Does city or state make a difference ? The effects of policy framing on public attitude toward a solar energy program

he transition in replacing fossil fuel with renewable energy occupies a prominent place on the climate policy agenda in the U.S. and around the globe (Feiock & Coutts, 2013; Ramaswami, Tong, Fang, Lal, Nagpure, Li,... & Chertow, 2017). However, the road to energy transformation in the U.S. has been quite bumpy, in part because it depends on altering individuals’ behavior toward participating in programs, investing in and using new technologies, and changing daily behaviors. One approach that is applied widely in efforts to change attitudes and encourage these actions is to inform citizens about the nature of the policy programs through particular policy frames (Skocpol, 1992; Levin, Schneider, & Gaeth, 1998; Kallbekken & Salen, 2011). By framing policy issues in a particular way, a government can strategically highlight the nature of a given policy program to shape the public’s perception of policy problems, direct resources available for political mobilization, and arrange the policy agenda for specific policy solutions (Mettler & Soss, 2004; Callaghan & Schnell, 2009; Mettler & SoRelle, 2014; Stokes & Warshaw, 2017). Consistent with this line of thought, framing can increase the public’s support for the policy program, which in turn translates into their willingness to alter their behavioral intentions. The nature of policy programs is multifaceted (Walker, Lee, James, & Ho, 2018). Policy outcomes play a profound role in an individual’s evaluation of policy programs’ effectiveness and result in positive or negative attitudes toward governmental agencies or the programs themselves (James & Olsen, 2017; Deslatte, 2019). A T V V Abstract: Replacing fossil fuel with solar photovoltaic (PV) technology is critical to the transition to renewable energy and thus a key feature of the contentious and often confusing policy debates surrounding climate change. Governments can frame such environmental issues in various ways, but consensus is lacking on whether economic or environmental benefits most effectively encourage pro-environmental attitudes and behaviors, such as supporting PV technology. In this study, we introduce a moderator—psychological distance between citizens and policy outcomes—to elaborate this relation. Based on the federalism literature, we suggest that different levels of government, as the policy implementers, represent a sense of distance. The construal level theory (CLT) is adopted, and we hypothesize that the congruency between psychological distance and the construal level of policy outcomes will increase policy support and willingness to pay for solar PV installation. The results of survey experiments on Amazon Mechanical Turk (MTurk) offer partial support for our theoretical expectations and add new insights. Practical and theoretical implications are discussed.

key question is whether economic (e.g., Spence, Leygue, Bedwell, & O'Malley, 2014;Grillos, 2017) or environmental benefits (e.g., Bolderdijk, Steg, Geller, Lehman, & Postmes, 2013;Raymond & Delshad, 2016) are the major determinants that persuade people to adopt pro-environmental attitudes and behaviors. To date, previous studies have provided inconclusive and sometimes contradictory findings with respect to how the perception of economic or environmental benefits influences citizens' attitudes and behaviors.
This article seeks to fill this lacuna in the efforts to influence citizens to act to support energy transition. Specifically, we investigate the conditions under which a specific policy-outcome frame will lead the public to support a solar PV program and to be willing to pay for the system themselves. We argue that, within the U.S. federal system, a governmental agency at a specific hierarchical level could elicit a sense of proximity/distance from the citizen. Further, building on construal level theory (CLT), we expect that a positive attitude would be elicited when the process of abstraction (from an object's concrete to abstract construal) matches the psychological distance between citizens and officials (Trope, Liberman, & Wakslak, 2007;Trope & Liberman, 2010). To investigate this proposition, a series of survey experiments are conducted through Amazon's Mechanical Turk (MTurk), and the empirical evidence partially supports the expectations we outline below. The theoretical and practical implications for policy framing to accelerate energy transition processes are then provided.

Policy Framing and Public Attitudes toward Sustainability
Extant research has demonstrated that policy framing plays an essential role in eliciting public support for diverse policy issues, such as climate change and energy (Chong & Druckman, 2007;Fletcher, 2009;Wiener & Koontz, 2010;Jones & Song, 2014). Policy frames could be either policy attributes or goals/outcomes (Levin et al., 1998). In this study, we focus on outcome framing for two reasons. First, outcome bias is prevalent (Baron & Hershey, 1988). Previous studies on environmental policy have emphasized the importance of policy outcomes when the government intends to encourage pro-environmental attitudes and behaviors (for a review, see Drew & van den Bergh, 2016). Moreover, studies in public policy and management have revealed that the disclosure of policy outcome information, analogous to performance information, can affect public perceptions of agencies or programs (e.g., James & Van Ryzin, 2016Nielsen & Moynihan, 2017;Deslatte, 2019). Hence, it is reasonable to examine the way different outcome framings affect perceptions of particular programs.
In environmental policy, it is inconclusive whether economic or environmental benefits elicit policy support more effectively. Some argue that individuals are willing to take actions when motivated by self-interest, such as acquiring economic benefits (Lindenberg & Steg, 2014;Spence et al., 2014) or facing lower financial barriers (Rabe, 2004;Lane & Potter, 2007;Caird, Roy, & Herring, 2008;Lanzini & Thogersen, 2014;Grillos, 2017). Alternatively, recent studies have found that climate concerns and benefits could outweigh the monetary cost of mitigation actions (Tobler, Visschers, & Siegrist, 2012;Bolderdijk et al., 2013;Steinhorst, Christian, & Matthies, 2015;Raymond & Delshad, 2016). We seek to shed new light on these contradictory findings by introducing an important moderator-the psychological distance between citizens and policy programs.
Evaluation of policy outcomes is constructed socially, and the public's sense of distance from policy programs can affect their opinion of governmental performance. Experimental studies have found that public assessment of policy programs is contingent upon party identity (e.g., James & Van Ryzin, 2017;Gromet, Kunreuther, & Larrick, 2013), especially when the source of information is either in-party (socially closer) or outparty (socially distant) (e.g., Slothuus & De Vrees, 2010;Lyons & Jaeger, 2014). However, partisan-motivated reasoning could be tempered by various issues that affect an individual's interests directly or indirectly (Mullinix, 2016), suggesting that the sense of distance from policy programs might influence their evaluation.
Extending this work, we focus on the implementation level (whether city or state government) and the corresponding policy outcomes (whether it produces environmental or economic benefits). Later, we will advance our hypothesis by integrating federalism studies with a psychological theory-the construal level theory (CLT)-to examine how the relation between economic-environmental benefits' framing and the city/state level affect policy attitudes toward solar PV policy.

City/State as Moderator
As noted previously, a policy program could be framed as either a higher-level abstract construal or a lowerlevel concrete construal. According to CLT, a higher level of construal focuses on an object's more primary and abstract property, while a lower level of construal relates to its relatively secondary and more concrete characteristics (Trope, Liberman, & Wakslak, 2007;Trope & Liberman, 2010). Consider an environmentally friendly product that could be framed as either reducing adverse environmental effects (a higher level of construal) or saving an individual money (a lower level of construal). Positive judgments and behaviors derive from the congruence or "fit" between the levels of construal and psychological distance, which is a subjective, egocentric perception based on either temporal, spatial, social, or hypothetical dimensions (Trope, Liberman, & Wakslak, 2007;Trope & Liberman, 2010). Using a high-level construal could ensure a more accurate description of an object at a distance, since the contextual information is limited. Fit could, therefore, elicit processing fluency (Lee & Aaker, 2004). In contrast, it becomes difficult to evaluate objects that include misfit information, such as assessing a product's economic appeals while thinking about life one year from tomorrow (Goldsmith, Newman, & Dhar, 2016).
In addition, people would "feel right" about the means to the end when the fit between the construal level and psychological distance is maintained. When the means used to attain the goal are suitable for the orientation of activities, a value or "functional advantage" of goal pursuit could be derived (Fessel, 2011;Higgins, 2000). People would then recognize that certain actions would be more feasible and effective in achieving the defined goals in a specific circumstance (Henderson, Wakslak, Fujita, & Rohrbach, 2011;White, Macdonnell, & Dahl, 2011). This insight has been applied to individuals' judgments and behaviors, such as intentions to purchase eco-friendly products (Goldsmith et al., 2016), pro-environmental behaviors and attitudes (e.g., Brügger, Morton, & Dessai, 2015;Chang, Zhang, & Xie, 2015;Sacchi, Riva, & Aceto, 2016), and decision-makers' evaluations (Burgoon, Henderson, & Wakslak 2013).
In the U.S. federal system, we argue that different levels of government could entail a sense of distance between the government and citizens and, because of this perception, citizens would prefer a particular level of government to take the main responsibility to achieve specific policy goals. Public perceptions of intergovernmental policy responsibility could be rooted in part in the nature of policy problems (Konsiky, Milyo, & Richardson, 2008;Schneider, Jacoby, & Lewis, 2010). Individuals may think federally when considering the geographic scale of environmental problems, (Jacobs, 2017;Konisky, 2011) or ensuring policy outcomes "close to the people" (Kelleher & Yackee, 2004). In addition, attitudes toward politics, such as partisan identification or political trust, could likewise affect preferences on the assignment of policy responsibilities (Wolak & Palus, 2010;Connolly, Klofstad, Uscinski, & West, 2019;Dinan & Heckelman, 2020;Wolak 2016). For instance, citizens are prone to defend state authorities against federal interventions, due to greater trust in the state government (Kam & Mikos, 2007). More importantly, citizens may prefer levels of governments of which they feel more in control. This occurs because more opportunities for civic engagement in a given jurisdictions are able to empower citizens (Craig, Niemi, & Silver, 1990;Ostrom, Tiebout, & Warren, 1961). Empowered citizens would then adopt a belief that "…individual political action does have or can have an influence upon the political process," i.e., external political efficacy (Campbell, Gurin, & Miller, 1954, pp.187). Higher external political efficacy suggests reduced power asymmetry and social distances between officials and citizens (Magee & Smith, 2013). As a result, we hypothesize: H1: Citizens' attitudes toward Solar PV policy and their willingness to pay to install solar PV will be influenced by psychological distances and economic/environmental benefits of the policy program. H1a: Citizens exhibit more support for the Solar PV policy and more willingness to pay to install the solar PV when the policy evaluation the state government publishes focuses on environmental relative to economic benefits. H1b: Citizens exhibit more support for the Solar PV policy and more willingness to pay to install solar PV when the policy evaluation the city government publishes focuses on economic relative to environmental benefits.

Experimental Design
To test these hypotheses, we designed a 2 by 2 factorial experiment with two factors: The implementing government and type of policy outcome. Experimental protocol is presented in Figure A1. The treatment is a constructed solar PV policy program, "Go to Solar," based on the Department of Energy 2007 policy report (see Appendix Table A1 for the complete text and wording). The sense of psychological distance is manipulated in two ways. First, the source of information indicates the level of government that is implementing the solar program. When it is the state government, our goal is to manipulate the perception that the policy program is distant from the respondents. In contrast, city government gives the perception of a proximate source. Secondly, the specific area the policy program affects is identified. If the information source is the state government, the influence will be framed as statewide. Otherwise, it will be citywide.
To evaluate the construal level's effect, we vary the policy outcomes presented randomly as economic or environmental. With respect to economic benefits, we provide the information that each solar PV panel could save one family $433 annually in energy expenses, and in the aggregate, a certain amount of money beneficial to the state/city ($433*5000 households=$2,165,000). Environmental benefits are the annual reduction in CO2 emissions and the aggregated reduction is described as equivalent to a corresponding number of tree seedlings grown in 10 years. 1 We have two dependent variables: policy support and willingness to pay. The respondents' tendency to support the "Go to Solar" program is measured on a 7-point Likert scale that ranges from strongly support (7) to strongly oppose (1) (M=5.94, SD=1,12, Skewness=-1.33). Respondents are also asked to express their willingness to pay 100% of the expense to install the PV system themselves. If they are unwilling to do so, the following question asks the minimum amount of financial subsidy that would be sufficient for them to change their mind, with responses that range from "10%" to "not willing to change my mind even with more than a 51%" subsidy. Combining these two, the measure of willingness to pay ranges from 1 ("not willing to change my mind even with more than a 51%") to 7 (pay 100% by themselves) (M=5.04, SD=2.29; Skewness=-0.47). All of the measures are presented in Table A2 in the Appendix.
In general, the respondents recruited from Amazon's MTurk (N=576) were attentive to our treatment. 2 Among them, 51% answer the two factual manipulation check questions corresponding to our two treatments correctly (see Table A4). 3 Later, we ran both a t-test and the OLS model with the respondents who passed the FMC questions. 4

Policy Support
Figure 1 presents the results of our hypothesis tests that congruence between outcome framings and psychological distances elicits more policy support for the solar PV program. The result does not support our expectation that manipulated environmental (vs. economic) benefits at the state (vs. city) bolster policy support. In fact, the opposite is the case. The respondents expressed a greater level of policy support when offered statewide than citywide economic benefits (t=2.65, Effect size=0.44). Further, the respondents prefer state governments to implement the program, which could produce economic benefits relative to environmental benefits (t=1.83, Effect size=0.31). With respect to city governments, the respondents appear to support the solar PV program when it is intended to achieve environmental benefits, but the difference is not statistically significant.
We also run a series of regression analyses to test the hypothesis formally, and Table 1 reports parallel results. 5 As noted, the implementing government and the type of benefits does not affect policy support solely, but jointly. Model 1 shows that the interaction effect is statistically significant and positive, but the main effects are not statistically significant, as we anticipated. Even after including the demographic variables of political party, race, income, and U.S. residents in model 2, the results remain the same. Among the control variables, as one might expect, Democrats are more likely to support the solar PV program relative to independents. 6

Figure 1 Two-way Interaction Effect in the Mean Rating of Policy Support
Notes: Circles describe the mean of policy support and bars describe 95% confidence intervals. Unlike the finding of the regression model on policy support, after demographic variables are included, the interaction effects between implementing governments and policy benefits become insignificant (see Table  1). We also found that white and U.S. respondents tend to report less willingness to pay. Overall, the results support the interaction between construal level and psychological distance in willingness to pay, but only in part. It is worth noting that the effect size of our policy framing overall is relatively low and, given the relatively smaller sample size, the analysis may be underpowered.

Follow-up Experiments
Because the findings are not fully consistent with our theoretical expectations, two follow-up studies were conducted to provide further explanations. 7 Some might argue that framing the proposed solar program explicitly as a mitigation action in our first experiment would by default arouse an expectation about environmental benefits. Hence, the program with economic appeals would become more attractive due to the promise of co-benefits in climate change policy. If so, we expect that once climate change framing is removed, the respondents would prefer the program with environmental appeals rather than that economic appeals.
The findings of the second experiment suggest the possibility that climate change framing could add value to the program with economic benefits. In the second experiment, we recruited 810 U.S. citizens from MTurk 8 and asked them to report policy attitudes toward a solar PV program with either economic or environmental benefits. Table 2 shows that the solar PV program accompanied by environmental (vs. economic) benefits gains more support. Specifically, when respondents receive the treatment of economic benefits, policy support decreases by .62 units. 9 However, respondents are indifferent to the specific level of government that should obtain these benefits. Considering willingness to pay for PV installation, the direction of the framing effects remains the same as the results of the first experiment. However, the differences are not statistically significant.

Figure 2 Two-way Interaction Effect for the Willingness to Pay
Notes: Circles describe the mean of policy support and bars describe 95% confidence intervals. Others might contend further that the psychological distance of local government may be social instead of geographic. Hence, we decided to account for this variance by measuring political efficacy. 10 Using treatments similar to those in Experiment 2, we again recruited 644 U.S. citizens from MTurk. The results largely replicate the findings of Experiment 2 and confirm the moderating effect of political efficacy. Respondents seem to perceive distance across both levels of government. On average, local political efficacy is greater than that of the state across all groups (see Table 3). After including three-way interaction terms among implementers, policy benefits, and local political efficacy in the OLS model, the coefficient is negative and statistically significant in the case of policy support and willingness to pay. Specifically, when the respondents feel the local government is closer to them, their support for the city clean air program becomes stronger (see the right panel in Figure 3 (a)). However, this effect is not as intense for the state program. With respect to the job creation program, the state government gains more support from respondents with greater local political efficacy, and the slope of this effect is slight in the case of the city program. The results are comparable in the case of willingness to pay (see Figure 3 (b)). When the respondents' local political efficacy increases, the city (vs. state) clean air program increases their willingness to pay as well. In contrast, willingness to pay is more likely to be motivated because of the state (vs. city) job creation program. Largely, these findings indicate that when the respondents feel closer to the city government, they prefer the city (vs. state) program with environmental appeals and the state (vs. city) program with economic appeals.

Figure 3 Three-way Interaction Effect on Policy Support and Willingness to Pay
(a)Policy support (b)Willingness to pay

Discussion and Implications
It has been long debated whether economic or environmental benefits are more effective in eliciting pro-environmental attitudes and behaviors. To shed light on the inconsistent findings in previous research, this article examines the effectiveness of policy outcome framing with respect to solar PV installation in the context of American federalism. We suggest that individuals use specific levels of construal when the economic/environmental benefits the solar PV system offers are highlighted. Further, people demonstrate policy support and willingness to pay by associating the policy benefits with a specific level of government that is psychologically farther or closer to them as an appropriate implementer.
Our experimental results suggest a more nuanced story regarding the effect of policy framing on public attitudes toward environmental policy. In general, outcome framing seems to change support for a solar PV program while having more mixed effects on willingness to pay. When the solar PV program is framed with co-benefits for reducing climate change impacts, policy support could be induced by presenting statewide (vs. citywide) economic benefits. Once the climate change framing is removed, environmental (vs. economic) benefits are more likely to influence policy support, regardless of which level of government implements the program. On the other hand, our findings also indicate that the willingness to pay to install a solar PV system seems to be elicited by presenting citywide (vs. statewide) economic benefits with climate change framing. However, this effect is insignificant when demographic variables are included in the regression model. Furthermore, without climate change framing, there is no significant difference in the influence of city/state framing on willingness to pay. Taken together, these findings suggest that outcome framing can shape public attitudes toward environmental policy programs to different degrees, depending on policy context. It seems that the public appears to care less about the distribution of intergovernmental policy responsibilities as long as the desired outcomes are outlined and achieved. Nevertheless, it is worth noting that the effect of such outcome-determined mindsets may not be overwhelming. There are heterogeneities within the treatment groups. Existing scholarship has demonstrated that political trust could motivate citizens' "safeguards of federalism" in the Federal-State relation (Kam & Mikos, 2007). Here, we expand this argument to the statelocal relation by measuring local political efficacy. Our results suggest that when the respondents believe local political efficacy is high, the city program that promises to deliver environmental benefits, and the state program that offers economic benefits, would be perceived as the preferable option. Specifically, citywide (vs. statewide) environmental benefits are more likely to encourage a willingness to pay on the part of people with higher local political efficacy. Taken together, although citizens may be ill-informed about the legal and administrative principles of governance arrangement, they appear to acquire a pragmatic understanding about the federalist system in their own ways and accordingly, choose appropriate implementers to serve their interests.
Finally, this article illustrates the value of the application of construal level theory to federalism studies. As we demonstrated, the psychological distance between government and citizens is important in the distribution of policy authorities across levels of government, and local political efficacy is by no means the only measure. Future work could continue to explore the ways in which different dimensions of psychological distance affect policy attitudes. Further, the message of policy outcomes in our experiments is more general. It may be the case that national and local officials could acquire more policy support by providing case-specific or general statistical statements on policy programs. Finally, the determinants of behavioral intention could differ from those of real behavioral change, so more work should be performed to unveil the motivation underlying energy transition.

Notes
1. This number is calculated based on the DOE greenhouse gas equivalent calculator: https://www.epa.gov/energy/greenhouse-gas-equivalencies-calculator. 2. The survey was conducted via Qualtrics in March 2019. Only MTurkers with a higher reputation (over 95% Approval Rate) are recruited, and are paid $1 after they finish the survey. After excluding responses from the same IP address, we have 576 observations. Demographic variables appear to be balanced across treatment groups, except for income level, political party, and race (see Table A3 in Appendix).
3. Although each treatment group has different correction rates, the Chi-squared test shows no significant difference in the distribution of the correction rate across groups (χ 2 4 = 4.40, p>.1). 4. Following some scholars' suggestions (Angrist, 2006;Aronow, Baron, & Pinson, 2019;Gerber & Green, 2012), we also employ the instrumental variables method and analyze the full samples. By and large, the findings are quite similar (see Table SA4 in the online supplementary material). 5. Given this dependent variable's skewness, we also run the ordered logit model for both policy support and willingness to pay to check the result's robustness (see Table SA3 in the online supplementary material). Largely, the results are identical. 6. Some may have a concern that political party affiliation affects the assignment of intergovernmental policy responsibilities and attitudes toward environmental policy. Hence, we conduct a subgroup analysis, and the results are relatively similar to the overall pattern (see Table SA1and SA2 in the online supplementary material). 7. Arguably, the solar PV program is targeted to reduce adverse environmental effects. That said, one might argue that if economic benefits are perceived to be the primary feature, policy support would be greater in the case of the statewide (vs. citywide) programs. The results of a pre-test (N=223 U.S. participants from Amazon's MTurk) support that economic benefits tend to be a concrete (feasible), but secondary feature of the program, while environmental benefits are primary, but abstract (See Supplementary B). 8. In this experiment, we employ two-wave data collection processes, because after collecting the first wave sample, we realize that we may not have sufficient statistical power. It appears that the results of the twowave sample may not exert great effects on our result, given the insignificant coefficient of the dummy variable for two-wave respondents in the regression analysis (see Table 2). Detailed information is reported in Supplementary C. 9. Similarly, this mean difference between groups shows that the respondents are more likely to support a solar PV program that could improve air quality, rather than one that increases job opportunities (MEnv/city=8.02, MEco/city=7.25, t=-2.93; MEnv/state=7.80, MEco/state =7.20, t=-2.16; see Table SC 11 in the online supplementary material). 10. We gauge political efficacy by asking two questions: The extent to which "each level of government cares what people like you think" and the extent to which "ordinary citizens like you can do a lot to influence the governments at different levels" (See Table SD2 in the online supplementary material). The Cronbach's alpha yields a reliability coefficient of 0.83 for the responses to state political efficacy and 0.86 for those to local political efficacy. Hence, we average the responses to these two questions. Taking the difference between local and state political efficacy, we are able to detect political efficacy's moderating effect.

1.Local Economic framing
The City mayor of one of your neighboring city published one solar policy evaluation report. Please read it carefully and answer the following questions:

Policy Background
To address the negative impacts of carbon emission, such as air pollution, sea level rise, flooding, the city government has launched a voluntary program, "Go to Solar". This program aims to increase the rate of photovoltaic (PV) installations to build energy-efficient homes.

Policy Program
Within the City, there are 5,000 voluntary households participating in the pilot program. Each household installs 40 photovoltaic panels in a 4.1-kW power system.

Policy Benefits
The city-wide impacts of "Go to Solar" program are positive and promising.
Each PV system provides a household around 6,400 kWh of electricity per year.
Each PV system saves a family $433 a year on their energy bills.
In aggregate, the installation of the PV system directs more than $2,165,000 dollars into city economy annually.

2.Regional economic framing
The State governor of one of your neighboring state published one solar policy evaluation report. Please read it carefully and answer the following questions:

Policy Background
To address the negative impacts of carbon emission, such as air pollution, sea level rise, flooding, the city government has launched a voluntary program, "Go to Solar". This program aims to increase the rate of photovoltaic (PV) installations to build energy-efficient homes.

Policy Program
Within the City, there are 5,000 voluntary households participating in the pilot program. Each household installs 40 photovoltaic panels in a 4.1-kW power system.

Policy Benefits
The city-wide impacts of "Go to Solar" program are positive and promising.
Each PV system provides a household around 6,400 kWh of electricity per year.
Each PV system saves a family $433 a year on their energy bills. In aggregate, the installation of the PV system directs more than $2,165,000 dollars into State economy annually.

3.Local Environmental framing
The City mayor of one of your neighboring city published one solar policy evaluation report. Please read it carefully and answer the following questions:

Policy Background
To address the negative impacts of carbon emission, such as air pollution, sea level rise, flooding, the state government has launched a voluntary program, "Go to Solar". This program aims to increase the rate of photovoltaic (PV) installations to build energy-efficient homes.

Policy Program
Within the State, there are 5,000 voluntary households participating in the pilot program. Each household installs 40 photovoltaic panels in a 4.1-kW power system. Policy Benefits The state-wide impacts of "Go to Solar" program are positive and promising. l Each PV system provides a household around 6,400 kWh of electricity per year.
Each PV system reduces around 4-5 metric tons CO2 emission per year. In aggregate, the installation of PV system leads to a City-wide reduction of around 25,000 metric tons annual CO2 emission reduction, equivalent to 413,380 tree seedings grown for 10 years.

4.Regional environmental framing
The State governor of one of your neighboring state published one solar policy evaluation report. Please read it carefully and answer the following questions:

Policy Background
To address the negative impacts of carbon emission, such as air pollution, sea level rise, flooding, the state government has launched a voluntary program, "Go to Solar". This program aims to increase the rate of photovoltaic (PV) installations to build energy-efficient homes.

Policy Program
Within the State, there are 5,000 voluntary households participating in the pilot program. Each household installs 40 photovoltaic panels in a 4.1-kW power system.

Policy Benefits
The state-wide impacts of "Go to Solar" program are positive and promising.
Each PV system provides a household around 6,400 kWh of electricity per year.
Each PV system reduces around 4-5 metric tons CO2 emission per year. In aggregate, the installation of PV system leads to a state-wide reduction of around 25,000 metric tons annual CO2 emission reduction, equivalent to 413,380 tree seedings grown for 10 years.    Table SB3: Importance and feasibility of policy outcomes of solar PV installation Table SB4: Paired t-test for the priority of policy outcomes as the reason for supporting solar policies Table SB5: Paired t-test for policy outcomes more likely to be obtained Table SB6: Assignment of policy responsibility to the state and city government Table SB7: Assignment of policy responsibility to the state and city government Table SC1: Treatments Table SC2: Variable descriptions  Table SC3: Descriptive statistics and randomization check Table SC4: Factual manipulation check Table SC5: Descriptive statistics and randomization check of the first wave data Table SC6: Factual manipulation check of the first wave data Table SC7: Policy attitude of the first wave data Table SC8: Descriptive statistics and randomization check of the second wave data Table SC9: Factual manipulation check of the second wave data Table SC10: Policy attitude of the second wave data Table SC11: Policy attitude of the first and second wave data Table SD1: Treatments  Table SD2: Variable descriptions  Table SD3: Descriptive statistics and randomization check Table SD4: Factual manipulation check Table SD5: Policy attitude across treatment groups