Social acceptance of new energy technology in developing countries: A framing experiment in rural India

Energy Policy 113 (2018) 466–477

Contents lists available at ScienceDirect

Energy Policy journal homepage: www.elsevier.com/locate/enpol

Social acceptance of new energy technology in developing countries: A framing experiment in rural India

T

⁎

Michaël Aklina, , Chao-Yo Chengb, Johannes Urpelainenc a

University of Pittsburgh, United States University of California, Los Angeles, United States c Johns Hopkins University, United States b

A R T I C L E I N F O

A B S T R A C T

Keywords: Social acceptance Technology Survey experiments Rural development Renewable energy India

While the literature on the social acceptance of new technologies focuses on industrialized societies, concerns about new technologies are often deeper and more widespread in developing countries. In a survey experiment with 3208 villagers from six major states of northern India, we examine the social acceptance of off-grid solar power as an alternative to grid extension. By randomly assigning different frames about this energy technology to the villagers, we study how concerns about the cost of sustainable energy, inequality of energy access, and the role of private business and the state shape individuals’ acceptance of off-grid solar power. We find that concerns about the role of private business and possible increases in rural-urban inequality are salient among the rural population. The rural population is concerned about (i) exploitative businesses practices and (ii) the inequality that solar power in villages is more expensive than conventional grid power in urban areas. These findings show that the social acceptance of new, sustainable energy technologies cannot be taken for granted in rural development and offer insights into the salience of different varieties of concerns.

1. Introduction How can we understand the social acceptance of new technology in developing societies? The literature on the social acceptance of energy technology is enormous but largely focuses on industrialized societies. The few studies that focus on the developing world – such as Yuan et al. (2011) on solar power, Amigun et al. (2011) on biofuels in Africa, and Mallett (2007) on renewable energy technology in general – are mostly descriptive and do not try explain patterns of social acceptance and opposition. And yet, if anything, new technologies have larger social effects in developing than in industrialized countries. On the one hand, new technologies hold greater potential to improve productivity and improve the quality of life in conditions of poverty and scarcity. On the other hand, new technologies also present a more significant change to societies that are less used to continuous technological change. Our research focuses on the introduction of off-grid solar power as a complement or substitute to traditional grid extension in rural India, where one-third of the population lives without a household electricity connection (Government of India, 2011). Some see solar energy as a promising solution to improve rural electrification in the country, but off-grid solar power is not without detractors. Compared to conventional grid electricity, the unit cost of off-grid solar power is much

⁎

higher, especially when one considers the subsidized grid electricity prices in rural India (Gambhir et al., 2012). Meanwhile, the Government of India does not have a consistent long-term program for the development of off-grid solar power, bringing uncertainty to the prospects of solar energy in the country. Although private entrepreneurs have begun to install solar technology on a commercial basis, studies show that India's rural population holds negative perceptions of the private sector (Urpelainen, 2016). With rural poverty being a lasting concern in Indian political debates (Drèze and Amartya, 2002) and, by some estimates, Indian households spending about 13.2% of their monthly expenditure on energy (Alkon et al., 2015, 3), theories of “relative deprivation” (Smith et al., 2012; Fontaine and Yamada, 2014) suggest that rural households react negatively to the possibility that they have to pay more for their electricity than their wealthier urban counterparts. We look at how different ways of framing a new technology influence public opinion toward it. In this context, to frame is “to select some aspects of a perceived reality and make them more salient” (Entman, 1993, 52). In a survey experiment, we randomly assign different descriptions of off-grid solar power to respondents to test how framing influences popular support for this technology as an alternative to traditional grid extension (Gaines et al., 2007). Because we focus on

Correspondence to: University of Pittsburgh, 4600 Posvar Hall, Pittsburgh, PA 15260, United States. E-mail address: [email protected] (M. Aklin).

https://doi.org/10.1016/j.enpol.2017.10.059 Received 7 May 2017; Received in revised form 22 August 2017; Accepted 29 October 2017 0301-4215/ © 2017 Elsevier Ltd. All rights reserved.

Energy Policy 113 (2018) 466–477

M. Aklin et al.

While distributed solar power can contribute to basic energy access in rural India, it is not without its weaknesses. Typically, the cost of purchasing one unit of electricity generated with off-grid solar technology is very high compared to the cost of grid electricity, especially after one considers the heavily subsidized electricity prices in India. The high unit price of solar electricity has two consequences. First, it means that distributed solar power is not ideally suited to generating large loads of power. Ultra-efficient technologies allow households to gain affordable access to basic services, especially lighting and mobile charging (Alstone et al., 2015), because the number of units of electricity generated is low, but the cost of solar electricity would be much higher if the households were to have a fridge or an air-conditioner. Even the cost of operating a fan would be relatively high. Second, the high unit cost of solar electricity inevitably prompts the question of equity. When poor rural households pay more per unit of electricity than their much wealthier urban counterparts, economic inequality increases. Some Indian non-governmental organizations, such as Prayas, are indeed calling for regulatory approaches to reduce inequality in electricity prices for the poor when decentralized options are used. As the Prayas Energy Group, an Indian non-governmental organization advocating for equitable electricity supply, puts it, “new policy-regulatory instruments for more equitable tariffs and innovative sustainable business models be put in place going forward” (Gambhir et al., 2012, 14).

a new technology to reduce energy poverty with potential to improve the quality of life, we concentrate on assessing the effects of negative frames. In other words, we consider the new technology a potentially valuable innovation and examine whether negative frames undermine its legitimacy among the general population in a developing society. Specifically, we test two sets of hypotheses on the origins of negative perceptions of new technology. The first pertains to the role of the identity of the actors that provide the technology. We examine how survey respondents react to frames that emphasize the role of the state or the private sector in promoting the use of off-grid solar power, expecting both frames to reduce popular support for solar power. The literature on Indian public opinion, both generally and in the context of energy technology, has found that the rural population distrusts both the state (Lal, 2006) and private business (Santhakumar, 2008; Aklin et al., 2014). The second pertains to the cost of off-grid solar power relative to subsidized grid electricity. We frame the issue first in terms of the cost implication, and then in terms of the rural-urban inequalities surrounding the introduction of off-grid solar power, again expecting both frames to reduce support for solar power. The results show that framing solar technology in light of private business involvement or rural-urban inequality provokes opposition against it among the study population. Both frames have a statistically significant negative effect on support for solar power, whereas frames emphasizing government intervention or higher prices have smaller and less significant effects. Specifically, putting the emphasis on privatization decreases support for solar power by 0.07 standard deviations, while highlighting rural-urban inequality reduces such support by 0.081 standard deviations.1 These effects, then, are substantively large. Additional subgroup analysis suggests that the results are mainly driven by households with established grid connection, households living in electrified habitations, households who have heard of solar power, and households who prefer micro-grids over regular grids.

3. Poverty, inequality, and new technology

Large swaths of rural India remain without access to reliable electricity. According to the 2011 Census of India, 400 million people lived in a household that did not use electricity as its primary source of lighting (Government of India, 2011). This amounts to 67% of the Indian population; in rural areas, 45% of households were without basic electricity access. Although the numbers have improved somewhat over the past years, as the government has invested in a major electrification drive called the Rajiv Gandhi Rural Electrification Scheme (Palit et al., 2014), hundreds of millions remain without basic electricity access. Even among households with a household electricity connection to the main grid, fluctuating voltage and frequent power outages reduce the value of such access (Chakravorty et al., 2014; Harish and Tongia, 2014). Distributed solar power offers a possible solution to this problem. While reforms of the Indian power sector remain mired in political difficulties (Kale, 2004; Joseph, 2010; Aklin et al., 2014), village-level solar power can be provided at the local level. Technology options range from solar home systems to micro-grids that distribute electricity to multiple households (Chaurey and Kandpal, 2010). A typical solar power system contains a solar panel, a battery for storage, and the wiring required to distribute the power. In the 2011 Census of India, fewer than 1% of all households reported using solar power as their primary source of lighting (Government of India, 2011), but the number is rapidly increasing. Rural India now has a large number of entrepreneurs who sell solar home systems, operate micro-grids for a monthly service fee, or lease out equipment to interested rural communities.

We expect an energy-poor rural population to see new energy technology both as an opportunity and a potentially disruptive force. Given that the argument for new technology as an opportunity is straightforward, the theory building focuses on the disruptive effects of technology. We analyze the “social acceptance” of new technology with an emphasis on the negative consequences of fears about disruptive effects, defining the term according to whether “broad majorities of people tend to agree with the idea of public support” (Wüstenhagen et al., 2007, 2685).2 In practice, our empirical approach examines people's support for government subsidies for new technology – in this case, off-grid solar power. Importantly, we do not examine people's individual willingness to pay for new technology, as such decisions are only partially related to the broader issue of social acceptance. We borrow from Assefa and Frostell (2007, 69), who consider “fear” or “concern” to be central elements of the social acceptance of new technologies. Defined as an “unpleasant feeling of perceived risk or danger, real or not,” the notion of fear captures the general public's concerns about the negative consequences of new technology. In the context of rural India, for example, fear could originate from the general public's concerns about the affordability of electric power. Hypotheses that frame solar off-grid power in terms of possible negative consequences increase the salience of the fearful reaction, and could thus turn the population against off-grid solar power. Our first set of hypotheses focuses on the possible negative effects of new technology. Drawing on a large body of literature on the social acceptance of new technologies, we hypothesize about the effects of high cost and inequalities. Although the role of these factors in shaping public opinion has not been studied in the context of widespread rural poverty, the theoretical expectations are clear: concerns with cost and inequality should reduce popular support for new technology. In economic sociology, this approach falls under the category of those that emphasize the characteristics of a technology in explaining the diffusion of innovations (Wejnert, 2002). The second set of hypotheses focuses on the role of the state and private actors in the introduction of new technology. New technologies

1 These estimates are based on models with village fixed effects. Results are virtually the same if state fixed effects are used instead.

2 Wüstenhagen et al. (2007) also offer more restrictive definitions focusing on “community acceptance” and “market acceptance.”

2. Solar energy in rural India

467

Energy Policy 113 (2018) 466–477

M. Aklin et al.

technologies can increase inequality in the short run by driving a wedge between economic opportunities available to the privileged and to the marginalized (Galor and Tsiddon, 1997). For example, Gartrell (1977) finds that income inequality within communities has promoted the adoption of new agricultural technologies in Andhra Pradesh, a result suggesting that inequality could actually contribute to innovation. In their analysis of the role of government policy in alleviating inequality in India, Drèze and Amartya (2002, 343) also emphasize that the new opportunities afforded by international openness could contribute to inequality unless policies and institutions allow the poor to benefit from these opportunities. In the case of off-grid solar technology, however, the implications of standard techno-economic models are not applicable because the primary users of new technology would be relatively poor rural households. Unlike many other technologies, off-grid solar power is a technology for the poor and the deprived – households that cannot afford a grid electricity connection. In this context, then, the question is whether these households would support a technology that is costly for the poor. To the extent that people are sensitive to their deprived status relative to rural and urban middle class and the elites (Gurr, 1970; Korpi, 1974; Smith et al., 2012), one would expect the poor to be more supportive of new technologies that do not add to inequality. Dugoua and Urpelainen (2014, 1728) show that, in five African countries, non-electrified households are more concerned about their situation if they live close to power lines: “relative deprivation both intensifies the perception of inequality and creates hope among unelectrified households for improvements.” In India, Santhakumar (2008) finds that distributional concerns are critical to understanding public opinion about electricity supply policies. His comprehensive analysis of public opinion about the privatization of electricity supply shows that the Indian population is particularly concerned about the distributional consequences of changes to the rules that govern the power sector. Here it bears emphasizing that we expect rural-urban inequality to have a negative effect on social acceptance. If people object to new technologies that exacerbate social inequities, the social acceptance of such technologies should decrease. It is still possible that people would individually use such technologies despite their aversion to them, but here our focus is on broad patterns of social acceptance in the collective consciousness of the society.

can be rolled out in a government-led program or by profit-seeking entrepreneurs. Studies show that the general public's acceptance of new technology depends to a large extent on the perceived trustworthiness and legitimacy of the suppliers of said technology. In the context of rural India, where public opinion about privatization is generally negative, such factors could play a major role. We test these hypotheses with the help of framing theory. We follow Entman (1993, 52), whose definition of framing is “to select some aspects of a perceived reality and make them more salient.” Empirically, the framing literature shows that decisions can be nudged by the way an issue is presented (Levin et al., 1998; Chong and James, 2007b, 2007a). When faced with a choice, individuals mobilize and confront various beliefs to determine their best action. Each belief can be thought of as a dimension related to the decision at stake. For instance, when considering the acquisition of a solar panel, an individual may take into account dimensions such as the financial, emotional, and social benefits and costs attached to this item. To test for the relevance of different dimensions, we can derive potential dimensions based on existing research and then test their salience through framing experiments. Levin et al. (1998) label this the “attribute framing” approach. As noted by Entman (1993), framing refers to both selecting some particular dimension and making it more salient. Thus, we can identify the most salient dimensions by emphasizing particular aspects of the new technology, such as its cost, and see how these aspects affect the respondents’ views of the technology. 3.1. Nature of technology When new technology becomes available, enthusiasm about the opportunities it provides is dampened by several factors. On the one hand, the new technology may, at least initially, be expensive and/or inadequate. On the other hand, the new technology may also raise concerns about increased inequality. The cost of new technologies plays an important role in social acceptance. When new technology is perceived to be of high cost, poor people's support for it should decline because of concerns about affordability. In a series of surveys in Germany, Zoellner et al. (2008) show that across three types of renewable energy technologies – solar, wind, and biomass – economic factors are consistently among the most powerful predictors of popular support for deployment: “if someone estimates the respective energy form to be cost effective, his or her overall evaluation of the energy form will improve.” Thus, cost considerations are a natural economic baseline for assessing the social acceptance of new technology – a yardstick for assessing the magnitude of other sources of resistance. This consideration should have more explanatory power in the context of poor rural communities. Given very low household incomes, the possibility that a large number of households would be willing to pay a high price for environmentally sustainable energy appears remote. While Aldy et al. (2012) report that American households are willing to pay extra for clean energy in a nationally representative survey experiment, it is unlikely that much poorer Indian households would put a premium on the environment. Affordability of energy is a major concern for households across the developing world (Winkler et al., 2011), and households that are lacking basic necessities will be particularly concerned about cost of living. Indeed, the 66th round of the National Sample Survey of India shows that, in 2010, the average household spent 13.2% of its monthly expenditure on energy (Alkon et al., 2015, 3). Given these considerations, it is useful to begin by evaluating whether framing a new technology as expensive reduces support for it in the society.

Hypothesis 2. Framing new technology as increasing rural-urban inequality reduces support for it. 3.2. Public and private provision The other dimension of new technology pertains to the identity of the actors that provide it. Sometimes new technologies are rolled out as a government service, with the state playing a key role in distribution. In other times, the distribution of the new technology is left to the private markets. The study of the respective roles of the state and the private sector has a long history in sociology and anthropology (e.g., Granovetter, 1985). The sociology of markets literature seeks to characterize the how the state and markets interact. In this context, Fligstein and Dauter (2007, 106) highlight the importance of “the role of local cultures.” We build off this approach by noting that people's experience with the state and markets may vary considerably across countries and affect their willingness to adopt new technologies. Both approaches have a possible downside relative to a neutral frame that does not draw attention to the identity of the actors delivering off-grid solar power. In the context of pervasive poverty and limited institutional capacity, emphasizing the role of the state in the delivery of off-grid solar power could turn the general public against the technology. When politicians fail to deliver on their electoral promises and there is a wide gap between people's expectations and the ability of the state to deliver government goods and services, frames that emphasize state leadership could provoke a backlash by

Hypothesis 1. Framing new technology as an expensive solution reduces support for it. Inequality is another significant concern in the social acceptance of new technology. In general, economic studies suggest that new 468

Energy Policy 113 (2018) 466–477

M. Aklin et al.

highlighting the possibility of failure. When the essential role of the state is emphasized in framing of off-grid solar power, respondents whose general confidence in politicians and the state administration is limited may react by turning against off-grid solar power.

and high levels of awareness of solar power should, all else constant, react more strongly to the frames above. Aware and interested people react more strongly to frames, because for them the new technology is a more salient possibility in the first place.

Hypothesis 3. Emphasizing the importance of government leadership in new technology provision reduces support for it.

4. Research design

The effects of emphasizing the private sector's role in solar power delivery could even result in hostility among the public. Public opinion research in developing countries usually finds a non-trivial percentage of respondents oppose and suspect the benefits of market economy and privatization. To name a few, the second round of Afrobarometer survey in 2004 asked respondents whether or not they agree with that “it is better for government to buy and sell crops, even if some farmers are served late.” More than 50% of respondents said so.3 As reported by Denisova et al. (2009), 46.7% of survey respondents in 28 post-Soviet countries opposed privatization and supported “renationalization” of previous privatized state enterprises. In a recent survey that includes 19 countries in Latin America, Latinobarómetro finds that only 37% of respondents thought privatization has been “beneficial” for their respective country.4 Research on public opinion in India also shows that most citizens have negative perceptions of economic liberalization (Chhibber and Eldersveld, 2000), and the privatization of electricity supply in particular (Santhakumar, 2008; Aklin et al., 2014). In a survey experiment in rural Uttar Pradesh, Aklin et al. (2014) show that while the provision of new information about the benefits of higher electricity prices has a strong positive effect on popular support for price reforms, it does not create support for privatization – the respondents remain adamantly opposed to privatization regardless of the framing of the issue. In the case of solar energy, a recent study in the state of Uttar Pradesh – one of our study states – shows that the inhabitants of non-electrified hamlets hold negative views of local private businesses (Urpelainen, 2016). According to these studies, concerns about the private sector relate to fears of exploitation, abusive business practices, and even outright scams. Therefore, framing off-grid solar power as a private sector effort could turn the general public against new technology.

We test our four hypotheses through a survey experiment conducted in India. Survey experiments consist of randomly administrating treatments to respondents. By treatment, we mean the intervention that we randomly administer to each respondent. Here, the treatment is a piece of information that frames the issue in a particular way (Chong and James, 2007b). The advantage of using an experiment over observational data is that the effect of confounding factors is ruled out by design. As a result, we can estimate the effects of treatments without worrying about control variables. Our experiment is based on a comprehensive energy access survey of 8568 rural households in 714 villages across six energy-poor states of Northern India – Bihar, Jharkhand, Madhya Pradesh, Odisha, Uttar Pradesh, and West Bengal (Aklin et al., 2016a, 2016b). These include four of India's five largest states. They are contiguous and all located in the north, where energy poverty tends to be more prevalent than in the southern states. According to the Census of 2011 (Government of India, 2011), the state-level percentage of electrified rural households in these six northern states ranges from 10.4% (Bihar) to 58.3% in Madhya Pradesh. Though these numbers underestimate the level of rural electrification in a time of rapid improvements brought about by the Rajiv Gandhi Rural Electrification Programme (RGGVY), electricity access remains scarce there. Moreover, according to the second round of the India Human Development Survey (IHDS-II), on average rural households in the states included in our survey with electricity connections only had 11.9 h of access to electricity on a typical day – less than onehalf of a continuous power supply. In Bihar, the average was as low as 6.2 h. In each state, we randomly selected one district per division (the highest administrative unit under a state).5 The likelihood of being selected is weighted by the relative share of the district's population in the state; more populated districts are, therefore, more likely to be chosen for our survey. In each district, we then randomly selected 14 villages. While our survey is not representative at the national level, it is representative at the state level. To pool our observations in our analysis, we built survey weights to account for the disparity in population size across the six states. For a full description of the survey design and sampling, see Section A1 in the appendix. Together with the main survey, we administered the survey experiment to one-half of the sampled villages, leaving us with 357 villages. Because we have a total of nine possible treatment combinations, we conducted the survey experiment with nine villagers within each village. After removing five villagers who declined to participate, the total sample consists of 3208 villagers. Our estimation equation takes the following form:

Hypothesis 4. Emphasizing the importance of private sector participation in new technology provision reduces support for it.

3.3. Heterogeneous framing effects To gain a deeper understanding of why people react negatively to frames that emphasize potential disruptive effects, we also theorize about how socio-economic status modifies framing effect. We identify four key cleavages in the society: ownership of conventional technology, potential access to conventional technology, attitudes toward technological alternatives, and prior awareness of new technology. In our context, ownership refers to having a household grid connection and access refers to living in a habitation that has been electrified. Because these two concepts are closely related, we expect them to have similar effects. Specifically, we expect people who have a grid connection or could easily gain access to one to react less strongly to all the frames above. Because these people are already in the realm of conventional technology, for them the relevance of new technology is less pronounced. In contrast, for people living in habitations without electricity, connectivity with the electric grid is not an alternative. In people who do not have access to the conventional grid, frames about solar off-grid power should be more powerful. The other heterogeneity we examine relates to attitudes and awareness. People who have positive attitudes toward new technology 3 4

yi = β1 Tiprice + β2 Tirural − urban + β3 Tiprivate sector + β4 Tipublic sector + γg + ϵi (1) where y represents the outcome variable, i refers to individual households, and γg , where g ∈ {v, s} , indicates village v or state s fixed effects. The T's denote our treatments and ϵ is the error term. We include sampling weights in the analysis throughout and report standard errors adjusted for clustered sampling by village. Although villages and respondents are sampled within each district proportional to the study population, we have to use sampling weights to correct for variation in the size of each district. 5 One exception is West Bengal, where we chose two districts per division to account for the larger size of the latter.

See http://www.afrobarometer.org/ for more details. See http://www.latinobarometro.org/lat.jsp for more details.

469

Energy Policy 113 (2018) 466–477

M. Aklin et al.

participate in the experiment, with each of them receiving one of the 3 × 3 = 9 possible combinations of the treatments. Doing so ensures that all treatment combinations are evenly distributed across all villages, preventing the treatment assignment from being confounded with observable socio-economic covariates as well as other unobservable ones. The balance statistics shown below reveal that our randomization was successful. The first two treatments relate to our hypotheses on the nature of new technology. Specifically, Hypotheses 1 and 2 predict that support for a new technology – solar power – decreases with cost and with increased rural-urban inequality. In other words, people are less willing to acquire a new technology when it is costly and when it has social consequences that are undesirable. By framing the issue on these two dimensions, we can identify their salience in one's decision-making process. We contrast the effect of these frames with a situation in which we do not mention anything about the cost and consequences of solar power. Next, we examine the role of the provider of the technology. Given low levels of trust in both the government and the private market, we expect references to their role in the provision of solar power to have a detrimental effect on people's level of support (Hypotheses 3 and 4). To stack the deck against our hypotheses, we focus on the strength of both the government and markets. Regarding governments, we emphasize its ability to reach remote areas. Distant areas are often ignored by markets because they offer little prospects for lucrative business. Our private sector frame highlights that the private sector, compared with its public counterpart, tends to be more innovative and creative when it comes to financing. There is one crucial caveat that we must consider. The private sector treatment emphasizes the ability of the private sector to innovate and solve credit issues, while the dependent variable asks whether villagers would like the government to prioritize subsidizing the grid over solar power. Besides suggesting distrust and hostility, an emphasis on the private sector's innovation potential could also result in villagers ascribing less importance to the government's role. If villagers believe that the private sector can lead the introduction of solar power in India, they might simply prefer the government to focus on extending the traditional electricity grid. In the Indian context, however, this scenario is hardly plausible. Our survey data show that people overwhelmingly prefer the government to run infrastructures over the private sector. When we asked our respondents who should operate the electric grid and who should be responsible for the distribution of LPG for cooking, in both cases over 90% of respondents chose the government over private corporations. As a result, it is highly unlikely that villagers would trust the private sector as the sole agent responsible for the deployment of off-grid solar power.

4.1. Dependent variable The key dependent variable is a question about the relative importance of government support for electric grid extension (traditional technology) and solar power (disruptive new technology). Specifically, we ask the following: “Do you believe the government should put more emphasis on subsidizing solar power than extending the national electricity grid?” We provided each villager with the following five options: “Only subsidize solar power,” “mostly subsidize solar power,” “both (solar power and electricity grid) are equally valuable,” “mostly extend the national electricity grid,” and “only extend the national grid.” We then assign each option a numerical value between 0 and 4. A greater value suggests a stronger preference for subsidized grid electricity extension, that is, more opposition to solar power as an alternative.6 A key feature of this survey experiment is the presence of a genuine trade-off. In our case, we explicitly specify the resource constraint faced by government officials and present the respondents two potential objectives for government subsidies: solar power and electric grid extension. Based on our pilot study, if we had instead asked about solar power without a trade-off, we would have received an overwhelmingly positive response. If people could increase access to solar power without any cost or trade-off whatsoever, testing hypotheses about social resistance to an alternative energy technology would be impossible. Fig. 1 shows that while nearly a half of our respondents report grid electricity and solar power are “equally valuable,” others are still somewhat divided between them. The total number of villagers choosing either “only subsidize solar power” and “mostly subsidize solar power,” is relatively equal to that of households choosing “mostly extend the national electricity grid” or “only extend the national grid.” In the experiment, we also included another related question on whether the villagers support government price regulation for solar electricity generated by private entrepreneurs: “Do you agree with the statement that the government should force private companies to charge low prices for solar power, even though entrepreneurs say they cannot provide solar power at a low price?” It turned out that popular support for such measure was overwhelming, with 2225 respondents reporting strong support for such a measure and very few opposing the measure. Therefore, there is little variation in the responses that could be explained by our survey treatments. 4.2. Treatments To test our four hypotheses, we consider the effects of four different treatments after exposing all respondents to a brief description of the current status of rural electrification in India:

4.3. Split samples

“Rural electrification in remote areas is an important problem. In addition to grid extension, the government of India is using solar power to provide electricity to remote villages. Both the government and private entrepreneurs are offering solar power for lighting and mobile charging in exchange for a weekly or monthly fee.”

We conduct four split-sample tests to explore the heterogeneous treatment effects and gain a better understanding of the causal mechanisms that drive our results. We report the exact wording of the questions we used in the footnotes. First, to evaluate the role of ownership, we split our samples based on whether or not a household has grid electricity.7 While there is nontrivial variation in the quality of the grid across and within regions, grid connection remains the most straightforward way to separate households that use conventional technologies to those that do not. By asking whether respondents use the grid for lighting, we ensure that the connection is good enough for this basic usage. Second, for access, we divide our observations into two groups based on the availability of grid electricity in the respondent's habitation.8 In

We summarize the four treatments in Table 1. We create these four treatments along two separate dimensions, each of which includes two treatments. To increase the statistical power and precision of our estimates, we carried out the treatment assignment at the village level. Within each village, nine households were randomly invited to 6 In practice, governments play a dominant role in grid extension in most countries, whereas off-grid solar power is a sector populated by many private enterprises. Even grid extension, however, is the responsibility of private utilities in some countries, such as Chile, and both franchisee models and quasi-independent rural cooperatives are common (Barnes, 2007).

7 8

470

The question we used is “Do you use grid electricity for lighting? [Yes/No].” The question we used is “Is grid electricity available in your habitation? [Yes/No].”

Energy Policy 113 (2018) 466–477

M. Aklin et al.

Fig. 1. Distribution of the main dependent variable among the respondents. Note that higher values indicate more resistance to solar subsidies.

Table 1 Summary of treatments. Dimension

Treatment

Statement Details

Implementation

HIGH PRICES

“ Although solar power has many advantages, some civil society organizations complain that the high tariffs that rural households pay for solar power are too high, since levels of income are low in many villages.” “ Although solar power has many advantages, some civil society organizations complain that the high tariffs that rural households pay for solar power are unfair, since urban households gain access to electricity at a much lower cost.” Empty.

RURAL-URBAN

CONTROL Public-Private

“ One important reason for why the government plays a critical role in the promotion of solar power is that the government has the resources to reach a large number of villages even in remote areas. “ One important reason for why private companies play a critical role in the promotion of solar power is that they can develop and deploy innovative technical and financial solutions. Empty.

PUBLIC SECTOR

PRIVATE SECTOR

CONTROL

4.4. Preference for government leadership

India, a rural habitation is a cluster of households within a revenue village. Doing so creates two groups of households. One group includes households connected with grid electricity and those without grid electricity but live in the habitations with connected households. The other group, then, includes unconnected households in habitations with no grid electricity. The notion of access is similar to ownership in the sense that it denotes the proximity of a household to conventional electricity technology (the grid). The appeal of distributed solar power is expected to be highest among those who are excluded from the grid for the foreseeable future. Thus, even unconnected households may be unresponsive to our frames if they may realistically expect a connection soon. Third, the main energy access survey asks the respondents to indicate their preferences between the regular electric grid and a microgrid, following a brief explanation of what a micro-grids.9 To evaluate the role of attitudes, we separate our observations based on the reported preference. We expect interest in new technologies such as micro-grids to make people more responsive to our frames, since it denotes a willingness to experiment. Finally, to examine the effect of awareness, we split the sample according to whether the respondent reports being aware of solar technology.10 Knowledge of solar power may have two effects. First, it may indicate a personal interest in new technologies. Second, it may be correlated to higher demand for such technologies. In any case, we expect people who are aware of the technology to be more supportive of solar power.

In addition to the main dependent variable, we also inquired whether or not respondents prefer government leadership in the promotion of solar power. The exact question is as follows: Do you believe that private companies should play a major role in the promotion of solar power in rural villages, or would you prefer that government lead the effort? The responses to this question also range between 0 and 4, with 4 indicating strong support for government lead. This question is useful because it allows us to focus specifically on the role of government policy. Among other things, this approach makes it possible to consider the alternative explanation that people react to the private sector treatment because they want a division of labor between the government and business. 4.5. Summary and balance statistics We provide the summary statistics of our dependent variables, both treatments, and subgroup variables in Table 2. On average, our sampled villagers placed equal weight on subsidies for solar energy and grid electricity while somewhat preferred that government led the promotion of solar energy. For the subgroup variables, 64% of our sampled households were connected with grid electricity while 87% of them lived in villages with such access. Only 40% of them, however, prefer micro-grid over electricity from the national grid. Finally, notice that more than half of sampled respondents had heard of solar power when we conducted the experiment. This suggests that most respondents were already aware of such an alternative to the traditional electricity grid, hence justifying the relevance of our questions. For more information about the data, see Section A2 in Appendix. The balance statistics are presented in Table 3 for all combinations of treatments: the randomization was successful. Across 11 standard socio-economic characteristics, the respondents in the nine treatment

9 The question we used is: “If the monthly bill for both options were the same, would you prefer your household to be electrified through a micro-grid system or through the regular grid? [Regular grid/Micro-grid/Don't know].” 10 The question we use is: “Have you heard about solar home systems? [Yes/No].”

471

Energy Policy 113 (2018) 466–477

M. Aklin et al.

0.085 respectively. The remaining two treatments, however, are found to be ineffective compared to the control group. In the appendix (Table A1), we show that the results remain similar if we control for a range of demographic covariates. This is not surprising, given that ours is a randomized experiment. These initial findings have two important implications. To begin with, focusing on the strength of the government has little effect on people's preferences on the trade-off between grid expansion and solar subsidies. Mentioning the ability of the private sector to innovate, however, increases support for the view that the government should focus on grid extension at the expense of solar subsidies. As discussed above, this result may have two possible causes. On the one hand, emphasizing the role of the private sector in solar power development creates social resistance to the technology because of the negative perception of the private sector. On the other hand, people may prefer the government to focus on grid extension and let the private sector

Table 2 Summary statistics of outcome variables, treatments, and subgroup variables.

Outcome variables Resistance to solar subsidies Preference for government leadership Treatment I: Implementation Rural-Urban High prices Treatment II: Public vs. private Public sector Private sector Subgroup variables Ownership of grid electricity Access in habitation Preference for micro-grid Awareness of solar power

Num. Obs.

Mean

St. Dev.

Min

Max

3208 3208

2.08 3.18

1.05 1.25

0 0

4 4

3208 3208

0.33 0.33

0.47 0.47

0 0

1 1

3208 3208

0.33 0.33

0.47 0.47

0 0

1 1

3208 3208 2931 3208

0.64 0.87 0.40 0.60

0.48 0.33 0.49 0.49

0 0 0 0

1 1 1 1

Table 3 Balance statistics for all 3 × 3 = 9 combinations of treatments. The p-value is derived from the ANOVA test. When p < 0.05 , we can reject the null hypothesis that all nine means are statistically equal. The lower panel tabulates the combination of treatments and control for all groups. Experiment Groups

Age Male Education Hindu Caste Monthly Expenditure (rupee/person) Monthly Expenditure (rupee/person) Total Land Owned Business Owner Awareness of Solar Polar Availability of Grid Electricity Use Grid Electricity

1

2

3

4

5

6

7

8

9

p-value

42.85 0.84 2.20 0.87 0.76 973.72 973.7 5.69 0.21 0.57 0.88 0.64

41.73 0.85 2.37 0.90 0.79 892.83 892.8 7.25 0.16 0.61 0.85 0.64

42.56 0.88 2.40 0.89 0.81 894.83 894.8 6.29 0.17 0.60 0.88 0.63

43.07 0.82 2.20 0.87 0.81 896.67 896.7 5.68 0.16 0.59 0.90 0.64

41.03 0.84 2.33 0.89 0.76 938.56 938.6 7.13 0.17 0.61 0.87 0.64

42.77 0.85 2.30 0.88 0.79 894.26 894.3 5.69 0.15 0.59 0.86 0.64

43.00 0.87 2.27 0.88 0.81 862.25 862.3 6.89 0.18 0.60 0.88 0.65

43.66 0.88 2.23 0.88 0.79 881.84 881.8 6.57 0.14 0.58 0.87 0.64

42.25 0.86 2.29 0.91 0.75 887.47 887.5 6.69 0.11 0.61 0.87 0.64

0.37 0.20 0.29 0.74 0.37 0.50 0.50 0.43 0.05 0.97 0.79 1.00

Public-Private Implementation

PUBLIC SECTOR

PRIVATE SECTOR

CONTROL

RURAL-URBAN HIGH PRICES CONTROL

Group 1 Group 2 Group 3

Group 4 Group 5 Group 6

Group 7 Group 8 Group 9

fund solar technology expansion. The second implication of our results is that the social acceptance of solar power does not decrease because of high costs, but because of relative deprivation in the form of rural-urban inequality. While the cost frame does not change social acceptance, an emphasis on inequality creates resistance to the alternative technology, as the rural population reacts negatively to a frame that emphasizes inequality relative to urban households.

groups are statistically indistinguishable from each other. Thus, we can interpret our framing effects as causal. 5. Findings We begin with a summary of the main results. We then continue our discussion with the subgroup analysis. Throughout, we offer a graphical presentation of the average treatment effects (Sections A3–A4 contain regression tables).

5.2. Subgroup analysis 5.1. Main results We next conduct analysis over four split samples. First, we consider the distinction between users and non-users of grid electricity. In our sample, 2057 out of 3208 households are connected to the electrical grids. Because we ask our respondents to consider the trade-off between grid electricity and solar power subsidies, households that already have a grid connection may have different views. Specifically, households with established grid connection were answering our questions by considering solar power as a more efficient substitute for grid electricity. In contrast, unconnected households may answer our questions from the perspective that treats solar power as the solution to the lack of electricity connection. The results are shown in Fig. 3(a) and (b). Among connected

The main results of support for national grid extension at the expense of solar subsidies are shown in Fig. 2. We find statistical support for two of our four hypotheses. First, emphasizing the role of the private sector in the provision of solar energy increases support for the extension of the national grid over off-grid solar power. The estimated effect is statistically significant regardless of the presence of village or state fixed effects. Second, framing the introduction of solar power in the context of urban-rural inequality, whereby villagers are reminded of the higher cost they pay for solar power than urban households, also significantly increases support for grid extension. In fact, both treatments have substantively similar effects, with point estimates of 0.073 and 472

Energy Policy 113 (2018) 466–477

M. Aklin et al.

some degree, this division will be correlated with other socioeconomic factors, as the households that are connected to the grid are more likely to be informed about new technologies. Interestingly, we find that respondents who have heard of solar energy, compared with their uninformed counterparts, show a very similar level of support for grid extension – the average scores are 2.088 and 2.077 for those who have heard of solar energy and those who have not, respectively. However, while the average level of support for grid extension does not differ a lot between the informed and uninformed households, we find different factors that contribute to each group's resistance toward solar energy. On the one hand, our key findings remain similar among households that are aware of solar power (Fig. 4(a) and 4(b)). The rural-urban treatment apparently increases support for the grid, and so does mentioning the role of the private sector in the solar industry. The two other treatments, similar to the main findings, appear to have no effect on their own. On the other hand, among households that were previously unaware of solar power, the treatment on the absolute cost of solar is the most effective at inducing resistance toward solar subsidies. Again, such a difference may suggest that our main findings are mainly driven by the villagers who possess the information to evaluate the merit of solar energy as a replacement to existing grid electricity. Finally, we examine whether self-reported preference over microgrid or grid supply of electricity shapes individual's attitude toward new technologies (Fig. 4(c) and (d)). In the main energy access survey, after a brief vignette explaining what micro-grids are, we asked respondents whether they had a preference for grid or local micro-grid electricity, holding price and the quality of supply constant.11 On average, we find that respondents who prefer the conventional electrical grid expressed less support for solar subsidies (the average response is 2.207). In contrast, those who prefer micro-grid are more so (the average response is 1.885). Among people who favor micro-grids, the results mirror our other findings, namely that people are particularly responsive to frames emphasizing rural-urban inequality and the role of the private sector. The effects are of a similar magnitude, although the confidence intervals are slightly wider than those seen in Fig. 2. Among people who prefer the regular grid, only the inequality argument seems to reduce support for solar subsidies. Given that the absolute cost frame has no discernible effect on our outcome, this points to the remarkable power of inequality to shape people' preference.

Fig. 2. Average treatment effect on support for grid extension at the expense of solar subsidies, with confidence intervals. The dependent variable ranges from 0 to 4, with higher values indicating a preference for grid extension (thus against solar subsidies). The mean is 2.083 and the standard deviation 1.053. For full regression output, see Section A3.

households, the pattern observed in the previous section – the anti-solar effect of the urban-rural divide and the concern about the private sector's involvement – largely remains. The rural-urban treatment is now statistically insignificant at conventional levels, but the point estimates are still similar, suggesting that the loss of statistical significance may be caused by a reduction of the sample size. However, the results are different for the sample of households that are not connected to the grid. Both the high price and the rural-urban inequality treatments generate higher support for grid extension over solar power subsidies, although the estimated average treatment effect is not statistically different from 0 in both cases. Since these people are usually poorer than the rest of the rural population, it is not surprising to see that costs – in both absolute and relative terms – play a vital role for unconnected villagers. As these households see investment in solar technology and a grid connection as substitutes to fill in the absence of electricity supply, they are particularly sensitive to the cost frame. In brief, the results based the connected households suggest that our results are mainly driven by the consideration in which solar energy is viewed as a new option to established grid electricity. Noticeably, we obtain a similar substantive conclusion when we split our sample by connected and unconnected rural habitations (Fig. 3(c) and (d)). The next two subgroup tests differentiate average treatment effects based on villagers’ broader knowledge and attitudes toward new energy options. First, we examine whether response patterns depend on prior awareness of solar energy. We split the sample between respondents who report being aware of what solar power is and those that do not. To

5.3. Preference for government leadership We now consider people's preference for leadership in solar power development, noting that there is not much variation in the responses. Among all 3208 respondents, 2064 villagers reported an unconditionally strong preference for government leadership. Nonetheless, the results offer interesting insights, as presented in Fig. 5. Discussing the identity of the technology provider tends to increase support for government management. The private sector treatment is positive and statistically significant, and the public sector one is almost significant at conventional significance levels. The most important implication of this finding is that the effect of the private sector treatment on the primary outcome variable cannot be attributed to a preference for a division of labor between the state and the market: Highlighting the private sector's role, in fact, increases support for government leadership in solar power development. Framing the issue with respect to the cost of the technology, however, increases the support for the private market, with high prices being significantly negative and the urban-rural treatment also negative and almost statistically significant. This, in turn, suggests that 11 In the energy access survey, the vignette is as follows: “Micro-grids are systems that generate electricity and distribute it at the local level. Electricity can be produced from solar energy, micro-hydro, diesel or some other source. The electricity can then be used by local users. It provides limited but reliable electricity supply compared to the grid electricity in rural India.”.

473

Energy Policy 113 (2018) 466–477

M. Aklin et al.

Fig. 3. Average treatment effect on support for grid extension at the expense of solar subsidies, divided by grid electricity users (Split Sample A) or availability of grid electricity in the habitation (Split Sample B), with confidence intervals. These models are estimated with village fixed effects. For full regression output, see Section A4 in the appendix. Note different ranges of x-axes.

considerations in affluent societies. Perhaps surprisingly, concerns about the cost of solar power, in the absence of inequality concerns, did not provoke such strong responses. The developing society context has added significance because new technologies may hold more promise while threatening to bring greater disruption than in the context of industrialized societies. Energy demand is growing rapidly in developing countries (Wolfram et al., 2012), and the future of the global climate depends, to a large extent, on the emissions trajectories of currently poor countries. From a global policy perspective, understanding the social acceptance of clean sources of energy is thus particularly important. If technologies such as solar power can reduce greenhouse gas emissions and eradicate energy poverty, understanding and mitigating the concerns of the rural society is important for facilitating the diffusion of new, welfare-improving innovations. Most of the literature on sustainable energy technologies reports largely positive popular views, with concerns being mostly of the “not in my backyard” (NIMBY) kind (van der Horst, 2007; McAdam and Boudet, 2012). In developing countries, however, the disruptive nature of new energy technology has other dimensions. Off-grid solar power is a new, sustainable energy technology that allows villagers to circumvent the limits of India's rural electrification programs and the power sector, but our study has shown widespread concern among respondents about the role of private business and rural-urban inequality. These results suggest that efforts to promote off-grid solar power through private sector development could lead to a public backlash unless government policy effectively deals with dishonest entrepreneurs looking to abuse villagers. Perhaps more significantly, rural-urban inequalities could have negative political consequences unless the government can support efforts to improve access to

individuals have varying perceptions over price and supply. This paradox is interesting because it suggests that individuals are aware of the ability of the private sector to keep costs down. From a public policy perspective, then, the main issue is how to improve non-price related regulations of the private energy sector. 5.4. Interactive effects of treatments While our theory does not focus on interactive effects, Appendix (Table A6) reports results from models that estimate the effects of all different combinations of treatments. The table shows that there is no clear pattern of interactive effects. Regardless of how the treatments are combined, the effects of the rural-urban inequality and the private business treatments are substantial and statistically significant. The size of these effects does not depend on the presence of other framing treatments, suggesting that there are no interactive effects. 6. Conclusion and policy implications The focus on a developing country context is an important contribution of our study. New technologies are rapidly shaping the socioeconomic development of nations, but the diffusion and penetration of said technologies depends on their social acceptance. In the Indian context, technologies such as off-grid solar power can enable households to step up to the first rung of the “energy ladder” (Masera et al., 2000; Cheng and Urpelainen, 2014). By identifying concerns about private business practices and rural-urban inequalities as factors that reduce support for off-grid solar power, we have shed light on salient dimensions of the problem. These factors are notable because they are different from the conventional NIMBY or cost-effectiveness 474

Energy Policy 113 (2018) 466–477

M. Aklin et al.

Fig. 4. Average treatment effect on support for grid extension at the expense of solar subsidies, divided by awareness of solar power (Split Sample A) and (Split Sample B), with confidence intervals. These models are estimated with village fixed effects. For full regression output, see Section A4 in the appendix.

Bank, which is already promoting branded solar products in Asia – and elsewhere in the world – on a small scale.12 At the same time, tackling rural-urban inequalities would probably require more elaborate solutions. Both in India and elsewhere, subsidized grid electricity in the countryside has resulted in limited hours of supply and widespread outages, as utilities choose not to serve their rural customers because of low electricity prices (Harish and Tongia, 2014; McRae, 2015). Although the general population prefers heavyhanded regulation of entrepreneurs, such an approach would probably be impractical. A better approach would allow entrepreneurs to set their prices but support the poorest households with vouchers or direct cash transfers. India and most other countries are at the early stages of creating offgrid solar markets. Over time, the dynamics of social learning could lead to different outcomes, some positive and others negative. As offgrid solar power becomes embedded in the society, people's distrust toward private entrepreneurs could diminish if policy and markets can help households identify high-quality products. On the other hand, a continued gap between the quality and cost of electricity service in rural and urban areas could worsen rural-urban conflicts, and lead to increased demands for policies that favor rural areas. Identifying potential future problems and addressing them with timely policy implications are thus important priorities for allowing off-grid solar technologies to gain social acceptance.

affordable electricity in rural communities. Our findings raise new questions for the research program on the social acceptance of new technology. If our results generalize beyond the case of solar power in India and concerns about the private sector and inequality generate resistance to new technology more broadly, then an important frontier for the sociology of new technology is the sources of these worries. Why, exactly, do people’s respond negatively to a private sector frame; why are ruran-urban inequalities more salient than high costs themselves? To what extent do these framing effects depend on the nature of the private sector, historical rural-urban cleavages, and other contextual factors? Answering these questions will require the collection of more original data, whether through panel surveys or through data from other sources such as market behaviors. Other applicable areas could include rural-urban conflicts over water resources (e.g., Narain, 2009), as urban consumers – commercial, residential, and industrial – use water and leave farmers and other rural users under stress. From a policy perspective, an important question is what governments in India and elsewhere can do to deal with these anxieties. Given that concerns about the private sector appear more significant than concerns about government intervention, a system of quality certification could be useful. If rural households can receive credible assurances about the quality of new technological goods and services, then the risk of popular backlash against these technologies is assuaged. Our results suggest that the Indian government could encourage the social acceptance of off-grid solar power by creating awareness about branded, quality-certified solar products. One simple way to achieve this goal would be to create a large-scale consumer awareness campaign in collaboration with the International Finance Corporation of the World

12

475

For more information about the campaign, see http://lightingasia.org/india/.

Energy Policy 113 (2018) 466–477

M. Aklin et al.

and political support for a US National Clean Energy Standard. Nat. Clim. Change 2 (8), 596–599. Alstone, Peter, Gershenson, Dimitry, Kammen, Daniel M., 2015. Decentralized energy systems for clean electricity access. Nat. Clim. Change 5 (4), 305–314. Amigun, B., Musango, J.K., Brent, A.C., 2011. Community perspectives on the introduction of biodiesel production in the Eastern Cape Province of South Africa. Energy 36 (5), 2502–2508. Assefa, Getachew, Frostell, Björn, 2007. Social sustainability and social acceptance in technology assessment: a case study of energy technologies. Technol. Soc. 29 (1), 63–78. Barnes, Douglas F. (Ed.), 2007. The Challenge of Rural Electrification: Strategies for Developing Countries. RFF Press, Washington DC. Chakravorty, Ujjayant, Pelli, Martino, Marchand, Beyza Ural, 2014. Does the quality of electricity matter? Evidence from rural India. J. Econ. Behav. Organ. 107, 228–247. Chaurey, A., Kandpal, T.C., 2010. A techno-economic comparison of rural electrification based on solar home systems and PV microgrids. Energy Policy 38 (6), 3118–3129. Cheng, Chao-yo, Urpelainen, Johannes, 2014. Fuel stacking in India: changes in the cooking and lighting mix, 1987–2010. Energy 76, 306–317. Chhibber, Pradeep, Eldersveld, Samuel, 2000. Local elites and popular support for economic reform in China and India. Comp. Polit. Stud. 33 (3), 350–373. Chong, Dennis, Druckman, James N., 2007a. Framing public opinion in competitive democracies. Am. Polit. Sci. Rev. 101 (4), 637–655. Chong, Dennis, Druckman, James N., 2007b. Framing theory. Annu. Rev. Polit. Sci. 10, 103–126. Denisova, Irina, Eller, Markus, Frye, Timothy, Zhuravskaya, Ekaterina, 2009. Who wants to revise privatization? The complementarity of market skills and institutions. Am. Polit. Sci. Rev. 103 (2), 284–304. Drèze, Jean, Amartya, K.Sen, 2002. India: Development and Participation, 2nd edition. Oxford University Press, New York. Dugoua, Eugenie, Urpelainen, Johannes, 2014. Relative deprivation and energy poverty: when does unequal access to electricity cause dissatisfaction? Int. J. Energy Res. 38 (13), 1727–1740. Entman, Robert M., 1993. Framing: towards clarification of a fractured paradigm. J. Commun. 43 (4), 51–58. Fligstein, Neil, Dauter, Luke, 2007. The sociology of markets. Annu. Rev. Sociol. 33 (1), 105–128. Fontaine, Xavier, Yamada, Katsunori, 2014. Caste comparisons in India: evidence from subjective well-being data. World Dev. 64, 407–419. Gaines, Brian J., Kuklinski, Quirk, Paul J., 2007. The logic of the survey experiment reexamined. Polit. Anal. 15 (1), 1–20. Galor, Oded, Tsiddon, Daniel, 1997. Technological progress, mobility, and economic growth. Am. Econ. Rev. 87 (3), 363–382. Gambhir, Ashwin, Vishal Toro, Mahalakshmi Ganapathy. 2012.Decentralised Renewable Energy (DRE) Micro-Grids in India: A Review of Recent Literature. Prayas Energy Group Report. Gartrell, John W., 1977. Status, inequality and innovation: the green revolution in Andra Pradesh, India. Am. Sociol. Rev. 42 (2), 318–337. Government of India, 2011. 2011 Census Report, Houselisting and Housing Census Data Highlights. . 〈http://www.censusindia.gov.in/2011census/hlo/hlo_highlights. html〉. Granovetter, Mark, 1985. Economic action and social structure: the problem of embeddedness. Am. J. Sociol. 91 (3), 481–510. Gurr, Ted Robert, 1970. Why Men Rebel. Princeton University Press, Princeton. Harish, Santosh M., Tongia, Rahul, 2014. Do Rural Residential Electricity Consumers Cross-Subside their Urban Counterparts? Exploring the Inequity in Supply in the Indian Power Sector. Brookings India, Working Paper 04-2014. Joseph, Kelli L., 2010. The politics of power: electricity reform in India. Energy Policy 38 (1), 503–511. Kale, Sunila S., 2004. Current reforms: the politics of policy change in India's Electricity Sector. Pac. Aff. 77 (3), 467–491. Korpi, Walter, 1974. Conflict, power and relative deprivation. Am. Polit. Sci. Rev. 68 (4), 1569–1578. Lal, Sumir, 2006. Can Good Economics Ever Be Good Politics? Case Study of India’s Power Sector. World Bank Working Paper 83. Levin, Irwin P., Schneider, Sandra L., Gaeth, Gary J., 1998. All frames are not created equal: a typology and critical analysis of framing effects. Organ. Behav. Human. Decis. Process. 76 (2), 149–188. Mallett, Alexandra, 2007. Social acceptance of renewable energy innovations: the role of technology cooperation in Urban Mexico. Energy Policy 35 (5), 2790–2798. Masera, Omar R., Saatkamp, Barbara D., Kammen, Daniel M., 2000. From linear fuel switching to multiple cooking strategies: a critique and alternative to the energy ladder model. World Dev. 28 (12), 2083–2103. McAdam, Doug, Boudet, Hilary, 2012. Putting Social Movements in Their Place: Explaining Opposition to Energy Projects in the United States, 2000–2005. Cambridge University Press, New York. McRae, Shaun, 2015. Infrastructure quality and the subsidy trap. Am. Econ. Rev. 105 (1), 35–66. Narain, Vishal, 2009. Growing City, shrinking hinterland: land acquisition, transition and conflict in Peri-Urban Gurgaon, India. Environ. Urban. 21 (2), 501–512. Palit, Debajit, Bhattacharyya, Subhes C., Chaurey, Akanksha, 2014. Indian Approaches to Energy Access. In: Halff, Antoine, Sovacool, Benjamin K., Rozho, Jon (Eds.), Energy Poverty: Global Challenges and Local Solutions,. Oxford University Press, New York. Santhakumar, V., 2008. Analysing Social Opposition to Reforms: The Electricity Sector in India. Sage, Thousand Oaks. Smith, Heather J., Pettigrew, Thomas F., Pippin, Gina M., Bialosiewicz, Silvana, 2012. Relative deprivation: a theoretical and meta-analytic review. Pers. Soc. Psychol. Rev.

Fig. 5. Average treatment effect on support for government leadership in solar power development, with confidence intervals. The dependent variable ranges from 0 to 4, with higher values indicating more support for government leadership. The mean is 3.178 and the standard deviation 1.255. For full regression output, see Section A3.

Acknowledgments We thank the Council on Environment, Energy and Water for their participation in this research project and MORSEL India for excellent data collection. This project was funded by the Shakti Sustainable Energy Foundation and ClimateWorks. Appendix A. Supplementary data A replication package is available at Harvard Dataverse, doi: 10. 7910/DVN/RXEVX4. Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j. enpol.2017.10.059. References Alkon, Meir, Harish, S.P., Urpelainen, Johannes, 2015. Household Energy Access and Expenditure in Developing Countries: Evidence from India, 1987–2010. Energy Sustain. Dev. 35, 25–34. Aklin, Michaël, Cheng, Chao-yo, Urpelainen, Johannes, Ganesan, Karthik, Jain, Abhishek, 2016a. Factors affecting household satisfaction with electricity supply in rural India. Nat. Energy 1, 16170. Aklin, Michaël, Cheng, Chao-yo, Ganesan, Karthik, Jain, Abhishek, Urpelainen, Johannes, Council on Energy, Environment and Water, 2016b. Access to Clean Cooking Energyand Electricity: Survey of States in India (ACCESS). Harvard Dataverse, pp. V1. http://dx.doi.org/10.7910/DVN/0NV9LF. Aklin, Michaël, Bayer, Patrick, Harish, S.P., Urpelainen, Johannes, 2014. Information and energy policy preferences: a survey experiment on public opinion about electricity pricing reform in rural India. Econ. Gov. 15 (4), 305–327. Aldy, Joseph E., Kotchen, Matthew J., Leiserowitz, Anthony A., 2012. Willingness to pay

476

Energy Policy 113 (2018) 466–477

M. Aklin et al.

developing countries. World Dev. 39 (6), 1037–1050. Wolfram, Catherine, Shelef, Orie, Gertler, Paul, 2012. How will energy demand develop in the developing world? J. Econ. Perspect. 26 (1), 119–138. Wüstenhagen, Rolf, Wolsink, Maarten, Bürer, Mary Jean, 2007. Social acceptance of renewable energy innovation: an introduction to the concept. Energy Policy 35 (5), 2683–2691. Yuan, Xueliang, Zuo, Jian, Ma, Chunyuan, 2011. Social acceptance of solar energy technologies in China–end users' perspective. Energy Policy 39 (3), 1031–1036. Zoellner, Jan, Schweizer-Ries, Petra, Wemheuer, Christin, 2008. Public acceptance of renewable energies: results from case studies in Germany. Energy Policy 36 (11), 4136–4141.

16 (3), 203–232. Urpelainen, Johannes, 2016. Energy poverty and perceptions of solar power in marginalized communities: survey evidence from Uttar Pradesh, India. Renew. Energy 85, 534–539. van der Horst, Dan, 2007. NIMBY or not? Exploring the relevance of location and the politics of voiced opinions in renewable energy siting controversies. Energy Policy 35 (5), 2705–2714. Wejnert, Barbara, 2002. Integrating models of diffusion of innovations: a conceptual framework. Annu. Rev. Sociol. 28, 297–326. Winkler, Harald, Simões, André Felipe, Rovere, Emilio Lèbre La, Alam, Mozaharul, Rahman, Atiq, Mwakasonda, Stanford, 2011. Access and affordability of electricity in

477