribution may see lower uptake where these households are less likely to make forward-looking housing decisions covering multi-year time horizons.

   Prior studies have found a negative relationship between age and solar uptake about the same number of times as compared to studies finding a positive relationship (Alipour et al. 2020). This empirical ambiguity could relate to numerous factors, such as omitted wealth variables or the absence of an approach to consider possible non-linear effects.

   Figure 1 suggests that there is a mostly monotonic and increasing relationship between age and solar uptake. This is evident for the full sample which is labelled as "All". There is a hint of a negative relationship at the upper end of the age distribution. There are lower values for solar uptake proportions for key groups. For households with respondents whose education level is below bachelor's level, there are lower solar uptake proportions for each age. This suggests that there might be an impact of education on solar uptake. However, income appears to have a larger impact compared to education. Housing values below median appear to be even more important than income in leading to lower solar-panel uptake.

   2.3 Race

   Race could be an important influence if there are different average preferences across racial groups. Further, it is possible that some racial groups may face additional constraints. For instance, it has been suggested that there is racial and ethnic injustice in solar uptake (Sunter et al. 2019). Another possibility is the absence of racial effects in cases where solar uptake is based on market transactions which may be independent of race.

   Variables related to race have been assessed for a possible impact on solar uptake in a minority of studies (Alipour et al. 2020). When included, race is often found to be a significant contributor to solar uptake variation (Bollinger and Gillingham 2012; Kurdgelashvili et al. 2019; Mildenberger et al. 2019; Sunter et al. 2019; Alipour et al. 2020). Prior studies tend to be based on aggregated data or focus on variables other than wealth.

   The possibility that race impacts on solar uptake is raised by Figure 2, which compares solar uptake for "White alone" and "Black alone" groups across income deciles. The solar uptake proportion is higher for the "White alone" group, regardless of the income decile. This might mean that studies that control for income may still find a significant impact of race on solar uptake. But Figure 2 only accounts for one economic variable, namely income, and does not control for other economic variables such as housing values.

   Figure 3 also considers the impact of race on solar uptake across deciles for housing values. Housing value is the respondent's estimate about how much the residential housing property would sell for, if it were for sale. Housing-value deciles (HVD) 5 to 10 are standard deciles, while HVD1 includes 38% of the sample without housing values (renters). Figure 3 shows that the solar uptake proportion is higher for the "Black alone" group in some cases. The mixed outcomes in Figure 3 suggest the possibility that race may not have a significant impact on solar uptake. Rather, apparent effects of race on solar uptake may be driven by higher proportions of...