RENEWABLE PORTFOLIO STANDARDS IN THE PAS T, PRESENT, AND FUTU RE: ADOPTION, EFFECTIVEN ESS, AND POST ADOPTI ON b y SOJIN JANG B.S., Indiana University Bloomington, 2009 M.P.A., Sungkyunkwan University, 2012 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy Public Affa irs Program 2018
ii This thes is for the Doctor of Philosophy degree by Sojin Jang has been approved for the Public Affairs Program by Todd Ely , Chair Christopher Weible , Advisor Deserai Crow Sanya Carley Date: May 12 , 2018
iii Jang, Sojin (Ph.D., School of Public Affairs) Renewable Portfolio Standards in the past, present, and future: Adoption, Effectiveness, and Post adoption Thesis directed by Professor Christopher Weible ABSTRACT Renewable Portfolio Standards (RPS) are state level poli cies that mandate a certain amount of renewable energy production. Despite the absence of federal law that requires the reduction of CO 2 emissions or increase in the electricity generation from renewable energy sources, s tates have volunta rily adopted and implemented them with expectations for economic, environmental, and political benefits. Over two decades of RPS history have spawned the copious literature on RPS adoption and effectiveness. This dissertation advance s the existing RPS lite rature by providing quantitative and qualitative analyses of RPS adoption, effectiveness, and post adoption decisions in an attempt to answer the question on the policy processes surrounding state level RPS: Why and how are state level RPS policies adopted and revised ? To what extent has RPS met its policy goals? In regards to RPS adoption ( C hapter two ), this dissertation pays attention to the roles of As fo r the effectiveness of RPS in state energy markets ( C hapter three ), this dissertation evaluates the effectiveness of RPS in achieving renewable energy generation expansion and electricity price stability. Electricity price stability is often touted as the economic benefits of RPS. Yet, less is known about the effectiven ess of RPS on electricity price stability while existing RPS evaluation literature has focused on RPS s effectiveness in renewable energy expansion and environmental effect s ( e.g., Carley, 2009; Fischlein & Smith, 2013; Prasad &
iv Munch, 2012; Sekar & Sohngen, 2014) . Thus, this dissertation adds knowledge to the existing RPS effective in renewable energy expansion and electricity price stability. Further, post RPS adoption decisions are studied using semi structured interviews with state legislators in West Virginia and Oregon ( C hapter fo ur ). Although considerable research has been accomplished to explain RPS adoption and effectiveness, less is known about the revisions of RPS after adoption. Through the interviews, this dissertation explores why and how RPS has taken divergent paths in We st Virginia and Oregon. There are three major lessons from this dissertation. First, a state is more likely to adopt RPS if it directly experienced large scale weather related crisis events, and politically and economically similar states had already ado pted RPS. Crisis events have been known as a major driver for policy change (Birkland, 2007; Boushey, 2012; Kingdon, 1995; Sabatier & Weible, 2007b) . Yet, policy adopt ion literature has paid limited attention to the roles of crisis events in triggering policy adoption. This dissertation reveals that one of the ways that states respond to directly experienced crisis events is to adopt a state level renewable energy polic y, RPS. Further, extant RPS adoption studies exclusively focused on the effects of RPS adoptions in neighboring or ideologically similar states as external factors (Carley & Miller, 2012; Carley, Nicholson Crotty, & Miller, 2016; Matisoff, 2008; Yi & Feiock, 2012) . This dissertation shows that states also imitate RPS adoptions in not only ideologically similar states but also economically similar states. Energy or environmental policy often incurs high economic a nd political costs, and the patterns of RPS imitation across states reflect concerns on such costs around RPS adoption and implementation. Second , this dissertation may be one of the very first stud ies effectiveness on electricity price st ability. The duration of RPS increases renewable energy
v generation in a state, and stabilizes electricity prices. However, a binary approach to RPS adoption is not a statistically significant indicator for either renewable energy generation or electricity price stability. Third, RPS is inextricably linked with politics and economy. W i thin this context , political ideology serves as the basis of RPS adoption and revision decisions by shaping the perceptions of policy effectiveness and costs associated with RP S implementations. The form and content of this abstract are approved. I recommend its publication. Approved: Christopher Weible
vi TABLE OF CONTENTS CHAPTER I. INTRODUCTION II. CRISIS OR CONFORMITY: EXPLAINING PATTERNS IN THE ADOPTION OF RPS ................................ ............ 8 Abstract ................................ ................................ ................................ ....................... 8 Introduction ................................ ................................ ................................ ................. 8 RPS through the lens of Diffusion of Innovation Theory ................................ ................. 9 Roles of crises in RPS adoption ................................ ................................ ................ 11 Horizontal diffusion of RPS adopti on ................................ ................................ ....... 16 Controls Other internal determinants of RPS adoption ................................ .......... 18 Analytical methods and Dependent variables ................................ ................................ . 24 Analysis ................................ ................................ ................................ ........................... 25 Conclusion and Policy Implications ................................ ................................ ............... 31 III. EVALUATING RENEWABLE PORTFOLIO STANDARDS: HAVE THEY KEPT THEIR ENERGY AND ECONOMIC PROMISES? ... .......... .... . . 35 Abstract ................................ ................................ ................................ ........................... 35 Introduction ................................ ................................ ................................ ..................... 35 Overview of RPS goals ................................ ................................ ................................ ... 37 Evaluation Framework ................................ ................................ ................................ .... 40
vii Dependent variables ................................ ................................ ................................ .. 40 Independent variables ................................ ................................ ............................... 41 Analysis ................................ ................................ ................................ ........................... 49 Conclusion and Policy Implications ................................ ................................ ............... 57 IV. AFTER THE ADOPTION OF RENEWABLE PORTFOLIO STANDARDS: GO GREENER OR BACK TO GREY ? ........................................................ .............. .... 60 Abstract ................................ ................................ ................................ ........................... 60 Introduction ................................ ................................ ................................ ..................... 61 Policy revision after adoption ................................ ................................ ......................... 62 Post RPS adoption ................................ ................................ ................................ ..... 64 Case selection ................................ ................................ ................................ .................. 66 Data collection ................................ ................................ ................................ ................ 68 Interview Questions ................................ ................................ ................................ .. 69 Analysis ................................ ................................ ................................ ........................... 70 Overview of RPS in West Virginia and Oregon ................................ ....................... 72 Cross case analysis of RPS revisions in West Virginia and Oregon ........................ 75 Conclusion and Policy Implications ................................ ................................ ............... 85 V. CONCLUSION ................................ ................................ ................................ ............... 90 RPS adoption in Chapter 2 ................................ ................................ .............................. 91 RPS effectiveness in Chapter 3 ................................ ................................ ....................... 94
viii RPS revision after adoption in Chapter 4 ................................ ................................ ....... 95 Overall Chapter Summary ................................ ................................ .............................. 97 Limitations ................................ ................................ ................................ ...................... 98 Future studies ................................ ................................ ................................ ................ 100 Summary of lessons ................................ ................................ ................................ ...... 101 REFERENCES ................................ ................................ ................................ ................... 103 APPENDIX A . Estimation Results for Model 1: Changes in p robabilities for imitation ....... ... .. .. .. . 115 B. Estimation Results for Model 2: Changes in probabilities for imitation .................11 6
ix LIST OF TABLES TABLE 2. 1 . Comparison of geographic and policy dimensions of crisis ................................ .................. 15 2. 2 . Measurement strategies for independent variables ................................ ................................ 22 2. 3 . Dyadic data structure ................................ ................................ ................................ ............. 25 2. 4 . Descriptive statistics ................................ ................................ ................................ .............. 2 6 2. 5 . Estimation results RPS adoption using dyadic Probit ................................ ......................... 30 3.1 . Measurement of electricity price stability ................................ ................................ .............. 41 3.2 . Descriptive statistics ................................ ................................ ................................ .............. 49 3.3 . Correlation between RPS and renewable energy capacity ................................ .................... 50 3.4 . Estimation results renewable energy generation ................................ ................................ . 53 3.5 . Estimation results electricity price stability ................................ ................................ ........ 56 4.1 . Interview questions ................................ ................................ ................................ ................ 69 4.2 . Interview codebook ................................ ................................ ................................ ................ 71 4.3 . Analysis of RPS revision decisions in West Virginia and Oregon ................................ ........ 8 3 5.1 . Chapter summaries ................................ ................................ ................................ ................. 9 7
x LIST OF FIGURES FIGURE 1.1 . RPS adoption timeline ................................ ................................ ................................ ............. 5 4.1 . Post adoption decisions ................................ ................................ ................................ .......... 66 4.2 . RPS adoption year, Population rank, and GSP rank of candidate states for case study ................................ ................................ ................................ ......................... 68
xi LIST OF ABBREVIATIONS EERS: Energy Efficiency Resource Standards MGPO: Mandatory Green Power Option PBF: Public Benefit Funds RPS: Renewable Portfolio Standards
1 CHAPTER I . INTRODUCTION Climate change is one of the most pressing environmental issues that ha ve deeply penetrated environment , economy, and politics at various levels . Although some scientists still have dissenting views on the consequences of climate change, they generally have acknowledge d human activities as a major contributor to climate chang e (Boykoff, 2007; Cavallo & Noy, 2010; Schipper & Pelling, 2006; van Aalst, 2006) . Recently, the rate of climate change has become much faster and the acceleration in climate change is largely attributed to the rapid increase in CO 2 emissions driven by the use of carbon based energy since the Industrial Revolution (EPA, 2015; Sekar & Sohngen, 201 4 ) . 1 Increase in CO 2 emissions and accelerated climate change can alter the economic, environmental, and social conditions of both major CO 2 emitting countries and relatively low emitting countries because of the cumulative and trans boundary nature of CO 2 . Many countries around the world ha ve promoted the use of renewable energy and/or set CO 2 reduction goals at the national level as they acknowledge a close relationship between the emissions from fossil fuel combustion and climate change (Sovacool & Barkenbus, 2007) . The use of renewable energy is of particular importance in the U.S. because it is the second biggest CO 2 emitter in the world and also the 11 th in CO 2 emissions per capita . 2 In the U. S. from 1990 to 2013, fossil fuel combustion for electricity generation accounts for 37% of CO 2 emissions. The second largest CO 2 emitting source is transportation, accounting for 31% of CO 2 emissions (EPA, 1 http://www3.epa.gov/climatechange/science/causes.html 2 Except for Luxembourg, all 9 countries with higher CO 2 emissions per capita are the Middle East countries that heavily rely their economy on oil and gas production (Qatar with the highest CO 2 emissio ns per capita followed by Trinidad and Tobago, Kuwait, Brunei Darussalam, Aruba, Luxembourg, United Arab Emirates, Oman, Saudi Arabia, and Bahrain). In case of Luxembourg, OECD (2013) notes that the low tax on road fuels is largely responsible for high CO 2 emission per capita in Luxembourg.
2 2016). E lectricity generation is the major source of CO 2 emission s in the U.S. , thereby calling for the substantial reduction of greenhouse gas emission s from electricity generation (Duane, 2010) . Nevertheless, t he U .S. has long been criticized for the passive responses against climate change because of the absence of the national level climate change action and President George W. Bush s rejection of the impl ementation of Kyoto Protocol . More recently, Trump administration announced the withdrawal from the Paris Agreement on climate change mitigation, and proposed the repeal of the Clean Power Plan, initially designed for reductions in greenhouse gas emissions from electricity generations. Despite number of state governments have been committed to the reduction of CO 2 emission s through the adoption of renewable energy policy. Among various policy options and programs, one of the most adopted is the Renewable Portfolio Standards (RPS) , which set the target amount of renewable energy production within a specific year . Iowa first adopted the RPS in 1983 and it has been widely adopted by 29 states and Washington D.C. as of 2015. The RPS mandates the certain portion of electricity coming from renewable energy sources such as solar, wind, geothermal heat, wave or tidal energy, and organic matter . Among these sources, the most widely used is wind and solar energy. State RPS policies are distinguished from one another in terms of policy design and implementation. These include the selection of target utilities, resource eligibility , applicability (geographic, industrial coverage), flexibility in account balancing, tradability of renewable energy credits (REC) and designation of administrative responsibilities for RPS target, RPS compliance monitoring and penalties for noncomplying entities (T. Berry & Jaccard, 2001) . The adoption of RPS policy is expected to bring environmenta l, social and economic benefits through renewable energy production ( T. Berry &
3 Jaccard, 2001; Bird, Chapman, Logan, Sumner, & Short, 2010; Cory & Swezey, 2007; Gonzalez, 2007 ) . I ncreases in the electricity costs in the short term and the need f or large scale investment in renewable energy production are subject to public criticism. In spite of such challenges, more than half of states ha ve already adopted RPS , and more states are expected to adopt RPS or make their renewable energy production go als more stringent . All 29 state RPS policies are different and operate in different contexts. The variability of RPS policies is found throughout the various components of RPS including the size and timing of renewable energy production, target utilities, presence of cost cap, types of eligible resources and facilities, tradability of renewable energy credits, compliance monitoring, and noncompliance penalty (T. Berry & Jaccard, 2001) . Yet, what best represents and characterizes state RPS policies are the stringency of RPS predicated upon the size and timing ( target year) of renewable energy production. States adopt RPS policies for various internal and external reasons. Recently, governments and businesses began to increase the use of renewable energy as a response to environmental problems and extreme weather events . Climate change is a long term and slowly developing phenomenon that cannot be directly felt. But, people and media often interpret specific weather related crisis events as related to climate change and learn the need for taking actions against climate change through experiences with those events. Although such interpretation is scientifically invalid, this is how humans perceive the threat of climate change and motivate themselves and others to engage in climate change mitigation or environmental protection measures. Although extant policy change literature has extensively discussed the roles of crisis events in explaining the facto r s for policy change , extant studies have paid less attention to crisis events as internal and external drivers of policy adoption. Further, recent diffusion scholars raised questions on the neighboring effects as a factor for policy diffusion and discuss ed
4 the need for exploration of external factors beyond neighboring effects (Baybeck, B erry, & Siegel, 2011; Shipan & Volden, 2012; Volden, Ting, & Carpenter, 2008) . In an attempt explore additional factors of policy diffusion across states, this dissertation examines the effects of policy adoptions in states that share similar political, economic, and energy market conditions. Studies on RPS adoption merit attention not only for the widespread adoption across states but also it opens up the avenues for the evaluation of RPS and examination of policy decisions after adoption. Although th e RPS is primarily a policy for ren ewable energy production, it is not the sole purpose of the RPS adoption . State policymakers promote and adopt the RPS not only for its immediate goal of renewable energy production, but also for its long term economic goals. To date, RPS evaluation literature has predominantly focused on the effectiveness of RPS policies in achieving renewable energy expansion and CO 2 emissions reductions (Carley, 200 9; Fischlein & Smith, 2013; Prasad & Munch, 2012; Sekar & Sohngen, 2014) . However, there is a dearth of research on the effectiveness of RPS policies in promoting electricity price stability, which is one of the main goals of RPS policies. When RPS supp orters discuss the adoption and expansion of RPS, they advocate such long term benefits of RPS for its positive and clean policy image. In this vein, policy scholars are tasked with probing into the extent to which RPS fulfilled the promised goals. The eva luation of RPS effectiveness provides critical insights into RPS for scholars and policymakers in states with and without RPS in different ways. RPS evaluation information serves as an indicator for the progress that the policy has made to date. States con sidering RPS adoption may refer to RPS effectiveness studies in making their future policy decision. Recently, however, the proliferation of RPS lost its momentum in 2014. Ohio first rolled back its RPS by freezing the renewable energy production goals fo r two years. In 2015, Kansas
5 revised their mandatory RPS into voluntary policy , and West Virginia repealed its RPS. Yet, a number of existing RPS states further heightened their renewable energy production goals. Vermont recently adopted the RPS in 2015 an d it is marked by its stringent goals of 75% of electricity from renewable energy sources by 2032. California, Hawaii, and Oregon amended their RPS in a direction that increases the renewable energy production to 50% or more of total electricity generation . However, the divergent paths of RPS policies after adoption are less studied and such recent trends are difficult to be capture d in the quantitative RPS study that analyzes a large N data spanning about two decades. Hence, a detailed case study is needed to explore the drivers that shaped and differentiated the direction of policy changes after initial adoption. As of 201 5 , 29 states and Washington D.C. have mandatory RPS . Figure 1 illustrates the states adopting mandatory RPS in a chronological order. Figure 1 . 1 . RPS adoption timeline This dissertation is composed of five chapters . This present Chapter 1 presents an overview of the dissertation and topic. Chapters two , three , and four provide the empirical analysis of the di f ferent aspects of RPS. Chapter five offers the summary of findings from the empirical analysis chapters.
6 Specifically, t he three empirical chapters (2, 3, and 4) are created to answer the following research questions: W hat factors explain patterns in the adoption of state RPS policies? (Chapter two ) Chapter 2 explore s the drivers of RPS policy adoption across 46 states from 199 8 to 20 09. To examine the RPS stringencies and its driving factors, this chapter use s dyadic a nalysis to examine the policy imitation across states . T his chapter contribute s to the existing RPS literature by examining the effects of internal and external crisis events on RPS adoption. As for diffusion of RPS policies across states, this chapter examin es the factors for policy i mitation. This chapter f inds that a state with frequent weather related hazard experiences is more likely to adopt a RPS policy. Further , state may take cues from politically and economically similar states. T o what extent have RPS policies achieved their intended energy and economic goals? (Chapter three ) Chapter 3 examine s the factors associated with the effectiveness from 199 8 to 201 2 . This chapter assesses the energy and economic effects of RPS. It analyze s the effectiveness of RPS and other energy policies on renewable energy generation and electricity price stability. This chapter f inds that an RPS policy is positively correlated with renewable energy generation. Further, it has been effective in reducing concerns on the increases in electricity prices. W hat factors shaped the state RPS policies after adoption? How have these factors contributed to policy decisions after adoption? (Ch apter four ) Chapter 4 explore s the revision of RPS after adoption using semi structured interview s and literature review . Due to the dynamic nature of RPS and full latitudes in RPS policy design, state policymakers have continuously amended their RPS to study of RPS policy decisions after adoption provide s a detailed account of contributors to recent
7 RPS amendments that evolved in opposite directions through the comparative case study of West Virginia and Oregon. West Virginia repealed the RPS in 2015 , whereas Oregon passed the bill in 2016 that introduced 50% of utilities from renewable energy by 20 40 and the banning of coal from electricity generation by 2035 . This chapter f inds that state government ideology shapes the views on environmental and economic issues and effectiveness of RPS, and it ultimately guid es the directions of policy revisions. Moreover, RPS repeal in West Virginia was led by state legislators while local government and citizen led effort fo r renewable energy expansion spurred the proposal and passage of RPS expansion bill in Oregon. through RPS adoption and revision , and its impact on state level energy prod uction and energy economy. By employing quantitative and qualitative approaches to understand policy processes and effectiveness of RPS, this dissertation provide s insights into how state governments have responded to weather related crisis events with RPS and imitated RPS adoption decisions in terms of meeting the goal s promised to citizens. Finally, interviews with state legislators provide explanations of why and how radically different types of RPS revisions took place in West Virginia and Oregon. This dissertation ends with overall findings, limitations, and future study tasks.
8 CHAPTER II. CRISIS OR CONFORMITY: EXPLAINING PATTERNS IN THE ADOPTION OF RPS A bstract In the U.S., states with renewable portfolio standards (RPS) set annual RPS schedules that require a certain share of electricity sales to come from rene wable energy. States adopt RPS policies for various reasons inside and outside states . Recently, governments and businesses began to increase the use of renewable energy as a response to environmental problems and extreme weathers. This chapter analyzes th e drivers of RPS adoption from 199 8 to 20 09 through the diffusion of innovation theory using dyadic analysis. This chapter found that a state with frequent weather related hazard experiences is more likely to adopt a RPS policy . As for external factors for RPS adoption, a state may take cues from politically and economically similar states. Introduction R PS policies are one of the most widely adopted climate change mitigation measures at state level (Engel, 2006) . Climate change is a long term and slowly developing phenomenon that cannot be directly felt. But, people and media often in terpret specific weather related crisis events as climate change and learn the need for taking actions against climate change through experiences with those events. Although such interpretation is scientifically invalid, this is how humans perceive the thr eat of climate change and motivate themselves and others to engage in climate change mitigation or environmental protection measures. In fact, extant policy change literature has extensively discussed the roles of crisis events in explaining the facto rs fo r policy change (Birkland, 2007; Boushey, 2012; Kingdon, 1995; Nohrstedt & Weible, 2010; Sabatier & Weible, 2007a) . Howe ver, policy innovation and diffusion studies have paid less attention to crisis events as a driver of policy adoption.
9 As for the diffusion of policy innovation, extant studies heavily focused on the effect of policy decisions in neighboring states to the likelihoods of policy adoption in a state (Shipan & Volden, 2012) . Although more recent policy diffusion studies began to probe into the effects of ideologically similar states, a state may take cues from the policy decisions of other s tates that share similar policy adoption motivation s and contexts where policy is adopted and implemented. Thus, this chapter investigates the possibility that a state refers to other states that have similar internal conditions or determinants relevant to RPS adoption besides geographical and ideological proximity to existing adopters. In sum, this chapter seeks to contribute to existing RPS adoption and diffusion literature by introducing two new important internal and external indicators for policy adopt ion, crisis events and shared internal conditions across states. T his chapter examines the drivers of RPS adoption from 199 8 to 20 09 through the theoretical lens of diffusion of innovation using dyadic analysis . D yadic analysis allows for the pairwise analysis of the effect of other states conditions and similarities in internal determinants on the adoption decisions between two states in a paired observation . This chapter found that a state with frequent weather related hazard experiences is mo re likely to adopt a RPS policy , and a state may take cues from politically and economically similar states. RPS through the lens of D iffusion of Innovation Theory The d iffusion of i nnovation t heory offer s a way to explain the timing and the design of polic ies adopted across multiple jurisdictions . The theory allows for the concrete explanation of policy adoption and diffusion mechanisms . 3 In th e diffusion of innovation theory , a n innovation to the states adopting it, no matter how old the (Walker, 1969 , p. 881 ) . Policy 3 This chapter considers the diffusion of innovation as a theory. Theories facilitate the understanding of related concepts or variables based on the set of assumptions or propositions and also seek to explain the causa l mechanisms between interrelated concepts (Kiser & Ostrom, 2000; Ostrom, 2010; Weible & Nohrstedt, 2012) .
10 innovation yields financial and political costs (Rose Ackerman, 1980) and states have often been judged according to the relative speed with which they have accepted new ideas (Walker, 1969) . Any pattern of successive adoptions of a policy across s imilar government units can be called diffusion (Eyestone, 1977 , p. 441 ) . Rogers ( 1995) presented the S shaped curve to describe an increment al learning process that the increased availability of policy information and reduced uncertainties with a new policy over time may increase the likelihood of policy adopt ions across states. A premise of p olicy diffusion is that the early policy adopters affect the policy decisions of potential adopters. M ore concretely , policy diffusion indicates policy decisions that are systemically conditioned by prior policy choices made in other [jurisdictions] (Simmons, Dobbin, & Garrett, 2006 , p. 787 ) . Interdependent decision making is critical to understand policy diffusion , and policy decisions continue to interact with one another (S. M. Brooks, 2007) . In the context of RPS adoption and diffusion across states, the diffusion of innovation theory states more capable of adopting the RPS relative to other states. Further, different policy diffusion mechanism can help RPS scholars to examine why the RPS h as been widely spread across states without pressure from the federal government. For this reason, c opious RPS literature delved into RPS adoption directly and indirectly through the theoretical lens of diffusion of innovation (Carley & Miller, 2012; Carley et al., 2016; Chandler, 2009; Huang, Alavalapati, Carter, & Langholtz, 2007; M atisoff, 2008; Nicholson Crotty & Carley, 2015; Yi & Feiock, 2012) . Yet, the determinants of RPS are not consistent across studies and common findings on the internal determinants of RPS is limited to the roles of state wealth (Carley et al., 2016; Chandler, 2009; Huang et al., 2007; Upton & Snyder, 2015; Y i & Feiock, 2012) , and CO 2
11 emissions or fossil fuel production (Carley & Miller, 2012; Carley et al., 2016; Lyon & Yin, 2010; Upton & Snyder, 2015) on RPS adoption. These limited consistencies in findings imply that RPS studies should explore determinants that are germane to the immediate need and operation of RPS rather than focusing on general indicators of state s internal conditions. Accordingly, the following sections are devoted to the discussions of effects of crises and existing relevant policies on the RPS adoption in an attempt to unearth the RPS adoption determinants that directly call for the switch to renewable energy and facilitate the operation of RPS. Roles of crises in RPS adoption C rises , such as large scale disasters , often play critical roles in triggering major or radical policy changes (Nohrstedt & Weible, 2010) . Further, p oliticians often rel y on a sense of crisis to catalyze the achievement of policy goals that they support (Keeler, 1993) . Policy scholars have long advocated for the roles of crises in facilitating policy c hange as in multiple streams framework, punctuated equilibrium and advocacy coalition framework. In multiple streams framework, Kingdon (1995) used the term focusing events to describe the role of abrupt and infrequent events, such as crises or disasters, on the agenda setting. Focusing events increase attention to policy issues, which policy actors se ek to place on the agenda by giving rise to media (Birkland, 2007) . Focusing events are viewed as exogenous shocks to the political system and draw political attention to a po licy problem that has been overlooked or neglected in punctuated equilibrium theory (Boushey, 2012) . In the advocacy coalition framework , shock or crisis event may contribute to the redistribution of political resources, disrupt the existing power structure among coalitions, and also promote major policy change (Sabatier & Weible, 2007a) . In spite of the important roles of focusing events or crises
12 for policy change as explicated in multiple streams, punctuated equilibrium and advocacy coalition framework , how focusing events or crises trigger the adoption of a new policy has not been extensively examined in the policy adoption and diffusion studies. Keeler ( 1993 , pp. 440 442 ) discussed mechanisms of how a crisis creates the contexts for policy change. First, the political party can gain advantage in suggesting a new policy as it emphasizes the role of policy as a solution or response to a crisis. Second, a crisis contributes to a sociopolitical environment favorable for governance that can add weight to the passage of reforms. Third, a crisis c ould trigger a social mobilization to advocate the need for reform. In particular, first and second description of the role of crisis speaks to the adoption of RPS as a climate change mitigation measure. Despite the arguments around the drivers of climate change, scientists began to reach a consensus on the CO 2 emissions from human activities as a major contributor to climate change (Boykoff, 2007; Cavallo & Noy, 2010; IPCC, 2007; Klein et al., . Am ong the various types and levels of environmental risks driven by climate change , weather related hazards are one of the most frequently discussed impacts of climate change. Warming climate is associated with the long term trends of weather related haza rds such as storms, floods, drought, and extreme temperatures (N. Brooks & Adger, 2003; IPCC, 2007; Klein et al., 2004; Mileti, 1999; Mirza, . In fact, p eople do not realize the immediate need for taking actions against climate change or severity of climate change largely because climate slowly develops over long t erm and cannot be directly experienced (Moser & Dilling, 2011) . However, people lea rn the signs and impacts of climate change through experiences in large scale weather related hazards. Similarly, Weber ( 2006) posited that catastrophic impacts of climate change and globally experienced
13 climate change impacts amplif y reactions to climate change risk . In other words, people often need to see or feel something to believe the need for action . By the same token, Ray, Hughes, Konisky and Kaylor ( 2017) found that individual experiences with e xtreme weather events are positively associated with civic support for climate adaptation policy. If there is one way that people learn the severity of climate change and need for taking actions for climate change mitigation, it would be through frequent l arge scale weather related hazards that may disrupt their normal lives. Scientifically, or within the environmental and natural science discipline, we cannot link each specific hazard event to climate change as a driver of the event since climate change is more about the long term trend s and patterns of weather systems. However, in the real world and media, the way non scientist average people understand climate change and weather often times, people perceive that individual weather related crisis event that we are experiencing now is directly related to climate change as portrayed in the headline s from news media as follows: Experts: Climate Change May Make Northeast Winter Storm s Worse (Thomas, 2015) CBS Boston. Feb 12, 2015 (Fountain, 2017) New York Times. Dec 13, 2017 Climate change, extreme weather already threaten 50% of U.S. military sites (Roth, 2018) USA Today. Jan 31, 2018 In turn, weather related cr isis events have increasingly served as a signal for the need to take action for climate change mitigation . Recently, insurance companies began to show interest in renewable energy as a growing number of scientific evidences suggest that the increases in frequency and magnitude of natural disasters are attributed to the explosive growth in the use of
14 conventional energy sources (Bull, 2001) . In fact, the House of Representatives in Hawaii drafted a bill that explicitly asks for the federal government and states to sw itch from fossil fuels to renewable energy as a response to various environmental and economic problems triggered by fossil fuel uses (Hawaii, 2017) . The bill describes the increasing number and scale of hurricanes affected by warming air and discusses the renewable energy as a solution to the climate related crisis. However, not much is known about the nuanced differences of crises and their roles in policy change. Accordingly, Nohrstedt and Weible ( 2010) suggested a typology of crisis with respect to its geographical and policy attributes. In their typology, the dimension of geographic proximity denotes the proximity of origin of event to the jurisdiction . 4 Another dimension of typology, policy proximity indicates the extent to which a crisis is relevant to the policy of subsystem s focus. Based on these criteria, four types of crises in the context of policy change were identified as following: immediate crisis (close geographic proximity, close policy proximity), geographic proximate crisis (close geographic proximity, distant policy proximity), policy proximate crisis (dist ant geographic proximity, close policy proximity), and vicarious crisis (distant geographic proximity, distant policy pro ximity) as summarized in Table 2. 1. 4 (Weible 2006, p. 98) . I replaced it with jurisdiction because this chapter exclusively focuses on the territorial scope.
15 Table 2 . 1 . Comparison of geographic and policy dimensions of crisis Close geographic proximity Distant geographic proximity Close policy proximity Immediate Crisis Example: Hurricane Katrina for the Louisiana crisis management subsystem Policy Proximate Crisis Example: 9/11 terrorist attacks for European security subsystems Distant policy proximity Geographic Proximate Crisis Example: Southern California wildfires for the California public health subsystem Vicarious Crisis Example: Swine flu crisis for counterterrorism subsystems Source: Nohrstedt & Weible (2010), p. 21 Yet , not all people directly experience severe weather extremes in their local areas or state s . The influence of hazards experiences in other states, policy proximate crisis, should also play a role in the RPS adoption since policymakers and citizens learn about th e severity of hazards and its relationship with climate change through media and social media networks although they are remote from the origin of such hazards . 5 Thus, the following hypothes e s are suggested: Immediate Crisis Hypothesis : Weather related c risis events inside a state will increase the likelihood of RPS policy adoption. Policy Proximate Hypothesis : Weather related crisis events outside a state will increase the likelihood of RPS policy adoption. Among different types of crisis events, this chapter focused on coastal and ice storms, hurricane, drought, flood, freezing, snow, and tornado as these are largely known to be associated with climate change in the long run (N. Brooks & Adger, 2003; IPCC, 2007; Klein et 5 This chapter exclusively focuses on the comparison of roles of immediate crisis and policy proximate crisis because crises that are distant in terms of po licy proximity (e.g., geographic proximate crisis and vicarious crisis) are very broad and difficult to limit the scope of distant policy proximity.
16 2006) . One may argue the need for counting wildfire toward climate related hazards. However, this chapter does not consider wildfire as climate related hazards largely due to the fact that humans activities are directly accountable for the starting of 84% of wildfires in the U.S. from 1992 to 2012 (Balch et al., 2017) . Following the occurrence of disaster events, two year window for changes in public behaviors or policies open (Mason, 2006; OECD, 2004) . By the same logic, c risis events in each state are measured by the number of emergency or disaster declaration by FEMA in the past two years. Specifically, immediate crisis is measured by crisis events occurring inside a state, and policy proximate crisis is measured by crisis events outside a state. Horizontal diffusion of RPS adoption When a state face s risks that can be addressed with a new policy, it tend s to take cues from other state s policy decisions (Balla, 2001) . By taking cues from other states that share similar characteristics, a state can mak e expectations about the performance of a new policy, thereby making a policy adoption decision with more certainty on the effect of policy . Among different state characteristics , policy decisions in geographically neighboring states are the most widely em ployed measure to explain horizontal diffusion since the inception of diffusion theory to date. The rationale for referring to neighboring states policy decision is that neighboring states would share political ideology, beliefs, customs, economy and cult ure. Although geographic neighbors can provide policy lessons to potential adopters, sometimes their influence is exaggerated (Shipan & Volden, 2012) . G eographically adjacent states do not always share these similarities and they may re fer to policy decisions in states that share specific attributes other than geographic borders. Therefore, this chapter investigates policy imitation by directly
17 measuring the effects of states that share demographic, economic, environmental, and political conditions on potential adopters policy decisions. 6 Imitation Policy imitation occurs as potential adopters refer to policy decisions in similar states (Karch, 2007) . The i mitation mechanism is based on contextual similarities across states , and the process of imitation provides the evidences of how a policy would work in a state and how a state responds to a new policy. T he study of the imitation mechanism contribute s to the policy diffusion literature by offering a hypothetical account of similar policy environment as a driver of policy diffusion (Karch, 2007). In explaining policy imitation beyond neighboring states, Grossback, Nicholson Crotty and Pet erson (2004) pioneered the measuring and analyzing the effect of policy adoptions in ideologically similar states on potential adopters . They argued that i deological information signals the potential responses of the electorate and other policy elites t o a new policy . Volden ( 2006 ) s study on the adoption and diffusio n of the Children s Health Insurance Program pioneered the inquiry of the impact of policy adoption in states that share similar internal attributes in addition to the emulation of successful policies. In the testing of Similar States Hypotheses , his study found that policy diffusion can be explained by similarities in political ideologies, per capita income, managed care structures, and budgetary considerations whereas no statistically significant effect of geographic neighbors is found. 6 Other horizontal diffusion mechanisms include coercion, learning, and competition. Although diffusion by coercion across the same levels of governments are possible, coercion by upper level government is more commonplace as the upper level government implements the measures to incentivize policy adoption or penalize non adoption (Frances S. Berry & Berry, 2007; Shipan & Volden, 2008) . For this reason, coercion is often described as a vertical influence (Frances S. Berry & Berry, 2007; Shipan & Volden, 2008, 2012) . In describing learning as a diffusion mechanism, Shipan and Volden (2008) other jurisdictions for the policy adoption in a jurisdiction that has not yet adopted the policy. Policy competition as a driver for policy diffusion involves economic spillover (Shipan & Volden, 2008) or economic competition between states for businesses and tax revenues (F. J. Boehmke & Witmer, 2004) . States may engage in the competition to attract private investments and businesses for potential growth in employment and tax revenue whereas they try to avoid attracting undesirable groups (Karch, 2007) .
18 Most RPS studies that examine horizontal diffusion rely on policy adoption in neighboring states (Carley & Miller, 2012; Chandler, 2009; Matisoff, 2008) or ideologically similar stat es (Carley et al ., 2016; Chandler, 2009) . However, such geographical proximity or ideological similarity may not be the only factors that states consider when adopting polic ies . States may refer to the policy decisions in other states that share similar economic, envir onmental, energy, and risk attributes as they would adopt RPS for various reasons. Thus, it is plausible that states would refer to decisions made by other states facing common challenges or characteristics to imitate the policy options as well as to predi ct the policy feasibility in a similar policy environment . In dyadic analysis that directly compares two states in the same observation , horizontal diffusion factors are simply calculated as the absolute difference in the internal determinants between two states. Similar States Hypothesis : A state is more likely to adopt RPS if other states that share similar conditions have RPS. Controls Other i nternal determinants of RPS adoption Net metering Net metering polic ies allow private utility industries to co nnect their renewable energy generators to the power grid and sell the power generated from the renewable energy sources for monetary benefit. As of 2015, 44 states have net metering policies. T he purpose of net metering and RPS is to promote the use of re newable energy sources for electricity generation. However, n et metering is limited in its scope, focusing on small scale generation , thus spurs the consumers renewable energy production to only a limited extent (Duscha, Held, & Rio, 2016; Forsyth, Pedden, & Gagliano, 2002) . Carley et al. (2016) and Yi and Feiock (2012) foun d that the existence of net metering policies are likely to promote the adoption of RPS. A state with a net metering policy may experience the increase and promotion of renewable energy
19 production at relatively low cost and small scale in advance, thereby increasing the likelihood of RPS adoption. Political ideology For political factors associated with state policy adoption, political ideology of state legislatures and governors are often employed to explain the relationship between politics and policy adoption. Historically, l iberal states are more likely to be favorable to the adoption of policies tha t are purposed to expand social welfare (William D. Berry, Ringquist, Fording, & Hanson, 1998; Fellowes & Rowe, 2004) or environmental actions (Carley & Miller, 2012; Chandler, 2009; Lyon & Yin, 2010; Yi & Feiock, 2012) whereas conservative states are more amenable to economic development policy. Liberal state government ideology is expected to have positive correlations with RPS. Population growth Demographic information , such as the size of state, education levels, and ethnicity of state citizens , can shape policy adoption decisions (Rogers, 1995) . Traditional policy innovation literature focuses on the size (population) of state as an explanatory variable for policy innovation (V. Gray, 1973; Walker, 1969) . In general, these researchers hypothesize that a larger (populous) state is more likely to adopt a new policy (Walker, 1969) . In terms of state population size and the RPS adoption, more rapidly growing stat es are under pressure for additional electricity generation to meet the increasing demand. RPS adoption literature suggests that the extent to which sta tes need to generate additional electricity is largely dependent on the change in the demand. The change in the electricity demand is better captured by change in the population rather than state s total population. For this reason, existing RPS studies tend to estimate the effect of population growth or change measure (M. J. Berry, Laird, & Stefes, 2015; Carley & Miller, 2012; Carley et al., 2016; Chandler, 2009; Huang et al., 200 7) . B ut , only M. J. Berry et al. ( 2015) found statistically significant positive correlation between
20 population change and the lik elihood of RPS adoption . Similar to previous studies assumptions on the effect of population growth on RPS, population growth would be positively linked with RPS. Fossil fuel consumption One of the important indicators that create the pressure for or aga inst the renewable energy production would be CO 2 emission levels. Scholars have used various measures such as CO 2 intensity and air pollutants (Matisoff, 2008) , nonattainment index and emissions from electric generation (Lyon & Yin, 2010) , coal based energy consumption for electricity generation (M. J. Berry et al., 2015) and CO 2 emissions per capita (Carley et al., 2016) . One of the RPS goals is to redu ce CO 2 emission levels by increasing the renewable energy production. RPS is often described as a measure to reduce CO 2 emission s and slow down the anthropogenic climate change. By logical sense, states with high CO 2 emissions should adopt stringent RPS policies to reduce CO2 emissions from electricity generation. However, high CO 2 emissions from power generation signal that a signi ficant share of state economy is driven by the extraction of fossil fuel sources as well as the production and sales of electricity generation from fossil fuel sources. Thus, empirical studies of RPS adoption often found the negative association between CO 2 emissions or coal dependence and the likelihood of RPS adoption (M. J. Berry et al., 2015; Carley et al., 2016; Matisoff, 2008) . I expect the negative association between the likelihood of RPS adoption and goals. State wealth Another critical indicator for policy adoption is state wealth since states with more resources can afford the costs and risks associated with the adoption of new policy (Walker, 1969) . State wealth has been consistently positively related to the likelihood of RPS adoption (Carley et al., 2016; Chandler, 2009; Huang et al., 2007; Upton & Sny der, 2015; Yi & Feiock, 2012) . However, s tate wealth is an ambiguous concept that needs additional specification. The
21 expansion of renewable energy production is a costly decision that calls for additional investment in electricity infrastructure and bearing of short term increase in the electricity production costs. In this light, affluent states are more capable of supp orting the RPS , and state citizens are more likely to afford the price tag attached to renewable energy production. Electricity price States with RPS tend to experience the increase in electricity prices for expansion of renewable energy production (Bird et al., 2010) . Notwithstanding the unavoidable costs associated with renewable energy production, renewable energy production has positive impacts on the energy supply as it promotes the energy security through the diversification of energy sources as well as the stable energy prices in the long term (T. Berry & Jaccard, 2001) . Fischer ( 2009) also discussed that RPS can either increase or decrease the electricity price, largely dependin g on the elasticity of renewable energy supply. In this vein, the mixed findings on the effect of electricity price on RPS adoption are not surprising because of the discrepancy between short term and long term effects. Lyon and Yin (2010) did not find sta tistically significant correlation between electricity price and RPS adoption whereas Carley et al. (2016) and Berry et al. (2015) found positive influence of electricity prices on RPS adoption and stringency measures. The expected relationship between ele ctricity price and RPS is inconclusive, yet it is worth examining how a state taps into renewable energy to respond to changing electricity prices. Renewable energy potential Renewable energy potential indicates the amount of electricity that can be gener ated from renewable energy given renewable energy sources and technical advancement (Lopez, Roberts, Heimiller, Blair, & Porro, 2012) . Renewable energy potentials are S tudies th at looked into the effect of renewable energy sources in each state focused exclusively on solar and wind energies
22 s ince the majority of renewable energy sources utilized in the U.S. is solar and wind energ ies (M. J. Berry et al., 2015; Lyon & Yin, 2010; Yi & Feiock, 2012) . Although these past studies use different indicato rs for renewable energy potentials and exhibit differences across the types of renewable energy potentials that affect the RPS adoption, they still found in common that the potential for renewable energy substantially affect the RPS adoption. Duration RPS duration is introduced in the analytical models since more states have adopted RPS over time . I control for the duration of RPS by counting the number of years since RPS was first adopted in the U.S. The number of states adopting RPS has increased gradual ly , and states accumulate knowledge and information on RPS from existing RPS states . The RPS duration variable is employed to account for such information accumulation and increasing adoption trends over time. Table 2 . 2 presents the measurement strategy for internal determinants and horizontal diffusion factors for the testing of hypotheses. Table 2. 2 . Measurement strategies for independent variables Independent variables Measurement Source Internal determinants Immediate crisis: Weather related hazards in state i Counts of weather related hazards event declared as emergency or disaster by FEMA in state i FEMA ( 2016) 7 Policy proximate crisis: Weather related hazards in s tate j Counts of weather related hazard event s declared as emergency or disaster by FEMA in state j FEMA ( 2016) Ne t metering D ichotomous variable indicating the existence of net metering in a state DSIRE (2017) 7 Some may view FEMA declared emergency or disaster as politicized measure of crisis events. Nevertheless, these are still one of the reliable measures for crisis events since FEMA is responsible for the nationwide management of large scale crisis events and has kept a historical record of large scale crisis events.
23 Table 2.2. Measurement strategies for independent variables cont d Independent variables Measurement Source State government ideology State government ideology score (0 100) in state i Berry et al. (1998 ) % CO 2 em issions per capita from electricity generation CO 2 emissions per capita from electricity generation in state i (metric tons) EIA (2016) Solar wind potential I ntegrated measure of technical potentials for wind and solar energy generation (G W ) in state i Lopez et al., ( 2012) Population g rowth A percentage change in population relative to in state i U.S. Census Bureau ( 2016) Real GSP per capita Real GSP per capita in state i U.S. Census Bureau ( 2016) Electricity price Annual average electricity price for total electric industry in state i EIA (2016) Duration of RPS in state i N u mber of years with RPS since 1997 DSIRE (2017) Horizontal diffusion Ideological similarity between state i and state j Difference in s between state i and state j Berry et al. (1998 ) Similarity in per capita CO 2 emissions from electricity generation between state i and state j Difference in per capita CO2 emissions from electricity generation between state i and state j EIA (2016) Similarity in r enewable energy potential between state i and state j Difference in r enewable energy potential between state i and state j Lopez et al., ( 2012) Similarity in p opulation between state i and state j Difference in p opulation between state i and state j U.S. Census Bureau ( 2016) Similarity in r eal GSP per capita between state i and state j Difference in r eal GSP per capita between state i and state j U.S. Census Bureau ( 2016) Similarity in energy price between state i and stat e j Difference in energy price between state i and state j EIA (2016) Neighbor Border sharing between state i and state j in the same observation
24 Analytical methods and Dependent variables E vent History Analysis (EHA) has been widely used for policy innovation and diffusion studies as it allows for the simultaneous analysis of internal and external determinants in the same model (Berry & Berry, 1990) . However, the conventional state year EHA has limitations in analyzing the effect of horizontal diffusion as it does not allow for the compa rison between previous adopters and potential adopters. On the other hand, dyadic EHA directly compares the policy decisions in two different states in a dyadic pair. More simply put, a researcher can compare a policy decision in a specific state with each respective state, rather than compare a decision in a state with average of all decisions made across all states . Recently, policy diffusion scholars have increasingly used dyadic EHA for more refined study of diffusion (Carley et al., 2016; Gilardi & FÃ¼glister, 2008; Nicholson Crotty & Carley, 2015; Volden, 2006) . Volden (2006) first employed dyadic approach to investigate different ance Program. A dyadic EHA allows the researcher to estimate whether the policy in state i ( follower or laggard ) changes to align with the policy in state j ( lea der) while controlling for the internal determinants of each state and the shared characteristics (Carley et al., 2016) . By directly comparing two states in a pair, the researcher can compare not only policy adoption decisions but also internal determinants of each state in a pair. Since dyadic EH A approach is used to analyze the alignment of policy decisions in state i and state j , the construction of DV is different from conventional EHA approach. The DV in i adopts in year t and state j in the same observation have RPS in or before t 1 . Table 2. 3 presents the DV and structure of dyadic analysis. Unlike in the state year EHA approach where a state is dropped from analysis once it adopts a new policy, d yadic
25 pair is dropped once state i and state j both adopted RPS and dyadic DV is coded 1. Thus, dyadic pairs after the occurrence of policy imitation will be dropped since no other imitation after t will be dropped. Tab le 2. 3 . Dyadic data structure ID ij State i State j Year Adoption i Adoption j Dyadic DV 1 AZ CT 2004 0 1 0 1 AZ CT 2005 0 1 0 1 AZ CT 2006 1 1 1 2 AZ DE 2004 0 0 0 2 AZ DE 2005 0 0 0 2 AZ DE 2006 1 0 0 Using Pro bit , t his chapter examines the drivers of RPS adoption across 46 states from 199 8 to 20 09 . 8 One may raise a concern on the time frame of this research that ends in 2009. Since 2009, Vermont is the only state that adopted RPS in 2015. Therefore, the study of RPS adoption un til 2009 would still offer valuable lessons for the indicators of RPS adoption. All independent va riables are lagged by one year , and standard errors are clustered by state dyads. Analysis Table 2. 4 presents the descriptive statistics of variables used fo r the model estimations . The first three rows exhibit data characteristics of three dependent variables. Further, the table provides the overall summary of crisis event count, relevant policies, socioeconomic factors as well as horizontal diffusion factors . The mean of imitation variable from 199 8 to 200 9 is 0.03 with the standard deviation of 0.1 8 . States experience 2.33 large scale climate related hazards in the past two years on average 8 Following the same logic with Carley et al. (2016), Iowa is excluded from this chapter because the RPS adopted in Iowa in 1983 is differen t from the current RPS adopted by other states. Further, MA, ME, and NV are dropped from state i since they are the first RPS adopters in 1997 and they do not have leader states from which they can imitate their policy decisions.
26 with minimum frequency as low as 0 incident and maximum frequency as high as 12 incidents every year. Net metering is measured using binary variable. The mean of net metering policy variable is 0. 39 with the standard deviation of 0. 49. Real GSP per capita ranges from $28,368 to $69,973. Table 2. 4 . Descriptive statistics Variables Obs Mean Std. dev Min Max DV Imitation 10141 0.03 0.18 0 1 Crisis Crisis inside a state (t 1, t 2) 10141 2.33 1. 80 0 12 Crisis outside a state (t 1, t 2) 10141 2.03 1. 80 0 11 Internal factors Net metering 10141 0.39 0.49 0 1 Population growth 10141 1.07 1.13 5.72 8.13 Real GSP per capita 10141 41578.82 7560.98 28368 69973 10141 44.21 22.24 6.51 90.79 CO 2 per capita from electricity 10141 15.15 17.86 0.01 97.26 Solar wind potential 10141 3271.80 2421.68 10.05 22466.8 Electricity price 10141 6.72 1.85 3.87 14.74 External factors Years with first RPS in the U.S. 10141 6.14 3.35 1 12 Neighbor 10141 0.75 0.26 0 1 Difference in population (logged) 10141 1.19 0.86 0 4.27 Difference in real GSP per capita 10141 9021.88 6844.3 7 0 37540 10141 27.68 19.32 0 84.78 Difference in CO 2 per capita 10141 10.78 16.68 0 95.63 Difference in solar wind potential 10141 3515.80 3963.26 0..59 22456.75 Difference in electricity price 10141 2.80 2.63 0 23.59 This chapter provides the empirical analysis of RPS adoption in models 1 and 2 as summarized in Table 2.5 . Models 1 and 2 analyze the lik elihood of RPS adoption in state i following RPS adoption decision in state j . Model 1 analyzed 2 1 , 481 observations , which include 4 9 state j as leader states and 46 state i as follower states. However, Boehmke ( 2009) raised a
27 concern that dyadic analysis may yield less than accurate fi ndings since late adopters appear to decision making processes merely because of the large data. His recommended solution for such bias is to limit the analysis t o the pairs that have a possibility of sharing the same policy decision. Thus , states that never adopted RPS are excluded from state j in model 2, leaving 10,125 observations with 28 state j and 46 state i in model 2 . Results are relatively consistent across two models except for the loss of statistical significance and reversed directions in electricity price and difference in state government ideology variables in model 2 . While the directions of coefficients for population growth and solar win potential rem ain constant across the analytical models, their statistical significance s ha ve improved in model 2. Since all internal determinants are measured using raw numbers except for net metering and EERS policy variables, the interpretation s of estimation results for weather related hazards, state government ideology, per capita CO 2 emissions, solar wind potential, population growth, real GSP per capita, and electricity price in Probit analysis are based on the effect of one unit increase in th ese indicators on the predicted probability of RPS adoption in state i following state j . The interpretation of net metering would involve the effect of existence of net metering in state i on the predicted likelihood of RPS adoption in state i following sta te j . Horizontal diffusion variables are interpreted differently from internal determinants. Neighbor variable indicates whether state i and state j are geographically neighboring states. The estimation of the rest of horizontal diffusion variables captures t he correlation between the absolute differences in indicators and the predicted likelihood of RPS adoption . In other words, the negative value of estimated coefficients for horizontal diffusion variables indicate that state i is more likely to adopt RPS fol lowing state j if state i and state j , share similar internal characteristics (i.e., the smaller the
28 differences between the two states are, the more likely state i is to imitate state j s RPS adoption decision). Appendices A and B provides the summary of chang es in probabilities for models 1 and 2, respectively. First, large scale weather related crisis events inside a state are positively related with the lik elihood of RPS adoption . However, policy weather related crisis events in other states come into effect in an oppo site direction. A state is less likely to adopt RPS if the counts of large scale weather rela ted hazards in other state (state j ) in a dyadic pair increase. The direction and magnitude of association between crises and RPS adoption indicates that direct experience with relevant crisis positively influences the adoption of RPS, whereas the negative relationship between weather crisis in other states and a state the effect of policy proximate crisis. Specifically, t he probability of RPS adoption for a state i with twelve hazards events in the past two years (maximum) is 2.38% more than that for the one with no hazard events in the past two years. Second, in terms of internal conditions , net metering , wealth, and liberal state government ideology is positively linked with the likelihood of RPS adoption. In other words, a state with net metering policy, high real GSP per capita, and liberal state government is more likely to adopt RPS policy. Also, a state with high renewable energy potential is more likely to adopt RPS in model 2. On the other hand, per capita CO 2 emission s from electricity generation is negatively li nked with the likelihood of RPS adoption . From this result, it can be speculated that the presence of large fossil fuel industries in a state may present challenges or oppose RPS adoption. The effect of electricity price on RPS adoption appears to be mixed . In model 1, a state that pays high electricity price is less likely to adopt RPS, but the relationship between electricity
29 price and RPS adoption is no longer significant in model 2. Population growth and solar wind potential are positively related to RP S adoption in model 2 . Third, state considering the adoption of RPS is likely to take cues from existing RPS states that share similar economic conditions . Neighboring state is not a statistically significant indicator of RPS adoption after controlling d ifferent types of similar states. As for the effects of horizontal diffusion factors, a state is more likely to adopt RPS if other state s that share similarities in per capita real GSP per capita and/or CO 2 emission s from electricity generation had already adopted RPS. Also, similar state government ideologies increase the likelihood of shared RPS adoption decisions in model 1. Intuitively, $1,000 less difference in real GSP per capita between state i and state j increases the probability of RPS policy imitat ion by 0.02%. Similarly, 1% less difference in per capita CO 2 emissions from electricity generation is associated with 0.05 % higher probability of RPS policy imitation in two states in a dyadic pair. However, estimated coefficients for differences in solar wind potential, and electricity price exhibit positive signs, ind icating the negative relationship between these similarities and RPS adoption decisions. T h e positive coefficients for these variables indicate that a state with less renewable energy potential tend s to refer to RPS adoption decisions in other states with higher renewable energy potential. Moreover, a state with low electricity price is likely to imitate RPS adoption decisions in states with high electricity prices.
30 Table 2. 5 . Estimation results RPS adoption using dyadic Probit Variables DV: RPS adoption in state i and state j Model 1 Model 2 Crisis Crisis inside a state 0.0544*** 0.0511*** (0.0129) (0.0150) Crisis outside a state 0.0483*** 0.0301** (0.0130) (0.0149) Internal factors Net metering 0.147** 0.369*** (0.0618) (0.0720) Population growth 0.0169 0.0434* (0.0218) (0.0226) Real GSP per capita (in thousands $) 0.0191 *** 0.0260 *** ( 0.0036 ) ( 0.0045 ) 0.00925*** 0.0153*** (0.00127) (0.00165) CO 2 emissions per capita from electricity 0.00694* 0.0139** G eneration (%) (0.00412) (0.00610) Solar wind potential ( in thousands GW) 0.0077 0.0053 *** ( 0.0088 ) ( 0.0017 ) Electricity price 0.0560*** 0.0212 (0.0124) (0.0179) External factors Years with first RPS in the U.S. 0.109*** 0.137*** (0.00971) (0.0127) Neighbor 0.0913 0.109 (0.0891) (0.103) Difference in population (logged) 0.0102 0.0455 (0.0326) (0.0396) Difference in real GSP per capita 0.0123 *** 0.0075 * ( 0.0039 ) ( 0.0045 ) 0.00414*** 0.000691 (0.00129) (0.00152) Difference in CO 2 emissions per capita 0.0299*** 0.0153** from electricity generation (0.00347) (0.00658) Difference in solar wind potential 0.0270 *** 0.0128 * ( 0.0067 ) ( 0.0071 ) Difference in electricity price 0.0536*** 0.0289*** (0.00871) (0.00943) Constant 3.782*** 5.188*** (0.165) (0.267) Observations Pseudo R squared 21,481 0.16 33 10,141 0.22 61 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
31 [Note] Model 1 analyzed 2 1 , 481 observations, which include 49 state j as leader states and 46 state i as follower states. For the purpose of robustness check, states th at never adopted RPS are excluded from state j in model 2, leaving 10,125 observations with 28 state j and 4 6 state i in model 2. Conclusion and Policy Implications Many countries ha ve promoted the use of renewable energy and/or set CO 2 reduction goals at the national level as they acknowledge a close relationship between the emissions from fossil fuel combustion and climate change (Sovacool & Barkenbus, 2007) . In tune with national and gl obal trends, scholars have advanced the explanation of RPS adoption drivers over the last decade. S cholars have reach ed some consensus on a few of internal determinants of RPS adoption , such as state wealth (Carley et al., 2016; Chandler, 2009; Huang et al., 2007; Upton & Snyder, 2015; Yi & Feiock, 2012) and state government ideology (Carley & Miller, 2012; Chandler, 2009; Lyon & Yin, 2010; Yi & Feiock, 2012) . Yet, t here remains much to be studied in terms of roles of crises, institutional contexts, and diffusion mechanisms . This chapter is set out to examine the r oles of direct and indirect crisis event experiences in the adoption of state RPS policies. Further, I investigated whether states refer to other states RPS adoption decisions that share similar internal conditions other than geographic borders. Although the theories of policy change explicitly discuss the role of focusing events or crisis in facilitating the context for policy change, the diffusion of innovation theory has paid limited attention to the effects of crisis events on the policy adoption. Usi ng the crisis typology suggested by Nohrstedt and Weible (2010), this chapter finds that immediate crises represented by large scale weather related hazards increases the likelihood of RPS adoption decision and the size of RPS goals, whereas weather relate d hazards experiences in other states negatively affect the likelihood of RPS adoption. Because of the intangible nature of climate change, people tend to downplay the importance of climate change mitigation and the use of clean energy.
32 Scientifically, we cannot attribute the occurrence of each of these large scale hazards to climate change. However, this is how non scientists and average citizens perceive and interpret the crisis experiences. This is more about human behavior and perception rather than the hard environmental and natural science. In the real world and our daily lives, crisis events signal the urgent need for solutions. Further, citizens and policymakers are able to turning this crisis experiences into opportunity for better future as they re spond to crises by adopting a renewable energy policy. Therefore, in this chapter , I find that a directly experienced crisis event plays significant role as an internal factor for policy adoption. However, the negative relationship between crisis events ou tside a state and RPS adoption can be traced to the nature of large N data for dyadic analysis. Only 342 dyadic pairs, equivalent to 18.24% of total dyadic pairs, have shared their RPS policy adoption decisions, while 81.76% of dyadic pairs did not engaged in the imitation of RPS policies . The multiplicative and time dependent nature of dyadic data drastically increased the number of pairs with dependent variables coded 0, marginalizing or introducing bias to the effects of external variables. Another possi ble reason for the negative coefficients for external crisis variable arises from the processing of information that occurred outside a state. Except for few hazards events that capture national level attention such as Hurricanes Katrina , Sandy, and Harvey, state governments may not have concrete information on the severity or frequency of large scale climate related hazards that took place in other states. In fact, the negative coefficient of external crisis opens up the avenue for additional resear ch on the specific types or attribu tes of crisis events that offer lessons to other states. For the study of policy diffusion, scholars have largely relied on the effect of policy adoptions in neighboring states. Although neighboring effects is the most frequently examined variable for the analysis of policy diffusion across multiple states, recent diffusion scholars
33 raised questions on the neighboring effects as a factor for policy diffusion and discuss ed the need for exploration of nuanced external or d iffusion factors beyond neighboring effects (Gilardi, 2016; Shipan & Volden, 2012) . Due to technological advancement and better access to information, state government may also look to other states that similar internal conditions or grapple with common issues. In this chapter, I found that politically and economically similar states are likely to share RPS adoption decisions. In fact, renewable energy policy is not just about using clean energy sources and reducing CO 2 emissions from electricity generation. T h ey are highly politicized and have an impact in local and state economy. Because the adoption of renewable policy incurs political and economic costs, it makes more sense for potential policy adopters to take cues from the policy decisions in other states that are politically or economically similar and reduce the uncertainties in cos ts associated with the policy. In doing so, state policymakers make better informed policy decisions. On the other hand, a state with the low levels of solar wind potential and electricity price imitates RPS policy adoption decisions in other states with h igh solar wind potential and electricity price. T h is implies that a state with fewer renewable energy sources and more fossil fuel supplies takes proactive measures for the switch to clean energy sources by imitating RPS policy adoption in states with more favorable conditions for renewable energy productions. T his chapter has several limitations. First , policy design elements of RPS , such as penalty for noncompliance and types of renewable energy allowed , are not taken into account in this chapter . Thorough analysis of RPS adoption that consider various aspects of RPS besides renewable energy production schedule would provide more concrete knowledge of policy adoption. Second, the number of observations for dyadic dataset is conspicuously higher than that for the dataset for traditional EHA due to the multiplicative nature of dataset in dyadic
34 analysis , resulting in numerous observations with 0 for the dependent variables . Accordingly, the statistical power of dyadic analysis is limited due to the rar ity of policy adoption or imitation activities where the dependent variable is coded 1. Third , sole reliance on secondary data is not sufficient to answer how each state set s their annual RPS goals and overcomes resistance from the utility industries. Many states struggled with the implementation of RPS during early RPS years because of the multiplicity of interests and conflicts around RPS such as local resistance to the renewable energy generating facilities and pressure from stakeholders for particular r enewable energy sources (Rabe & Mundo, 2007) . However, various interests and conflicts ar ound RPS and its goals are less likely to be identified by the analysis of state level secondary data. Therefore , future studies may consider a mixed method approach to RPS adoption and goals for the exploration of various contextual factors that shape sat es RPS policies specific to each state.
35 CHAPTER III. EVALUATING RENEWABLE PORTFOLIO STANDARDS: HAVE THEY KEPT THEIR ENERGY AND ECONOMIC PROMISES? A bstract Renewable p ortfolio s tandards (RPS) are adopted by numerous states in the U.S. to achieve various goals including energy, environmental, and economic goals. To date, RPS evaluation literature has predominantly focused on the effectiveness of RPS policies in achieving renewable energy expansion and CO 2 emissions reduction. However, there is a dearth of research on the effectiveness of RPS policies in promoting electricity price stability, which is one of the main goals of RPS policies. Accordingly, this chapter examines whether RPS has contributed to the improvement of electricity price stability from 199 8 to 201 2 in 50 states. This chapter finds that an RPS polic y is positively correlated with renewable energy generation. Further , it has been effective in reducing electricity prices over time, contrary to the renewable energy opponents concerns on the increases in electricity prices. Introduction Knowledge gained from policy evaluation can inform various stages of policy processes and add political justification and legitimacy in shaping the future direction of government (Sanderso n, 2002) . Given its value and useful ness in assessing government , policy evaluation requires rigo r in precis ely measur ing and gathering data about problem attributes and different aspects of the design and goals of policies. Accordingly, p olicy evaluati on involves two primary tasks : the assessment of achievement of intended policy goals and of the extent that the observed effects can be attributed to the evaluated policy or program (Wollmann, 2007) . The goal of this chapter is to conduct both of these tasks in the evaluation of r enewable p ortfolio s tandards (RPS).
36 In the U.S., RPS policies are adopted by states to achieve multiple goals related to energy, e nvironmental, and economic benefits (Chen, Wiser, Mills, & Bolinger, 2009; Cory & Swezey, 2007; Davies, 2014; Holt & Wiser, 2007; Hurlbut, 2008) . Yet, RPS policies may not be sufficient by themselves to achieve their broad policy goals . In fact, other policies may be more effective than RPS in achieving these goals. In this vein , a careful examination of the relationship between different energy policies and clean energy development should be conducted to assess the effectiveness of a policy under study (Dong, 2012) . The evaluation of RPS effectiveness provides critical insights into RPS for scholar s and policymakers in states with and without RPS in different ways. RPS evaluation information serves as an indicator for the progress that the policy has made to date. States considering RPS adoption may refer to RPS effectiveness studies in making their future policy decision. T his chapter evaluates the extent to which RPS policies have achieved the manifested goals of increas ing renewable energy generation, and stabilizing energy price s . Although renewable energy generation is one of the most extensivel y studied topics in the RPS literature, it suffered some methodological issues such as an overreliance on binary measures of RPS adoption and endogeneity issues associated with RPS stringency indicator (Carley, 2009; H. Yin & Powers, 2010) . In an effort to address these measurement issues, I employed duration variables for RPS and other relevant policy instruments to measure the effect of their experiences over time. Further, RPS s effectiveness in energy price stability has been less studied compared to its energy and environmental benefits ( e.g., Carley, 2009; Menz & Vachon, 2006; Prasad & Munch, 2012; Sekar & Sohngen, 2014; Yin & Powers, 2010) . This chapter is expected to serve as the building b locks of the studies of energy policies on electricity price stabilities.
37 This chapter finds that RPS polic y duration is positively correlate d with renewable energy generation, whereas binary RPS adoption indicator is not related to renewable energy gener ation . As for the effectiveness of RPS policies in achieving electricity price stability, RPS is not statistically significant indicator for electricity price stability. Rather, Energy Efficiency Resource Standard and Mandatory Green Power Option are more effective in stabilizing electricity prices in a state. Overview of RPS goals States with RPS policies set annual schedules for the mandatory share of renewable energy production or sales, mostly expressed in the percentage of state s total electricity pr oduction or sales except for New York and Texas. Annual RPS schedules substantially vary across states, ranging from Missouri , setting a target of 15% renewable energy sales by 2021 to Hawaii , planning to achieve 100% renewable energy sales by 2045. RPS po licies have been one of the most widely adopted policies among state level environmental and renewable energy policies (Shrimali, Jenner, Groba, Chan, & Indvik, 2012) . RPS policies are employed by sta tes for many reasons, including to increase renewable energy production, reduce CO 2 emissions, and stabilize the energy economy (Davies, 2014; Hurlbut, 2008; Johnson, 2014) . Since the 1990s, state governments have adopted RPS policies at a rapid pace and competitively set higher renewable energy goals. A multitude of RPS goals have played critical roles in garnering support from citizens and mitigating objections from industries and other stakeholders. Increasing renewable energy generation is the immediate goal of RPS policies . The en ergy goal of RPS policies is the most important among other goals since other expected environmental and economic goals of RPS policy can be achieved by a switch in the energy generation from fossil fuel s to renewable energy. As for the RPS effectiveness on renewable
38 energy generation, the literature tends to show a positive relationship between RPS and renewable energy generation. For example, Carley (2009) found positive effect s of RPS policies on the amount of renewable energy generation but not on the share of renewable energy out of total electricity generation. Fischlein and Smith (2013) took a more detailed approach to RPS by using annual RPS schedule and other detailed design attributes of RPS policies to estimate the effect of RPS policie s on renewable energy generation . They found that the stringent RPS schedules and the presence of unbundled renewable energy credits (RECs) increase the share of renewable resources . Although Fischlein and Smith (2013) s approach to RPS stringency is sophisticated and captures the variability of RPS across states, scholars raised concerns on using RPS stringency as an independent variable because of endogeneity issues (Prasad & Munch, 2012; Yin & Powers, 2010) . Scholars also examined the effectiveness of RPS in achieving CO 2 emissions reductions , and they consistently found that RPS contributes to the reduction of CO 2 emissions (Chen et al., 2009; Kydes, 2007; Palmer & Burtraw, 2005) . Although Prasad and Munch (2012) found no statistically significant relationship between the presence of RPS policy i n a state and CO 2 emissions, the years of experience with RPS policy was negatively related to CO 2 emissions. Eastin (2014) examined the effectiveness of RPS on CO 2 emissions and state s air quality. The negative correlation between RPS adoption and CO 2 em ission s and the positive correlation between RPS adoption and better state air quality indicate that RPS policies can be used as a part of climate change mitigation measures . Sekar and Sohngen (2014) assessed the effectiveness of RPS on CO 2 emissions from 1997 to 2010. Controlling for regional effects and the structure of state economy, average temperature in January and July, and energy prices, their findings indicate d that RPS reduced CO 2 emissio ns by 4% by 2010.
39 The economic goals of RPS policies include economic development throu gh job creation (Bird et al., 2010; Rabe, 2008; Wiser et al., 2016) . Another critical economic benefit of RPS is electricity price stability (Cory & Swezey, 2007; Heeter et al., 2014; Rader & Hempling, 2001) . In the state of Washington, the declaration of policy in Energy Independent Act explicitly describes electricity price stability as one of expected benefits of RPS (Washington State Legislature, 2006) . 9 Knapp (2012) studied the effect of RPS on state level employment and found that RPS adopters are more likely to lose jobs as opposed to non adopters. However, the dependent variable in his study is measured by the gross number of employed people in a state and it does not necessarily capture the employment effect specifically from RPS. Yi ( 2013) examined the effects of state and local clean energy and climate policies on the number of green jobs. In measuring state and local clean energy and climate policies, he used an index to capture various policies adopted in one variable. The study found that energy and climate policies are positively related to the number of gree n jobs. Although this study focused on the creation of green jobs as effects of multiple energy and climate policies, the scope of study is limited in the year of 2006 and metropolitan areas in the U.S. In fact, it is very difficult to measure the employme nt effect as it requires data specific to the renewable energy industry and the fossil fuel industry. Once RPS states increase renewable energy production and reduce the use of fossil fuel resources, the genuine employment effect is realized only if the nu mber of job s created in the renewable energy industries exceeds those lost in the fossil fuel industry. However, employment effect s cannot be analyzed for now because of data availability issues. The best available 9 plentiful local resources will stabilize electricity prices for Washington residents, provide economic benefits for Washington counties and farmers, create high quality jobs in Washington, provide opportunities for training apprentice workers in the renewa ble energy field, protect clean air and water, and position (Washington State Legislature, 2006)
40 industry specific data is utility industr y employment data but it only gives overall employment measures, rather than energy or resource specific employment information. Another important economic goal of RPS policies is energy price stability , which serves as a critical driver for RPS adoption ( Hurlbut, 2008; Lieberman & Doherty, 2008; Rader & Hempling, 2001) . RPS would improve energy price stability by increasing renewable energy generation, which will in turn make energy prices more predictable, and energy intensive industries will subsequen tly become less susceptible to the volatile nature of energy prices. However, RPS evaluation literature lacks the analysis on the effect s of RPS policies on electricity price stability in spite of the growing importance of improvement in energy price stabi lity and mounting tensions and conflicts around the supply of oil and gas. Given that energy price stability is a critical part of RPS goals and the significance of electricity price stability on the national and global economy, this chapter seeks to disce rn the effect of RPS policies on the improvement of electricity price stability. Evaluation Framework Dependent v ariables This chapter examines the effectiveness of RPS in increasing renewable energy generation and electricity price stability from 199 8 to 201 2 in 50 states . Since more than 50% of state RPS policies became effective on or after 2007, the estimation of 10 year electricity price stability would yield less accurate findings on the relationship between RPS and electricity price stability. Ac cordingly, this chapter . For the calculation of electricity price stability, this chapter employs relative % change in the annual electricity prices . The measurements of electricity price stability are summarized in Tabl e 3.1.
41 Table 3.1 . Measurement of electricity price stability DV Measurement E lectricity price stability = electricity price in year t F or example, electricity price s for total electricity industry in Hawaii in 2002 and 2003 were 13.39 and 14.47 cents per kwh , respectively . In the observation for Hawaii in 200 3 , electricity price stability variable is calculated as: Using th e equation presented in Table 3. 1 , electricity price stability DV for Hawaii in 200 3 is 8.07%. Data for electricity price were collected from EIA ( 2016) . Independent variables RPS policy I first utilize d a binary indicator of the presence of RPS in the analytical models for the estimations of effects of RPS policies . M ost RPS effectiveness studies relied on a dichotomous variable as an explanatory variable for RPS , as it serves as a primary indica tor for RPS policy in a state (Carley, 2009; Eastin, 2014; Knapp, 2012; Sekar & Sohngen, 2014) . In case of RPS specifically, all RPS states have time variant renewable energy sales goals. These goals increase annually or biannual ly at a varying rate. In fact , Yin and Powers (2010) captured the
42 heterogeneity of annual RPS goals for the measurement of RPS in their study of RPS effectiveness. However, they acknowledge the possibility of endogeneity issues in their study. Further, Pra sad and Munch (2012) also attempted to use the s ame variable as Yin and Powers (2010), but they did not include the estimation results in the ir study due to the endogeneity issue in the model. Instead, Prasad and Munch relied on the policy duration measure s in their study of RPS and its effectiveness in CO 2 emissions reduction. F o llowing Prasad and Munch (2012) s approach, this chapter will also use RPS duration variable to represent RPS policy in a state in addition to a binary indicator of RPS adoption. T he duration of RPS is a salient aspect of program evaluation because of the cumulative nature of the policy effect. Jacobs ( 2014) posited that su fficient time is one of the most critical elements of amendments for policy success after initial adoption. RPS duration accounts for not only experiences with RPS, but also annual increments in RPS goals to some extent. Following their approach, this chap ter also uses the measure of RPS duration in addition to the binary indicator for the presence of RPS in each state. Controls This chapter seeks to evaluate the effectiveness of RPS polic y on the promotion of electricity price stability . O ther programs relevant to renewable energy or the operation of RPS, electricity supply and demand, state s that affect electricity price stability are controlled in an attempt to exclusive ly focus on the relationshi p between RPS polic y and its promised benefit of electricity price stability. Relevant policies to RPS To date, states have competitively adopted RPS particularly over last ten years. RPS policies have also gained support from citizens for multiple benefi ts as seen in the RPS adoption in Colorado, Washington, and Missouri as a result of ballot initiatives. However, RPS polic y
43 alone may not be sufficient to achieve its policy goals. In fact, RPS goal of renewable energy generation and electricity price stab ility can be also met , pursued , or affected by other energy relevant policies because states use multiple energy and environmental policy instruments to tackle energy and environmental problems . For this reason, policy evaluation is conducted not only to i nvestigate the effectiveness of policy in achieving its intended policy goals but also to examine if the observed effects are driven by the policy or program subject to evaluation (Wollmann, 2007) . T herefore , a careful examination of the relationship between different policy instruments and energy policy goals should be conducted to assess the effectiveness of a policy under study without bias (Dong, 2012) . All policy variables are measured in two ways. Consistent with the measurement of RPS policy in a state, I will use binary variables for each policy to indicate the presence of policy in a state, and policy duration variables representing the number of years with each policy in a state. Net m etering States have adopted and tried various measures to encourage the use and production of renewable energy and net metering, which is one of the options for commercial and residential consumers directly engaging in renewable energy production. Renewable energ y producing consumers can reduce their utility bills using the credit received by their in house renewable energy generation. As of 2016, 44 states have net metering policies. As more households are able to procure electricity on their own through this pro gram, net metering is thought to contribute to increases in total renewable energy generation in a state as well as declines in price fluctuations. However, the findings from empirical studies are contrary to such assumptions on the positive effects of net metering. Yin and Powers (2010) analyzed the relationship between net metering and renewable energy capacity and found a negative correlation between them, although it is not statistically significant. In line with their findings,
44 Shrimali et al. (2012) f ound a statistically significant negative effect of net metering on renewable energy capacity . Data on the presence of net metering in a state are collected from DSIRE ( 2017) and state utility laws. Energy Efficiency Resource Standards ( EERS ) EERS require the reduction in energy use or demand on an annual basis , thereby encouraging more efficient use of electrici ty. The coexistence of RPS and EERS indicate the increase in the amount of renewable energy generation and reduction in electricity demand at the same time. Since the achievement of annual RPS goals is calculated as % of total electricity sales, EERS helps the achievement of annual RPS renewable energy sales goals by reducing the total electricity demand or sales, which is the denominator of % RPS goals . For this reason, EERS is often considered complementary to RPS . Therefore, EERS would indirectly increas e the share of renewable energy generation and possibly contribute to electricity price stability in a state. As of 2016, 22 states have EERS in place. State EERS data are collected from DSIRE (2017) and state utility laws. Mandatory Green Power Option (M GPO) MGPO provides customers with the option of choosing electricity produced from renewable energy sources. Some states adopted mandatory green power option when they implemented RPS (Menz, 2005) . This option may be conducive to the operation and further expansion of renewable energy as more customers opt fo r clean energy sources. Yin and Powers (2010) found the positive correlation between MGPO and renewable energy generation capacity in addition to the positive effect of RPS policy on renewable energy generation capacity. Shrimali et al. (2012) also showed the positive relationship between MGPO expansion would be led to the promotion of electricity price stability by increasing renewable
45 energy generation in a s tate and reducing the reliance on fossil fuel sources for electricity generation. State MGPO data are collected from DSIRE (2017) . Public Benefit Funds (PBF) PBF provides financial support for environmental actions and programs with tax and fees collected from electricity use. As of 2016, 19 states have adopted well as CO 2 emission s reductions. Previously, a number of studies that evaluated the effectiveness of RPS in achieving its energy or environmental goals probed into the roles of PBF as well since PBF facilitates the operation of renewable energy (Prasad & Munch, 2012; Shrimali et al., 2012; H. Yin & Powers, 2010) . In the RPS evaluation studies, Prasad and Munch ( 2012) found that the substantial reductions in CO 2 emissions were correlated with the availability of PBF in a state. However, Yin and Powers (2010) and Shrimali et al. (2012) have not found a statistically significant relationship between PBF and renewable energy capacity or generation except for its positive effect on biomass energy generation capacity in Shrimali et al. (2012). In regards to electricity price stabi lity, PBF is expected to make positive contribution to electricity price stability as it supports renewable energy expansion in a state. PBF data was collected from DSIRE ( 2017) and state utility laws. Deregulated electricity market In the regulated electricity market, electricity markets are vertically integrated and customers rely on single electricity provide r for their electric utilities. On the other hand, electricity providers engage in price competition in the deregulated electricity market, and customers choose their own electricity providers. However, it should be noted that states with deregulated elect ricity markets have only partially deregulated their electricity provision structures. Carley (2009) indicated that deregulated electricity market is linked with
46 a ssociation with the amount of renewable energy generation and negative association with the share of renewable energy generation imply that electricity deregulation increases renewable energy generation through competition. However, price competitions in t he deregulated electricity markets would lead electricity suppliers to choose fossil fuel sources for electricity generation for the cost advantage. In regards to the electricity price stability in a state with deregulated market, I expect that deregulatio n would be positively linked with electricity price stability as electricity suppliers would strive to provide electricity at better prices and even at the fixed price for certain period of time. As of 2017, there are 16 states with deregulated electricit y markets. Data for deregulated electricity markets are collected from Electric Choice ( 2017) . Deregulations are measured in two ways, which include binary indicator for deregulated electricity market and duration of deregulated market. Socioeconomic and energy market attributes Electricity price A state with high electricity price is expected to less likely to increase renewable energy generation because an increase in renewable energy generation would require additional investment in renewable energy generation facilities. At the same time, high electricity price may indicate that a state is with fewer resources for electricity generation, making it favorable to renewable energy expansion. Electricity price is also an important determinant of electricity price stability. Since electricity price st ability in this chapter represents % of changes in electricity prices, the measure of electricity price stability would be different across states because one cent change in electricity price would give different % changes depending on the base electricity price. For instance, one cent change in a state with low electricity price indicates higher % change in electricity prices whereas that in a state with high electricity price would give
47 relatively lower % change in electricity prices. In this vein, I expe ct that electricity price is positively associated with electricity price stability. Electricity price data is collected from EIA (2016). State government ideology Liberalism is more likely to be linked with pro environmental decision than conservative co unterparts (Dunlap, Xiao, & McCright, 2001) . Liberal state legislature would more actively engage in environmental issues and show high interest in increasing renewable energy generation. State government ideology data is obtained from (William D. Berry et al., 1998) . Population growth rate A state with large population or requires additional electricity generation . However, how each state will meet the increased demand varies by state. Increasing deman d may not be solely met by renewable energy productio n and could hinder the achievement of RPS policy goals by increasing the use of nonrenewable energy. However, it is also likely that a state makes extensive effort to meet the demand by increasing renewa ble energy production, especially when interest in environmental protection and renewable energy is high. Thus, the size of state in terms of population may have an impact on RPS effectiveness in multiple directions. This chapter will use population growth rate and electricity generation per capita to control for the size of state. I will calculate the population growth rate by % change in population relative to previous year . State population data is collected from U.S. Census Bureau ( 2016) . Renewable energy generation capacity Electricity generation capacity indicates the maximum electric output an electricity generator can produce under specific condition (EIA, 2017b) . A state with high renewable energy generation capacity will find it easier to expand renewable energy production and also to achieve the various goals of RPS policies . In this light, renewable energy generation capacity serves as an intervening variable in this analysis. The adoptions of
48 RPS and other relevant policies do not directly affect states renewable energy generation. Rather, states should increase renewable energy generation capacity to generate additional ele ctricity from renewable energy sources. This chapter use s % renewable energy generation capacity out of total electricity generation capacity in a state. Data for the calculation of the share of renewable energy generation capacity is obtained from EI A ( 2016) . Electricity generation from fossil fuel sources Active fossil fuel industries in a state may hinder renewable energy expansion to protect their businesses. If fossil fuel production is a critical economic driver in a state, state government may be reluctant to increase renewable energy generation. At the same time, s tates that rely a large part of electricity generation on fossil fuel sources are susceptible to fluctuation s in electricity price because of the high price volatility of coal, natural gas, and oil. Furthermore, these states may encounter barriers in expanding renewable energy generation due to the resistance from existing industries, which would hinder the diversification of energy portfolio, a critical driver for electricity price stability . Electricity generations from fossil fuel sources are measured by % electricity generation from coal, natural gas, and oil, respectively. T h e data for electricity genera tion from fossil fuel sources is collected from EI A ( 2016) . F or the estimation s of renewable energy generation and electricity price stability , this chapter uses fixed effects regression models with controls for year effects. Further, standard errors of all models are clustered by states. Each state has unique culture, history, regulations, law, and economy that shape policy decisions as well as electricity industry. Further, y ear specific effects such as inflation rate, global financial crisis, and election should be also taken into account. All independent variables are lagged by one year. One may raise a concern about the employment of the same analytical models for renewable energy generation and electricity
49 price stability. The same models for two different RPS goals are used dues to the primary objectives of policy evaluation that seeks to identify the type of policies directly relevant to the observed effects. Analysis Tab le 3.2 presents the descriptive statistics of variables used for this chapter . Each state generates 3. 67 % of electricity from renewable energy every year on average and the standard deviation is 4.71 %. S tates experience 2.81 % change in electricity prices o n average. Average duration of RPS in a state is 1.72 years. N et metering is the most widely adopted policy , as the mean s of binary net metering policy variable and net metering duration variable are the highest among all relevant policy variables includin g RPS . As for electricity generation fossil fuel sources, the mean of state s coal production is the highest followed by natural gas and oil production. Table 3.2 . Descriptive statistic s VARIABLES Observation Mean Standard Deviation Dependent variables % share of RE generation E lectricity price stability 750 750 3.67 2.81 4.71 6. 22 Independent variables Relevant policies (binary) RPS 75 0 0. 31 0. 46 Net metering 75 0 0. 61 0. 49 EERS 750 0.15 0.36 MGPO 750 0.07 0.25 PBF 750 0.27 0.44 Deregulation Relevant policies (duration) RPS 75 0 75 0 0.28 1.72 0.45 3 . 27 Net metering 75 0 5.55 6 . 98 EERS 750 0.59 1.77 MGPO 750 0.27 1.19 PBF 75 0 1 . 71 3.40
50 Table 3.2. Descriptive statistics cont d VARIABLES Observation Mean Standard Deviation Deregulation 75 0 2 .1 4 4.06 Socioeconomic condition Electricity price 75 0 8.19 3.15 State government ideology 75 0 46.60 24.03 % population growth 75 0 1.06 1.17 % renewable energy capacity 75 0 3.38 4.43 % electricity from coal 75 0 47.25 29.98 % electricity from natural gas 75 0 17.58 21.13 % electricity from oil 75 0 4.02 11.77 In an attempt to detect the issues with multicollinearity, I examined the correlations between policy variables as well as renewable energy capacity as shown in Table 3.3 . Overall, correlations between two independent variables are weak, except for moderat e correlations between RPS and net metering, RPS and PBF, and PBF and deregulation. Further, the magnitude of correlation between RPS and renewable energy capacity is 0.26 for RPS adoption and renewable energy capacity, and 0.28 for RPS duration and renewa ble energy capacity, respectively . The correlation coefficients of 0.26 and 0.28 indicate weak positive relationships. Therefore, the correlation between independent variables should not be a critical concern in this analysis. Table 3.3 . Correlation betw een RPS and renewable energy capacity 3.3 .A : Binary variables for policy adoption, Renewable energy capacity in % RPS NM EERS MGPO PBF Dereg. RE cap. RPS 1.00 NM 0.43 1.00 EERS 0.52 0.22 1.00 MGPO 0.18 0.21 0.19 1.00 PBF 0.43 0.35 0.28 0.04 1.00 Dereg. 0.38 0.24 0.19 0.01 0.58 1.00 RE cap. 0.26 0.28 0.25 0.32 0.15 0.03 1.00
51 3. 3. B : Policy duration variables, Renewable energy capacity in % RPS NM EERS MGPO PBF Dereg. RE cap. RPS 1.00 NM 0.46 1.00 EERS 0.49 0.17 1.00 MGPO 0.13 0.25 0.11 1.00 PBF 0.50 0.38 0.32 0.10 1.00 Dereg. 0.34 0.12 0.16 0.05 0.49 1.00 RE cap. 0.28 0.38 0.24 0.32 0.16 0.03 1.00 Table 3.4 shows the estimation results for state level renewable energy generation. Model 1 includes binary policy variables, and model 2 examined policy duration variables or the length of time a policy has been in place, instead of binary indicators for policy adoption. In model 1, net metering policy is negatively r elated to the share of renewable energy generation. W hen policy durations are taken into account as in model 2, the duration of RPS is associated with increases in renewable energy generation shares, whereas the durations of net metering and deregulated el ectricity markets decreases renewable energy generation in a state. Each a dditional year of RPS experience is correlated with an additional 0.0738% of electricity generation from renewable energy sources. EERS, MGPO, and PBF are not statistically significa nt indicator s of renewable energy generation in both models 1 and 2. The effects of states socioeconomic and energy market conditions are consistent across models 1 and 2. In both models, electricity price and renewable energy capacity increase renewable energy generation. In other words, a state with higher electricity price and/or additional renewable energy generation capacity is likely to generate larger share of its electricity from renewable sources . However, population growth is negatively related to renewable energy generation, implying that additional demand may not be met by the use of renewable energy. Further, electricity generation from coal is linked with a reduction in renewable energy
52 gener ation. Thus, states that produce coal or rely on coal for electricity generation are less likely to generate electricity from renewable energy sources. State government ideology, and electricity generations from natural gas and oil are not statistically si gnificant.
53 Table 3.4 . results renewable energy generation (1) % RE generation ( 2 ) % RE generation RPS (binary) 0.0470 (0.156) Net metering (binary) 0.242* (0.132) EERS (binary) 0.0531 (0.168) MGPO (binary) 0.335 (0.230) PBF (binary) 0.0727 (0.178) Deregulation (binary) 0.0865 (0.266) RPS (duration) 0.0738** (0.0337) Net metering (duration) 0.0599*** (0.0220) EERS (duration) 0.0359 (0.0386) MGPO (duration) 0.0740 (0.0493) PBF (duration) 0.0360 (0.0299) Deregulation (duration) 0.0494* (0.0274) Electricity price 0.0953*** 0.0977*** (0.0320) (0.0360) Government ideology 0.00263 0.00308 (0.00256) (0.00258) Population growth 0.0777* 0.0755* (0.0414) (0.0411) Renewables capacity 0.793*** 0.805*** (0.0174) (0.0184) Electricity from coal (%) 0.0201* 0.0209* (0.0118) (0.0119) Electricity from natural 0.00393 0.000411 gas (%) (0.00972) (0.00968) Electricity from oil (%) 0.000167 0.00319 (0.0145) (0.0148) Constant 1.571** 1.640** (0.789) (0.774) Observations 750 750 R squared 0. 842 0. 844 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
54 Table 3.5 displays the analysis results of RPS effectiveness in electricity price stabilit y . Models 1 and 2 examine the relationship between the energy policy adoption variables and energy price stabilities , while models 3 and 4 probe the statistical relationship between how long energy policies ha ve been in effect and electricity price stability . M odel s 1 and 3 contain the output s of regression analys e s with state and year effects controls . However, the Wooldridge test for autocorrelation indicates that these traditional regression analyses suffer from autocorrelation issues because the F statistic and p value for model 1 are 14.417 and 0.0004, and those for model 3 are 12.372 and 0.0010 , respectively . The null hypothesis of no first order autocorrelation is rejected, and Cochrane Orcutt estimations in models 2 and 4 are introduced as remedies for the autocorrelation issues detected in models 1 and 3. The Cochrane Orcut t method introduces a lag definition and drops the first observation in the iterative method (Stata, 2013) . 10 Across all models , higher values of dependent variable accounts for higher change in electricity price, there by indicating instability in electricity prices. In other words, negative coefficients in these models indicate that independent variables are positively associated with electricity p rice stability. In model s 1 and 2 , RPS adoption dummy variable s are not statistically significant indicators of electricity price stabilities . In these models , net metering adoption variables are positively related to the changes in electricity prices, im plying that net metering contributes to increases in electricity prices. On the other hand, the existence of EERS is negatively linked with % changes in electricity prices, thereby reducing electricity prices over time. Specifically, the existence of EERS in a state is associated with 4% decline in electricity prices. However, the MGPO, PBF, and deregulation variables are not statistically significant. As for other controls of 10 Model 1 and 3 uses original dataset with 750 observ ations. Cochrane Orcutt drops 50 observations in models 2 and 4, leaving 700 observations for estimations.
55 socioeconomic and energy supply characteristics, electricity price and renewable energy capacity are positively correlated with the dependent variable, indicating that a state with high electricity price s and renewable energy capacity is more likely to experience a high rate of changes in electricity prices. State government ideology, population growth rate, and electricity generation from fossil fuel resources are not statistically significant. In models 3 and 4 , which take into account the duration s of energy policies, RPS and EERS are effective in reducing electricity prices. Each a d ditional year with RPS in a state is associated with 0.88% decline in electricity prices in model 4. EERS is only statistically significant in model 3, and each one year increase in EERS duration decreases electricity prices by 0.38%. Net metering, MGPO, P BF , and deregulation ar e not significant ly related to electricity price stability. Consistent with the results from models 1 and 2, electricity pric e and renewable energy capacity negatively affects electricity prices stability, while state government ideo logy, population growth, electricity generation from fossil fuel resources do not have statistically significant relationships with electricity price stabilities. In sum, RPS contributes to electricity price stability when taking into account the years of experience with RPS, and EERS also effectively reduce s electricity prices over time. On the other hand, the effects of net metering, MGPO, PBF, and deregulation r emain in question. Interestingly, state s paying high electricity prices experienced greater changes i n electricity prices than those paying less for electricity, and states that have increased renewable energy capacity experienced drastic increases in electricity prices.
56 Table 3.5 . Estimation results electricity price stability DV: Electricity price stability (1) Base model (2) Cochrane Orcutt (3) Base model (4) Cochrane Orcutt RPS (binary) 0.687 0.766 (0.805) (0.872) Net metering (binary) 2.167*** 2.708*** (0.752) (0.851) EERS (binary) 4.095*** 4.310*** (0.870) (0.933) MGPO (binary) 0.460 1.020 (1.178) (1.284) PBF (binary) 0.941 0.587 (0.945) (1.093) Deregulation (binary) 0.236 0.00216 (1.394) (1.817) RPS (duration) 0.645*** 0.888*** (0.176) (0.208) Net metering (duration) 0.116 0.103 (0.127) (0.165) EERS (duration) 0.381* 0.119 (0.203) (0.242) MGPO (duration) 0.0738 0.204 (0.252) (0.283) PBF (duration) 0.0972 0.234 (0.153) (0.181) Deregulation (duration) 0.0374 0.0723 (0.142) (0.172) Electricity price 2.098*** 2.236*** 2.262*** 2.921*** (0.206) (0.221) (0.219) (0.271) State gov t ideology 0.00686 0.00789 0.00188 0.0149 (0.0133) (0.0144) (0.0136) (0.0157) Population growth 0.00874 0.0786 0.0659 0.0486 (0.291) (0.302) (0.289) (0.308) Renewables capacity 0.216** 0.223** 0.172* 0.357*** (0.103) (0.111) (0.103) (0.127) Electricity from coal 0.0217 0.0406 0.00388 0.0197 (%) (0.0623) (0.0758) (0.0628) (0.0804) Electricity from natural 0.0175 0.0173 0.0319 0.0139 gas (%) (0.0523) (0.0638) (0.0517) (0.0667) Electricity from oil 0.0791 0.0957 0.0501 0.104 (%) (0.0750) (0.0915) (0.0771) (0.0991) Constant 21.38*** 24.36*** 12.80*** 29.35*** (4.655) (5.414) (4.342) (5.940) Observations 750 7 0 0 750 7 00 R squared 0. 353 0. 351 0. 355 0. 327 Standard e rrors in parentheses *** p<0.01, ** p<0.05, * p<0.1
57 Conclusion and Policy Implications RPS policies have been employed by many U.S. states in pursuit of meeting multiple goals through renewable energy production (Carley, 2011; Carley & Browne, 2013; Leon, 2012) . While considerable research has been accomplished to evaluate the effectiveness of RPS, the achievement of electricity price stability through RPS has received less attention from scholars compared to energy and environmental goals ( e.g., Carley, 2009; Prasad & Munch, 2012; Sekar & Sohngen, 2014; Yin & Powers, 2010) . Further, the existing studies of RPS and renewable energy generation did not control for the effects of other energy releva nt policies on renewable energy generation thoroughly. This chapter seeks to contribute to the existing RPS effectiveness literature by investigating the effects of RPS and other energy relevant policies on renewable energy generation and electricity price stability. RP s contribution to increases in state level renewable energy generation has been largely known and discussed to date, and it also holds true after taking into account states experiences with RPS. However, not all energy policies increase renewable energy generation in a state. For instance, net metering, EERS, MGPO, and PBF are related to state level renewable energy generation to varying extents. Net metering is adopted for renewable energy generating residential customers. EERS supports the implementation of RPS by reducing the total electricity demand. MGPO provides customers with options to use electricity generated from renewables. PBF offers funding support to clean and efficient energy use. However, none of these policies increase st ate level renewable energy generation, and net metering is negatively associated with renewable energy generation in a state. Although the positive relationship between RPS and renewable energy generation is similar to previous studies (e.g., Carley, 2009; Shrimali et al., 2012), an interesting lesson of this study is that RPS is the only effective policy for increasing
58 renewable energy generation. Therefore, a state seeking to expand renewable energy generation should consider the adoption of RPS, more tha n any other policies relevant to clean energy. T he duration of RPS is linked with declines in electricity prices al though a binary variable for RPS is not related to electricity price stability . Moreover, s tate s with EERS ha ve experienced decreases in el ectricity prices , while increase s in renewable energy capacity leads to higher electricity prices. The increase in electricity prices in state s with high renewable energy capacity may be attributed to costs incurred by upfront investment in facilities and R&D for renewable energy production . Despite the negative relationship between renewable energy capacity and electricity price stability , the positive contribution of RPS duration to electricity price stability reflects the betterment of renewable energy efficiency and production costs experienced over time. T he efficiency and production cost of renewable energy ha ve been significantly improved . Recently in Colorado, the median bid price for wind and storage turn is lower than the operating costs of existi ng coal plants (M. Gray, 2018) . S olar energy prices fell by 69% between 2010 and 2016, an d the price of wind power generation decreased by 18% in the same period (IRENA, 2018) . Although we are still in the infant stage s of renewable energy production, the continued expansion of renewable energy production facilities w ill achieve an economy of scale. Therefore, further declines in ren ewable energy investment costs are expected as the installation of renewable energy production facilities increase over time (IRENA, 2018) . A state with EERS is required to use less energy, ultimately increasing the predictabilit ies of electricity consumption s and electricity prices. Under current circumstances where renewable energy production is still quite expensive and under progress , reduction in e lectricity demand is a cost effective way to stabilize electricity prices . EERS place s fewer burdens on custom ers and incur s less cost for state government and utility providers than other policies. Therefore, this
59 study implies that electricity price stability is not only achieved by RPS , which seeks to diversify energy sources for electricity generation , but als o by EERS, a cost effective measure for both utility providers and administrators (ACEEE, 2017) . Although this chapter attempts to contribute to existing literature by examining the policy factors for electricity price stability, it has limitations in that it did not explicitly take into account technological advancement or cost improvement in renewa ble energy generation, which might make a difference in the estimation results. Another limitation of this study stems from that it heavily focused on clean energy policies as explanatory variables but did less on other detailed market characteristics that are specific to each state. The additional examinations of RPS compliance mechanism s , utility provision structures, and decision making processes would provide a more comprehensive explan ation of the state level clean energy markets. Nevertheless, this chapter s findings on the effectiveness of different energy policies set the stage for future study of the optimal mix of policy instruments for different policy goals. Energy policies are often implemented as a package to address environmental and economic challenges at the same time. However, governments often do so without concrete knowledge o f the effects of different policy packages. Accordingly, future studies would compare the effecti veness of different combinations of energy policies for renewable energy expansion, CO 2 emissions reductions, and economic growth to better inform policymakers considering energy policy adoptions for environmental and economic purposes.
60 CHAPTER IV. AFTE R THE ADOPTION OF RENEWABLE PORTFOLIO STANDARDS: GO GREENER OR BACK TO GREY? A bstract Policy revision after adoption is a critical part of policy process for adjusting a new policy to the internal and external conditions of a city or state. Nevertheless, p ost adoption decisions received little scholarly attention despite the frequent revisions after adoption, particularly for long term policy such as Renewable Portfolio Standards (RPS). Post adoption decisions greatly shape the effectiveness and feasibility of RPS. All states with RPS have different mandatory annual renewable energy sales targets. State legislators can revise RPS policies to better meet the policy energy production goals have become more stringent and 50% or more electricity is projected to come from renewable energy sources in these states. In contrast, Ohio, Kansas, and West Virginia retracted or even repealed their RPS policies. Less is known, h owever, about the determinants of post adoption decisions relative to the amount of empirical and theoretical knowledge accumulated on the drivers of policy adoption. Accordingly, this chapter explored the drivers of post adoption decisions that were not i dentified through quantitative data analysis in the existing studies. I examined and compared the cases of RPS repeal in West Virginia and RPS expansion in Oregon through semi structured interviews of state legislators, secondary data, and literature revie w . Findings indicate that state government ideology frames views on environmental and economic issues and the effectiveness of RPS, and ultimately guide the directions of policy revisions. Further, W e st Virginia s RPS repeal effort was initiated by Republ ican state legislators, because they were determined to terminate the RPS since the beginning of the new legislative session. On the other hand, RPS expansion in Oregon was the
61 output of strategic negotiation between two parties and largely driven by botto m up efforts from cities and counties, citizens, and environmental organizations. Introduction Numerous scholars have advanced the policy adoption literature (Berry & Berry, 1990; Carley & Miller, 2012; Grossback et al., 2004; Volden, 2006) . They excavated a number of factors that affect the likelihood of policy adoption using various analytical methods. However, policy adoption is only a part or another beginning of the policy process , and legislators continue to revise adopted policies as they gain experiences with the policies . Policy revision after adoption is an integral process for adjusting a new policy to the internal and external conditions of a city or state. Yet , policy revisions after adoption have received little scholarly attent ion despite the frequent revisions after adoption , and existing policy revision studies rely on the binary measure for policy revision without taking into account the nuanced differences across policy revisions in states ( e.g., M. J. Berry, Laird, & Stefes, 2015; Carley, Nicholson Crotty, & Miller, 2016) . Particularly for long term polic ies such as RPS , post adoption decisions greatly s hape the effectiveness and feasibility of RPS. State legislators would revise the RPS to better meet the different ways and these revisions differentiated the size a nd scope of RPS. California, Hawaii, and Oregon set their renewable energy sales goals as high as 50% or more. On the other hand , Ohio and Kansas scaled back their existing RPS and West Virginia even retracted the ir RPS. Apparently, existing RPS states hav e revised their RPS in multiple directions at varying extents. Even if these states adopted their RPS in a similar timeline, divergent paths after adoption has not been extensively studied and quantitative method reveals its limitation in explaining such
62 n uanced differences in motivations and post adoption decisions. Therefore, these methodological limitations in quantitative research and the lack of rigorous studies on RPS revisions after adoption led to the exploration and comparison of the factors for RP S repeal in West Virginia and RPS expansion in Oregon through a detailed cross case analysis in an attempt to answer the following research questions. What factors shaped state RPS policies after adoption? How did these factors contribute to the formulati on of policy decisions after adoption? Policy revision after adoption Post adoption decisions may take the form of expansion, retraction, renewal, rollback, repeal , or minor modification (Karch & Cravens, 2014) . Importantly, policy e xpansion, retraction , or repeal indicates critical changes in the direction and degree of policy revisions after adoption. Particularly for RPS, states have full latitude in design and implement ation . Such flexibility ha s allowed states to constantly amend implementation schedules, goals, and rules in a lded a broad spectrum of post adoptio n decisions . The empirical studies of post adoption converge towards two literature strands: First, scholars have studied the post adoption in a binary fashion by simply looking into whether there was an amendment(s) after policy adoption without taking i nto account the forms and degrees of changes in the policy (e.g., M. J. Berry, Laird, & Stefes, 2015; Karch & Cravens, 2014) . In terms of empirical findings, we know little about the determinants of post adoption decisions that take different directions and forms. The empirical studies of post adoption rely on factors for policy adoption to explain the deter minants of post adoption decisions . Yet , all we know is that policy
63 adoption and post adoption decisions are influenced by different factors. Boehmke and Witmer (2004) study on state Indian gaming industries revealed that policy innovation and expansions are influenced by different factors. Specifically, they found that social learning, measured by policy adoption in neighboring states, 11 does not have a significant effect on policy expansion whereas it has a positive effect on policy innovation. Karch and Cravens (2014) studied the adoption and modification of Three Strike laws and found that conservative political ideology and ethnicity of residents have statistically significant relationship s with adoption of the laws whereas the modification of the law is more influenced by the financing necessity, shifting ideologically environment, and stakeholders interested in maintain ing status quo. Second, those who considered the direction of policy evolution tend to focus on policy repeal or termination (Kr ause, Yi, & Feiock, 2015; Ragusa, 2010; Ragusa & Birkhead, 2015) based on deLeon s description of factors for policy termination . deLeon (1978) illuminated the three criteria for termination. First, financial imperatives gave a rise to policy ter mination. Budget constraints influenced American legislators to reduce programs wherever the cuts are possible. Second, program inefficiency or ineffectiveness triggered policy termination. Failure to deliver a desired service in a timely manner raised a q uestion on the policy and it is likely to bring about policy termination. Third, political ideology influences policy termination. Historically, conservatism is in favor of less government spending and it is likely to encourage policy termination. Ragu sa (2010) studied the repeals of landmark laws from 1951 to 2006. He found that divided government influences the policy processes in complementary ways: it hinders the reverse of enacted policies at the same time promoting the durability of legislation over the long term. C. R. Berry, Burden, and Howell (2010) investigated 11 Policy expansion in this study is measured by the number of new and alternative compacts agreed in a year by each state.
64 the post enactment histories of all federal domestic programs established between 1971 and 2013. Their findings suggest that changes in the partisan composition of Con gresses have a strong influence on program durability and sizes. Recently, Krause, Yi , and Feiock (2015) examined the policy termination offered by deLeon (1978). They confirmed the hypothesized effect of political ideology and interest grou p pressure as well as perceived program ineffectiveness on the termination of loc al greenhouse gas initiatives. Post RPS adoption State level RPS polic y is one of the most widely adopted state level renewable energy policies. The proliferation of RPS has spawned copious literature on the adoption (Carley & Miller, 2012; Matisoff, 2008; Nicholson Crotty & Carley, 2015; Yi & Feiock, 2012) and evaluation of RPS (Carley, 2009; Prasad & Munch, 2012; H. Yin & Powers, 2010) . Yet , mere adoption and evaluation do not represent the complete progress of RPS policies. Particularly, the long term nature of RPS policy and its effect ha ve facilitated changes in the policy details. Nevertheless, the drivers behind major changes whether it is to expand or scale back RPS have been understudied despite its implications for the legislators in the pot ential and previous RPS adopters. In this vein, I will identify the factors that bring to bear on the post RPS adoption decisions and examine how these factors contributed to the formulation of such decisions through the case studies of RPS repeal in West Virg inia and expansion in Oregon. After the initial adoption, numerous proposals to increase, rollback, or repeal the RPS have been introduced, but they rarely survive in the legislature and very few are enacted each year . 12 Since 12 26 bills for the RPS rollback and 29 bills for the RPS increase were introduced in 2013; 14 RPS rollback bills and 13 RPS increase bills were proposed in 2014; and 26 RPS rollback bills and 29 RPS increase bills were suggested in 2015. Among a number of b ills for RPS scale back and expansion, RPS increase bills were only enacted in four
65 its inception in early 19 90, the passage of RPS rollback in Ohio to freeze its goal for two years is the first RPS rollback in the two decades of RPS policy history. Subsequently, Wisconsin also participated in this RPS reduction waves by allowing some utility providers to be exem pted from RPS requirements beginning in 2015. Further, Kansas withdrew its RPS mandate and displaced it with nonbinding goals in 2015. In the same year, West Virginia became the first state to repeal the RPS policy. A growing number of states passed the RP S rollback bills since 2014. However, a number of RPS states continued to make their RPS policies more stringent by expanding the scope of RPS application and/or setting higher goals. California, Hawaii, Maine, Oregon, and Vermont strive to produce more el ectricity from renewable energy sources than other states in the nation. California is one of the leading states for renewable energy production . California s RPS goal is to produce 50% of electricity from renewable energy by 2030. Vermont enacted the RPS in 2015 aiming to produce 55% of electricity from renewable energy sources by 2017, increasing by an additional 4% every 3 years until reaching 75% by 2032. Hawaii is also one of the leading supporters of renewable energy. It mandated that 100% of net elec tricity sales should come from renewable energy by 2045. Recently, the governor of Oregon signed the bill that proposed 50% of utilities from renewable energy by 2030. State RPS policies continue to evolve and there is no consistent direction in this evolu tion. Berry, Laird, and Stefes (2014) first studied the RPS policy amendments after adoption. However, this aggregative measure can be problematic because it equates the adoption and amendment. Further, the focus of this model is exclusively on the frequen cy of amendments without consideration of detailed account of the amendments. states (Colorado, Maryland, Minnesota, and Nevada) in 2013. None of the bills subject to the reduction of RPS has been enacted until 2013. 2014 and 2015 are the watershed y ears in the RPS history.
66 Case selection T h is chapter employed a case study approach to RPS policy revision as it allows for the examination of contextual factors (R. K. Yin, 1994) . Among various RPS cases across states, this chapter focus es on deviant cases in opposite directions . Gerring (2007) described a deviant case as an outlier case , and studying deviant cases reveal new information on the topic of interest. In this chapter, I seek to identify the context within w h ich post adoption decisions were made in the cases of West Virginia and Oregon, two cases that are placed at the two ends of a continuum as shown in Figure 4 . 1. Figure 4.1 . Post adoption decisions West Virginia wa s the first state to repeal its RPS in more than two decades of RPS history , and selected as the case for this chapter. A lthough a number of states including Illinois, Michigan, New Hampshire, and Texas have propose d bills for RPS repeal in 2015, none of them ha ve Ohio and Kansas first scaled back their RPS policies in 2014, but RPS repeal in West Virginia better meets the criteria of a deviant case because it is an out lier case due to the fact that policy repeal is very rare . Oregon was selected as a corollary case to West Virginia due to its deviant characteristics in the opposite direction becoming more stringent after initial adoption . To select Oregon, I first identified four deviant cases of RPS expansion that evolved in a positive direction in terms of policy stringency. Cali fornia, Hawaii, Oregon, and Vermont are marked by its RPS production goals that are set as
67 50% or higher after adoption. 13 Cases were furt her narrowed down based on the criteria of RPS adoption year. I selected the cases of RPS policies that were adopted up to two years before or after 2009, when West Virginia adopted RPS. California adopted the RPS in 2002 and the duration of RPS in Califor nia is twice as long as the one in West Virginia. Further, California is the most populated state in the nation whereas West Virginia ranked the 38 th in state population t Virginia in terms of GSP. For this reason, California makes a less adequate case for the comparative study with RPS in West Virginia. Hawaii is also excluded for similar reasons. It adopted the RPS in 2004, five years before the RPS adoption in West Virg inia. Although the population and GSP of Hawaii appear to be akin to West Virginia, Hawaii is inherently not like any of continental states for its island settings that largely shaped its unique, social, political, and economic conditions. Vermont passed i ts RPS in 2015, when West Virginia repealed the RPS. Thus, the RPS in Vermont is too young for the comparison and its stringent RPS goal is not the results of the policy expansion after adoption. Rather, they began with such an ambitious goal from the outs et. Finally, the RPS in Oregon is selected for case study due to the relative proximity to West Virginia in terms of enactment years but the evolution in opposite direction. Oregon first employed its RPS in 2007 and the 27 th populous states in the U.S. and 25 th in GSP as of 2015. Figure 2 illustrates the year of RPS adoption, population, and GSP of the candidate states for this study relative to those of West Virginia. 13 California: 50% of retail sales by 2030 Hawaii: 100% of its net electricity sales by 2045 Oregon: 50% of retail sales by 2040 Vermont: 75% of retail electricity suppliers by 2032
68 Figure 4.2 . RPS adoption year, Population rank, and GSP rank of candidate states for ca se stu dy The examination of two opposite deviant cases will shed light on the factors that led to the bifurcated paths of RPS policies in two states RPS has been repealed and expanded in certain directions . Data collection First, I conducted literature review for the case overviews and histories of RPS policies in West Virginia and Oregon. Due to the paucity of empirical studies on RPS policies in West Virginia and Oregon, the case study analyses draw on n ews media, state government websites, and energy reports for background information on RPS in West Virginia and Oregon. Second, I identified a list of state legislators who directly sponsored the recent RPS decisions in West Virginia and Oregon, and conduc t ed interviews with these sponsors of the recent RPS revisions . In West Virginia, 11 state legislators sponsored the RPS repeal in 2015, while 13 state legislators and 2 representatives in Oregon sponsored the expansion of RPS in 2016. However , one of the state legislators that sponsored RPS expansion in Oregon passed away in 2016. Also, another senator member and RPS expansion sponsor retired in 2016. In West Virginia, one of the sponsors of RPS repeal did not re run for the election in 2016. Among
69 23 pote ntial interviewees, I completed t otal 8 interviews, 4 with RPS repeal sponsors in West Virginia and the other 4 with RPS expansion sponsors in Oregon . All interviews were conducted over the phone. While the total number of interviews is limited , these RPS revision sponsors in West Virginia and Oregon are legislators who were directly engaged in the policy processes , drafted bill s , and voted on them . T h us, they have the best understanding of RPS revision bills and most information about their passages, which allows for researchers to gain critical knowledge on RPS revisions. Interview Questions Interview questions are motivated by existing theories of policy innovation , diffusion, and repeal. I nterviewees were asked about their views on RPS and recent RPS revisions as well as the perceived effectiveness of these policies . Table 4.1 presents the list of interview questions. The f irst question asks the motivation or reason to support recent RPS revision decision. In West Virginia, the recent RPS decision would be policy repeal, whereas it would indicate policy expansion in Oregon. Table 4.1 . Interview questions Explanatory variables Interview Question Relevant policy theories/literature Internal determinants What was your motivation or reason to support recent RPS (repeal or expansion) decision? Were there any pressures from stakeholders for recent RPS decision? Policy innovation, Policy termination/re peal What helped or facilitated the implementation of RPS until before the recent RPS decision? Policy termination/repeal As a state implements RPS, what kind of difficulties did a state and/or state government experience? Policy termination/repeal
70 Table 4.1. Interview questions con d Explanatory variables Interview Question Relevant policy theories/literature To what extent has RPS been effective and how has it affected the recent (repeal or expansion) decision? Policy innovation, Policy termination/repeal One of the long term effects of RPS is an increase in employment and its contribution to state economy. However, the linkage between economy and renewable energy appears to be more complicated than we expected. What do you think of the relationship between state economy and RPS, and how has it affected the recent RPS revision? Policy innovation, Policy termination/repeal How has overall state legislature reacted to RPS until the enactment of recent RPS revision? Policy innovation, Policy termination/repeal Horizontal diffusion Was there any city, county, or state that you benchmarked for the recent decision? If so, why and how did you benchmark their policy decision? Policy diffusion Analysis This chapter conducted a cross case analysis of RPS repeal in West Virginia and expansion in Oregon using semi structured interviews and document analysis . A c ross case analysis increases generalizability or transferability of research , and enhance understanding and ex planation of cases (Miles, Huberman, & Saldana, 2014) . I employed protocol coding for the analysis of interview transcripts . Protocol coding uses a predetermined set of coding and this type of coding is appropriate for the study of RPS amendments because interview questions are grounded on deLeon (1978) s study of policy termination factors. Further, subcoding is introduced to capture the details of primary protocol coding (Miles et al., 2014) . The codebook for the analysis of interview tran scripts is shown in Table 4.2. Since there is not much known about post adoption decisions relative to policy adoption, findings from this
71 qualitative study may reveal the unknown or less studied facto rs that influence post adoption decisions. In terms of the consistency of information collected from int erviewees , the point of saturation was reached where interviewees provide similar or same answers to the interview question and their responses are consistent with information portrayed in the state government websites and other media sources . Table 4.2 . Interview codebook Super code Motivation : mentions of motivation to support recent RPS revision (repeal or expansion) decision Sub code Economic Environmental Political Super code Stakeholders: mentions of any pressures from stakeholders for recent RPS revision decision Sub code Yes No Super code Positive implementation experience : mentions of positive experiences with RPS implementation in a state Sub code Economic Environmental Political Super code Negative implementation experience : mentions of negative experiences with RPS implementation in a state Sub code Economic Environmental Political Super code RPS effectiveness : mentions of RPS effectiveness in relation to recent RPS revision decision Sub code Effective Ineffective Super code : mentions of the impact of RPS on state economy Sub code Positive Negative Super code Overall legislature reaction to RPS revision : mentions of the overall Sub code Overall agreement Conflict/Disagreement Super code Benchmarking of others for RPS revision decision : mentions of cities, counties, or states for the reference of RPS revision
72 Overview of RPS in West Virginia and Oregon West Virginia In 2006, West Virginia adopted RPS called Alternative and Renewable Energy Portfolio . Before the repeal in 2015, large investor owned utilities in West Virginia were mandated to produce 20% of electricity from alternative or renewable energy sources by 2025. West Virginia had been the only state with RPS among the nation s top five coal producers even though stringency and eligible technologie s. West Virginia permits both renewable and alternative energy sources for achieving its RPS goals. In addition to renewable energy sources, alternative energy sources eligible for West Virginia RPS included advanced coal technology, coal bed methane, natu ral gas, fuel from gasification or liquefaction, synthetic gas, integrated gasification combined cycle technologies, waste coal, and tire derived fuel (Small, 2015) . It did not set the requirements on the amount or portion of electricity generation that must come from renewable energy sources. The only requirement is that electricity generation from natural gas shoul d be no more than 10% of the standard (Small, 2015) . The classification of RPS in West Virginia varies by resear chers and institutes. The Center for the New Energy Economy and EIA view West RPS as equivalent to mandatory RPS whereas Lawrence Berkeley National Laboratory, Institute for Energy Research, and Interstate Renewable Energy Council do not conside r the West Virginia RPS as mandatory Mills, Rabe, and Borick (2015) acknowledge it as a voluntary RPS. For West Virginia, the inclusion of alternative energy technologies and no minimum requirement of renewable energy production for the compliance with RPS goals may have facilitated the implementation of RPS because utility providers could take advantage of existing
73 economic and industrial infrastructure, largely built upon coal mining and production businesses. By increasing the use of alternative energy through the RPS implementation, West Virginia aimed at not only the reduction in CO 2 emissions but also better utilization of byproduct of coal mining and other industrial activities that underlie Wes leading industries in West Virginia have opposed the unusually coal friendly RPS relative to the RPS in other top coal producing states such as Illinois, Montana, Pennsylvania, and Texas. Oregon Oregon produces the most of electricity from conventional hydroelectric power and natural gas. Conventional hydroelectric power is not eligible to meet RPS requirement in Oregon and it is responsible for 42% of electricity generation in Oregon in 2017. Coal and natural gas ac count for 45% of total electricity generation while 13.4% of electricity comes from nonhydroelectric renewable resources (EIA, 2017a) . In 2015, West Virginia ranked the 22 nd in the total net electricity generation, producing 5,730,000 MWh and Oregon generated 5,632,000MWh, recorded as the 23 rd in the nation (EIA, 2015a) . Oregon is known as one of eco friendly states not only for government environment as well as Kiernan (2015) compared the environmental quality and the eco friendliness of policies in 50 states and Oregon ranked the first in eco friendly behaviors of governments and citizens. 14 Oregon has 404 electric vehicle charging stations and total 1,000 charging outlets (EIA, 2015a) . Oregon enacted the RPS in 2007 as a part of the Oregon Renewabl e Energy Act of 2007. The bill mandated that 25% of utilities should come from renewable energy. Specifically, they laid out different goals for different sizes 14 Measured using the metrics of the number of green (LEED) buildings per capita, % of energy consumption from renewable sources, energy consumption per capita, energy efficiency scorecard, gasoline consumption per capita, water consumption per capita per da y, number of alternative fueled vehicles per capita, green transportation (percentage of the population that walks, bikes, carpools, takes public transportation or works from home), and % of municipal solid waste recycled.
74 of utility providers. 15 Oregon does not produce crude oil or coal but natural gas is produced in the Mist F ield in northwestern Oregon (EIA, 2015a) . In terms of industry structure, the Oregon Office of Economic Analysis found that the state is marked by the highest concentrati ons in the forest sector and high technology manufacturing from 2012 to 2022 (Lehner, 2014) . In 2010, former governor of Oregon, Te d Kulongoski (D) took the initiative in creating a renewable friendly state for business and industry by introducing the tax incentive and assistance programs, innovation investment, and loan programs in addition to the RPS adoption (Alpern, 2010) . Oregon now houses a number of world s leading data centers including those for Amazon, Apple, Face book, Google , and other major companies that require a vast amount of electricity supply. Greenpeace has once castigated these large data centers in Oregon for their large electricity consumption. However, it recently praised the data centers in Oregon for the increase in the use of renewable energy or relocation of facilities to the renewable energy generating areas (Rogow ay, 2014) . In contrast to West Virginia that encountered the large objection to RPS, industries in Oregon have made effort to increase the use of electricity generated from renewable energy sources. Recently on March 11, 2016, Governor Kate Brown (D) s igned the Senate Bill 1547, which was proposed to increase the renewable energy production goal and eliminate coal from the energy sources in Oregon. The bill mandated that 50% of retail electricity must be generated from renewable energy sources by 2040. The same bill also included the plan to ban coal fired electricity by 203 5 . This is the first ban on coal fired electricity in the nation and this decision will further accelerate the electricity production from renewable energy. 15 Large utilities: 25% by 2025 Small utilities: 10% by 2025 Smallest utilities: 5% by 2025
75 C ross case analysis of RP S revisions in West Virginia and Oregon Motivation All four West Virginia legislators discussed the economic reasons as motivations for RPS repeal. More specifically, all respondents raised concerns on increases in energy prices as a result of expansion in alternative and renewable energy in West Virginia. According to EIA (2015b) , West Virginia was ranked as the fifth in the nation for total energy production in 2012, producing 4.7% of the nation's total electricity . It was the largest coal producer east of the Mississippi River and the second largest in the nation after Wyoming; the state accounted for 12% of the U.S. total coal production in 2012. In 2013, 4 4% (51 million short tons) of the coal that was mined in West Virginia was shipped to other states, and 33% (38 million short tons) was exported to foreign countries. Coal fired electric power plants accounted for 95.5% of West Virginia's net electricity g eneration in 2014, and renewable energy resources (primarily hydroelectric power and wind energy) contributed 3.5%. West Virginia typically generates more electricity than it consumes. T he c oal industry is one of the main contributors to the economy of Wes t Virginia. Not just the state consumes coal powered electricity, but also it makes a profit from the sales of coals to other states as well as foreign countries. One interview respondent (WV1) 16 specifically pointed out the need for the protection of coal economy as a major reason to support RPS repeal: We are a coal producing state, we produce coal efficiently, and coal is burned efficiently. In fact, it is the most competitive product. I had a hard time understanding limiting its capability being purchas ed. I am glad that the majority of state legislature agrees with the termination. It benefit West Virginia citizens, and it doesn t benefit West Virginia economy. (WV1) In fact, West Virginia has spearheaded the protection of coal based industries and 16 (State initials and numeric code) indicates a randomly assigned ID for interview respondents in West Virginia or Oregon.
76 economy. As the governor of West Virginia, Earl Ray Tomblin (R) signed the House Bill 2015, which suggested the repeal of RPS , he stated as follows: In this statement, Governor Tomblin implied that the support for RPS from business and industry in 2009 was no longer available and the RPS was economy. Interview respondents in West Virginia viewed that the switch from coal to alternative or renewable energy led to increases in energy prices because they believed tha t coal is the cheapest energy source in West Virginia. In turn, increases in energy prices and the availability of cheap coal are considered the major reasons that state legislators supported the idea of RPS repeal. On the other hand, a major reason for the support for RPS expansion in Oregon is environmental protection through the use of clean energy sources. As Governor Brown signed the bill 1547 for RPS expansion , she said as follows: The prohibition of coal fired electricity in recent RPS bill is contrary to West Portfolio, the understand economic drivers and factors change over time, and the Act as it was passed in 2009 is no longer beneficial for our state. After it passed both houses of the Legislature with overwhe lming bipartisan support, I have signed House Bill 2001, repealing the West Virginia "Knowing how important it is to Oregonians to ac t on climate change, a wide range of stakeholders came to the table around Oregonians' investments in coal and renewable energy resource mix of the future. Now, Oregon will be less reliant on fossil fuels and shift our focus to clean energy. I'm proud to sign a bill that moves Oregon forward, together with the shared values of current and future generations."
77 challenge the Clean Power Plan. Another differen ce recent RPS policy decisions is that they placed a n emphasis on different effects of RPS. Governor Tomblin of West Virginia approved the RPS repeal as he expressed concerns on state economy. On the other hand, Governor Brown of Oregon explicitly discussed the importance of taking actions against climate c hange and the expansion of renewable energy production . All interview respondents in Oregon articulated the need for taking actions for climate change mitigation. In 2004, Oregon set short and long term goals for greenhouse gas emission reductions, ultimat ely seeking to reduce emissions 75% below emission levels in 1990 by 2050. O n e respondent recalled (OR 3) , All of our work on climate action in these days is motivated by the fact that we are not meeting our goals in a state Climate science is suggesting we need to be more like 80% (greenhouse gas emission reductions) by 2050. So , we have known this for a while. But we just confirmed last year got a report from global warming commission and from our Oregon climate research in stitute. You know that our previous goals are not high enough and we are not even on track to meet those goals ( greenhouse gas) that was a lot of what motivated RPS. In a similar vein, t he website of Oregon state government describes RPS policies as a pa rt of solutions for addressing climate change (Oregon Department of Energy, n.d.) , and the recently released report from Oregon Global Warming Ass ociation discussed that RPS has contributed to 30% decline in greenhouse gas emissions in Oregon since the adoption in 2007 (Duncan, 2016) . Stakeholders While all interview respondents in West Virginia said that there w as no pressure from stakeholders on recent RPS repeal decision, those in Oregon shared mixed experiences with stakeholders. They mentioned the support from environmental groups for RPS expansion and opposition from businesses. In general, renewable energy advocates, people from existing
78 industries relevant to fossils, and renewable energy lobbyists participated in political processes to speak for or against the expansion of RPS. However, respondents did not specify who they were and how they engage in polic ymaking processes. While environmentalists or renewable energy advocates create pressure for more stringent , clean energy policy and often say Oregon is not doing enough, business groups think that the state is trying to move too fast. Another respondent (CUB) and Sierra Club. The respondent influential in getting the investors owned utilities to the table to talk about this. Initial effort to (OR3) Andy Magg i, Oregon Sierra Club Chapter D i rector added, come together to advance real climate solutions as we move away from coal and toward more clean energy . (Hales, 2016) Implementation facilitators As for West Virginia s e xperiences with RPS, respondents pointed out the liberalism in state government as a facilitating factor for RPS implementation . All the state legislators that sponsored the RPS termination were Republicans. Until before 2016 when RPS was terminated, West Virginia state legislature has long been dominated by Democrats and three of the governors were also Democrats since 2001. The overall political landscape has suddenly changed in 2016 as the majority of state legislature has been occupied by Republicans wh o advocated the protection of coal industries. In other words, shift in the majority of state legislature in 2016 means that RPS, once largely supported by Democratic legislators, is no longer viewed as necessary in West Virginia. As for the factors that f acilitated RPS implementation in West Virginia, o ne respondent recalled, D emocratic party controlled the legislature in 2009. Liberal presence in the legislature ... that change d in 2014 and more
79 conservative legislat ors were elected . (WV1) Since the Republicans became the majority of both houses for the first time in last eight decades, RPS repeal was priority one for the first legislative session at the Republican led state legislatures (Detrow, 2014) . Continuing in this vein, Pantsios ( 2015) viewed that the repeal of RPS in West Virginia is associated with a bipartisan politics as the Senate unanimously passed the repeal bill and the House has 95 legislators voting for RPS re peal, whereas 4 legislators, all Democrats, disagreed with the RPS repeal bill. It is expected that conservative states would continue to resist renewable mandates for ideological reasons and the abundance of coal in their states (Brownstein, 2016) . Although the roles of political ideology was not explicitly discussed as a driver of RPS repeal, one of state legislators discussed that he decided to run for election to repeal RPS (Fried, 2015) . Similar to West Virginia, Oregon s RPS gained support from the counties with liberal i deology. At the same time, respondents mentioned that vibrant discussions about the need for renewable energy outside government, ambitious renewable energy goal in Portland , and ballot initiative for renewable energy expansion have facilitated the switch to clean energy in Oregon. Recently, the city of Portland and Multnomah County recently set 100% renewable energy goals by 2050 (Sickinger, 2017) , and this go al is far more stringent than the statewide mandatory RPS goal, which is 50% by 2040. Before the passage of RPS expansion bill, Renew Oregon, an Oregon based environmental advocacy group, filed a ballot measures to ban coal fired electricity and increase r enewable energy production in 2015 (Kullgren, 2015) . In addition to the dominance of liberal ideology, the bottom up effort at the local level has created the context in favor of further rene wable energy expansion in Oregon. Implementation barriers When interviewees were asked about challenges with RPS implementation in West Virginia , they mentioned the issues with administrative costs and
80 hurdles as well as increased cost of energy (WV1, WV2) . However, they mostly refused to provide answers to the question or details of implementation barriers. Oregon has also encountered challenges as they implemented RPS. The challenges include conservatism in a state, resistance from existing industries, a nd underestimated costs associated with renewable energy production . While liberalism is dominant in the most part of Oregon, opposition usually comes from either southern or eastern Oregon, conservative parts of the state (OR2) . They still encounter resistances from existing industries about renewable energy issues (OR1, OR2) . Similar to the challenges experienced in West Virginia, a respondent in Oregon also mentioned the implementation costs that are higher than anticipated (OR4) . Perception of RPS effectiveness All interview respondents in West Virginia viewed that RPS was ineffective in West Virginia due to higher electricity prices , and one respondent added that RPS was only effective in increasing energy prices . (WV3) For the effectiveness of RPS in Oregon, the majority of respondents agreed that RPS has been effective although o ne respondent posited that it is too early to tell the effectiveness of RPS . (OR1) Among those who said RPS was effective, they have differen t views on the extent to which and in what areas RPS has been effective. Although RPS appears to be effective in Oregon, respondents claimed that the progress is slow because of the conflict surrounding renewable energy in terms of the tradeoff between eco nomic development and environment (OR2) . Further, despite the positive views on RPS effectiveness, RPS is perceived as merely a part of climate change mitigation measures and clean energy expansion effort where all energy or environmental policies in place are complementary to one another (OR3) . State economy and RPS Along with negative views on the effectiveness of RPS, a ll respondents shared their skeptical views on RPS and renewable energy as economic drivers in West Virginia.
81 They believed that the implementation of RPS caused the closing down of coal mines, unemployment, and increase in electricity prices . In discussing the negative impact of RPS on state economy, one interview respondent brought a political discussion on the energy economy during the Obama administration by arguing that there was no linkage between RPS and state economy during the eight years of Obama s presidency (WV3). Given that West Virginia is a major coal producer and a large part of state economy relies on the coal production, state industries. Contrary to the perceived ne gative effect of RPS on West Virginia s economy , all respondents in Oregon agreed that RPS is po sitively related to the growth of state economy. They purported that RPS has contributed to state economy by increasing clean energy jobs, investment in windmills and better access to solar panels and R&D in wind power. They agreed that such contribution t o state economy has affected recent RPS expansion decision in Oregon. When Oregon first adopted RPS, people concerned about the possibility of RPS negatively affecting state economy. However, Pacific Power, the major electric power company in Oregon, only expects 1% increase in the electricity price until 2028 (Bade, 2016) . W ind farms brought economic benefits to the state, and companies like Google and Facebook have built server farms powered by hydroelectri city in Oregon (WV3). By the same token, another respondent advocated the RPS s effect on the diversification of the state economy by mentioning the two of the nation s largest wind power energy institute headquarters in Oregon and how they have contribute d to the diversification of economic drivers (OR4) . By expanding clean energy industry and increasing jobs in renewable energy research and installation , interview respondents in Oregon believed that RPS would enhance the stability and growth of the state economy.
82 Legislature s reactions to recent RPS revision decision The state legislature of West Virginia experienced overall agreement on the RPS in general except for four Democratic votes that reject the RPS repeal in the House. Given the pr edominance of Republicans in both houses and their strong wills to repeal the RPS since the beginning of legislative session, state legislators did not experience challenges in pushing their agendas for the protection of coal based economy. In contrast wi th overall agreement in recent RPS repeal decisions in the West Virginia state legislature, all interview respondents in Oregon mentioned conflicts along partisan lines as Democrats seek to bring in additional environmental protection or clean energy progr ams and Republicans want to slow down these progressive measures. In fact , it appears that conflicts around recent RPS expansion decision were more than the support or opposition of clean energy. During the legislative session, RPS expansion was merely one of the bills that wait for being passed. In turn, two of the respondents recalled that the passage of RPS expansion bill has also to do with the passage of other bills such as electric transportation funding package and clean fuels program . One respondent shared other conservative state legislators saying that they would not provide support for other bills if they push renewable energy policies too hard (OR2). F u rther, Democrats dropped few other bills that they had promoted previously , and instead support ed other Republican led bills such as delisting gray wolves from the endangered species to gain support for clean energy bills from Republicans (Roberts, 2016) . N egotiations for the passag e of bills took place in the state legislature, and Democrats had to be strategic in terms of the type of the bills that they continue or cease to pursue in order to pass the RPS expansion bill. Thus, the passage of RPS expansion was not just about Democrats supporting clean energy, but it should be also understood as a part of political processes within the context of entire legislative session. Benchmarking For the question on the benchmarking o r learning from other states or localities
83 regarding RPS repeal decision, all respondent discussed that they have not referred to other cities Similarly, most of re spondents in Oregon said either no or they do not remember or they do not know the answer to the question. However, one respondent assumed that Oregon legislature may refer to policies in California and Oregon, but these states are the ones that they look at for other matters in general, and the respondent is not certain about what aspects of their RPS Oregon took cues from (OR3) . Table 4.3 provides the summary of cross case analysis. Table 4. 3 . Analysis of RPS revision decisions in West Virginia and Oregon WV: RPS repeal Oregon: RPS goal 50% Motivation Economic difficulties It[coal] is most competitive product, I have a hard time understanding limiting its capability being Glad that the majority of state legislature agree with the termination. [RPS] d oesn't benefit W est V irginia citizens, d oesn't benefit W est Virginia economy . (WV1) RPS is not necessary. It e ventually increase s energy price s. (WV3) Environmental protection Warming climate . .. We do need to protect our env ironment . (OR 1 ) Stakeholders No pressure from stakeholders Environmental groups for RPS expansion, opposition from businesses Implementation facilitators Liberalism in the state legislature Part of it was democratic party control of the legislature in 2009 . (WV1) Governor Manchin (D) opposed carbon produced energy . (WV2) Liberal counties, ambitious renewable energy goal in Portland, environmental groups Key player was our CUB (cit i zen utilities board/ consumer utilities board). They were very influential in getting the investors owned utilities to the table to talk about this. Initial effort to get coal out of the mix was really coming from Sierra C lub . (OR3)
84 Table 4.3. Analysis of RPS revision decisions in West Virginia and Oregon con d WV: RPS repeal Oregon: RPS goal 50% Implementation barriers Economic difficulties Actual costs of energy for the rate payers have gone up . (WV1) Energy prices went up . (WV4) Conservative counties Economic difficulties Resistance from existing industries Obviously, the opposition usually comes from either southern or eastern O regon. (OR2) A fair amount of incentive was much higher than anticipated public benefit dedicate d to and there were some significant problems and grew exponentially in terms of what the credit was one time and it was relatively underutilized and overall went through the ceiling (OR4) RPS effectiveness Ineffective It was e ffective in raising people's electricity price s . (WV3) Effective Yes. [But] t he battle always is those people that would say this is a tradeoff between economic development and environment , s o we need to go slow and moderate . I don t buy that. (OR2) I think everything goes together. I think there is not one thing alone that makes difference. But RPS clearly makes difference so opposed to the market . Market has just made it easier to move to compliance. (OR3) RPS and state economy Negative impact of RPS on state econo my [ between RPS and state economy ] for the whole entire eight years when Obama was the president. how to develop energy polic ies . (WV3) Positive impact of RPS on st ate economy At the time when the RPS was first adopted 10 years ago, we were hearing that this will hinder Oregon economy. We have not seen it at all. You know we have parts of the state, thanks to wind farms, they were able to weather the recession with out any T hose initial fears about the negative effects of RPS have proven to be unfounded . (OR3)
85 Table 4.3. Analysis of RPS revision decisions in West Virginia and Oregon con d WV: RPS repeal Oregon: RPS goal 50% RPS and state economy We have two of the largest energy institute headquarters here in terms of wind power and I do think that it has led to the improvement of economy. Agriculture was important for economy in Oregon in 1970, and it is still important. But, the more the diversifie d the economy, the state is better in the long term . (OR4) Legislature reaction Overall agreement on the repeal decision Legislature sent clear message that this bill has to be repealed. There was pretty much agreement in the legislature . WV(4) No o ne had issues with terminating RPS . WV(1) Conflict along partisanship Some of the [conservative] folks said, if you push this environmental stuff too far, then we are gonna back away from support for bipartisan huge [electric] transportation funding package . OR(2) Benchmarking No benchmarking No benchmarking Actually I don't remember. I can't really say. I assumed we looked at regionally C alifornia and Washington. But that's generally what we look at as we work on. But I don't really remember what aspects of RPS we have looked at . (OR3) Conclusion and Policy Implications When West Virginia repealed its RPS in 2015 and Oregon set 50% RPS goals in 2016, both governors released public statements about their decisions. As for West Virginia, Governor Tomblin discussed changes in economic conditions and RPS is no longer beneficial for the state economy. On the other hand, Governor Brown in Oregon advoc ated the need for clean energy for current and future generations. In their statements, it seemed that West Virginia decided to repeal RPS for economic reason, and Oregon set ambitious RPS goals for environmental
86 protection. However, what led state legisla tors or governors to discuss economic or environmental issues for their recent RPS revision decisions is missing in these public statements and other large N RPS revision studies ( e.g., M. J. Berry et al., 2015; Carley, Nicholson Crotty, & Miller, 2016) . T h rough the literature review and interviews with state legislators in West Virginia and Oregon, this chapter identified the factors for initial m otivation for recent RPS revision decisions , supports and challenges throughout the implementation of RPS , and perceived effectiveness of RPS in these two states . Although West Virginia s RPS repeal seems largely based on the state legislators strong will to protect the coal based state economy at the surface, what triggered or set the stage for such discussion and agenda was the shift in dominant political ideology in West Virginia. Once conservative legislators occupied the majority of state legislature seats, they reportedly sought to repeal RPS, claiming that electricity price has soared during the years with RPS and West V i rginia suffered unemployment problems in coal mines. That is, the shift in dominant political ideology from liberalism to conservatism gave an impetus for RPS repeal with the rationale of reducing electricity prices and creating jobs by revitalizing coal industries. Further, the proposal and passage of RPS repeal was largely top down as state legislators took the lead in the repeal of RPS, and the RPS repeal bill was the first bill they passed with ambitions as they enter the first legislative session with Republican majorities in both houses in last eight decades. T h is ideological shift in the state legislature may be com parable to the Republican Revolution where the prevalence of conservative ideology in the political culture opens up the opportunities to oppose climate policies and science (McCr ight & Dunlap, 2011) . Oregon has been known to set 50% RPS goal for environmental purposes although most of Oregon s electricity still comes from non RPS eligible resources such as fossil fuels and
87 hydroelectric power. State legislators in Oregon viewe d that warming climate change is real and they are not doing enough to reduce greenhouse gases. Historically, Oregon has been one of the most liberal states and has not experienced a dramatic shift in the dominant political ideology in the state legislatur e . Most liberal state legislators firmly believed that global warming is real, and Oregon is not doing enough to slow down global warming. RPS gained support from the liberal parts of the state, while it has been challenged by conservative politicians and counties. In an effort to cope with the conflicts around RPS expansion, Democrats engaged in strategic negotiations with Republican legislators. The RPS expansion bill was merely a part of multiple bills waiting to be passed in the legislative session. Thu s, RPS supporting legislators had to be strategic in promoting RPS expansion since they have to consider the passage of other important bills that need support from Republicans. A l though liberal legislators played critical roles in passing the RPS expansio n bill, it should be also noted that the passage was the output of political negotiations where Democrats voted for other bills led by Republicans to secure support for RPS expansion. Another critical finding of this chapter is the roles of bottom up influence in the state level policy revision. Unlike RPS repeal in West Virginia which was initially led by state legislators, local governments and civic groups effort for clean energy catalyzed Oregon s ambitious renewable energy goals and withdrawal from coal fired electricity. For the effectiveness of RPS and its relationship with state economy, interview respondents in Oregon discussed RPS s positive contribution to state economy such as job creatio n, investment in wind power , and diversification of economic drivers. Oregon s RPS initially seeks to reduce greenhouse gases and promote the use of clean energy . Oregon has also enjoyed economic benefits of RPS as they pursue these environmental goals. S t ate legislators
88 increasingly began to perceive economic benefits of RPS as a critical factor for the stabilization or growth of state economy. deLeon (1978) discussed the three determinants of policy termination, which include cost reduction, policy ineff ectiveness, and conservative ideology. Although dominant conservative ideology was found to be linked with policy termination in Berry et al. (2010) and Krause et al. (2015), they did not examine how the ideology shapes the contexts for policy termination and how it affects the rest of termination determinants. The findings in this chapter demonstrated that a shift to conservative state legislature itself is a critical fac tor for policy termination, and it also shapes how state governments interpret policy effectiveness and economic impact of the policy in a state. In a state where the existing policy became more stringent after adoption, the prevalence of liberal ideology serves as a critical factor for policy expansion as it permeates the views on policy e ffectiveness and standpoint in climate change issues. Political ideology frames the way policy is perceived and interpreted, determining the fate of existing policies. This chapter employed interview method for data collection with eight state legislators in West Virginia and Oregon. Although interviews with state legislators revealed how politics play a role in the post adoption decisions, the number of interviews are very limited and it may affect the reliability of this chapter . Most interview responses were quite consistent in each state, there fore the threat to reliability is lim ited. State legislators directly engage in policy revision and make final decisions on the future direction of existing policies. Moreover, the number of legislators who direct ly work for the passage of RPS revisions bills is very small . Given the limited number of RPS revision sponsors in West Virginia and Oregon as well as the consistency in information provided by respondents, this chapter draws meaningful lessons on RPS revi sion and policymaking.
89 Another limitation arises from the nature of case study that lacks external validity (Stoecker, 1991) . Multiple case studies would address this issue but this chapter exclusive focus on RPS repeal in West Virginia and expansion in Oregon would raise a question on the external validity of this chapter, although the focus on RPS repeal in West Virginia an d expansion in Oregon still draws meaningful lessons on policy revisions after adoption due to the paucity of policy revisions in opposite directions and the attainment of point of saturation . Future studies on post adoption decisions of RPS will involve m ultiple cases in number and also in terms of types of post adoption decisions such as policy retraction and stasis in addition to policy repeal and expansion.
90 CHAPTER V . CONCLUSION State governments have made substantial effort to reduce CO 2 emissions in various sectors and RPS policies has been widely adopted as a part of the climate change mitigation effort o ver last two decades . Compared to the late 1990 s where RPS policies were first introduced in Massachusetts , Maine, and Nevada, renewable energy technology has significantly improved and the cost of renewable energy generation has fallen sharply over time. At the same time, the U.S. has experienced drastic changes in economic and political environment, which largely affected the prospects of RPS an d other energy relevant policies. Given these circumstances, this dissertation shed s light on the factors that affected RPS adoption, effectiveness, and revision s after adoption by answering the following research question s : Chapter 2: W hat factors expla in patterns in the adoption of state RPS policies ? C hapter 2 focuse s on the roles of weather related crisis events and the patterns of states taking cues from other states RPS adoption decisions. Chapter 3: T o what extent have RPS policies achieved their intended energy and economic goals? Chapter 3 on RPS effectiveness evaluate s RPS and other energy policies effectiveness in achieving the goals of renewable energy generation expansion and electricity price s tability. Chapter 4: What factors shaped state RPS policies after adoption? How did these factors contribute to the formulation of policy decisions after adoption? Chapter 4 on RPS revisions explore the factors that gave rise to drastic policy revisions af ter adoption by probing into the cases of RPS repeal in West Virginia and 50% RPS goal in Oregon.
91 This current Chapter 5 summarize s overall lessons from this dissertation, the findings from the empirical chapters, present s policy implications and study li mitations, and set s future study agenda. This dissertation contributes to the existing literature of policy processes and state level renewable energy policy by offering lessons for policy adoption, effectiveness, and revisions after adoption of RPS. As for the factors for RPS adoption, state governments may adopt RPS as a response to the increasing signs of climate change perceived through weather related hazard crisis events. Further, states tend to imitate RPS adoption decisions in other states that sh are political and economic similarities. State policymakers were able to promote RPS not only for environmental benefits and but also for economic benefits. In regards to RPS effectiveness in renewable energy expansion and electricity price, this dissertat ion found that RPS contributed to renewable energy expansion and the improvement of electricity price stability . Moreover, interviews with state legislators revealed that RPS adoption and revision are likely to be driven and shaped by state government ideo logy that affect states views on environmental protection and economic development. Therefore, state government ideology may determine the prospects of RPS for adoption and post adoption decisions. RPS adoption in Chapter 2 As for the adoption of RPS p olicies, this chapter seeks to explain the internal and external factors affecting RPS adoption in a state. Specifically, I focused on the roles of weather related crisis events on RPS adoption decisions . Crisis events have been considered as one of the main drivers of policy change in the theories explicating policy change such as multiple streams framework, punctuated equilibrium theory, and advocacy coalition framework . Nevertheless, how crisis
92 events trigger the adoption of a new policy has not been extensively examined in the policy adoption and diffusion studies. In an attempt to investigate the roles of crisis events internal and external to states, I measured weather related crisis events in two ways. One is immediate cris i s represented by large scale weather related hazards inside a state, and the other indicator of crisis events is policy proximate crises measured by large scale weather related hazards outside a state. I found that frequent large scale weather related crisis events increase the like lihood of RPS adoption , whereas weather related hazards experiences in other states negatively affect the likelihood of RPS adop tion. Scientifically, the link between climate change and each weather related crisis event is still unclear. Climate change is a long term phenomenon, and has more to do with long term weather trends. However, the results from this chapter suggest that no n scientists and lay people might perceive the severity of climate change through specific weather related hazard events. That is, t hey might connect weather extremes to climate change . These perceptions might also be bolstered by the news m edia and their depict ion of climate change as one of the main driver s of extreme weather events. F u rthermore, states tend to take cues from other states that share similarities in political and economic conditions. Policy diffusion studies have mostly relied on policy adoptions in neighboring states for policy diffusion under the assumption that neighboring states share similar economic and political conditions. However, sole reliance on neighboring states as policy diffusion factors provide an incomplete explanation for diffusion of policies across states. Also, state governments can easily obtain policy information in all other states with ease thanks to the advancement of technology and better accessibility to information (Shipan & Vold en, 2012) . Recent policy diffusion studies found the positive effect of policy adoptions in ideologically similar states on the likelihood of potential adopters policy decisions (Carley et al ., 2016;
93 Chandler, 2009; Grossback et al., 2004) . In fact, renewable energy policy is not just about using clean energy sources and reducing CO 2 emissions from electricity generation. They are highly politicized and have an impact o n local and state ec onomy. Because the adoption of renewable policy incurs political and economic costs, it makes more sense for potential policy adopters to take cues from the policy decisions in other states that are politically or economically similar and reduce the uncert ainties in costs associated with the policy. In doing so, state policymakers make better informed policy decisions. This chapter contributes to the diffusion of innovation and RPS literature by examining the roles of crisis events and policy adoption decis ions in other states that share similar internal conditions besides geographic proximity. Like with other policy change literature, crisis events play significant roles in policy adoption as they signal the need for policy solution to the crisis, and peopl e turn direct crisis experiences into an opportunity to prepare for better future. Further, this chapter reveals that states imitate policy adoptions in other states that share more than geographic borders. R ecent policy diffusion scholars have been conce rned that existing policy diffusion studies have over relied on neighboring state effects for policy diffusion studies and states may not be simply dependent on policy decisions in neighboring states (Baybeck et al., 2011; Shipan & Volden, 2012; Volden et al., 2008) . Grossback et al. ( 2004) claimed that potential policy adopters may reduce the uncertainties surrounding a new policy by imitating policy decisions in other states that share similar poli tical ideology. Renewable energy policy certainly comes with both political and economic price tags as it requires upfront investment in renewable energy industries, and conservative and major fossil fuel producing states like West Virginia and T e xas have repealed RPS or introduced the bill to remove renewable energy mandates. Possibly due to political and economic implica tions of RPS , states
94 share RPS adoption decisions if they are politically or economically similar. States can make more informed predicti ons of adoption, implementation, and effectiveness of RPS by looking at RPS in other states that are similar in terms of ideology, wealth, and energy industry structure. RPS effectiveness in Chapter 3 The immediate goal of RPS is to increase renewable energy generation. However, RPS is not just about renewable energy . P olicymakers expect economic benefits as well when adopting and promoting RPS in their states. In fact, one of the well known RPS economic goals is electricity price stability by increasi ng renewable energy generation through RPS adoption. Yet, RPS evaluation literatur e has predominately focused on renewable energy generation ( e.g., Carley, 2009; Fischlein & Smith, 2013) , w hile less attention has been paid to economic goals of RPS. Accordingly, this chapter investigated and compared the effectiveness o f RPS in renewable energy g eneration and electricity price stability. Although the major focus of this chapter is to evaluate the effectiveness of RPS, all p olicy evaluation s involve two primary tasks : the assessment of achievement of intended policy goals and the extent that the ob served effects can be attributed to the evaluated policy or program (Wollmann, 2007) . Accordingly, other energy relevant policies effects were also evaluated in an attempt to accurately estimate the effective ness of RPS on its intended energy and economic goals. Among five different energy policies examined in the same model, RPS was the only policy that is positively correlated with renewable energy generation . N et metering, EERS, MGPO, and PBF are not rela ted to increases in state level renewable energy generation, although these policies may complement or facilitate the implementation of RPS. As for electricity price stability , RPS and EERS were found to contribute electricity price stability. A state with EERS is required to use less energy, increasing the predictabilities of electricity consumptions and
95 electricity prices. Although a state with high renewable energy production capacity experienced higher electricity prices , the positive contri bution of RPS duration to electricity price stability reflects the improvement of renewable energy efficiency and production costs over time. R eduction in electricity demand by EERS is a cost effective way to stabilize electricity prices especially under c urrent electricity markets where the share of renewable energy generation is still very low . EERS place s fewer burdens on customers and incur s less cost for state government and utility providers than other policies. Policies like net metering, RPS, and PB F requires heavy investment from either customers or state governments to initiate and implement the policy. Therefore, the finding s impl y that electricity price stability is not simply achieved by diversifying energy sources for electricity generation. Ra ther, it should be also cost effective for both suppliers and customers. RPS revision after adoption in Chapter 4 Policy revision after the adoption is a critical part of policy process for adjusting a new policy to the internal and external conditions o f a city or state. Nevertheless, post adoption decisions have received little scholarly attention despite the frequent revisions after adoption. Par ticularly for a long term policy like RPS , post adoption decisions have shaped and changed the prospects of RPS to a great extent. In some states like California, Hawaii, and Oregon, renewable energy production goal has become more stringent , and 50% or more electricity is required t o come from renewable energy sources in these states. In contrast, Ohio, Kansas, and West Virginia retracted or even repealed their RPS policies. However, less is known about the determinants of post adoption decisions relative to the amount of empirical and theoretical knowledge accumulated on the drivers of policy adoption. This cha pter explored the drivers of RPS revisions through the cases of RPS repeal in West Virginia and 50% renewable energy goal
96 in Oregon. Due to the rarity of such cases of policy repeal and progressive renewable energy expansion at the state level , I conducted interviews with state legislators to probe into why and how RPS took divergent paths in these two states. The major motivation for state legislators to support RPS repeal in West Virginia was to avoid increases in energy prices and the prote ction of coal economy . State policymakers viewed that RPS was not effective in West Virginia, and only effective in increasing energy prices. Although the justification of RPS repeal was to protect West Virginia economy, what facilitated such justification for RPS repeal was initially the dominance of conservative ideology in West Virginia in 2015. Republicans took the initiative in improving state energy economy by protecting coal based economy and reducing renewable energy, and they believe such decisions would lead to more jobs and less energy costs in West Virginia. In fact, when RPS was adopted in West Virginia, Democrats took control of state legislature and state government was able to pass RPS adoption bill with support from Democrats. In the case of RPS repeal in West Virginia, I found that not only a shift to conservative state legislature itself is a critical indicator for policy termination, but also political ideology shapes how state governments interpret policy effectiveness and economic impact of the policy in a state. energy, and Oregon has enjoyed economic benefits of RPS as they pursue these environmental goals. From the interviews, it was evident that the m otivation for RPS expansion in Oregon was not economic, but environmental. However, state policymakers began to perceive economic benefits of RPS as a critical factor for stabilization or growth of state economy. Further, local government and citizen led e ffort for renewable energy expansion spurred the proposal and passage of RPS expansion bill in Oregon. In Oregon where state government liberalism is
97 prevalent, environmental protection is one of the most consistently pursued values in the state level poli cymaking . Also, pro environment policies are believed to improve state economy by increasing employment and investment in the green industry. In sum, t his chapter adds the knowledge to the existing RPS lite rature by exploring how state government ideology shapes views on overall state level e nergy policies , economy, and perceived effectiveness of RPS. Overall c hapter s ummary Table 5. 1 provides a summary of research questions, hypotheses, methods, and findings from each chapter. Overall, Table 1 shows that this dissertation confirmed hypotheses on the roles of weather related crisis events inside a state on RPS adoption. However, results indicate that weather related crisis outside a state decrease the likelihood of RPS adoption in a state. Similar states h ypothesis was partially confirmed. States with similar political and economic attributes are likely to share RPS adoption decisions . In regards to RPS effectiveness, RPS increases renewable energy generation and contributes to electricity price stability . Post adoption decisions in West Virginia and Oregon were examined by using semi structure interviews and literature review, and it was found that state government ideology is a key to RPS revision decisions. Table 5.1 . Chapter summaries Chapter 2 on RPS ad option RQ W hat factors explain patterns in the adoption of state RPS policies? Hypotheses Weather related crisis events inside a state will increase the likelihood of RPS policy adoption. [Confirmed] Weather related crisis events outside a state will increase the likelihood of RPS policy adoption. [Confirmed in an opposite direction] Methods A state is more likely to adopt RPS if other states that share similar conditions have RPS. [Partially confirmed] Quantitative, dyadic analysis using Probit
98 Table 5.1. Chapter summaries cont d Chapter 2 on RPS adoption Findings Weather related crisis events inside a state increase the li kelihood of RPS policy adoption, whereas those outside a state do not. A state is more likely adopt RPS if other states that share similar political and economic conditions have RPS. Chapter 3 on RPS effectiveness RQ T o what extent have RPS policies achieved their intended energy and economic goals? Hypotheses RPS will increase renewable energy generation in a state. [Confirmed] RPS will contribute to electricity price stability in a state. [ Confirmed ] Methods Quantitative, fixed effects regression Findings RPS increases renewable energy generation in a state. RPS contributes to state level electricity price stability by reducing electricity prices. Chapter 4 on post RPS adoption decision RQ What factors shaped state RPS policies after adoption? How did these factors contribute to the formulation of policy decisions after adoption? Hypotheses Exploratory study Methods Qualitative, interview and document review Findings State government ideology shapes adoption and revision decisions for state level energy policies, energy economy, and perceived effectiveness of RPS. L i mitations No research is without limitations. Accordingly, I highlight a few of the more important ones for this dissertation. One of the major limitations of this dissertation is that it did not take into account policy design elements of RPS, such as RPS goals, penalty for noncompliance and types of renewable energy allowed. Thorough analysis of RPS adoption that considers various aspects of RPS policy design would provide more concrete knowledge of policy adoption. A second limitation relates to the measurement of crisis events. R eliance on the count of FEMA declared emergency or disasters for the indicators of crisis events may yield biased understandings of the roles of cri sis events because this dissertation did not consider the extents of economic or social damages caused by these events. However, FEMA declared emergency or
99 disaster is still one of the most publicly accessible measures of crisis events. Further, the estima tion of economic or social costs associated with hazard events vary by entities and long term recovery cost is difficult to predict. Given the challenges with the cost estimation of each hazard, the count of FEMA declarations would still be a viable option to measure the occurrence of large scale crisis events in the U.S. A third limitation of this dissertation has to do with chapter 3 as the analytical models for electricity price stability in the chapter did not control for supply and demand of electricity market in a rigorous manner. Supply and demand are critical determinants of electricity prices, and may affect the patterns of changes in electricity prices. A lthough the analytical models introduced indirect measures for supply and demand by controlling for population growth rate and electricity prices, other factors affecting supply and demand such as regulations, weather conditions, and technology are overloo ked in this chapter. A fourth limitation relates to chapter 4 and the number of interviews conducted. Although interviews with state legislators revealed how political ideology play s a role in the post adoption decisions, the number of interviews are very limited and it may affect the reliability of findings. However, even though only four interviews were conducted in each state these interviewees were the senators, that is people directly engaged in the policy revision, were the ones with actual authority to vote on the revision, and, thus, most knowledgeable about what happened. Thus, while the sample is small, there are still meaningful lessons on how politics shape policy revisions given this informed popula tion. As states accumulate additional experiences with RPS, they will continue to revise their RPS policies in multiple directions in the coming years. A dditional case studies that cover a broad range of revisions can improve the reliability as well as va lidity of the findings discovered
100 in chapter 4 . Potential candidates for additional case studies should be Kansas, Ohio, Hawaii, and California. Kansas withdrew RPS mandates and displaced it with nonbinding goals in 2015. Ohio fro ze its RPS goal for 2015 a nd 2016 is the first RPS rollback in the two decades of RPS policy history . In contrast, California set 50% renewable energy sales goal by 2030 and Hawaii targets 100% renewable energy sales by 2045. The addition of these cases would offer more concrete ex planation of policy revisions. Future studies This study was comparative but within the same country. There are great opportunities to expand the science of diffusion and insights on RPS if we take a more comparative perspective that spans countries. T his dissertation provides the foundation laying knowledge for international comparative studies of RPS and citizen's roles in climate change mitigation measures. Renewable energy policy in the U.S. is often described as bottom up policymaking , whereas the national government took the lead in the renewable energy policy in China. Design and implementation of renewable energy policy in Korea have characteristics of renewable energy po licies in both U.S. and China. In the future research, I will examine how different types of regimes and governance affect RPS policy adoption, implementation, evaluation and effe ctiveness . The second future research is on RPS termination agenda setting. Although West Virginia is the only state that terminated its RPS policy, a number of states have made attempts to terminate or retract RPS. In Kansas, Minnesota, North Carolina, Ohio, Texas, and Virginia, multiple bills are in committee for the consideration of repeal or retraction, while Maryland, Missouri, and Wisconsin are t rying to include non renewable energy source as RPS eligible sources. For this study, the integration of quantitative analysis, interviews, and media analysis
101 would provide critical insights into why and how states weaken or repeal their RPS policies in th e era where people directly suffer from climate change and global level effort for climate change mitigation is in progress. In regard to my third research agenda on the adoption of climate change mitigation measures, s tate level policy adoption studies r explain the likelihood of policy adoption. However, the bottom up movements among citizens to ask for additional environmental protection measures have grown as citizens are more educated and have better acce ss to information regarding environmental issues than before. Active citizen engagement in environmental issues has elicited a number of additional climate change mitigation measures, not only at the local level but also at the state and national levels. H owever, existing state level environmental policy studies have paid little attention to the roles of citizens despite their growing importance in the state and national level policies. Surveys and interview s with different environmental advocacy groups and policymakers would help identify the types of effective coalitions, the promotion of policy learning, how coalitions reached out to government and other citizens, and conflict management and persuasion strategies . Summary of lessons This dissertation see ks to answer the question of why and how are state level RPS policies adopted and revised? To what extent has RPS met its policy goals? State governments respond to directly experienced weather related crisis events with RPS adoption, and consider politica l and economic implications of RPS when adopting RPS. RPS is more than a renewable energy policy. It is a part of state level solutions for climate change mitigation and green economy. Acknowledging the growing importance of RPS and energy management at th e state level, this dissertation investigated adoption , effectiveness, and post adoption decisions of RPS.
102 State governments respond to directly experienced weather related crisis events with RPS adoption, and consider political and economic implications o f RPS when adopting RPS. Apparently, RPS did increase states renewable energy generation and impro ved electricity prices. At the same time, EERS also contributed to electricity price stability by reducing electricity demand. T h roughout the policy processe s surrounding RPS, state policymakers often engage in debates, largely driven by differences in their political ideol ogies. Dominant state government ideology plays a critical role in the perception of economy, costs and effectiveness of RPS, which affect policy processes of RPS. Indeed, RPS offers avenues where we can witness interactions among political ideology , economy, and environmental views.
103 REFERENCES ACEEE. (2017). State Energy Efficiency Resource Standards (EERS) . Retrieved from http://aceee.org/files/pdf/policy brief/eers 02 2014.pdf Alpern, P. (2010). Portland Cultivates Future as Hub for Manufacturing A lternative Energy. Retrieved from http://www.industryweek.com/articles/portland_cultivates_future_as_hub_for_manufactu ring_alternative_energy_22925.aspx Bade, G. (2016, August 8). Pacific Power: Oregon 50% RPS will barely raise customer rates thru 2028. U tility Dive . Retrieved from https://www.utilitydive.com/news/pacific power oregon 50 rps will barely raise customer rates thru 2028/424037/ Balch, J. K., Bradley, B. A., Abatzoglou, J. T., Nagy, R. C., Fusco, E. J., & Mahood, A. L. (2017). Human started w ildfires expand the fire niche across the United States. Proceedings of the National Academy of Sciences , 114 (11), 2946 2951. Balla, S. J. (2001). Interstate Professional Associations and the Diffusion of Policy Innovations. American Politics Research , 29 (3), 221 245. Baybeck, B., Berry, W. D., & Siegel, D. A. (2011). A Strategic Theory of Policy Diffusion via Intergovernmental Competition. The Journal of Politics , 73 (1), 232 247. Berry, C. R., Burden, B. C., & Howell, W. G. (2010). After Enactment: The Lives and Deaths of Federal Programs. American Journal of Political Science , 54 (1), 1 17. Berry, F. S., & Berry, W. D. (1990). State Lottery Adoption as Policy Innovation: An Event History Analysis. The American Political Science Review , 84 (2), 395 415. Berry, F. S., & Berry, W. D. (2007). Innovation and Diffusion Models in Policy Research. In P. A. Sabatier (Ed.), Theories of the Policy Process (pp. 223 260). Boulder, CO: Westview Press. Berry, M. J., Laird, F. N., & Stefes, C. H. (2015). Driving energy : the enactment and ambitiousness of state renewable energy policy. Journal of Public Policy , 35 (2), 297 328. Berry, T., & Jaccard, M. (2001). The renewable portfolio standard: Design considerations and an implementation survey. Energy Policy , 29 (4), 263 277. Berry, W. D., Ringquist, E. J., Fording, R. C., & Hanson, R. L. (1998). Measuring Citizen and Government Ideology in the American States, 1960 93. American Journal of Political Science , 42 (1), 327 348.
104 Berry, W. D., Ringquist, E. J., Fording, R. C., & Hanson, R. L. (2010). Measuring Citizen and Government Ideology in the American States, 1960 93. American Journal of Political Science , 42 (1), 327 348. Bird, L., Chapman, C., Logan, J., Sumner, J., & Short , W. (2010). Evaluating Renewable Portfolio Standards and Carbon Cap Scenarios in the U.S. Electric Sector . Golden, CO. Birkland, T. A. (2007). Agenda Setting in Public Policy. In F. Fischer, G. J. Miller, & M. S. Sidney (Eds.), Handbook of Public Policy Analysis (pp. 63 78). Boca Raton, FL: CRC Press. Boehmke, F. J. (2009). Policy Emulation or Policy Convergence? Potential Ambiguities in the Dyadic Event History Approach to State Policy Emulation. The Journal of Politics , 71 (3), 1125 1140. Boehmke, F. J ., & Witmer, R. (2004). Disentangling Diffusion: The Effects of Social Learning and Economic Competition on State Policy Innovation and Expansion. Political Research Quarterly , 57 (1), 39 51. Boushey, G. (2012). Punctuated Equilibrium Theory and the Diffus ion of Innovations. Policy Studies Journal , 40 (1), 127 146. Boykoff, M. T. (2007). Flogging a dead norm? Newspaper coverage of Flogging climate change in the United States anthropogenic and United Kingdom from 2003 to 2006. Area , 39 (4), 470 481. Brooks, N., & Adger, W. N. (2003). Country level risk measures of climate related natural disasters and implications for adaptation to climate change . Tyndall Centre Working Paper 26 (Vol. 26). Norwich, UK. Brooks, S. M. (2007). When Does Diffusion Matter? Explai ning the Spread of Structural Pension Reforms Across Nations. The Journal of Politics , 69 (3), 701 715. Brownstein, R. (2016). The Winds Are Changing for Renewable Energy . Retrieved from https://www.theatlantic.com/politics/archive/2016/07/are the winds changing for renewable energy/490250/ Bull, S. R. (2001). Renewable energy today and tomorrow. Proceedings of the IEEE , 89 (8), 1216 1226. Carley, S. (2009). State renewable energy electri city policies: An empirical evaluation of effectiveness. Energy Policy , 37 (8), 3071 3081. Carley, S. (2011). The Era of State Energy Policy Innovation: A Review of Policy Instruments. Review of Policy Research , 28 (3), 265 294.
105 Carley, S., & experiences. Wiley Interdisciplinary Reviews: Energy and Environment , 2 (5), 488 506. Carley, S., & Miller, C. J. (2012). Regulatory Stringency and Policy Drivers: A Reassessmen t of Renewable Portfolio Standards. Policy Studies Journal , 40 (4), 730 756. Carley, S., Nicholson Crotty, S., & Miller, C. J. (2016). Adoption, reinvention and amendment of renewable portfolio standards in the American states. Journal of Public Policy , 1 28. Cavallo, E., & Noy, I. (2010). The Economics of Natural Disasters: A Survey (IDB Working Paper Series No. IDB WP 124). Washington, D.C. Chandler, J. (2009). Trendy solutions: Why do states adopt Sustainable Energy Portfolio Standards? Energy Policy , 37 (8), 3274 3281. Chen, C., Wiser, R., Mills, A., & Bolinger, M. (2009). Weighing the costs and benefits of state renewables portfolio standards in the United States: A comparative analysis of state level policy impact projections. Renewable and Sustainab le Energy Reviews , 13 (3), 552 566. Cory, K. S., & Swezey, B. G. (2007). Renewable Portfolio Standards in the States: Balancing Goals and Rules . Technical Report (Vol. NREL/TP 67). Golden, CO. Davies, L. L. (2014). Evaluating RPS Policy Design: Metrics, G aps, Best Practices, and Paths to Innovation (No. S.J. Quinney College of Law research paper No.101). Salt Lake City, UT. Detrow, S. (2014). Some states use laws and rules to slow growth of renewable energy . Retrieved from http://governorswindenergycoalit ion.org/some states use laws and rules to slow growth of renewable energy/ Dong, C. G. (2012). Feed in tariff vs. renewable portfolio standard: An empirical test of their relative effectiveness in promoting wind capacity development. Energy Policy , 42 , 476 485. DSIRE. (2017). N.C. Clean Energy Technology Center. Retrieved January 15, 2017, from http://www.dsireusa.org/ Duane, T. P. (2010). Greening the Grid: Implementing Climate Change Policy Through Energy Efficiency, Renewable Portfolio Standards, an d Strategic Transmission System Investments. Vermont Law Review , 34 (10), 711 780. Duncan, A. (2016). Electricity and Coal Transition Plan . Dunlap, R. E., Xiao, C., & McCright, A. M. (2001). Politics and Environment in America: Partisan and Ideological Cleavages in Public Support for Environmentalism. Environmental Politics , 10 (4), 23 48.
1 06 Duscha, V., Held, A., & Rio, P. del. (2016). Analysis of interactions between ren ewable support and climate policies . Eastin, L. J. L. (2014). An Assessment of the Effectiveness of Renewable Portfolio Standards in the United States. The Electricity Journal , 27 (7), 126 137. EIA. (2015a). Oregon: State Profile and Energy Estimates. Ret rieved from http://www.eia.gov/state/?sid=OR EIA. (2015b). West Virginia: State Profile and Energy Estimates. Retrieved from http://www.eia.gov/state/?sid=WV EIA. (2016). Detailed State Data. Retrieved November 1, 2016, from https://www.eia.gov/electrici ty/data/state/ EIA. (2017a). Oregon State Profile and Energy Estimates. Retrieved May 1, 2018, from https://www.eia.gov/state/?sid=OR#tabs 4 EIA. (2017b). What is the difference between electricity generation capacity and electricity ed May 1, 2018, from https://www.eia.gov/tools/faqs/faq.php?id=101&t=3 Electric Choice. (2017). Map of deregulated energy states. Retrieved December 29, 2017, from https://www.electricchoice.com/map deregulated energy markets/ Engel, K. H. (2006). State and Local Climate Change Initiatives: What is Motivating State and Local Governments to Address a Global Problem and What Does this Say about Federalism and Environmental Law? The Urban Lawyer , 38 (4), 1015 1029. Eyestone, R. (1977). Confusion, Diffusion, and Innovation. The American Political Science Review , 71 (2), 441 447. Fellowes, M. C., & Rowe, G. (2004). Politics and the new American welfare states. American Journal of Political Science , 48 (2), 362 373. FEMA. (2016). Disaster Declarations by State/Tribal Government. Retrieved from https://www.fema.gov/disasters/state tribal government/ Fischer, C. (2009). Renewable Portfolio Standards: When Do They Lower Energy Prices? The Energy Journal , 31 (4), 81 100. Fischlein, M., & Smith, T. M. (2013). Revisiting renewable portfolio standard effectiveness: Policy design and outcome specification matter. Policy Sciences , 46 (3), 277 310. Forsyth, T. L., Pedden, M., & Gagliano, T. (2002). The Effects of Net Metering on the Use of Smal l Scale Wind Systems in the United States . Golden, CO.
107 Climate Change. The New York Times . Retrieved from https://www.nytimes.com/2017/12/13/climate/hurricane harvey climate change.html Fried, R. (2015). West Virginia: First State to Repeal Renewable Portfolio Standard! Retrieved from http://www.sustainablebusiness.com/west virginia first state to repeal renewable portfolio standard 52721/ Gilardi, F. (2016). Four Wa ys We Can Improve Policy Diffusion Research. State Politics and Policy Quarterly , 16 (1), 8 21. Gilardi, F., & FÃ¼glister, K. (2008). Empirical Modeling of Policy Diffusion in Federal States: The Dyadic Approach. Swiss Political Science Review , 14 (3), 413 4 50. from https://www.carbontracker.org/colorados renewables Gray, V. (1973). Innovation in the States: A Diffusion Study. The American P olitical Science Review , 67 (4), 1174 1185. Grossback, L. J., Nicholson Crotty, S., & Peterson, D. A. M. (2004). Ideology and Learning in Policy Diffusion. American Politics Research , 32 (5), 521 545. Hales, R. L. (2016). Oregon passed a 50% renewable port folio standard . Retrieved from https://theecoreport.com/oregon passes 50 percent renewable portfolio standard/ Hawaii, S. of. House Concurrent Resolution, Pub. L. No. H.C.R. No. 113 (2017). Retrieved from http://www.capitol.hawaii.gov/session2017/bills/HC R113_.PDF Heeter, J., Barbose, G., Bird, L., Weaver, S., Flores Espino, F., Kuskova Burns, K., & Wiser, R. (2014). A Survey of State Level Cost and Benefit Estimates of Renewable Portfolio Standards . Golden, CO. Hirschi, C., & Widmer, T. (2010). Policy Change and Policy Stasis: Comparing Swiss Foreign Policy toward South Africa (1968 94) and Iraq (1990 91). Policy Studies Journal , 38 (3), 537 563. Holt, E. A., & Wiser, R. H. (2007). The Treatment of Renewable Energy Certificates, Emissions Allowances, and Green Power Programs in State Renewables Portfolio Standards . Huang, M. Y., Alavalapati, J. R. R., Carter, D. R., & Langholtz, M. H. (2007). Is the choice of r enewable portfolio standards random? Energy Policy , 35 (11), 5571 5575. Hurlbut, D. (2008). State Clean Energy Practices: Renewable Portfolio Standards . Golden, CO.
108 IPCC. (2001). Climate Change 2001: Impacts, Adaptation, and Vulnerability . Geneva, Switze rland. IPCC. (2007). Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change . (M. L. Parry, O. F. Canziani, J. P. Palutikof, P. J. van de r Linden, & C. E. Hanson, Eds.). Cambridge, UK: Cambridge University Press. IRENA. (2018). Renewable Power Generation Costs in 2017 . Jacobs, D. (2014). Policy invention as a process of evolutionary tinkering and codification: the emergence of feed in tar iffs for renewable electricity. Environmental Politics , 23 (5), 755 773. Johnson, E. P. (2014). The cost of carbon dioxide abatement from state renewable portfolio standards. Resource and Energy Economics , 36 (2), 332 350. Karch, A. (2007). Emerging Issues and Future Directions in State Policy Diffusion Research. State Politics & Policy Quarterly , 7 (1), 54 80. Karch, A., & Cravens, M. (2014). Rapid Diffusion and Policy Reform: The Adoption and Modification of Three Strikes Laws . State Politics & Policy Quarterly (Vol. 14). Keeler, J. T. S. (1993). Opening the Window for Reform: Mandates, Crises, and Extraordinary Policy Making. Comparative Political Studies , 25 (4), 433 486. Fr iendly States. Kingdon, J. (1995). Agendas, Alternatives, and Public Policies . New York, NY: Harper Collins. Kiser, L. L., & Ostrom, E. (2000). The Three Words of Action: A Metatheoretical Synthesis of Institutional Approaches. In M. D. McGinnis (Ed.), Polycentric Games and Institutions: Readings from the Workshop in Political Theory and Policy Analysis (pp. 179 222). Ann Arbor, MI: University of Michigan Press. Klein, R. J. T., Nicholls, R. J., & Thomalla, F. (2004). Resilience to natural hazards: How useful is this concept? EVA Working Paper No. 9 . Potsdam, Germany. Knapp, C. A. (2012). The Effect of Renewable Portfolio Standards on State Level Employment: An Ex Post Analysis . Krause, R. M., Yi, H., & Feiock, R. C. (2015). Applying Policy Termination Theory to the Abandonment of Climate Protection Initiatives by U.S. Local Governments. Policy Studies Journal , 44 ( 2 ), 176 195 .
109 Kullgren, I. K. (2015, October 6). Environmental group files ballot measures for coal free power by 2030. The Oregonian . Retri eved from http://www.oregonlive.com/politics/index.ssf/2015/10/group_launches_ballot_measure.ht ml Kydes, A. S. (2007). Impacts of a renewable portfolio generation standard on US energy markets. Energy Policy , 35 (2), 809 814. Lehner, J. (2014). Industry Structure and Growth. Retrieved from https://oregoneconomicanalysis.com/2014/09/10/industry structure and growth/?structure?and?growth/ Leon, W. (2012). Designing the Right RPS: A Guide to selecting goals and program options for a Renewable Portfolio Stan dard . Washington, D.C. Lieberman, D., & Doherty, S. (2008). Renewable Energy as a Hedge Against Fuel Price Fluctuation . Commission for Environmental Cooperation . Retrieved from http://www3.cec.org/islandora/en/item/2360 renewable energy hedge against fuel price fluctuation en.pdf Lopez, A., Roberts, B., Heimiller, D., Blair, N., & Porro, G. (2012). U.S. Renewable Energy Technical Potentials: A GIS Based Analysis . Golden, CO. Lyon, T. P., & Yin, H. (2010). Why Do States Adopt Renewable Portfolio Standards ?: An Empirical Investigation. The Energy Journal , 31 (3), 133 157. Mason, B. (2006). Community disaster resilience: a summary of the March 20, 2006 workshop of the disasters roundtable . Washington, D.C.: National Academies Press. Matisoff, D. C. (2008). The adoption of state climate change policies and renewable portfolio standards: Regional diffusion or internal determinants? Review of Policy Research , 25 (6), 527 546. McCright, A. M., & Dunlap, R. E. (2011). The politicization of climate change and polarization 2010. The Sociological Quarterly , 52 (2), 155 194. Menz, F. C. (2005). Green electricity policies in the United States: Case study. Energy Policy , 33 (18), 2398 2410. Menz, F. C., & Vachon, S. (2006). The effectiveness of different policy regimes for promoting wind power: Experiences from the states. Energy Policy , 34 (14), 1786 1796. Miles, M., Huberman, M., & Saldana, J. (201 4). Qualitative Data Analysis (3rd ed.). Thousand Oaks, CA: SAGE Publications.
110 Mileti, D. (1999). Disasters by Design: A Reassessment of Natural Hazards in the United States . Washington, D.C.: Joseph Henry Press. Mills, S. B., Rabe, B. G., & Borick, C. ( 2015). Widespread Public Support for Renewable Energy Mandates Despite Proposed Rollbacks . Ann Arbor, MI. Mirza, M. M. Q. (2003). Climate change and extreme weather events: can developing countries adapt? Climate Policy , 3 (3), 233 248. Moser, S. C., & Dilling, L. (2011). Communicating Climate Change: Closing the Science Action Gap. In J. S. Dryzek, R. B. Norgaard, & D. Schlosberg (Eds.), The Oxford Handbook of Climate Change and Society (pp. 161 174). Oxford, UK: Oxford University Press. Nicholson Cro tty, S., & Carley, S. (2015). Effectiveness, Implementation Capacity, and Policy State Politics & Policy Quarterly , 1 20. Nohrstedt, D., & Weible, C. M. (2010). The Logic of Policy Change after Crisis: Proximi ty and Subsystem Interaction. Risk, Hazards & Crisis in Public Policy , 1 (2), 1 32. management. Disasters , 30 (1), 64 80. OECD. (2004). Educational Facilities and Risk Man agement Natural Disasters: Natural Disasters . Paris, France: OECD. Oregon Department of Energy. (n.d.). Addressing Climate Change. Retrieved from http://www.oregon.gov/energy/energy oregon/pages/climate change.aspx Ostrom, E. (2010). Beyond Markets and S tates: Polycentric Governance of Complex Economic Systems. American Economic Review , 100 (June), 641 672. Palmer, K., & Burtraw, D. (2005). Cost effectiveness of renewable electricity policies. Energy Economics , 27 (6), 873 894. Pantsios, A. (2015). War on Renewables Claims Victory in West Virginia. EcoWatch . Retrieved from https://www.ecowatch.com/war on renewables claims victory in west virginia 1882011401.html Prasad, M., & Munch, S. (2012). State level renewable electricity p olicies and reductions in carbon emissions. Energy Policy , 45 , 237 242. Rabe, B. G. (2008). States on steroids: The intergovernmental odyssey of American climate policy. Review of Policy Research , 25 (2), 105 128.
111 Rabe, B. G., & Mundo, P. A. (2007). Bus iness influence in state level environmental policy. In M. E. Kraft & S. Kamieniecki (Eds.), Business and environmental policy (pp. 265 297). Cambridge, MA: MIT Press. Rader, N., & Hempling, S. (2001). The Renewables Portfolio Standard: A Practical Guide. National Association of Regulatory Utility Commissioners , (February). Ragusa, J. M. (2010). The Lifecycle of Public Policy: An Event History Analysis of Repeals to Landmark Legisl ative Enactments, 1951 2006. American Politics Research , 38 (6), 1015 1051. Ragusa, J. M., & Birkhead, N. A. (2015). Parties, Preferences, and Congressional Organization: Explaining Repeals in Congress from 1877 to 2012. Political Research Quarterly , 68 (4) , 1 15. Ray, A., Hughes, L., Konisky, D. M., & Kaylor, C. (2017). Extreme weather exposure and support for climate change adaptation. Global Environmental Change , 46 , 104 113. Roberts, D. (2016, March 17). Oregon just used its Democratic majorities to pa ss a wildly ambitious clean energy bill. Vox . Retrieved from https://www.vox.com/2016/3/17/11252280/oregon clean energy bill Oregon Rogers, E. M. (1995). The Diffusion of Innovations . New York, NY: Free Press. Rogoway, M. (2014). Oregon data centers win praise from Greenpeace, a former skeptic. Retrieved from http://www.oregonlive.com/silicon forest/index.ssf/2014/04/oregon_data_centers_win_praise.html Rose Ackerman, S. (1980). Risk Taking and Reelection: Does Federalism Promote Innovation? The Journal of Legal Studies , 9 (3), 593 616. Roth, S. (2018, January 31). Climate change, extreme weather already threaten 50% of U.S. military sites. USA TODAY . Retrieved from https://www.usatoday.com/story/weather/2018/01/31/climate change extreme weather military defense department trump global warming wildfires droughts/1079278001/ Sabatier, P. A., & Weible, C. M. (2007a). The Advocacy Coalition: Innovations and Clarifications. In P. A. Sabatier (Ed.), Theories of the Policy Process (2nd ed., pp. 117 166). Boulde r, CO: Westview Press. Sabatier, P. A., & Weible, C. M. (2007b). The Advocacy Coalition Framework: Innovation and Clarifications. In P. A. Sabatier (Ed.), Theories of the Policy Process (pp. 189 220). Boulder, CO: Westview Press. Sanderson, I. (2002). Ev aluation, Policy Learning and Evidence based Policy Making. Public Administration , 80 (1), 1 22.
112 Sapat, A. (2004). Devolution and Innovation: The Adoption of State Environmental Policy Innovations by Administrative Agencies. Public Administration Review , 64 (2), 141 151. Schipper, L., & Pelling, M. (2006). Disaster risk, climate change and international development: scope for, and challenges to, integration. Disasters , 30 (1), 19 38. Sekar, S., & Sohngen, B. (2014). The Effects of Renewable Portfolio Standards on Carbon Intensity in the United States (No. RFF DP 14 10). Washington, D.C. Shipan, C. R., & Volden, C. (2008). The Mechanisms of Policy Diffusion. American Journal of Political Science , 52 (4), 840 857. Shipan, C. R., & Volden, C. (2012). Pol icy Diffusion: Seven Lessons for Scholars and Practitioners. Public Administration Review , 72 (6), 788 796. Shrimali, G., Jenner, S., Groba, F., Chan, G., & Indvik, J. (2012). Have State Renewable Portfolio Standards Really Worked? Synthesizing Past Policy Assessments to Build an Integrated Econometric Analysis of RPS effectiveness in the U.S. SSRN Electronic Journal . Sickinger, T. (2017, June 1). Portland, Multnomah County set 100 % renewable energy goal. The Oregonian . Portland, OR. Retrieved from http:/ /www.oregonlive.com/portland/index.ssf/2017/06/portland_multnomah_county_set. html Simmons, B. A., Dobbin, F., & Garrett, G. (2006). Introduction: The International Diffusion of Liberalism. International Organization , 60 (4), 781 810. Small, L. (2015). Wes t Virginia Sticks With Coal Despite Trends in Favor of Cleaner Energy. Retrieved from http://www.eesi.org/articles/view/west virginia sticks with coal despite trends in favor of cleaner energy Sovacool, B. K., & Barkenbus, J. N. (2007). Necessary but Insufficient State Renewable Portfolio Standards and Climate Change Policies. ENVIRONMENT , 49 (6), 21 30. Stata. (2013). . College Station, TX. Stoecker, R. (1991). Evaluating ad rethinking the cas e study. The Sociological Review , 39 (1), 24 50. Thomas, S. (2015, February 12). Experts: Climate Change May Make Northeast Winter Storms Worse. CBS Boston . Boston, MA. Retrieved from http://boston.cbslocal.com/2015/02/12/climate change snow noreaster expe rts may make northeast winter storms worse/
113 U.S. Census Bureau. (2016). Population and Housing Unit Estimates. Retrieved November 1, 2016, from https://www.census.gov/programs surveys/popest/data/data sets.html Upton, G. B., & Snyder, B. F. (2015). Rene wable energy potential and adoption of renewable portfolio standards. Utilities Policy , 36 , 67 70. van Aalst, M. K. (2006). The impacts of climate change on the risk of natural disasters. Disasters , 30 (1), 5 18. Insurance Program. American Journal of Political Science , 50 (2), 294 312. Volden, C., Ting, M. M., & Carpenter, D. P. (2008). A Formal Model of Learning and Polic y Diffusion. American Political Science Review , 102 (3), 319 332. Walker, J. L. (1969). The Diffusion of Innovations among the American States. The American Political Science Review , 63 (3), 880 899. Washington State Legislature. Declaration of policy (200 6). Retrieved from http://app.leg.wa.gov/RCW/default.aspx?cite=19.285.020 Weber, E. U. (2006). Experience based and description based perceptions of long term risk: Why global warming does not scare us (yet). Climatic Change , 77 (1 2), 103 120. Weible, C. M. (2006). An Advocacy Coalition Framework Approach to Stakeholder Analysis: Understanding the Political Context of California Marine Protected Area Policy. Journal of Public Administration Research and Theory , 17 (1), 95 117. Weible, C. M., & Nohrstedt, D. (2012). The Advocacy Coalition Framework: Coalitions, Learning, and Policy Change. In E. Araral, S. Fritzen, M. Howlett, M. Ramesh, & X. Wu (Eds.), Handbook of Public Policy (pp. 125 137). New York: Routledge. Wiser, R., Barbose, G., Heeter A Retrospective Analysis of the Benefits and Impacts of U.S. Renewable Portfolio Standards . Retrieved from https://emp.lbl.gov/sites/all/files/lbnl 1003961.pdf Wollmann, H. (2007). Policy Eval uation and Evaluation Research. In F. Fischer, G. J. Miller, & M. S. Sidney (Eds.), Handbook of Public Policy Analysis (pp. 393 402). Boca Raton, FL: CRC Press. Yi, H. (2013). Clean energy policies and green jobs: An evaluation of green jobs in U.S. metro politan areas. Energy Policy , 56 , 644 652. Yi, H., & Feiock, R. C. (2012). Policy Tool Interactions and the Adoption of State Renewable Portfolio Standards. Review of Policy Research , 29 (2), 193 206.
114 Yin, H., & Powers, N. (2010). Do state renewable portf olio standards promote in state renewable generation . Energy Policy , 38 (2), 1140 1149. Yin, R. K. (1994). Case Study Research: Design and Methods . Thousand Oaks, CA: SAGE Publications.
115 APPENDIX A Estimation Results for Model 1: Changes in p robabilities for imitation Model 1 from: to: dif ference : from: to: dif ference : Marginal effect x=min x=max min >max x=0 x=1 0 >1 Crisis inside a state 0.0033 0.0194 0.0161 0.0033 0.0038 0.0006 0.0007 Crisis outside a state 0.0064 0.0011 0.0053 0.0064 0.0056 0.0008 0.0007 Internal factors Net metering 0.0039 0.0059 0.0021 0.0039 0.0059 0.0021 0.002 Population growth 0.0034 0.0066 0.0033 0.0045 0.0047 0.0002 0.0002 GSP per capita 0.002 0.0188 0.0168 0.0003 0.0003 0 0.000 3 ideology 0.0015 0.0144 0.0129 0.0012 0.0013 0 0.0001 CO 2 emissions 0.0062 0.0007 0.0054 0.0062 0.0061 0.0001 0.0001 Solar wind potential 0.0044 0.0072 0.0028 0.0044 0.004 5 0.0001 0.0001 Electricity price 0.008 0.0001 0.0079 0.0141 0.0122 0.0019 0.0008 External factors RPS d uration 0.0007 0.0239 0.0231 0.0005 0.0007 0.0002 0.0015 N eighbor 0.0046 0.006 0.0014 0.0046 0.006 0.0014 0.0013 Diff. in population 0.0045 0.0052 0.0006 0.0045 0.0047 0.0001 0.0001 Diff. in GSP 0.0064 0.0016 0.0048 0.0064 0.006 2 0.000 2 0.000 2 Diff. in gov t ideology 0.0065 0.0023 0.0042 0.0065 0.0064 0.0001 0.0001 Diff. in CO 2 emissions 0.014 0 0.014 0.014 0.013 0.001 0.0004 Diff. in solar wind 0.0036 0.0189 0.0152 0.0036 0.003 9 0.000 3 0.0004 Diff. in elec price 0.0031 0.0711 0.068 0.0031 0.0037 0.0006 0.0007
116 APPENDIX B Estimation Results for Model 2: Changes in p robabilities for imitation Model 2 from: to: dif ference : from: to: dif ference : Marginal effect x=min x=max min >max x=0 x=1 0 >1 Crisis inside a state 0.0064 0.0302 0.0238 0.0064 0.0073 0.001 0.0012 Crisis outside a state 0.0104 0.0041 0.0063 0.0104 0.0096 0.0008 0.0007 Internal factors Net metering 0.0059 0.0158 0.0099 0.0059 0.0158 0.0099 0.0088 Population growth 0.0038 0.0194 0.0156 0.0078 0.0088 0.001 0.001 GSP per capita 0.0033 0.051 0.0477 0.0003 0.0003 0 0.0006 0.0016 0.0487 0.0471 0.0011 0.0012 0.0001 0.0004 CO 2 emissions 0.0153 0.0002 0.0151 0.0154 0.0148 0.0005 0.0003 Solar wind potential 0.0055 0.0862 0.0807 0.0055 0.00 64 0 .0009 0 .0013 Electricity price 0.0075 0.0138 0.0063 0.006 0.0063 0.0004 0.0005 External factors RPS d uration 0.0011 0.058 0.057 0.0007 0.0011 0.0004 0.0033 N eighbor 0.0086 0.0116 0.0029 0.0086 0.0116 0.0029 0.0026 Diff. in population 0.0102 0.006 0.0042 0.0102 0.0091 0.0012 0.0011 Diff. in GSP 0.0106 0.0049 0.0057 0.0106 0.010 4 0.0002 0.0002 Diff. in gov t ideology 0.0084 0.0098 0.0014 0.0084 0.0084 0 0 Diff. in CO 2 emissions 0.0137 0.0001 0.0135 0.0137 0.0131 0.0005 0.0004 Diff. in solar wind 0.0078 0.0166 0.0088 0.0078 0.00 81 0 .0003 0.0003 Diff. in elec price 0.0071 0.0382 0.0311 0.0071 0.0077 0.0006 0.0007