Assessing potential for policy feedback from renewable energy incentive programs

The study examines whether and how financial incentive programs for residential solar photovoltaics (PV) generate policy feedback through the electorate that could entrench and expand decarbonization policies. Drawing on policy feedback theory, the paper asks if beneficiaries of solar incentives form effective political constituencies, and how volume, partisan composition, and geographic distribution across electoral districts shape their feedback potential in a polarized political environment. The context is the prominence of consumer-facing incentives in U.S. energy transition policy and the Inflation Reduction Act (IRA). The authors propose that effective feedback depends on: (1) existing policy preferences of partisan elites, (2) the number of beneficiaries, (3) the partisan makeup of beneficiaries, and (4) their distribution across electoral districts. They apply this framework to New York State’s PV incentive program, assessing both the scale of adoption and the partisan balance of adopting households across U.S. Congressional, New York State Assembly, and State Senate districts.

Policy feedback theory posits that enacted policies reshape political attitudes, interests, and behaviors, thereby influencing subsequent policy contests. Historical cases such as Cold War defense spending and the Social Security Act illustrate how broad, geographically dispersed material benefits can entrench supportive constituencies and make retrenchment politically costly. The literature distinguishes resource effects (material benefits and traceability of program benefits that heighten beneficiaries’ stakes) from interpretive effects (political learning and perceptions shaped by program experience and elite cues). In a polarized and nationalized U.S. political landscape, elite party positions heavily guide local officials, and negative partisanship can limit cross-party uptake of program legitimacy even when local benefits exist. Renewable energy incentive programs have played key roles in scaling EVs and residential PV. Prior work finds modest partisan gaps in PV adoption (slightly favoring Democrats) and greater political activity among adopters, but often neglects the political geography of adopters across actual electoral districts. The paper integrates these strands, emphasizing how partisan elite preferences and the spatial distribution of beneficiaries condition feedback potential.

Data integration: The study links three administrative datasets at the household level using a probabilistic record-linkage algorithm (Enamorado et al., 2019): (1) NYSERDA records of residential PV projects receiving state incentives; (2) a statewide property file compiled from public tax assessment rolls (ATTOM Data Solutions) identifying residential properties and ownership; and (3) the New York State voter registration file providing party registration for each registered voter. Geographic coordinates for properties/projects are used to assign households to U.S. Congressional, NY State Assembly, and NY State Senate districts. Precinct-level 2020 presidential election returns (MIT Election Data and Science Lab) are aggregated to obtain district Republican/Democratic vote shares; party of the locally elected official (2020) is identified per district. District-level covariates (population density and its square, homeownership rate, median income) come from the 5-year 2020 American Community Survey. Sample: The NY program supported 142,046 household PV installations statewide. Due to data quality and conservative linkage choices, 91,459 projects (69% of projects with complete address information) are reliably matched to unique property records; household-level analyses use these matches. For the small share (3.3%) of projects without street addresses, Zip codes assign projects to districts; when Zip codes span districts, projects are fractionally allocated. Analytic strategy: (1) District-level volume analysis models the count of PV installations per district as a function of Republican presidential vote share, with and without controls (population density and its square, homeownership rate, median income), using negative binomial regression across each district type. Visualization includes OLS trend lines for descriptive plots. (2) Partisan composition analysis constructs three matched datasets (one per district type) by pairing each PV-adopting household with a randomly selected non-adopting household from the same district. Differences in partisan shares between adopters and non-adopters are summarized by district (plots limited to districts with ≥200 installations for display). A linear probability model regresses an indicator for PV adoption on household-owner party affiliation shares (Democrat and Republican, each split by primary participation; unaffiliated; Independent; minor parties: Working Families, Conservative, Green, Libertarian; other), using unregistered owners as the reference, to quantify how representative adopters are of district electorates. The goal is descriptive composition, not causal effects. Measurement notes: Conservative matching prioritizes accuracy over coverage to minimize party-affiliation misclassification. The linkage errors are assumed unrelated to partisanship; unmatched cases reduce sample size but should not bias partisan comparisons. District assignment uses precise coordinates when available; fallback Zip-based assignment may introduce minor measurement error.

- Volume across districts: Descriptively, higher Republican presidential vote share is associated with more PV installations across U.S. Congressional, NYS Assembly, and NYS Senate districts; many of the highest-deploying districts are Republican-led. However, negative binomial models show that after adjusting for population density (and its square), homeownership rate, and median income, the positive association disappears and becomes negative (statistically significant only for NYS Assembly). A curvilinear density effect is observed, with suburban, middle-density districts showing the highest installation volumes; dense urban and very rural districts have fewer installations. Homeownership rates positively predict installations (significant in NYS Assembly). - Partisan composition within districts: PV adopters are disproportionately Democrats relative to randomly sampled non-adopters from the same district. In most districts—including the majority of Republican-led Congressional, Assembly, and Senate districts—Republicans are underrepresented among adopters. Republicans are underrepresented in 76% of all districts, including 63% of Republican-led U.S. House districts, 63% of Republican-led NYS Assembly districts, and 55% of Republican-led NYS Senate districts. In many districts the Democratic overrepresentation among adopters exceeds 0.10. The largest observed district-level difference shows adopters at 0.46 Democratic versus 0.22 among non-adopters (difference 0.24) in a Republican-led Congressional district. - Linear probability model estimates (relative to unregistered owners): Democrats have much higher probabilities of adoption. Primary-voting Democrats: +0.15 to +0.17; non-primary Democrats: +0.10 to +0.11. Republicans are the least likely among registered partisans to adopt, with small positive coefficients vs. unregistered (reflecting that unregistered owners are least likely overall), and little difference between primary and non-primary Republicans. Independents and unaffiliated owners fall between Republicans and Democrats. Left-leaning minor parties (Working Families, Green) exhibit among the highest adoption probabilities, and Libertarians are also high, comparable to Working Families in some models; Conservative Party affiliates resemble Republicans. - Despite the Democratic skew, Republican adopters remain a substantial minority in many Republican-led districts: they comprise at least one-third of adopters in 5/8 Republican-led Congressional districts, 22/43 Republican-led NYS Assembly districts, and 9/20 Republican-led NYS Senate districts. Implication: There is strong potential for a sizable pro-solar constituency in Republican-led districts due to volume, but the within-district Democratic skew may complicate positive policy feedback among Republican elected officials.

The findings address whether consumer-facing renewable energy incentives can generate supportive constituencies that help entrench and expand decarbonization policy. Politically, Republican-led districts often have higher volumes of PV beneficiaries (largely due to favorable housing stock and suburban density), suggesting potential leverage where elite Republican preferences are least supportive. Yet within those districts, beneficiaries skew Democratic, potentially limiting pressure on Republican incumbents whose primary electoral threats come from co-partisans. This pattern raises the risk that incentives are interpreted as disproportionately benefiting Democrats, reinforcing polarization and dampening cross-party policy reinforcement. Nonetheless, the presence of a nontrivial share of Republican adopters indicates opportunities for bipartisan constituencies, particularly if program communications emphasize economic benefits and local gains. For national-scale policies like the IRA, managing interpretive effects—through framing, visibility of benefits, and equitable geographic distribution—will be crucial to avoid backlash and to cultivate supportive coalitions in pivotal districts.

The study contributes a framework for assessing policy feedback potential that integrates four elements: partisan elite preferences, the volume of beneficiaries, the partisan composition of beneficiaries, and their distribution across electoral districts. Applying the framework to New York’s residential PV incentive program shows that Republican-led districts house large numbers of beneficiaries, but adopters within districts skew Democratic. This mixed pattern suggests both opportunity (large beneficiary pools in pivotal districts) and challenge (limited co-partisan pressure on Republican officials). The work refines prior evidence on adopters’ political activity by showing stronger primary-voting effects among Democrats than Republicans, implying a somewhat less potent Republican pro-solar constituency. Future research should extend analyses beyond New York to states with different partisan balances and policy designs, directly measure changes in political attitudes and behaviors among beneficiaries to close the policy feedback loop, and examine the role of organizations and movements in mobilizing beneficiaries into effective policy coalitions.

- External validity: Single-state focus (New York), a predominantly Democratic state; dynamics may differ in states with different partisan contexts and program designs. - Behavioral outcomes not measured: The study does not test whether receiving benefits changes political attitudes or behaviors; it focuses on constituency composition and distribution. - Data and linkage constraints: Only 69% of projects with complete addresses were matched to unique properties (91,459), prioritizing accuracy over coverage; unmatched cases reduce sample size but are assumed unrelated to partisanship. About 3.3% of projects lacked street addresses and were assigned by Zip code, introducing minor geographic measurement error. - Observational design: District-level associations may be confounded by unmeasured factors beyond the included covariates; LPM results are descriptive of composition, not causal effects. - Visualization thresholds: Some descriptive plots exclude districts with fewer than 200 installations, potentially down-weighting patterns in low-adoption districts.