DETERMINANTS OF STATE LEVEL RENEWABLE ENERGY POLICY by DANIEL T HATTRUP B.A. / B.S., Regis College of Regis University, 1994 M.B.A, Regis University, 2002 M.A., University of Colorado at Denver, 2006 A dissertation submitted to the F aculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy Public Affairs Program 2014
ii This dissertation for the Doctor of Philosophy degree by Daniel T Hattrup ha s been approved for the Public Affairs Program by Paul Teske, Dissertation Chair Chris Weible, Examination Chair Don Klingner Dan Rees November 20, 2014
iii Hattrup, Daniel T (Ph.D., Public Affairs) Potential Determinants of State Level Renewable Ener gy Policy Dissertation directed by Dean Paul Teske. ABSTRACT Renewable energy remains in the forefront of environmental awareness, public policy and national security. Almost all of the variation in the renewable energy generated could be associated with the states themselves, rather than any policy success. There was little variation in energy generated across time or within the states. The initial conclusion regarding state level total renewable energy generation is that it is a story of legacy federal programs, including many hydroelectric power plants. Narrowing the view to solar and wind generation instead of total renewable energy provides new insights : the predictive power of the states sui generis is eliminated; providing a tax break or lowering the marginal cost for renewable energy successfully increased wind and solar energy generation. Most of the growth has appeared on the Predicting why a policy was adopted was difficult. A s most of the policies did only variables with consistent predictive power toward adoption were the presence of the policy in neighboring states and average wage levels The form and content of this abstract are approved. I recommend its publication. Approved: Paul Teske
iv TABLE OF CONTENTS CHAPTER I INTRODUCTION ................................ ................................ ................................ ... 1 II LITERATURE REVIEW ................................ ................................ ........................ 8 Theory ................................ ................................ ................................ ...................... 9 Rational Choice Theory ................................ ................................ ......................... 10 Measures of Success ................................ ................................ .............................. 13 Innovation Diffusion and Adoption ................................ ................................ ....... 16 Role of Government in Innovation Diffusion and Adoption ................................ 20 Supply versus Demand versus Consumption ................................ ......................... 25 Studies on Renewable Energy ................................ ................................ ............... 28 Rational Choice Theories ................................ ................................ ....................... 30 Modern Economic M odels ................................ ................................ ..................... 32 Political Philosophy and Alternative Energy ................................ ......................... 37 Policy Innovation and Adoption ................................ ................................ ............ 39 Policy Success and Failure ................................ ................................ ..................... 43 Policy Tools Taxes ................................ ................................ .............................. 46 Policy S pecific R esearch ................................ ................................ ........................ 50 Literature Review Conclusion ................................ ................................ ............... 55 III METHODOLOGY ................................ ................................ ................................ 57 OLS Models ................................ ................................ ................................ ........... 61 Event History Analysis ................................ ................................ .......................... 64 Hypotheses and Predictions ................................ ................................ ................... 65
v IV ANALYSIS ................................ ................................ ................................ ............ 73 Overall Results ................................ ................................ ................................ ....... 74 Demographic R esults Full Model Regressions ................................ ................... 79 Policy Review and Results ................................ ................................ ..................... 83 P o licy R eview and R esults Full M odel R egressions ................................ .......... 84 Policy Review and Results Solar and Wind Regressions ................................ ... 85 Demographic and Policy Analysis Conclusions ................................ .................... 90 Event History Analysis ................................ ................................ .......................... 91 V DISCUSSION ................................ ................................ ................................ ........ 96 Research Questions Discussion ................................ ................................ ............. 98 Hypothesis Analysis and Results Discussion ................................ ...................... 102 VI FURTHER RESEA RCH ................................ ................................ ..................... 107 VII CONCLUSION ................................ ................................ ................................ .... 114 REFERENCES ................................ ................................ ................................ ............. 159 APPENDIX A REGRESSION RESULTS ................................ ................................ .................. 128 B DESCRIPTIVE STATISTICS ................................ ................................ ......... 159 47
CHAPTER I INTRODUCTION Power generation using renewable sources has mo ved from being a topic of environmental interest to one of economic and national security. While small steps have been taken by industry to use more renewable energy sources, the major changes in energy production have come after policy changes and incent ives created by the government. This dissertation will look at the interaction of government policy, market forces, technological innovation and entrepreneurial actions in order to further the literature regarding adoption and growth in renewable energy g eneration. It will explore the changes in renewable energy generation based on policies, legislative makeup, consumer demographics and characteristics T raditional energy producers burn fossil fuels such as natural gas, coal and oil. Energy produced fr om fossil fuels contributes to carbon dioxide emissions, and exhausts a limited fuel source. In contrast to these limited fuel sources, renewable energy sources include solar, wind, water, geothermal heat and biologic products, such as switchgrass, ethano l, and waste. include nuclear power. This paper will only be looking at renewable energy, and not examining current efforts in the nuclear industry. R egulations, incentives and policy mandates can all contribute to the use of renewable energy fuels instead of fossil fuels. Policy in the 1970s focused on conservation of energy, as renewable energy technology was still being developed. Concurrent with the Reagan era orientation toward market based systems and culminating with New Public Management programs, policy approaches have moved
2 from federal funding and administration to the laboratory of the state legislatures. Federal programs have funded research and development, as well as p roviding tax credits for people and companies who install renewable energy production produc ts such as solar panels or wind powered turbines While states have also used these same types of policies, much of the innovation in policy has occurred at the st ate level. Among these innovations are renewable portfolio standards (requiring a certain amount or percentage of renewable energy), fuel disclosure rules that inform the population where their energy comes from, and mandates either allowing renewable gen eration in zoned areas or requiring certain percentages of green building materials. Modern interest in alternative energy policy began after World War II, when the military precursor s to the Atomic Energy Commission were reorganized and changed its mis sion from a military focus (creating atomic weapons) to using nuclear technology to generate civilian power. The innovation and diffusion of nuclear power showcases many of the issues still found in sustainable energy debates: choice of type of competing technologies for energy production competition versus monopoly, government subsidies for building reactors government funded R&D (P. deLeon, 1979) A similar study identifies political agendas and technological lock in as reasons why solar power was not adopted as an alternati ve energy source. While nuclear was preferred to solar w ithout the intervention of the government in funding research and development, neither of these alternative energy industries may have survived competitive pressures fr om established fossil fuel burning utility companies (Laird, 2001) In America, t he oil shocks of the 1970s brought alternative energy sources to the forefront of political energy debate, recognizing that American oil depend ence created an
3 economic risk (Laird, 2001) Policies were created by the government to reduce energy use and develop alternative fuel sou rces, including subsidies, price caps and even fe deral energy conservation The 1980s saw continuation of many of these federal policies although they were not expanded. A n ew generation of political risk s came about in 1990 with the invasion of Kuwait by Iraq. to Western nations, and the invaders were routed by a United Nations coalition. Current threats to the stability Strait of Hormuz and uprising / overthrow of established governments in oil producing Arab nations. The political and military actions undertaken to mitigate the risk of loss of oil supplies and ensuing economic instability of the industrialized (and even industrializing world, such as China) has led American energy policy to be characterized as a dependence on imported oil matched to a willingness to sec ure oil by force, if necessary (Hart, 2006) It is important to note that the use of military power to secure energy and/or resource supplies is an active energy policy decision. It creates an implicit subsidy for oil production a nd fossil fuel consumption, which is often not incorporated into analysis of the production costs of fossil fuel energy vis vis renewable energy. One reason for this may be the difficulty in defining what should be considered part of the subsidy (costs of direct military actions, foreign aid to oil producing states, other trade deals, etc). Congress allocated $94 billion to the war effort in Iraq for 2009 (Belasco, 2009) which is 18% of the annual trade in crude oil (at close of 8/14/2009) (Landy, 2009) A second reason may be that not incorporating the military costs of securing fossil fuels serves to lower the price of these fuels, both for populist reasons as well as more favorable
4 comparisons to renewable. This security is an externalized cost that the oil producers (or consumers) need not factor in when looking at the cost of fossil fuel based energy. The costs of the Iraq wars, as well as the naval presence in the Indian Ocean and Gulf region continue to accumulate and continue to subsidize the oil industry. In addit ion to paying for the costs of energy, consumer income plays an additional two roles for this analysis. The first is the substitution effect, arising from a marked change in the environmental awareness of the general population (often linked to publicatio Silent Spring in 1962 and Earth Day in 1970 ) Environmentally aware consumers have exhibited a willingness to pay a premium for (i.e. those that are produced in an environmentally friendly way (Carson & Mouldon, 1991) ) Consumers will make an active choice to continue purchasing these goods, even in an economic downturn (Carrigan & Pelsmacker, 2009) The second role is commonly known as the income effect goes up, s/he increases consumption as well. This income based demand for energy creates another mechanism for government policy to function, by creating incentives for increases in consumption of renewable energy. Decreases in overall energy use, tied to increases in consumption of renewable energy, would change the ratio of renewable to non renewable sources. Gatersleben, Steg and Vlek (2002) find that household energy use is tied to house size and income, rather than environmental awareness. In addressing the various producti on and consumption issues, different levels of government have enacted a mix of rules and regulations, standards, and market based incentives. A preference for command and control models over market models or vice
5 versa, starts in p olitical and economic philosophy. This preference is evidenced by market mechanisms (decoupling of utility profits from energy generation, cap and trade emission markets, renewable portfolio standards), or public policy implementations (regulatory actions such as technology fo rcing or green building codes). While post World War II actions focused on the role of the federal government, many of the recent innovations in policy have used market incentives to meet the policy goal. While the government plays a key role in creating energy polic y a central question remains: which is more effective government mandates or market forces? terpretation, as it requires evaluation on multiple levels. Standards for performance need to be define d and measurement criteria established. The most basic analysis is cost benefit: is renewable energy less expe nsive than fossil fuel energy? Even this simple model requires one to a ppropriately i ncorporat e the external costs of fossil fuel energy (pollut ion, aforementioned military costs, etc) that pricing is a decades long debate with only limited resolution (Baumol, 1972; Coates, 1960; Parry & Bento, 1999) Moving beyond the simple cost analysis, environmentally friendly behavior can be valued based on social norms regarding purchasing green goods. As Shiller (1984) explained, soci al pressure and preferences can affect d ecision making ; and ideas of environmental justice and commitment to future generations are ranked as important considerations in resolving environmental con f licts (Clayton, 2000) Policy evaluation requires community agreement on the desired goals; these financial, social or environmental goals can be in conflict.
6 This paper will look at the drivers of adop tion of state level policies. The first are these policies effective at increasing generation of renewable energy, and thus they speak for themselves for adoption? The mechanism for this could be either fin ancial subsidies / incentives, or increases in demand that allow for economies of scale to evolve. If the policies are not simply being adopted? Are there are social an d economic influences to the adoption process? These and other questions will be explored. To deal with these many issues, t his paper will explore the economic and political theories that underlie the various policies; and look at some determinants of w hy the different policies were adopted in some states, but not in others. Existing studies of renewable energy are reviewed for previously used methodology and results, and built upon to create the methodology used for testing the effect of policies. The paper will then test these various policies and demographic markers against state level renewable energy generation data for changes in generation. The policies will be regressed against different state level variables to potentially explain their adopti on. The implications of reaching for policy makers. If a certain group or type of policy or incentive is more effective that others, those c ould be adopted in other states to furth er renewable energy consumption; environment al advocates can use the demographic data to study which approaches may be most easily passed in their state. Contemporary events and public opinion continue to focus the national and international community on renewable energy as a force to reduce human i mpact on the environment and climate change. P olitical agendas from both established parties and
7 non governmental organizations strive for a high level of awareness of energy policy solutions. In the U.S., President creases in fuel economy for gasoline powered vehicles and tied economic recovery funds to alternative energy vehicles (Vlasic, 2012) In Europe, the executive office of the Euro pean Union has used a bully pulpit to call for continued progress in finding alternatives to traditional fuel sources (Exergia Consultants, 2012) With these future plans in mind, let us first look to the past to see what has been trie d and researched before.
8 CHAPTER II LITERATURE REVIEW The literature review is split into theoretical underpinnings of how one can describe and understand the policy levers, and the ways that the effects of these levers are measured. State level polici es can be categorized as regulations or financial incentives. The use of a financial incentive is a market based policy that focuses on rational choice theory, where a market participant is able to gather information regarding his or her options and imple ment that in a market choice of (non )consumption or (non )production. Regulations are generally used in a command policy, where the market is seen as ineffective or where a minimum standard must be implemented by market participants. Both incentives and regulations fit into Rogers (2003) diffusion of innovation model, which will be applied to this discussion. Finally, political views, institutions and demographics will be reviewed fo r their influence on environmental policy and energy consumption choices. Much of the literature around renewable energy and policies has followed from the policies enacted (which were primarily at the federal level prior to the 1980s) and the data availa ble for research. A considerable number of case studies of various locat ions, countries and policies has been supplemented by pro forma projections of possible energy generation and costs. Empirical and cross location research has been rather limited, an d this dissertation will contribute to the literature with both an empirical focus, as well as using the fifty states as a natural experiment with variance among the states. How the various incentives and regulations fit into the theory will be introduced in the first section
9 of this literature review, and many of the studies examining the success of the policies will be included in the second section. Theory This section will first look at the ground level theories used to describe the market for renewab le energy, transitioning from a pure market view toward one with government intervention. From this high level approach, a specific discussion follows regarding the evolution and application of renewable energy policies, intertwined with the theoretical a nd political impetus for the changes. The history of renewable energy policy mirrors that of the broad changes in American political views, from the command and control models effective during the Second World War, to high levels of federal action in the 1970s through to the diffusion of innovation to the states beginning in the 1980s. A second mirroring is that of the shift from technocratic expertise in government programs toward the use of market forces to create e conomic e fficiency in these programs a doption and execution. In an adoption of innovation approach, the U.S. was one of the few countries with the knowledge to harness nuclear energy in the 1940 50s Even here, there are distinctions, as the U.S. took a fairly central approach to plant des ign, whereas other countries (e.g. France) had multiple designs for power that competed with each other (P. deLeon, 1979) While the percentage of US federal f unding for nuclear power declined significantly from 1948 to 2012 (49% to 25% of expenditures), most of this was made up for by spending increased energy efficiency (9.7% to 16.4%) and electric systems research (4.4 to 15.1%), with renewable energy growing from 11.6% to 17.1% (Ratner, Belkin, Nichol, & Woehrel, 2012) Other countries have seen more significant growth in
10 their expenditures on nuclea r and renewable energy research on a relative basis, and can be considered to be sharing the burden of the research and development costs for much of our renewable innovations. This will be discussed further in depth in the discussion around innov ation model. Rational Choice Theory This section will discuss Rational Choice Theory (RCT) a primary framework that can be used to discuss individual, market actors and political decision. It will first look at the different subsets of RCT and how it is applied to the market. Some of the weaknesses of applying RCT will be identified, as well as show examples of how the government enacts policies using a rational choice approach. RCT is commonly seen in economic theory through cost benefit analysis, whi ch asks if the benefits of an action outweigh the costs of it. Its roots on the individual level r un back to Hobbes and Rousseau; those roots grow into as the fundamental building block of society. I n the economic realm, RCT traces its history to the Physiocrats and Wealth of Nations Their shared idea that p eople mak e decisions that make themselves better off underpins trade in a society This has formed the foundation for types of incentives and in government policy: subsidies are based on the idea that if a good costs less, people will be more likely to purch a se it. A difficult assumption to validate in this model is that consumers have perfect information when deciding. However, e ven if a per son does not have perfect information about a decision or its consequences, t here is a rich literature that discusses choices (Green & Shapiro, 1994; Jones, 2002; Knight, 1921; Simon, 1997, 1947)
11 T is that individuals attempt to maximize their personal utility (happiness), without attempting to define what that happiness is. D emand side management policies public benefit funds and power consumption management programs typically appeal to the thin version of RCT. By providing incentives to use energy efficient appliances (or even energy efficient light bulbs), the government tries to sway consumer behavior. Tax c by lowering the cost of a good, the government hopes to overcome the difference in price between traditionally (usually cheaper) and alternative ly fueled (usually more expensive) goods. sion of RCT often assumes that what makes people happy is a belief or action shared across a group of individuals (Green & Shapiro, 1994) While the thin version simply looks at behaviors, the thick version looks at beliefs and desires that led to that decision (Elster, 1994) energy as a result of their belief system whether environmentalism or based in religious disclosures, and mandatory options for green power. In either version individuals choose between different options, attempting to make themselves better of f. Evidence that people value the environment is available both from what they buy (Carson & Mouldon, 1991) and what they say (Guber, 2003) Recent governmental organizations such as Repower America and in dividuals like T. Boone Pickens alon g
12 with alternative energy companies in the market ; as well as increased optio ns in the market for hybrid and electric cars. These groups combine market influences (new jobs opportunities and new markets) with an overall ethos of stewardship for the Earth. The application of RCT to consumer and producer decisions will be discussed further, under the policy review section of the literature review. Looking toward the macro view, ap plying RCT to renewable energy creates an intersection between economics and p olitics. c lassic al economic s assum es perfect competition and free entry and exit from the market. When looking at energy production, both of these assumptions must be relaxed in order to effectively model the choices m ade by consumers (and producers). These can be seen It is both costly and risky to develop a new in dustry or product (Islas, 1997). Current industry actors will make incremental cha nges to their existing products and processes, while new (usually smaller) firms will attempt to revolutionize the market (Acemoglu & Cao, 2010) With incomplete knowledge of the risks and return, the market will function in efficiently, decreasing the success of supply side financial incentives. Information asymmetries are when one party (either the producer or consumer) knows more about a good or service than the other party. In the case of renewable energy, power generatio n disclosure rules and public education funds attempt to close this knowledge gap (DSIRE, 2013) basis An alternative view of this knowledge gap is provided by Simon (1997, 1947) His comments on the satisficing nature of individual choice indicate (in a thin RCT view) that power consumers should not care about where their energy comes from, provided
13 that the lights come on when one flips a switch. In a thick RCT view, there is room to explore the motivation beyond turning on a light, usin g natural lighting alternatives, or changing out the light bulbs for fluorescent lights or light emitting diodes (LEDs) that will consume less energy. While satisficing the need for energy, consumers can begin to express their membership in a social cons ciousness and understanding of their carbon footprint. Viewing renewable energy production as an incomplete market allows us to draw upon transaction cost theory to explain the lack of suppliers and/or availability of substitutes for a specific good. T he hurdles of creating infrastructure and creating market awareness of substitutes can make otherwise viable energy projects look unprofitable to corporations, and they are rejected by business decision makers. These infrastructure and market hurdles are evidence of high barriers for market adoption of innovation in the renewable energy industry. A traditional remedy to an incomplete market is government intervention. This intervention can be either producing the energy, creating a demand for it, or offe ring subsidies / buy in tariffs for the new market. In addition, technology forcing (or production mandates) are also tools that the government can use to help solve for the incomplete market. Measures of S uccess As mentioned above, a common measure for success in RCT is cost benefit analysis. At the individual level choice level, it is probably the primary method ? Smith (1776) indicate d that the market will move to an economically efficient point through the amalgamation of individ ual choices. Influences on that individual decision can be traced
14 back to Veblen (2000, 1899) where consumption choices are impacted by the decisio n social group. North (2003) note d that b elief structures get transformed into societal and economic structures by institutions both formal rule s and informal norms of behavior. Bridging individual choice and societal structures, theorem (1963) can be oversimplified to the idea that aggregating conflicting individual choices on a s ocial level will not maximize social utility. A plan for social utility is either imposed by another group, or is a compromise from maximum individual utility. Olson (1965) discussed the need for incentives to promote choices that will create significant benefits for the group, but not necessarily to the individual making the choice. measure of cost. Over the short and/or long term, did the policy stimulate a market into of access for the good (if valued by the society) and then adoption of the good, incremental increases in renewable energy and increases in percent of population buying this good. Even these incentives and choices for benefits to the group can be disputed, based on political philosophy or priority of the problem at hand. As a result, decision making and policy analys is needs to incorporate additional levels of scope beyond individual cost benefits. Klingner, Nalbandian and Llorens (2010, p. 90) offer ed three measures: productivity, effectiveness and responsiveness. Productivity is the output versus the inputs, and similar to cost benefit analysis; effectiveness is tied to productivity, with a quality measure include d, ensuring that services exceed a certain standard for the
15 goal we set out to accomplish worthwhile in light of the other goals we might have Defin ing and prioritizing the goals of the citizenry can be accomplished at the local level. During this process, three issues arise: the scope of the problem, the scope and cost of the solution, and how the problem is defined Both the media and the wealth a nd numbers of advocates on either side of a policy question play an important role in shaping the definition of the problem and potential solutions (Pitkin 1967) In addition, policy entrepreneurs can help define the agenda, including defining the problem and response among the citizenry and policy makers (Mintrom, 1997) Public participation in policy making will help it reflect their ideals, while building trust in the regulation of the policy (Rowe & Frewer, 2000) In addition, advocacy groups or other non governmental organizations can bring together individuals of like minds (Yang & Callahan, 2007) This can be for common beliefs ( e.g. Sierra Club, Greenpeace), or market opportunity tied to environmental awareness ( e.g. for wind energy) (Andrews & Edwards, 2004) Measuring the responsiveness of a program or policy is difficult, requir ing explicit value judgments; and policy analysis will often focus on more easily defined cost benefit analysis (Klingner et al., 2010, p. 91; Moynihan et al., 2011) Expanding on the Miller Stokes model, Thomassen (1994) found that some representatives are able and willing to vote their conscience on social issues. Shanks and Strand (1994) identified and economic characteristics in their catchment partisan identification of policy predisposition, expressed preferences for policy issue and consistency in evaluation of the
16 representative. Eulau and Karps (1977, p. 242) operationalize d policy responsiveness as when a representative and the constituency agree, the representative is responsive, regardless if this is in the best interest of the voters. Thus, the more stable a district is, the more likely a representative can make a value judgment without fear of being voted out in the next election. Bowler and Donovan (Bowler & Donovan, 2002) show ed that legislatures should be more responsive to citizens in state s with the initiative process. ve an additional effect, in that they can unbundle an issue from representative voting. In a state with initiative capabilities, if a topic is of enough actions (Besley & Coate, 2000) Thin rational choice has set the foundation for individual choice making, while thick rational choice indicates that individuals may make decisions based on beliefs that extend beyon d simple economics. The interaction of individuals in groups leads to deeper analysis, and discussion of multiple viewpoints (Nonaka, 1991) Decision making in groups moves away from cost benefit analysis as a result of the beliefs held by members of the group, and the success of a decision or policy can be measured both by effective productivity, mind, we now look to how different decision makers l earn about policy inventions. Innovation Diffusion and Adoption A prominent model of the factors and actors in adopting innovations comes from Rogers (2003) His model looks at the stages of the diffusion process, the factors that influence each stage, as well as the levels where decisions are made. The process is one
17 of innovation, which is communicated via channels, with time and social systems leading to adoption or rejection. The next deeper level i n this process can be summarized as innovations have the following characteristics : relative advantage, compatibilit y, complexity, trialability and observability. Communication is the method of transferring knowledge (or the innovation) from one person to the next, and the success of communication is tied to the level of homophily (sameness) between the actors communic ating. Time is demarcated in the process in five steps: knowledge, persuasion, decision, implementation and confirmation. Finally, all of these pieces are governed within a social system, structured with three main types of innovation decisions: optional (made by individuals), collective (made by consensus of the members of a system) and authority ( made by those possessing power) (Hoffmann, 2011) Describing or adopting an innovation can be seen as an iterative p rocess, rather than a discrete moment in time (Knott & Wildavsky, 1980) This process has been used in multiple studies, including the effect of research on policy making (Landry, Lamari, & Amara, 2003) If an innovation is initially rejecte d, it can still be revisited The decision to reconsider can be influenced by new or additional information, changes in values or political support (Fesler & Kettl, 1991) Tying together the process vi sually is an often seen graphic, which shows the interconnections of the stages.
18 Figure 1: Flowchart of Innovation Adoption and Diffusion model. Reproduced from (European I nformation Society, 2012) When reviewing an innovation, actors will look at relative advantage and compatibility with existing technology among others. As an example, Islas (1997) described how using gas turbines for short term supplemental energy creation was adopted quickly by the market ; it allowed for a comparative advantage (easy to start/stop to capture higher spot mar ket energy rates), and was compatible with existing technology. The incremental gains allowed for greater profit with a low level of complexity; it was thus quickly adopt ed by existing industry and did not requir e a radical departure from the norm (Acemoglu & Cao, 2010) Prior to the turn of the 21 st century, innovation in renewable energy was the purview of niche companies in a small market (Islas, 1997) Only when sustainable
19 processes became a marketing advantage (a separate innovation) did existing energy companies project adequate reward for allocating resources into developing renewable energy technology. There was no relative advantage to companies to move away generating energy from fossil fuels (in fact, there would be added costs), so the Customer demand can stimulate innovation in p roduction (Griffith, Huergo, Mairesse, & Peters, 2006) so as consumers communicated their desire for more environmentally friendly goods, this collective will changed the perception of renewable energy, persuading the producers to reconsider their prior decision to reject renewable energy. From a thin RCT point of view, t he initial non entry of companies into to the renewa ble energy market follows the pattern forecast by advocates of path dependency W hen a technological standard is first adopted (in this case, fossil fuel power generation), it is a rational choice for other companies to also adopt this standard. By incre asing the demand for a single technology, economies of scale can be achieved by the suppliers of the technology. This lowers costs for all of the companies. Even if the technological standard is no longer adequate, appropriate, or efficient, the difficul ty of achieving the already achieved economies of scale in the other technological deme snes will prevent adoption of the new technology (Arthur, 1989; David, 1985; North, 1990; Piers on, 2000) Radical changes in the market are possible, but infrequent (Acemoglu & Cao, 2010; Schumpeter, 1950) With a thick RCT view, the collective will of consumers acting either as individ ual decision makers or as adherents to stakeholder interest groups, may make it profitable for a company (operating in a thin RCT mode) to create the environmentally
20 friendly process / products that the consumers are demanding. In addition, other companie s (e.g. Interface carpet, Patagonia Sportswear ) actively market themselves as sustainable The sense of stewardship to the environment and society outweighs the bottom line economic calculations of minimal cost with maximum revenue to produce a good. How ever, even within collective action, consumer behavior can be inconsistent or influenced by different motivating factors (Gatersleben et al., 2002) The Yale Center for Environmental Law and Policy has published surveys regarding American attitudes toward the environment, breaking down the population into 6 primary groups: alarmed, concerned, cautious, disengaged, doubtful and dismissive (2009; 2010) These surveys show that self identifying Republicans tend to be doubtful or dismissive of environmental concerns, while Democrats are the greatest proportion of people alarmed, concerned or cautious about th e state of the environment (2009, p. 25) In confirmation of received wisdom, the Alarmed and Concerned are supportive of almost all policies, with the exception of cap and trade and a gasoline tax (two of the (2009, p. 17) However, even when environmental (2009, p. 62) As a result, while support for environmental regulation may be strongest in left leaning legislatures, it would not be surprising to see Republican or split congresses enact market based policies. Role of Government in Innovation Diffusion and Adoption If innovations are not adopted either at the optional or collective level an authority innovation model may co me into play for a variety of reasons Foster (1978)
21 b uilt (Walker, 1969) seminal work in innovation across regions of states, with some states leading as early adopters, and others as laggards. Decisions by state level authorities combine citizen responsiveness, political philosoph y and citizen homophily into the policy process (Rabe, 2006) Political philosophy and homogeneity of the legislature determines the kinds of levers used, either market based or rules and regulatio ns. The effect of regulation on directed market / technological innovation is an open debate. Calel and DeChezlepretre (2011) present ed historic examples where government action was modeled and found to increase innovation; they use a difference in difference model to show that the regulations did not change the innovation trend. The conversion of the tacit knowledge to explicit and actionable knowledge can take a variety of paths (Nonaka, 1991, 1994) The primary system drives representatives to reflect the beliefs of the voters; and in open primaries, the median view of the constituency (Gerber & Morton, 1998) Legislators rely on legislative staff research, as well as personal and constituent life experiences when taking a stance on policies. The staff is influenced by the executive, study commissions and what is happening in other states (Gray & Lowery, 2000) V arious policies can bridge between the market and regulatory perspectives. A n example is technological forcing the government tells the energy producers that they are required to use a certain amount of renewable energy in their ener gy production. Two options are available in this technological forcing: allowing the industry to choose among renewable energy sources portion to come from specific sources. By forcing a certain amount of demand for a
22 does not dictate the production process, rather it harnesses market forces of efficiency grow into more demand (Field & Field, 2005) To help energy consumers grow this demand, g overnment policies target methods such as subsidies and tax incentives lower these transac tion costs, thus bringing renewable energy and fossil fuel derived energy costs more in line with each other. This is a simple reflection of th in rational choice theory as relates to pricing if something is less expensive, people are more likely to purc hase it. In addition, via the thick lens, as people value environmentally friendly products (including electricity) more highly, a marginal consumer may deci de to pay a little bit (i.e. environmentally friendly) energy rather than fossil fuel (Carson & Mouldon, 1991; Guber, 2003) These incentives and subsidies may help a consumer meet his willingness to pay (the maximum price a person is willing to pay for a good) for green electricity. As discussed above, g overnment environmental policy options can be centralized or decentralized ; they can attempt to support one solution or multi ple solutions to energy problems. The underlying rational choice framework here is public choice modeling, as en promoting renewable energy, there is potential for different producers to engage in re nt seeking behavior. If producers are able to influence the legislative bodies, certain alternative energy sourc es may be favored over others, leading to barriers to entry for technologies (Sawin, 2004) While legislators state that think tanks and the like do not influence their decisions, lobbyists are ranked in the top five sources of information on a policy (Gray &
23 Lowery, 2000) In many countries, the policies can come from federal, state or local levels. Each of these levels interacts with the framework to add to the complexity o f the analysis. Some of the other inputs and considerations facing policy makers are included in chart below. This literature review continues with a discussion of the inputs, both internal and external to the state where the policy is being enacted. F e sler and Kettl (1991, p. 48) discuss ed an organization as a system of inputs, throughputs and outputs, along with a feedback mech anism. By combining that simplified model with environmental policy and research, a rough model of the policy construction, design and outcomes can be derived. A graphic representation of that model follows : Figure 2: Inputs of Policy Making Pr ocess Policy Characteris tics Design Financial Regulatory Market driven Implementation & Administration Technology( ies) Public/Private Target Audience Outcomes Definitions Metrics External Inputs Internal Inputs Academic research Federal requirements Lobbying groups Neig hboring States Citizen preferences Demographics Initiative process Legislative staff Natural endowment Political philosophy Policy Evaluation Success Adaptations Modifications Time frame Continue policy? Feedback Met goals? Desired results? Changes to policy? Review of Inputs
24 Looking to the past, s tudies of n uclear (P. deLeon, 1979) solar (Laird, 2001) and ethanol (Schmitz, Moss, Schmi tz, Furtan, & Schmitz, 2010) power have show n the importance of both political support and fina ncial incentives in creating a nascent alternative energy industry. For nuclear power, federal requirements led to policy design of centralized mandates and su bsidies. With solar and ethanol technologies there was little market or consumer awareness of the potential (few internal inputs, influencing design), solar policies focused on subsidies; ethanol received a supply side subsidy and mandate to be blended i nto gasoline. I nitial positive perceptions in the power of nuclear energy for America were negated by the fallout surrounding the Three Mile Island incident. This bias against nuclear power has continued after the Chernobyl explosion and expanded greatly throughout the world after the Fukushima disaster in 2011. As a result of this change in citizen preferences other renewable technologies hav e been demanded by the market. This feedback renewable energy, as there is more c onversation (additional information, change in preferences) around the subject To help create a new market, many recent incentives for renewable ener gy development have used market forces and unconventional processes for creating energy. An example of t his is net metering programs where consumers can purchase and install solar panels and wind turbines on their property. Excess energy is sold back to the utility company, thus creating a distributed energy production grid. There remains a risk that politi cal actors may be able to have one or two c ompeting renewable energy power source (Frondel & Peters, 2005) Such a technological path dependenc y means adopting a single technology leads to quicker cost savings and efficiencies of scale for the producers of the
25 existing market players for the prefe rred energy source (Pierson, 2000; Schwartz, 2000) This c an also lead to poor social outcomes as localities may not be able to afford the renewable option that best suits their natural climate endowment. Supply versus Demand versus Consumption T he policy pendulum swings between central and distributed forces frequently in American politics (Stillman II, 1999) and the debate in renewable energy is no different. It started with a simple problem solution: the future of energy was going to be nuclear power the plants were designed and defined by government contractors, and the orders went out As alternatives began to appear, the clarity of goals, processes and consistency of the situation confronting political actors changed, and contributed to the morphing of the effective organizational scheme (L. deLeon, 1996; Miller, 1992; Scott, 2003; Scott & Davis, 2007) When evaluating renewable energy policies across the past four decades we witness the transition from hierarchical (command and control) to more anarchic (and market based) policy administration. Another pendulum has been economic analysis : early analysis looked at supply driven mechanisms l ater economic analysis has focused on managing consumption to either grow an economy or achieve economies of scale, which make it (m ore) profitable for a corporation to produce a good. While many economists focus on demand side focus on production issues. Environmental policy has followed this same pa ttern. Early efforts were focused on developing the supply and technology required for renewable energy production.
26 conservation. Later programs have created incentives and sub sidies for the consumption of renewable energy. The current trend in renewable energy is toward distributed generation, where individuals install solar panels or wind turbines on their house or land, bring ing the supply demand supply argument full cir cle. As different technologies were developed, the means of meeting the goal of increasing the use of renewable energy became less clear. As o ther ideas were mooted, the pendulum of administration began to swing from a solitary answer to a set of multi ple solutions. It i s appropriate here to look at a community based model of innovation in policy and technology (L. deLeon, 1996) A marked change in regulation was heralded by President Clinton in 1995. Standards would be set by the government, but companies would be allowed to find the most efficient way to meet those standards (Clinton, 1995) State level agencies joined together to advocate for their role in protecting the environment and to share information and experiences with different policy initiatives (Environmental Council of the States, 2008) There are marked advantages of having individual states innovate new function as innovators who are able to sweep away ineffective policies and create new framework s of regulation and production (Eisinger, 1988; Schumpeter, 1950) State government officials will be more aware of the specific needs of both their citizenry and in dustry (Berry & Berry, 1999; Gray & Lowery, 2000) Finding the appropriate level of government for regulation is a process of matching by size if it is a local issue, local governments s hould regulate, state level issues are regulated by the states (Butler & Macey, 1996)
27 ural resources. Iceland, with its volcanic terrain, uses a considerable amount of geothermal power for its needs. Some areas of the United States have relatively consistent winds and those states have built wind farms to harness this phenomenon. A theo retically efficient way to address the desire to produce more renewable energy is to focus on multiple state and local level solutions, based on the comparative advantages of localities, states and countries. This addresses issues of responsiveness to the environmental concerns, without attempting to prescribe single solutions for different regions. It is important to expect rent seeking behaviors within rational choice theory, where political advocacy groups, existing businesses, and other int erest groups attempt to skew the policy to favor one technology versus another, or one set of policy options against another. This may lead to poor societal outcomes, where local knowledge and understanding would make the system more productive and often less expensive than a centrally imposed solution. Examples of the potential rent seeking behavior are hydrogen fuel cells for automobiles or corn based ethanol regulations in the United States (Economist, 2013b, 2013c) Another advantage of innovating at the state level is that it has a lower risk by addressing what may be a high priority in one state, other states (with different prioritization) can learn what imitated or policies may be further enhanced with additional elements or to the policy Examples include welfare reform in Wisconsin (Mead, 2005) and emission
28 control in California (Kirchstetter, Singer, Harley, Kendall, & Traverse, 1999) where these policy innovations have been adopted regionally or nationally. Disadvantages to this devolved system are often related to resources or externalities. Smaller governments do not have the resources available at the federal level (Peterson, 1995) e conomic development (Berry & Berry, 1999) A lack of environmental regulation (which raises costs to producers) can be seen as a comparative advantage in drawing new n, with the effect of greater pollution (Butler & Macey, 1996; Peterson, 1995) Studies on Renewable Energy The research in environmental policy spans po litics, economics, public management, organizational behavior and sociology among others. While many of these research articles will be discussed, there is considerable overlap as the boundaries between the fields are often not clearly delineated. In pri or research, many frameworks have been explored for their explanatory power. These include technological innovation, adoption and diffusion, public choice modeling, institutional change and transaction cost theory. Method ologies for research have include cost benefit analysis, social cost analysis, game theory, endogenous growth and learning by doing. As substantial empirical data has been scarce prior to this time, this paper will take a fairly straightforward empirical approach of looking at production and consumption data vis vis the incentives and programs enacted over a period of time. As the innovation, adoption and diffusion literature suggests non economic variables can influence the
29 decision making process, demographic, group and political var iables will also be included in the review. At this point, it is important to note that much of the literature on environmental policy and energy policy has been in the form of case studies. While there are multiple comparative studies across countries, these are again in the form of limited case studies, individual cases, and attempts at extrapolating their efficacy are made. A seminal study that empirically lo oks at renewable energy a cross the fifty states is an article by Menz & Vachon (2006) As this dissertation has an empirical focus, that article will be discussed more than many of the other parts of the literature. This section will attempt to partition the theories and present the development from historic to current thought. It begin s with applications of rational choice theory, followed by classic al economic theory, tying in politic s to correct market failures. Classic al economic models set out multiple assumptions that are infrequently, if ever, seen in reality. This discussion is not designed merely to point out these flaws, but rather often designed to create more competition in a market, which should push it to more efficiency. Modern economics looks at the costs of changing the st atus quo, either with products or process changes. In economic terms, these are referred to as transaction costs and path dependency. Transaction costs are the costs of finding someone to trade with (search costs), ne gotiating an agreeable exchange (contracting costs), and ensuring that both parties live up to their obligations (enforcement costs) (North, 1990; Williamson,
30 1987) Path dependency is a separate issue dealing primarily with the increasing returns to scale as more people adopt a certain technology or process (Arthur, 1989; Pierson, 2000) P olitics is double edged when reviewing the effect of government policies on path dependency With enough political capital (e.g. through public initiat ives, responsiveness to popular beliefs), collective and authority driven innovatio ns can force industry to change, and send the market into new arenas, either lowering barriers for new innovators (Acemoglu & Cao, 2010) or to ward more socially desired outcomes (Berry & Berry, 1999) Turned on its head by industry political action committees, public choice models can be used to show the exact opposite -that government can actually contribute to path dependency as well (Gray & Lowery, 2000) Finally, the current literature o n renewable energy is surveyed. This is very much an emerging field, and there is still considerable debate regarding what and how it should be analyzed. Multiple surveys and case studies are used in limited areas, describing policies successes and failures both within the United States and internationally. The literature in Europe is more robust and re flects earlier adoption of many of the policy innovations in renewable energy. Little empirical work has been published regarding the efficacy of policies within the United States. This dissertation will contrib ute to the empirical literature and build o n established research, while redefining some of the key variables required to test the effectiveness of state level policies. Rational Choice Theories C lassic al economic assumptions include perfect information between the buyer and seller, free entry and exit from a market, as well as a large number of buyers and
31 sellers. As mentioned in the introduction, it is not likely that one will ever fin d perfectly competitive markets; foundational economic theories of market failures first focused on the monopoli stic tendencies of corporations. Another assumption that must be relaxed is the postulate that prices decline as economic profit opportunities are exploited and competition increases. An established counterfactual to competition in the market is capital This leads to the view that if the market doe s not provide alternatives to monopolies it is a market failure requiring government intervention. This is the basis of the establishment and regulation of natural monopolies. The microeconomic model of the market is based on rational choice individual consumers assess the utility they will get from consumption of one good versus another good and pick the one that will provide them the mos t utility. Corporations look at competing projects (or bundles of projects) and pick the one that is expected to give the highest return to their investment. With the adven t of bounded rationality (Simon, 1997, 1947) it is impossible to perfectly select the good s bundle that will return the highest amount of utility. Both corporations and consumers use an iterative process to lea rn and refine their expectations (Arrow, 1962; Arthur, 1994; Knight, 1921; Lindblom, 1959) Imperfect information and thus ac ting/forecasting under risk leads to the adoption of multiple criteria decision making models (Arthur, 1999; dePalma, Meyers, & Papageorgi ou, 1994; Holz, Kuczera, & Kalma, 2006) These models allow for different outcomes, based on the goals extending the f undamental microeconomic cost benefit analysis. In addition, a well discussed issue with cost benefit analysis is the externalization of costs from a producer or consumer to society in general. Finally, as
32 groups interact socially, different beliefs and desires begin to weight their analysis of decision making (Julnes, 2000; Yang & Callahan, 2007) While classic al economics provides a foundation for modeling actions in the marketplace, it is somewhat oversimplified. There are those who hold strong convictions that the market wil l come to an efficient point ; c ontrasting this, a considerabl e literature in modern economics (and even neo conservative economics) looks at the discrepancies between classic al economic theory and what is observed in reality. The proposed explanations of these discrepancies form the basis of modern economics. Moder n Economic M odels One of the primary changes in economics came when the focus changed from the long term to the short term. This short term view implies markets are constantly in a state of disequilibrium of information and pricing signals With imperfec t information, the market fails resulting in incomplete markets due to information asymmetries (Akerlof, 1970) Viewing the lack of choice in alternative energy as a market failure can be done through path dependency (Arthur, 1989; David, 1985; Pierson, 2000) and transactio n costs (North, 1990; Williamson, 1987) Applying a path dependency lens, the choice of what good to produce (or technology to use) begins with estimated future returns to that inves tment I f the returns (further innovations) or other market participants (competitors) then innovations are less likely to happen as the returns to the investment may not be fully captured by the creator (Young, 1928) If there are increasing returns to standardizing ( i.e. all selectors pick one technology), then the market will become path
33 dependent (Arthur, 1989) The rational choice of an organization is to keep its options open as regards t echnology as well as suppliers (Thompson, 1967) ; however, alternative options have not been exploited in the energy generation industry. While an begin to make decisions, that organization risks being left behind if it does not also act. The organization must balance the early adoption advantage against potentially being late to the market and finding itself locked out as a result of barriers to entry, or worse, betting on the wrong technology. The other side of t his market failure is using a transaction cost lens for cost benefit analysis A subset of transaction costs are s earch costs In environmental terms, an example is bas ic research what renewable good can be used for fuel? Once that is determined, cost benefit analysis requires an estimate of potential return These estimates are problematic: as questions about market adoption, amount of demand, as well as possible infrastructure development all remain to be answered Research shows that in this envir onment, companies will tend to invest in current research streams, not in radical innovation (Acemoglu, Aghion, Bursztyn, & Hemous, 2010; Acemoglu & Cao, 2010) I (via carbon taxes) an d providing research subsidies can jumpstart the innovation, while not harming the long term growth prospects (Acemoglu et al., 2010) When looking at alternative energy through the innovation, adoption and diffusion framework, one sees that the United States was an early adopter (Rogers 2003) for nuclear power and solar More recently, the European Union countries have continued to innovate while the US has decreased investment (on a percentage basis) in
34 renewable en ergy sources (Belkin, 2008) Due to the l ocal desire for green products and environmental awareness, the EU may be carrying a significant proportion of the research and development costs for renewable energy, showing the implementation of a consensus / collective adoption of green innovation. Th e value attached to protecting future generations, or the environment as a whole cannot be easily estimated, yet these are also important goals for many Americans (Center for Climate Change Communication, 2009, 2010) An alternative to the market based production of renewable energy supplies is that Nedergaard (2006) discusse d agricultural production and his m odel is easily transferred to the energy production market. When free entry and exit from a market are denied (for whatever reason, including high start up and shut down costs for a business), then producers receive a higher return to their innovation (Acemoglu & Cao, 2010) and will rationally use political power to encourage policies that are favorable to them. An example of twinning these lenses are renewable energy sources for automobiles While there are gasoline stati ons on every corner, early ado ption of alternative fuel vehicles has been limited to fleet s that have central refueling stations, which lowers complexity and raises compatibility for fueling (Granovskii, Dincer, & Rosen, 2006; National Energy Policy Group, 2001) T he risk and uncertainty of switching fuel technologies has been blamed for the failure of innovation adoption solely based on market incentives (Pierson, 2000) Rece nt policy that encour ages fuel cell technology reflects the power of established companies in protecting their position (van den Hoed & Vergragt, 2004)
35 Moran and Sherrington (2006) use a cost benefit analysis to evaluate a Scottish wind farm. The market costs identified include capital investment, operation and maintenance, infrastructure (electric grid) c osts, and rent on the land. Benefits include revenues, lower fuel costs, and additional stability within the GDP. Added to these traditional market forces are non market impacts: avoided carbon emissions, pollution farm. Willingness to pay and declines in land values are used as measures for the non market forces. The authors find that while there are definite costs to building the wind farm, the wind farm is net positiv e in social welfare; in addition, it is preferred option rather than building a fossil fuel fired power plant in the same location. A general equilibrium model is used by Schumacher and Sands (2006) to describe the growth in use and change in pricing for power created by a clean coal generation method, natural gas, carbon sequestration and wind power A carbon dioxi de emission 55 wa s required to make both carbon sequestration and wind power competitive in the open market. Kobos, Erickson and Drennen (2006) used development rates and progress rates, along with cost curves and installed capacity. As a technology is m ore developed, there are more adopters (higher progress) which indicates that the costs should fall as the installed capacity increases. The authors looked for found that domestic R&D in th e United States is insufficient to explain the drop in prices for wind power in the US; rather, collaboration between American manufacturers and foreign (primarily
36 European) producers allowed America to piggy back on European development and adoption. L und (2006) look ed at various energy source s and energy efficient products and their corresponding adoption rates. H e fitted comparisons of different technologies to an S shaped diffusion curve, which allows for exponential growth as the technology is initially adopted and then slowing growth as it matures. He fou nd that government mandates for technology adoption as well as financial s ubsidies impact the penetration rate of the different energy sources. Also, products that have competitive pricing or cost advantages (an example are compact fluorescent lights) do not require subsidies in order to be adopted. Lund did not address the need for education within the mar ket before these products will be adopted. An institutional framework is used by Breukers and Wolsink (2007) in three case studies of wind power adoption in En gland, the Netherlands and Nordr hein Westphalia (a state in Germany). They use d capacity building to look at the relationships between organizations, the policy domains in which they work and an advocacy coalition framework to describe the policy community of interested policy groups. They found that w ind power ha s three areas. While th ree different methods ( subsidies, public private partnership and grass roots actions ) have been used to achieve the goal of additional wind power, none of the three areas have worked at the local planning level to gain buy in for the programs. As a result, there contin ues to be local opposition to wind projects. These articles give a brief introduction to the economics of environmental policy. Even while taking a strictly economic approach, they control for political actions.
37 Involving the local citizenry can increas e their support and investment in the policy change (Busenberg, 2001) Moran and Sherrington (2006) show ed effective engagement and popular opin ion for a project and Breukers and Wolsink (Breukers & Wolsink, 2007) show ed the opposition. Before moving any further into the current literature, it is appropriate to review philosoph ies as a driving force of political and organizational behavior actions. Political Philosophy and Alternative E nergy Smith (1776) postulated that government interv ention in the economy, with few exceptions, interferes with the maximization of societal utility. From this beginning, the traditional role of government has been to provide public goods (national security, street signs) as well as correct other market fa ilures (externalities, information asymmetries, monopolies and incomplete markets). When politics and the market intersect, it is a advantage in the market via pol itical power or favorable policies. Recent models of government action discuss capture models for regulators, as well as framing a plethora of unintended consequences from the regulation of industry (Tietenberg, 1980) These have led to tradeoffs between regulation and more market based initiatives in an attempt to use the self maximizing behavi or of market organizations to govern their own actions (Bertram, 1992; Lewis, 1996) Also, efforts at contract enforcement, and repeated interactions between market participants, have made business more effective at self regulation through market mechanisms (Ostrom, 1990) Measuring political processes is often contingent upon the goal that is being pursued. If an agency is looking to show an effective use of taxpayer dollars, it will use a
38 cost benefit a nalysis between two options. However, as Pierson (2000) note d one of the characteristics of political life is th at it has multiple equilibria there are multiple goals that can be pursued. More appropriate measures for political goals are using the multiple criteria decision making processes however, these have an inherent underlying value system. Defining the the various political entities. The intersection of market and political forces can be seen within the discussion regarding renewable energy. It is important to note that the policy decis ions and most coalition groups do not reflect an absolutist principle as resolving political issues requires balancing cost benefit analysis with political or social requirements, regardless of cost effectiveness or risk of the danger. ition to CBA [cost benefit analysis] it therefore appears to fail as a comprehensive decision (Pearc e, 1998) This section looks at historical frames of reference for alternative energy development efforts, highlighting political and social pressure for change. Laird (2001) describe d the debate in the U.S. over solar versus nuclear energy sources in three parts: pre crisis, crisis, and post crisis, using the oil embargo of the 1970s as the crisis. The language of crisis was used as early as the Truman administration, as resource independence was a requirement for the goals of economic growth and national security. There was no sense of urgency as oil supplies seemed secure. Politically, development was primarily based around nuclear energy, due to its instituti onali zation within the military and the connections between the military and government development agencies. The oil shock of the 1970s changed priorities to stop
39 gap measures to prevent continued economic problems. Post shock, interest turned to finding add itional solutions for energy production. Oil use declined, and alternatives such as natural gas, nuclear, wind, and hydroelectric sources were used to meet energy needs. The discussion around solar showed that a punct uation (oil shock and environmental i sm ) was policy. The experience since the Brundtland Commission in 1987 is that incremental policy making is on going, yet has not achieved the stated goals of renewable energy sources. A signi what shocks are required to either (a) create the market incentives required for the market to adopt a new technology or (b) politically mandate adoption of renewable energy sources? Th is body of theoretical literature sets the foundation for policy analysis T he next section will dig into a number of recent articles that span case studies and global surveys of energy policy, primarily European in nature. As mentioned before, Europe vi ews government and the markets as collaborators more so than the United States, which tends to look more to the market for solutions. In addition, a political mandate that has forced reduction of carbon emissions (if not increases in renewable energies) i s the Kyoto Protocol. As most European nations signed and ratified the treaty, they are under an onus to develop new technologies to meet their energy needs. Policy Innovation and Adoption Alternative energy innovation and adoption has been attempted int ernationally, with different degrees of success. This is an important point if the ma rket adopts new technologies, there is no reason for government intervention. If the benefits of using a
40 new technology are not evident to the market or do not pass th e yardstick of a cost benefit analysis it will likely not adopt the new technology. If the new technology is considered socially desirable, even if not economically efficient, collective or authoritative action is required. tors] believe that classical economics offer some but not complete explanation s for development successes and failures. Most accept that non economic objectives like environmental sustainability, political participation, economic justice and social equit y are as worthy objectives as is (Klingner, 2009, p. 14) In framing the adoption process, it is impo rtant to look at the interactions of market and governmental forces. This section will look at a variety of market and government forces that worked for the adoption (or non adoption) of each of the technologies. previously institutionalized technologies is highlighted throughout muc h of this literature. Multiple factors are at work when an innovation is adoption, including government policies, market incentives, and the characteristics of the actors within the market (either providing or demanding the goods). When originally cons idered, the steam turbine was adopted over the gas turbine. However, gas turbine entrepreneurs were able to show profitable returns for adopting a hybrid system of gas and steam turbines. As an example of a market driven innovation, returns were expected to be positive for the gas turbine (in its niche markets), and it was adopted through market forces without political intervention (Islas, 19 97) This is (1989) prediction that non increasing returns to a technology allow for an entrepreneur to ex ploit an opportunity.
41 A hybrid example of the nexus of market and government policy is Danish wind power a s studied by Buen (2006) focusing on interactions between policy incentives, market interactions and innovation. Danish government tax credits and investment opportunities in the wind farms have created a greater sense of local participation and acceptance o f the projects. This has led to an increase in installed capacity and demand for wind power. In addition, when California tax policy created an incentive for wind farm investment, the Danish firms were able to scale up production to meet this new market. In the almost exclusively authority adopted realm is nuclear power. The massive costs r elative to the uncertainty of the market requires government gua rantees for plant construction. Historical case studies regarding t he adoption of nuclear power are split by popular opposition and consequent government commitment to the various program s (Economist, 2012) While initial cost benefit analysis of nuclear power showed it was cheaper than fossil fuels (especially during the oil crisis) (P. deLeon, 1979) incidents like Three Mile Island and Chernobyl polarized public opinion against widespread adoption. The issue of nuclear waste disposal adds to both the economic and political c osts of nuclear power (Roques, Nuttall, Newbery, deNeufville, & Connors, 2006) With this in mind, environmental groups used the courts to block development in Germany and the (Economist, 2006; Hatch, 1995) Current debates about nuclear power look at grea ter adoption throughout Europe as carbon permit costs Kyoto compliance, and other pricing pressures r aise the costs of carbon based power generation (Atkinson & Tietenberg, 1987; Roques et al., 2006;
42 Tietenberg, 1980) America remains formally unconstrained as it is not a signatory to the Kyoto treaty. Roques, et al (2006) use d a discounted cash flow analysis to revi ew the cost per unit of production (benefit) of vari ous types of power technologies, finding that without significant government subsidies, the market is unlikely to develop alternatives to standard fossil fuel power generation In addition, regulatory hu rdles are being eliminated in some areas (Mesler, 2001) and market forces are being relied on over command and control models for technology choi ce (Roques et al., 200 6) Despite these changes, technologically and economically feasible projects are not being funded. Faiman, Raviv and Rosenstreich (2007) show ed that at a 5% interest rate, the southwestern states of the United States could replace their current energy needs exclusively through the use o f solar power. In addition, these projects would be priced competitively wi th current power generation and would fund further growth through their profits, rather than requiring additional investment. Looking at mobile power generation, automotive techno logy is heavily dependent upon augmenting internal combustion engines with either electric motors (hybrids) or substituting crops derived fuel for refined oil. In order to meet the goals of greenhouse gas emission reduction, biodiesels are found to be cos t ineffective in reducing those emi ssions (Frondel & Peters, 2005) However, w hen looking at other externalized costs (health care, pollution), biodiesel is a valuable tool for use in non road diesel engines (W assell Jr & Dittmer, 2006) A technological substitute for the internal combustion engine is hydrogen fuel cells. These are for automotive use (van den Hoed & Vergragt, 2004) Despite this lock in, development
43 is talked about, rather than actually researched and implemented (Solomon & Banerjee, 2006) These examples show the interplay of market and policy forc es in explaining adoption or non adoption of technologies. While there are incentives for incremental development (niche profit potential, subsidies, tax credits), political and market thought typically will not change unless there is a significant breakp oint (eg oil crisis, environmental disaster) that leads to a policy mandate (Jacobsson & Lauber, 2006; Pierson, 2000; Schwartz, 2000) This section has provided examples of different weightings between market and government forces for implementing renewable energy policy. In the United States, the preference is for more m arket oriented programs, whereas in Europe a considerable amount of the pressure comes from the people and government. With this in mind, one turns to an examination of policy studies that have show n which polices are effective, as well as attempting to f ind common themes in those policies that failed. Policy Success and Failure The definition of policy success depends upon the underlying reasons why the policy was adopted. Since the 1970s (and more recently), reducing reliance on foreign energy sources has been a driving factor, and public benefit funds and other conservation efforts were the result. More recently, talk has shifted toward energy independence, and this goal is being met through increased natural gas reserves as well as increases in renew able energy sources for the long term. With natural gas, previously inaccessible fracking sent into the well to break up the rock and free up the gas. Recent projections show the
44 United States becoming a net energy exporter by 2020, while also increasing the use of renewable energy (U.S. Energy Information Administration, 2012) Goals can be positivist in nature, but also go beyond to post positive desires of the citizenry and philosoph ical desires that may be in conflict with strict market driven decisions. Lund (2007) look ed at 20 case studies of renewable energies and products to determine the change in installed capacity or absolute number of installations to estimate the efficiency of policy in driving these changes. In addition, he estimate d short term and long term costs of the technology, including subsidies until it reaches bearing h dependency literature. He foun support) are effective, catalytic measures were ude lower Lund attributed this difference t o stronger market sensitivity to subsidies, as well as the ability to target end users and active stakeholders. He also showed that the efficacy of the subsidy is tied to the size of the financi al support provided by the government. The Netherlands ha s been dynamic in their use of policy experiments to increase sustainable fuel use. Taxes, demand side subsidies, and supply side feed in tariffs have all been attempted in the last twenty years. Van Rooijen and van Wees (2006) and Agnolucci (2007) conclude d that a changing investment environment was detrimental to installation and growth of renewables. In addition, they recommend ed using specific targets and goals to signa l governm ent commitment to the programs and creating mandates when voluntary actions are insufficient to generate sufficient action by market
45 organizations. The primary issue was not who was targeted, but rather a perceived lack of credibility in meeting Dutch renewable energy targets. Harmeli nk, Voogt and Cremer (2006) extend ed this style of analysis to the entire European Union. They consider ed direct policy instruments, such as investment support, feed in tariffs, and mandates. The ir overall success measure is attainment of European Union commitment to a 2 2% share of electricity coming from renewable energy sources. Growth in energy sources under current policies are estimated and compared to the putativ assumptions, only 6 of the nations will meet their goals, w ith 3 others coming close. They concluded that the current mix of government and market forces were not going to meet the goal, and call ed for additional government intervention. An in Jacobs s on and Lauber (2006) In this primarily descriptive res earch, the authors identify the need for entry and exit of new competitors to extend the knowledge base in an industry, the importance of advocacy coalitions and breakpoints in practice (e g the Chernobyl disaster) to determine the success of government p olicies. While government ongoing success is not simply a function of additional government intervention, but also exogenous factors such as equilibrium punctuations. A bala nce of government and market forces is important for the success of g overnment policy. Buen (2006) survey ed N orway, which was unsuccessful in creating its wind industry as it had a surplus of hydroelectric pow er, and the government mandated windfarm technology was designed for a scale incongruent with the conditions
46 in Norway. Sweden also mandated a single type of blade technology for their windfarms. This type of blade was subject to competition from other manufacturers and found to be less efficient. As a result, while Sweden has increased wind use, their adoption has not been as quick as that by either the U nited States or Germany (Astrand & Neij, 2006) These articles indicate that an ap propriate balance of influence is to u se government resources to overcome market barriers and create local support, while allowing innovation and market forces to drive the actual product development and implementation. These macroeconomic views shift naturally to lookin g at the specific poli cy tools for the next step of this analysis. Policy Tools Taxes United States Energy policy has moved through different stages in the 1970s, 80s and 1990s. Policies in the 1970s included development of alternative fuels (primarily nuclear), minor develo pment of renewable energy (primarily solar), with a greater focus on conservation. Much of the legislation enacted during the 1970s authorizing tax credits was allowed to expire in the 1980s, primarily due to conservative political beliefs that energy pol icy should be set by the market, not by the government (Carpenter & Chester, 1984) In the 1990s, a steady supply of oil out of the Arab Gulf region was again threatened and energy policy began to look at incentives for renewable fuel source development. Recently, biofuels and other ethanol based solutions have attracted the attention of the f ederal government. Carpenter and Chester (1984) surveyed over 13,000 hous eholds in ten western states regarding awareness and use of the solar energy improvement tax credit enacted in the 1970s. They control for urban/rural location, income, education, marital status, age of
47 respondent, age and type of house as well as number of days a house required heating. Many respondents self report that they would have made conservation improvements regardless of the tax credit. However, the regression estimates indicate that the federal tax credit did, in fact, provide an incentive to install solar panels or solar water heaters. Those most likely to make improvements are married, have college degrees, higher incomes and live in older houses. A model to predict installation of solar panels is proposed by Durham, Colby and Longstreth (1988) Surveys from 800 respondents provi ded data on household size, education, income, perception of the seriousness of the energy crisis as well as if the household had installed a solar hot water heater. This was combined with the size of the state tax credit, cost of energy and amount of sol ar energy available in the state. The state tax credit and energy prices were found to be significant; the level of education and number of household residents were less significant (7% and 8% level s of significance respectively). The econometric success however, did not predict the actual results the model predicted 5 installations, while there were actually 47 in the survey. The authors attribute d this to incomplete knowledge of energy costs and usage, the year of installation and what the househol d s. Walsh (1989) used a two period economic model to analyze the effect that a tax credit has on residential conser vation efforts. The enacted policies included a federal tax credit, five state level tax credits, and four states which allowed for a tax deduction for conservation efforts. His mode l estimated changes in utility between a period prior to conservation im provement and a period after the improvement. He control led for income, perceived cost savings from the tax credit, and a preference for immediate
48 use/consumption of income (a discount rate). He f ou nd that the tax credits are not effective policy measure s. Proposed reasons for this are that the savings are too little, the credit is too difficult to claim, lack of information or understanding regarding the tax credit. The effect of fuel taxe s is measured via long term elasticities, as well as reduction i n carbon dioxide emissions in a recent study by Thomas Sterner (2007) In his comparison of the United States and low fuel tax countries relative to high fuel tax countries (Europe and others), he show ed that a raising of tax levels in the United States to those in Europe would cut carbon emissions by half. This is not seen as a replacement of the permit system under the Kyoto Protocol, b ut rather as a supplement that will work in the long run to reduce significant amounts of carbon pollution. The political viability of raising taxes to this level is not discussed. Johnson (2006a) compare d Assembly Bill 1493 to a feebate 1 system. The mandates and permit system were shown to minimize cost (which is economically effic ient) while the feebate system wa s seen to minimize emissions (which is assumed to be socially preferred). He later (2006b) discusse d the efficiency of the Swedish nitrogen dioxide feebate program. Keys to the success of feebates are accurate categorization of the polluting vehicles and setting appropriate levels of These issues of setting proper levels is seen in the current pricing for carbon credits, which have dropped in valu e from over $25 to a recent low of $7 (Economist, 2013a) 1 A feebate is a fee / rebate system. In this case, it is a revenue neutral emissions tax levied on all taxpayers in California. Those that then purchase / drive c ars which have lower emissions will receive a rebate, which is funded by those whose vehicles have emissions above an accepted level.
49 A case study of three price support mechanisms in Ireland is analyzed using an EU designed environmental model ing p rogram by Huber, et al. (2006) In it, quota obligations, feed in tariffs and competitive tender schemes we re evaluated as ways of generating renewable energy source derived power. The feed in tariffs and co mpetitive ten der schemes hav e comparative advantages to each other and absolute advantages over quote obligations. The authors note d superior knowledge and an ability to change the design after the policy is legislated) can be a significant hurdle in the administration of the policy. Gouchoe, Everette and Haynes (2002) surveyed six states and examined their financial incentive programs for renewables. They offered eight rec ommendations, including flexible programs, provide education, incorporate cross state cooperation, have generous incentives and make the application process thorough, but convenient. Mayes, Gielecki and Poling (2005) analyze d non hydro renewable energy policies in four co untries and California. They fou nd that political commitment, the structure of public policies and the ability of public opponents to delay and/or derail renewable projects all generati on. A natural transition from these large scale or macroeconomic studies is to look at the specific tools that have been used, and to see their efficacy in creating the desired outcomes. The section that follows looks at the somewhat disconnected litera ture that looks at policy levers from a variety of perspectives.
50 Policy S pecific R esearch Public Benefit Funds (PBFs) are government funds used to promote different ener gy policies. Some are used to lessen the cost of energy use for low income families, subsidize purchase s of energy efficient products or to lower the price of renewable energy. According to Nadel and Kushler (2000) there are no set standards for establishing a public benefit fund, nor to the use of the money in the fund. Some attempt to reduce consumption, others stimulate demand for renewable energies, in both cases, these are described as demand side management techniques. Their descriptive work indicates that PBFs can have a positive effect on lowering peak energy demand, as well as increasing sustainable energy use. While many PBFs target solar energy, Forsyth, Tombari an d Nelson (2006) analyze d the cost kilowatt capacity) in a distributed setting. They find that small wind wa s more cost effective than photovoltaic energy (which receives a significa nt amount of the subsidies for distributed energy generation). Bolinger and Wiser (2006) look ed at influe nces in centralized renewable energy generation. Their case study of projects funded by the Clean Energy States Alliance f ou nd that state level public benefit funds expenditures are substantial and directed primarily at wind power (60% of funds). The amo unt of power generation capacity is still growing, although at a slower rate than before. Different incentives we re being created and often targeted at power production, rather than facility construction. This wa nto federal funds, and will also create an incentive for maximizing renewable power generation.
51 Moving from power generation to conservation models, Loughran and Kulick (2004) model ed public benefit fund efforts at conservation and total energy use reduction Unsurprisingly, policy attempts to get people to conserve energy are correlated w ith a decline in overall energy use. One concluding comment from the authors is of particular he energy efficiency investments are also those most likely to take advantage of the policy incentives (such as rebates and subsidies). The authors do not make any suggestions on how to more ef fectively target these programs, nor comment on the causality of the investments. A meta analysis of conservation programs was completed by Gillingham Newell and Palmer (2006) Their goal was to determine the cost effectiveness of the programs. These conservation efforts are found to be cost effective, even without counting reductions in greenhouse gases and other air pollu tants. While these programs contribute d to the reduction in energy use, they are only a single element of the solution to energy security problems. term. PBFs attempt to m eet various goals of education, conservation and use of renewable energy. The focused case studies have shown the potential of PBFs to be effective in meeting their stated goals, however, they do not create a causal relationship between PBFs and the actio ns taken by consumers. Some studies have actually shown that those who use renewable en ergy or install solar panels were likely to use green power regardless of policy incentives.
52 While specific policy options have been reviewed above, it is important to remember that these policies do not exist in a vacuum. There are interaction effects between the different policies, and they can build on each other. Lindn, Carlsson Kanyama & Eriksson (2006) studied 600 Swedish households and f ou nd that while tax credits and rebates are effect ive policy tools, it is also important to focus on educational and informational campaigns to change consumer behavior. Menz and Vachon (2006) use d a least squares regression to evaluate multiple state level energy policies in relationship to wind power generation Using data drawn from the US Department of Ene rgy, they attempt ed to cor relate wind power and multiple concurrent policies Menz and Vachon open ed the door to an interesting field of study. Their dependent variable s were a four part measure for wind power capacity: the installed wind capacity at the end of 2003, the absolute growth in capacity since 2000, the absolute growth in capacity since 1998 and the number of large wind energy projects (those larger than 25 megawatts). Their independent variables include d five policy options: renewable portfolio standards (mandated requirements (so consumers know where their energy comes from non renewable or renewable sources), mandatory green power offerings, public benefit funds and retail choice of electric generation These variables were either binary (0/1) or were measured by the number of years the policy had been in place. These measures allowed for existence of the projects, as well as capturing time trends that may come from longer standing policies.
53 They fin d that mandatory green power option was the sole significant explanatory variable in all of their regressions. Their dependent variables measured first the production capacity (i e a snapshot of how many megawatts of wind energy utilities could produce i n 2003) and secondly two measures of growth in capacity of wind power. Renewable portfolio standards were significant in the capacity and long term growth regressions. In addition, they f ou nd that retail choice in selecting a power company ha d a negative effect on the adoption of wind power. Disclosure rules and benefit funds were found to be not significant in all of their regressions. binary measurement of the independent variables is parsimonious and effective for three of their vari ables (generation disclosure requirements, mandatory green offering and retail choice). However, both the renewable portfolio standards and public benefit funds are too narrowly defined. Portfolio standards typically mandate a percentage amount of energ y consumed to be derived from renewable energy sources. R ather than simply indicating that RPS exist in a state, a more effective measure for RPS is the required ratio between the renewable and non renewable energy sources. If the RPS sets the standard a t 10 per cent, then the influence of concurrent policies will be captured in an increase above the 10:90 ratio mandated by the RPS. While the RPS may been enacted at the time of the study, the requirements for renewable power were not yet in force. The benefit funds offer rebates and incentives to consumers of energy, as well as subsidizing the construction of some renewable energy plants. The amount of authorized financial reso the subscription rate of the fund (ie, how much money is actually
54 being disbursed to consumers) or PBF expenditures are all better measure s of its effectiveness than a binary acknowledgement of existence or age of the program. A final no wind renewable resources (often, photovoltaic / solar). As the dependent variable was only wind power, it is unsurprising that PBFs were found to be not significant; however, this m ay not be a result generalizable to all renewable energy. Many of t he limitations that surrounding their effort s are still extant, and these In limiting the dependent variable to wind power, their study ignores the potential for other renewable energies. Solomon and Banerjee (2006) sh ow ed that nations typically adopt the renewable energies that are in the greatest abundance for their country. Buen (2006) and Astrand & Neij (2006) attribute d government policy failures on attempts to force specif ic technologies / adoption of non plentiful renewable resource on Norway and Sweden respectively. It is not a great extension of this literature to assume that American states will have varying amounts of renewable resources to exploit. Each state, in ec onomically rational terms, will use the most abundant renewable resource. Limiting the dependent variable to wind power ignores this basic tenet of economic decision making. The negative sign on the retail choice variable contradicts findings that consum ers are willing to pay more for energy produced from different renewable sources as compared to fossil fuels (Borchers, Duke, & Parsons, 2007) The Menz and Vachon paper is a foundational empirical work in the impact of various policies on renewable energy production. By expanding the operational definition of some of their variables, it is possible to have a greater data set, as well as
55 more generalizable r esults. This paper will build upon their research to continue to push the frontiers of our knowledge with regards to policy and renewable energy production. Literature Review Concl usion This review briefly steps through the theoretical and applied pieces of the policy generation, innovation, adoption and implementation. The primary framework used within the market is rational choice theory weighing costs and benefits, adopting th ose innovations that bring a benefit to the actor. As information moves from one person to another and group to group, positions change and other factors begin to be considered. This leads to an innovation adoption and diffusion model that subsumes the b asic rational choice model into a more complex decision matrix. The innovation model looks at the political and group considerations for innovation adoption, adding more layers to the decision making process. The individual actor (person or organization) may adopt an innovation, or when the collective will is strong enough, a consensus can pressure for adoption informally or formally (through authority adoption measures). These innovations can take the form of policies focusing on market incentives or gov ernment rules and regulations, nudging the market actors to a certain set of desired actions. The articles above provide a wide ranging look at the literature in the renewable energy field. Common trends are that most country level case studies and poli cy studies are primar il y descriptive a nd often attempt to associate interest groups with policy development and implementation. Research on the policy market nexus has typically used quantitative models to descri be price and adoption changes.
56 Many of t he published studies in this field have been qualitative studies and Renewable energy research is very much an emerging field, and it has benefited from the policy inno vations at the state level. These innovations have opened the door for a considerable amount of variation between the states, which makes it a field ripe for empirical study. The descriptive nature of many of the studies leads to a selection bias, as t hose projects that are notable successes or failures are studied for best practices and policy learning. With the maturing of these polic ies and experience with renewable energy, it is time to move beyond the case study and engage in empirical studies of the efficacy of these policies. By l ooking at the policies across the states, and correlating the change in all renewable energy sources to government policy this paper will create the first in depth examination of the wide ranging impacts of the se diffe rent policies. This dissertation contributes to current research by examining the role of various government policies in the growth of renewable energy sources at the state level. Are some incentive policies effective in growing renewable energy capacity, while others are not? What demographic elements influence adoption of policies? Does Democratic control of the statehouse lead to more rules and regulations in the environmental realm? These questions are interrelated and provide further information to guide policy makers in adopting incentive schemes.
57 CHAPTER III M ETHODOLOGY The literature review listed many different inputs and considerations that go into the policy process. This paper test s some of these, specifically the external input influenc e of neighboring states, as well as all of internal inputs (with the exclusion of the legislative staff variable). Citizen preferences and political philosophy are subsumed in two different variables, partially under citizen initiative process as well as the election of a Democratic legislature. High level policy design characteristics become subgroups of the policies (financial or regulatory policies). Success of the policy is defined as increasing generation of renewable energy for many policies, or re duction in total energy (for demand side management policies). This dissertation will further the literature by expanding the types of policies studied, the mechanisms or policy levers they use, as well as being one of the first to use panel data to look at changes in generation after policy adoption In addition, it will add renewable energy policy as a class to a burgeoning literature around the various determinants of policy adoption across the states. As a feedback mechanism, it will answer a narrow question of policy success (as defined above) and if there are some policies that could be enacted to meet those goals. It is among the first to take a wide reaching empirical approach, complementing the descriptive nature of the Nadel and Kushler (2000) article. By looking at all renewable energy sources it will address a limitation of Menz and Vachon (2006) as t hey focus ed exclusively on wind power It is important to note th at legislatures and constituencies may have other goals in mind for their policies that do not simply revolve around increasing the proportion of renewable
58 energy or decreasing total energy consumption Those goals and policies can be considered in future research and are outside the scope of this venture. The primary research area a What drives the adoption of renewable energy policies The research questions to be addressed are: 1. Are financial incentive policies effective at stimulating dema nd for renewable energy ? 2. Do government mandates or market forces increase production of renewable energy? 3. Do the demographics of a state help deter mine demand for renewable energy ? 4. What demographic indicators influence the adoption of renewable energy p olicy? The goal of these research questions is to identify why policies have been adopted. At the most basic analysis, if a financial incentive policy is found to increase RCT model. Success for a government mandate policy would be a consistent and replicable increase in renewable energy generation, across multiple states which adopted at different times. A market policy success would be one where either the governme nt has created enough demand (i.e. completed an incomplete market) or market demand was sufficient to attract new entrants and create the market (e.g. the solar panel market is now rebates and feed in tariffs). The second pair of questions dig further into the theory to ask what is dr iving policy adoptions. Rather than measures of success, these are a first cut at understanding the characteristics that drive some states to adop t renewable energy policies, while others do not.
59 In addition, t here are hypotheses that can not be measured with the current data and time in force of the policies. The first of these revolves around the arguments for efficiency of scale effects if eno ugh producers are required to generate renewable in the literature review. However, this can not be measured until there are a number of producers that are generating re newable energy. A related hypothesis is that of where government mandates may initially raise energy prices as a result of new industries / technologies being utilized and then as suppliers enter the market and more energy is produ ced, market forces will lower that price. T esting t his theory runs into issues with a lack of pricing information, as well as the fact that many utilities are viewed as natural monopolies and subject to review by Public Utility Commissions which may opera te with different policies. The research questions will be investigated in two ways: ordinary least squares (OLS) will be used to measure the effect of policies on the generation of renewable energy (the first three questions ). The final question ( adoptio n of various policies ) will b e tested using the tool of event history analysis (Berry & Berry, 1999) The OLS fixed effects model uses a dependent variable of the net generation of renewable energy as reported to the US Energy Information Administra tion 2 Independent variables are broken down by demographics, state characteristics and policies. Demographic variables included were population, income and unemployment rate. State characteristics were binary variables : decoupling of power generation f rom 2 Efforts were also made to use the year to year change in renewable energy production as well as the percentage of renewa ble energy to conventional energy sources, but these regressions had little to no predictive power, mostly due to a lack of variation in the dependent variable.
60 utility profits, if the state legislative houses were both controlled by the Democratic Party in each state were drawn from the Database for State Incentives for Renewable Energ y. Year and state variables were included to capture fixed effects occurring in specific years or within the individual states. Within the event history analysis, the pol icies themselves are the depende nt variable. The independent variables are the popu lation, income, unemployment rate, adoption of a similar policy in a neighboring state and a year variab le. To control for linear increases in knowledge or likelihood of adoption, a trend dummy variable is also included in the regression (Buckley & Westerland, 2004) In addition, clustering models were run on the states and adopted policies, independent of geographic proximity (Shipan and Volden, 2012). Table 1: Regressions used in this dissertation Dependent Variable Independent Variables OLS model (Research questions 1 3) Generation of Renewable Energy, by state State, demographics, characteristics, policies (and reduced models limiting which of these variables) Event History Analysis (Probit models research question 4) Adoption of Individual Policies State characteristics, adoption of policy in neighb oring states These regressions build the foundation for further exploration with the goal of creat ing more effective policies. Ramifications include the ability to target expenditures to those policies that are shown to have a significant effect, as thi s research will provide
61 the first empirical backing for what have been qualitatively researched, anecdotally successful programs. OLS Models The dependent variable for the policy regressions is the net generation of energy (in megawatt hours) as reporte d to the US Department of Energy Information Agency The DOE provides some information regarding the use of renewable energy (hydroelectric and biomass) going back to 1949. By 1982 all major sources of renewable energy (hydro, biomass, geothermal, solar and wind power) were widely available and producing a substantial amount of energy (defined by the DOE as at least one half a trillion British Thermal Units (BTUs), a standard measure of energy). The policy information was collected from the Database of St ate Incentives for Renewables & Efficiency (DSIRE ) DSIRE is a public pri vate partnership between North Carolina State University and the Department of Energy which tracks the resources and goals of state and local financial incentives and regulations for renewable energy. T he Initiative and Referendum Institute at the Univers ity of Southern California tracks states that have initiatives for statutes or constitutional amendments ; referenda (to remove a law) were ignored for the purposes of this study. The state level data for decoupling utility profit from the generation of en ergy was derived from the Center for Climate and data for population, unemployment, average wage and control of legislature variables. All data is publicly available via the internet or direct ly from DSIRE. This analysis relies on panel data from the 50 states, with policy information available from 2001 through 2009, and energy generation data available over the same
62 period of time. State and year fixed effects are inc luded in the model to control for characteristics endemic to either the year or the specific state being studied. A reduced form equation for this is: The initial dependent variable is an integer, as described above net renewable energy produced in megawatt hours. Subsequent analysis showed that more information was available by adding regressions that had dependent variables of a percentage of renewable to total energy and the total amount of solar and wind energy. The policy va riable is a binary variable, 1 if the policy was in force in that year, a 0 otherwise. The breakdown of the incentives versus regulation policies are listed in the appendix. The State variable is a dummy variable to capture state and local level characte ristics. The demographic variables are a mix of integers and binary variables. The population number is the annual estimate; unemployment and income are the simple average of the quarterly estimates from the US Census Bureau. Democratic Legislature is a binary variable where 1 indicates that Democrats controlled both houses of the legisla ture, 0 otherwise. The citizen s initiative variable was set to 1 if either the ability to add a referendum item or to directly pass a law via ballot was present in t he state. Decoupling profits was noted as a 1 if the utility was allowed to decouple profits from either electric or gas generation ( decoupling of the individual energy source was also tested separately and found to have no effect). The state control v ariable will function to capture any state to state variation. In econometric terms, these controls will be used to capture any state level omitted variable bias. The omitted variable bias is the effect that other variables, which are not included
63 in the model, could have on the outcome. Another way of saying this is that there may simply be differences between Californians, Kansans and Virginians when it comes to use of renewable energy. Rather than attempting to control for differences between these p opulations as populations, the state binary variable will attempt to identify that those differences exist and their impact, without explaining what the differences actually are. A year fixed effect variable is included in the model, and is designed to c apture any year specific changes in renewable energy generation. These changes would typically be a discrete event, such as an extra hot summer (requiring higher energy generation) or perhaps a focusing event (q.v. multiple streams and po licy window polit ical theory). When looking at the fixed effects, it is important to note that the effect of being in Oregon are assumed to be consistent across time, and to hold true for the average Oregonian. The year fixed effect is similar it is different each yea r, but consistent across the United States. A noteworthy item was found when using the fixed effect models. Many environmental issues cross over man made geographic borders and weather systems can be multiple years in duration. This became apparent with a multi year drought affecting multiple states in the Tennessee Valley Authority region. The year fixed effects would also not capture the effect (it is not consis tent through time.) The option of interacting year and state would not work, as that would take all of the degrees of freedom out of the model. The policy matrix controls for other policies that may be enacted in a state. The number of state level polici
64 393 policies, categorized into 33 separate types h ave grown to 1142 policies in 40 different categories in 2009. A trend in the political environment has been to use market forces to change behav ior through a series of incentives and coordination with the utilities. As a result, many of the new policies in the past 5 years are classified as solar panels portfolio standard mandates for green energy generation. The list of policies is quit e extensive and, as a result, clustering models were employed to see if the number of policies (or the states where they were enacted) could be grouped to lower the number of variables. The clustering used was run through two models, the first optimized t o two clusters, which split by policy adoption rates (i.e. spilt between policies that were frequently adopted, and those that were not.) The second model was forced to split into 4 clusters, this resulted in clusters split again by rate of adoption, not by similarity of policy type. These four clusters were the ext remes (frequent and no adoption) and then two smaller clusters of lower adoption rates. Event History Analysis The final research question will be addressed using a probability model, with the policies themselves taking on the role of the dependent variable. As they are a binomial, probability models lend themselves well to the discrete change from when a policy is not enacted to when it is enacted. The variable detailing adoption in nearby s tates is a percentage of the number of states bordering the state in question that have adopted the policy, divided by the total number of bordering states.
65 While the practice of clustering by neighboring states has been expanded to non geographic, as recommended from Shipan and Volden 2012); when policies were clustered in this analysis, it did not contribute to the explanatory power. The nature of environmental issues crossing state lines, but not having as much influence across the coun try, does lend itself toward a neighbor based clustering. In addition, neighboring states have an incentive to cooperate, especially when reaching goals such as emission reduction, as the effects are additive across the state borders for cleaner regional air (Carley, 2011). Finally, when reviewing the descriptive statistics around the 40 policy categories in this st udy, it was found that almost half were adopted in less than 10 states, and another 5 were fairly stable in their rate of adoption (i.e. they d id not increase/decrease in count by 10% over the 9 year study period). One policy appears only for one year, and is something of an anomaly due to program funding While these policies will be included in the regressions, the event history analysis will mostly focus on the 15 policies that had high rates of adoption, as well as changes (mostly increases) in the number of states that adopted the policies. Hypotheses and Predictions With a large number of variabl es ( 50 state variables, multiple policy and demographic variables ), very high R 2 and adjusted R 2 value s were expected and found Quite a bit of variation was absorbed with this many variables so it may bias the ability of individual policies to stand out as significant. As a result of the large n umber of variables, a step wise analysis was conducted to see if a reduced model may be more effective. The analysis began with year and state variables, demographics, and then
66 grouped the types of policies, as well as splitting the different groups to se e their individual predictive power. Starting with the reduced models may introduce an omitted variable bias; however, there is a logical basis for starting with smaller models and moving toward the expanded regression. The hypotheses have been written as general questions here, and will be written in the discussion section (Section V). For now, the paper looks at trends and v ery broad strok es of a brush. The discussion section will further draw out consistencies in the results, as well as opportunities where the data may not be convincing, either in conflict with traditional theory or inconsistencies across other regressions or when new variables are introduced. The different models use d to tests these hypothes es started with all of the variables and then were pared down after results initial analysis of results. The full model includes all of the potential variables dependent variable of total renewable energy generated in a state, all energy policies, state s, demographics and characteristics. The state model eliminates policies, characteristics and demographics. Likewise, policies models only look at policies, exclusive of the other potential independent variables.
67 Table 2: Summary of models, with Depen dent Variable and Independent Variable Independent Variables Model Name Dependent Variable States Policies Demographics Decouple Profits Year Full Model Total Renewable Energy Limited State Total Renewable Energy Limited Decoupl ing Total Renewable Energy Limited Demographics Total Renewable Energy Full Solar Total Solar Energy Full Wind Total Wind Energy Full Solar + Wind Total Solar and Wind Energy Limited Solar T otal Solar Energy Limited Wind Total Wind Energy Limited Solar + Wind Total Solar and Wind Energy The mo dels will be used as follows to answer the research questions : Q 1: Are f inancial policies effective at stimulating production of renewable energy ? Financial policies can be enacted that drive the marginal cost of renewably generated electricity below the willingness to pay from the citizens. In a purely thin RCT mode l, that willingness to pay would be the marginal cost of non renewably generated energy. With a thick RCT model, the willingness to pay may be higher for renewable energy than non renewable If costs are reduced for renewable energy, will more people purchase renewable energy? Both the full and limited models were used to test this hypothesis. Significant positive coefficients on the financial polices (vis vis other policy types, as defin ed by DSIRE) will indicate that the policies are successful in stimulating production Failure to find these results may be indicative of either a lack of effect or a lack of deterministic data. Q 2 : Are g overnment mandates effective at increasing produc tion of renewable energy?
68 the market response is sufficient to lower costs to a point where consumers gravitate toward purchasing renewable energy, it would validate the thin version. Arguments can government level otherwise the policies / mandates would not have been enacted. This paper will not split that hair. T he full model will be used to eva luate this, focusing on regulations such as mandates, licensing and certification. T here are specific polic ies such as renewable portfolio standards and set asides which are of special interest here These policies will be broken out for specific review, as concurrent policy goals of RPS are to increase renewable energy use and technological stimulus (push) to e ither develop new technologies / transfer in from other markets, making renewables more cost effective. A significant coefficient on the RPS poli cy variable is necessary and sufficient only for the first goal. The second goal must be measured by an increase in the percentage of renewable energy or an increase above and beyond t he mandated increase in the RPS, which can not be tested at this time a s many of the RPS have production deadlines set out into the future. Q 3 : Will u se of market forces increase the generation of renewable energy? Again, this is a mixed theory question levers to increase the production of r DSIRE splits policies into fina ncial and regulatory categories; however some of these are specifically around market choice. An example of such a policy is the mandatory green power choice. After full model regressions are run, an initial review of the results will
69 show if any of the financial policies are significant and how they function in choice models Earlier research has shown that the green power choice model to be significant if people are w illing to pay more for renewable energy, this lends credence to a thick RCT view of green energy. Rejection of this hypothesis may also be indicative of a lack of data over time, education, or other information failure. This will be more fully explored i n the Discussion section. It is important to note that many regulatory policies are in effect to ensure that the financial policies can be successful (e.g. net metering as well as solar and wind access policies ensure that distributed generation incentiv es can be implemented.) The potential interaction effect of the policies is noted, but not tested in these regressions. Q 4 : Do s tate populations have different levels of receptiveness to ward the use and generation of renewable energy? This question fa lm. There are complications with causality with this natural resource endowment, relative pricing of fossil fuel to renewable energies contribute to the choice of renewable energy; however it is firmly at end of the scale. This question is investigated in a fairly straightforward manner using a fixed effects model. State level dummy variables are used as a proxy to measure willingness to use renewable energy. Underlying this model is the assumption that renewable energy options are available to consumers and that the consumers are aware of their options in energy use. Presence of policies that contribute to education or choice mandates may be highly correlated with statistically significant state level variables. Finding many policies that significantly lower the cost of renewable energies
70 Q 5: Do s tate level demographics variables such as average wage and unemployment aff ect renewable energy generation? As noted in the literature review, green energy is often viewed as a luxury good. In a thin RCT viewpoint, if the economy declines relative to the size of population, it would make sense to see a corresponding decrease in the use of renewable energy by th e population. However, if greater forces are at work (leading to a thick RCT view), then it may make sense to see the demographics have no effect on the renewable energy measures. Population is included in this, and is assumed to be both significant and positively correlated to use of energy the more people in the state, the more energy it is going to need. Q 6: Will d iffere nces in state level demographic variables impact adoption of state level renewable energy policies? Changing from RCT theory, th is is in the realm of the innovation adoption and diffusion model. To review this, the individual policies become the dependent variable, and the independent variables are characteristic s across states that did / did not adopt the policies. With over 30 policies to review, there will obviously be a need to group the policies that did have significant input from demographics, as well as those that did not. Again, this is more of a research question that has been applied to each of the policies, not a sing le view on all demographics and all policies. Based on the effectiveness of the mod el as currently described, additional regressions could be run in future research in force, as well as the lead time for renewable energ y generation construction may have an effect on the policy efficacy If required, time series analysis can be used to lag the effect of the policies. However, it is expected that there will not be sufficient data to do a robust time
71 series analysis. An alternative to this is assigning values corresponding to years in effect for the policies (i.e. a policy that is 5 years old would have a value of 5), however, there is insufficient information regarding time in force for many of these policies to give acc urate valuations for all of the data. The government policies are an attempt to change the behavior of the marginal consumer. While many people will simply not change their pattern of consumption, there are those who are indifferent to green versus conv entional power and may purchase green power if given an incentive to do so. This is supported in the literature (Carpenter & Chester, 1984; Darby, 2006) which points out that a policy may not change the minds of those set against renewable energy, however, it may raise awareness an d acceptanc e among others P ublic benefit funds may well serve as a mechanism to educate and convince the marginal decision maker to use renewable energy rather than non renewable energy. In this same vein, renewable portf olio standards may occur more fre quently in more The alternative perspective (i.e. the states with the highest level of education would have RPS) is theoretically contraindicated in that the mandate would not be required, as the educated population will have already internalized the costs of traditional energy generation and collectively selected renewables for their energy supply At this stage, the re are less marginal decision makers, which would make PBFs less effective. A separate research topic within this thread would be adoption of RPS correlated to different measures of environmentalism in the states. Other ideas about other research topics such as these
72 (which are intentionally not covered in this dissertation ) are available in section VI Further Research. The significance (or lack thereof) of individual policies in these regressions will help to inform future policy decisions. At the curr ent stage of renewable energy policy development, there are a considerable number of policy innovations going on. Not all of these innovations will be successful. Separating the wheat from the chaff will allow future polic ies to be more focused on what h as been found to empirically work. The variables included in this research provide a sweeping look at market and command and control policies. Early development of alternat ive fuels (such as nuclear) was heavily controlled by the government. This paper will allow for additional insight into the centralized versus decentralized policy debate.
73 CHAPTER IV ANALYSIS The various questions asked earlier in this dissertat ion require multiple approaches in an attempt to answer them. An initial group of ordinar y least square regressions were run to look at the effect on generation of renewable energy, starting with the full model and then looking at constituent parts states, demographics policies and a combination of policies and demographics. Probabilistic event history analysis was completed adoption as the dependent variable to look at other predictors of policy adoption. Event History Analysis provides a look at the common characteristics and events that led up to an event or t ransition in this case, adoption of a policy. This section will draw general conclusions from looking at the two sets of regressions (OLS and EHA) looked at together, and then will dive into each of the regressions separately in an effort to answer the hypotheses and summarize any inconsistencies between the regression results along with potential reasons for these inconsistencies. When the regressions were run, initial results pointed to almost all of the variance being absorbed by the state level fix ed effects. In addition, there were no noticeable changes in megawatt hours generated from year to year. Delving deeper into the data, the largest amount of renewable energy comes from hydroelectric sources (551mwh in 2000 and 546mwh in 2009). The overa ll regressions tell the story of hydroelectric power. As many of the hydroelectric projects were funded by the federal government, it made more sense to evaluate projects that are within reach of the average consumer. The greatest changes in renewable e nergy generated came from wind power (11.2 vs
74 147.8 mwh ) and solar (80% growth .99 to 1.8mwh). Both of these power generation sources are explicitly the targets of state level renewable energy policy. As a result, the regressions were re run, only looki ng at solar, wind and the combined solar and wind output. As displayed above the hypotheses, throughout this section, a few terms are used repeatedly and regularly to define the regressions / models all of the policies, all o f the states and all of the demographics against generation of renewable energ only the states as independent variables with their constituent parts regressed against total renewable energy generated. Changes to the dependent variable are found in the + Combining them should be understood intuitively + regresses all policie s, demographics and states against the combination of solar and wind power generated as the dependent variable. As above, the reason for splitting out solar and wind from total energy is that they are the primary sources of renewable energy that an indivi dual can affect through their consumption choices. Overall Results As noted above, the primary finding is that studies of all renewable energy sources are mostly studies of hydroelectric power. The small growth in wind is outweighed by the sheer magnitud e of hydropower. It is interesting that 10 states in the West/Midwest are generating over 20% of their renewable e nergy from solar and wind power this will be commented on later in interpreting the results. We move on to the individual results.
75 Populat ion and wages Population is the most consistent predictor of renewable energy generation. The initial logic of more people in a state leading to more energy in the state seems to hold true for renewable energy as well. When looking at the solar and wind power models, the growth in these sources was primarily focused in the Midwest and Pacific coast as discussed before. These states (in particular Oregon and Colorado) saw influxes of over half a million on small population bases during the study time T exas, which grew by 3 million and also saw a large increase in wind power, has a negative sign in front of its fixed effect coefficient. This seeming disparity (large growth in population and wind power) begged for a review of population and the state fi xed effects in the solar and wind regressions, it appears that much of the growth in solar and wind power generated is attributed to population changes (coefficient of 3vs 2.7 for total renewable energy). The state fixed effects are then negative for stat es that may have a substantial growth in solar or wind power, but less than those states that had higher percentage increases in population and solar + wind power. While this can be looked at as a foible of the math, it could also indicate that there are policies, values or communities that make those states a preferred destination for environmentally aware citizens, thus why they experienced higher population growth than other states. Another commonly appearing variable was two sides of a coin average wage showed up as a positive indicator while the unemployment rate was negative. Only one of these two variables would appear as s ignificant in any regression, which could make one wonder if there is an issue of multicollinearity They were not highly c orrelated (r of
76 .228), so while they may be capturing the same effect, they certainly do not closely match each other in the observations. Initiative Process While population in a state seems fairly causal for use of energy; the causality of the initia tive process and the use of renewable energy was less clear. A potential outlier effect was noticed, as 75% of the states with an initiative process are west of the Mississippi and the top three renewable energy states (which generate more than the next 7 states combined) are all on the Pacific coast. These states were removed from the dataset to evaluate the stability of the effect. When Washington is removed, the initiative variable remains significant, although with less effect (approximately one thir d of when Washington is included); removing all of the top three renewable energy producing states the variable is dropped from th is forced limited regression. In addition, when including the initiative process in the solar and wind regressions, the ini tiative variable is dropped from the regressions, lending more credence to the idea that it may be a correlation, rather than a causal relationship. A case study of initiatives within the three states would be an interesting way to investigate the questio n of causation versus correlation. States, sui generis A theoretical basis for including the states as fixed effects has already been discussed perhaps there is something different about the citizens of states: Californians, Kansans and Virginians may not be a homogeneous group. In reviewing the initial results, a second reason comes to mind: there is a small number of states that generate the vast majority of the hydroelectric power. Thus, this state fixed effect is really more of an
77 indicator of hav ing power generating dams in the state. As discussed in the overall results, the state level effec t in the solar and wind regressions becomes an indicator of those states that have a high er growth rate in solar or wind power, relative to the growth in the i r population. Continuing to look at the st ates themselves, excluding all other measures and regressing only the states as binary variables against the renewable energ y generated, they explained 97.9 % of the variance in the differences in energy generated Looking at the states with population added in and this measure increased to 98. 1 %. As a result, there is a strong argument to be made that almost all of the differences in total renewable energy generated can be attributed to the states themselves. T his can be a result of natural endowment combined with historical policies (e.g. New Deal and subsequent programs for hydroelectric power). Figure 3 offers a graphic look at the country relating population (color of state) to amount of renewable energy generated in the state (size of dot, actual 20 0 9 generation listed) shows that while there is some direct correlation between population and renewable energy generated, there is significant variance across the states as well. California is a large state ( population wise), but generates less renewable energy (1.45mwh/ per capita) than Arkansas (2mwh ). As an opportunity for renewable market expansion, it helps to look at the percent of renewable energy relative to total energy generated for the country as we ll. Figure 4 illustrates current saturation of the market, with Idaho and Washington leading with 86% and 75% respectively, but the 3 most populous states only having 26%, 6% and 24% (CA, TX, NY, respectively) There
78 appears to be opportunity for growth in renewable energy at the expense of fossil fuels, rather than renewable energy technologies competing with each other. Figure 3 P opulation and T otal R enewable E nergy G enerated Figure 4 Population and R atio of Renewable E nergy to T otal Energy Ge nerated. Moving from this 2009 snapshot to a year to year analysis, the data suggests that most of the variance in generation has come in three areas: hydroelectric power has some
79 variance in the middle of the decade, solar has slowly been growing over th e last decade, and wind power has increased significantly relative to its baseline. Th e growth in solar and wind may be the result of many of the policies attempt ing to influence the individual and small suppliers of distributed energy to the grid and an y effect may be obscured by the changes in hydro power To test this a second set of regressions were run to focus only on solar and wind energy renewable energy. Homeowners can install residential so lar systems and landowners can have wind farms in addition to solar. The other sources of renewable energy that were eliminated from the regressions were geothermal, hydroelectric, biomass an d wood products. Each of those systems would require a fairly s ignificant investment, well beyond the means of the average consumer. Population, the initiative process, and the very nature of the states themselves are consistent predictors of total renewable energy generation. Although much of the change in the amo unt of renewable energy generated is found in the growth in wind power, multiple questions need answering regarding the effectiveness of the different policies they can be answered through a review of both the combined totals of renewable energy and the exclusively solar and wind regressions. However, be fore looking at the policies, the various demographic variables will be discussed. Demographic R esults Full Model Regressions P opulation and average wage / unemployment were consistent ly in the full and reduced total renewable energy models Initiative was dropped from the regressions, as no states changed whether they allowed initiatives during the study period. Each of the remaining variables will be discussed in this sect ion. First on the list is t he year fixed
80 effect variable, designed to measure the influence of added awareness, education or desire to purchase renewable energy over time. A positive change in this variable could reflect verall preference for renewable energy over traditionally produced energy, either for a sense of growing environmentalism, stewardship or another reaso n. Individual Years E ach year of the study was included as a dummy variable to look at year to year vari ations in generation. In theory, i f the weather was unseasonably warm or cold, one could expect a change in the amount of energy used. The earlier year variables ( 03 and 04) were consistently insignificant for the full model, whereas 02 and 06 08 we re significant and consistent in effect. Focusing on the story within the year variable, there is a significant drop in the middle of the decade, with recovery to 2001 levels in 2008 and 2009. Looking beyond the regressions, the drop may be attributed t o a drought in the region served by the Tennessee Valley Authority. As the drop in hydro power generated by TVA was considerably larger than the increases in wind power, the consistent growth trend seen in wind in the Midwest is hidden. It also obscures any national changes in preference for renewable energy. Democratic Control of Legislature regu licies, a Democratic legislature is more likely to enact these energy policies. More c onsistent in effect, but is less frequently seen in the regressions was the variable for having a
81 Democrat controlled legislature. When it did appear, it was a positiv e contributor to renewable energy generation; it was only present in some of the limited total energy regressions and the full solar model. The magnitude of the coef ficient was fairly small: ~2000m wh in the limted regressions, 37.64mwh for solar. Interpr eting this leads to a conclusion that conservation and use of renewables is not necessarily tied to a political philosophy Reviewing the infrequency of appearance in the regression results combined with the conservative (Center for Climate Change Communication, 2009, 2010) it appears that political voting at the state level does not change the choice for renewable energy in a meaningful manner. Decoupling Utility Profits from Generation The final state characteristic is decoupling profit from generation. This is a market based mechanism to create incentives for utilities to innovate, by allowing them to retain the profits from their success. Historically, utilities operated formula where their costs were met, plus a certain profit for the company, set by the public utility commission. With the advent of many market mechanisms (e.g. tradable energy, renewable credits), many states have freed their utilitie s from generating in a service area (which may be inefficient) and allow them to generate and transmit power from low cost areas to high demand consumers. Both decoupling separately and combined were evaluated: t he natural gas decoupling had no effect on the generation of renewable energy, whereas decoupling electrical generation did. Scrutinized in the full models, decoupling did not have a significant effect on the generation of renewable energy. However, in the limited models ( limited models without states, as well as in the solar, wind and solar + wind), it was consistent and significant.
82 Relative to the effect from other characteristics and policies, the decoupling was associated with an 8 9,000 megawatthour increase in total renewable energy gener ation. Looking only at solar and wind, it was ac cretive 118m wh for solar, 484 for wind c ombining nearly linearly to 597m wh in the combined model. This is consistent in both magnitude and effect. A s a check on the previous regressions, an additional reg ression was run using total renewable energy as the dependent variable with state demographics and c haracteristics as the independent variables This tests for an omitted variable, checking if the characteristics selected are equivalent to testing as a st ate as a whole. The results indicate that there may be a considerable number of omitted variables in the state characteristics, as the R 2 for this characteristic model was o nly .3282, indicating explained over three times the variance as th is limited model did. This difference is apparent even when excluding the top 3 or 5 renewable energy producing states eliminating the top 10% of renewable energy producing states and re running produces estimates tha t explain 70% of the variance, still explaining over twice as much as the characteristics on their own. This is an area for potential future research When reviewing all of the characteristics as a whole, a few non surpr ises: population and average income are positively correlated with energy generation. It is burden. Free market advocates will be pleased to see the significance o f decoupling profits and allowing companies to decide production and work toward innovation.
83 Policy Review and Results The analysis now turns to reviewing the policies and is a major focus of this dissertation. The policy review is split into two major sections: the first is a review of the policies as they impact generation of renewable energy, both in full and reduced models, including the reduced solar + wind only models. The second section reviews the determinants of policies. The overall effect of the policies found to be significant will be discussed, and then the details of individual policies presented, finished with a summary of the section. The policies affecting renewable energy fall into two broad groups financial and regulatory policies. While some goals are explicitly designed to add to renewable energy generation (renewable portfolio standards, solar/wind access requirements); many of the public benefit funded financial incentives and programs are available for either increas ing renewab le energy generation or reduction of all energy generation. As a result, when discussing these policies, and their effects, the expected impact could be either a positive (added renewable energy) or a negative (reduction of energy used). The overall resu lt is that financial policies were able to make some impact, especially those that either defined a sales tax refund or other tangible benefit, combined with those that provided a guaranteed buy in tariff for renewable energy. It appears that adding certa inty of benefit helps consumers decide if they want to purchase a renewable energy system. It also makes this author think that it is dominated by companies that are able to quantify the benefits, and then appropriately price products that allow them to s ell the renewable systems to consumers, while selling credits to the utilities.
84 Po licy Review and Results Full M odel R egressions Looking at the full models (including all demographics, states, and policies), only two policies show up consistently: net m etering and property tax exemptions. Net metering is a requirement that buildings with a renewable energy source have a bi directional meter so when generating more energy than the building is using, it will turn the meter backward and feed energy into the grid, rather than taking it out. Property tax exemptions typically function to compensate homeowners for the additional cost of renewable based energy by exempting the added cost from property taxes. The expected result for net metering is that it wou ld be positive feeding excess back into the grid should lead to people producing more distributed energy. The regression results for net metering were in conflict: in the full regression, it was approximately 1 1 25 0m wh in the solar, wind and solar + wi nd regressions, it has a positive effect on the generation in a state. The discrepancy in the sign may be the result of Net Metering being highly correlated (.56) to interconnection regulation, as well as correlated (~.25) to renewable power standards, p ublic benefit funds and property tax exemptions. The coefficients netted out to approximately 1.2mwh, which would make sense that these items, jointly, lead to an increase in renewable energy. The property tax exemption appears in the three full models focusing on renewable energy generated, consistent in effect of an additional 1450 1550mwh in the regressions. The exemption lowers taxes on a building with renewable energy to the same level as a comparable building without renewable energy sources. A r elated (but separate) policy (property tax assessments) appears to be significant in the solar
85 regressions. The assessment policy allows for the same effect, but is administered at a local level and may not be consistent across the state. A combination of state level tax exemptions was the final policy to appear in the full model regressions, and it only appeared in one of the three full models. The effect was to add 933mwh, so it is signed appropriately for the theory that lower taxes on the equipment will lead to more people deploying the system. However, it does appear in the wind and solar + wind regressions, consistent in magnitude in a range around 425 mwh and contributing to the amount of renewable energy generated. With this in mind, for the li mited application in solar and wind, the state combination of tax relief does appear to have a positive effect in generation. Again, almost all of the variance in the data was absorbed by the state variables in the full regressions, so it is not surprisi significant effect on the generation of renewable energy. Of those that did appear, financial incentives seem to work to a small extent (at a minimum, people are not hurt on taxes by adding renewable ener gy). Policy Review and Results S olar and W ind R egressions The policies that made a difference for solar and wind regressions are those that one would expect a private consumer to be interested in. Selling energy back, getting a tax break, and having so me level of assurance about the competence of the installer all add to the generation of distributed renewable energy. Some of the policies that were significant in all or most of solar + wind regression s and signed in accordance with theory include rene wable demand side management programs, public benefit funds, appliance efficiency standards, mandatory
86 uti lity green power options equipment certification, research outreach, and the aforementioned net metering laws. Some of the policies are only signifi cant in either the total renewable energy regressions or the solar and wind regressions. Appearing in the total regressions are construction design standards interconnection and state bond programs (signed negative, rather than positive). Polices in sol ar and wind, but not the total energy pieces, include contractor licensing (solar), energy standards for public buildings, generation disclosure, a community based project in Minnesota, production incentives, property tax assessments, sales tax refunds, an d state combination of grants and loans. Renewable demand side management and public benefit funds typically work toward the same goal: reducing the amount of power used through the use of more efficient appliances, light bulbs or even just remembering to turn the lights off when you leave a room. Both of thes e programs appeared consistently with magnitudes of 8500 10500 and ~5300 mwh reduction in all of the regressions that included population (the regression of policies only, without population, saw a smaller effect of 5898 and 4432 mwh respectively). As renewable energy is a portion of total energy consumed, it follows that renewable energy would drop with effective DSM and PBF policies and programs. Appliance efficiency standards require that any appliance sold in the state meet certain criteria. This could have two complementary effects: more efficient appliances could reduce the amount of energy used (similar effect to PBS and other demand side management) and the added awareness of the amount of energy used could trigger a
87 traditional to renewable) The results from the regressions show that these standards are correlated with an increase in total renewable ener gy (in the 10 14 mwh range) and in wind (~480 mwh). Mandatory utility green power offerings are fairly self explanatory. Utilities are required to offer consumers the ability to source their power from renewable generation. While the actual electricity renewable source, it is a good way for consumers to indicate their willingness (and actually complete the action) to buy green energy. With the solar and wind regressions, it was consistently positiv e and relatively large in magnitude, eclipsed only by tax refunds and one off policies in most cases. The consistency and significance of the coefficient on ratio nal choice theory carry some weight in the decision making process of utility customers. Equipment certification is the first of the consumer protection regulations (these occur in the solar and wind regressions more frequently) This policy requires tha t the equipment must be built to a certain standard for safety and performance, which helps prevent inferior products from entering the market, adding to the trust that regular citizens could have in purchasing the equipment. That holds true in the wind m arket this policy increases generation by 360 400 mwh, and in the total renewable energy regressions by a factor of almost 10 (3000 3900 mwh). The research outr each policy leads to a natural conclusion if the state subsidiz ing renewable energy resear ch (public, private or in partnership), it makes sense that the amount renewable energy generated would increase. The effect was seen more
88 consistently in wind power (increases of ~500 mwh when population is taken into account) and in the 3 4000 mwh incre ase for total renewable energy. This effect of this policy will likely only accrue to first adopters if there is a finite amount of research that can be done on renewable energy generation; assuming there is considerably more research that can be done, on e would still expect a diminishing return to a later adoption of this policy. The final po licy that was consistent in the solar + wind regressions is net metering. As discussed above, it was negative in the full (states included) regressions, but positive in the rest of them, adding credibility to the idea that people will respond to the idea that they can sell excess capacity back into the power grid. Limiting the effect of this policy are terms and conditions for many of the utility programs that cap th e size of a renewable installation that they will subsidize. Apparently, natural monopolies do not like too much competition. When looking at policies relevant to the solar + wind power regressions, contractor licensing begins the list, and is similar t o equipment certification as a consumer protection policy. By regulating contractors it instills confidence in the general population that the installation will meet a minimum standard, and this is correlated with an increase of approximately 90 mwh in so lar power for the states with such licensing. One program of particular interest is in Minnesota, and referred to as the Community Based Energy Development program targeting wind power. Enacted in 2005 and in place a couple of years later, it is associa ted with an increase in over 2,000 mwh in wind power in that time. While a fairly complicated program, it incorporates floors for feed in rates and heavily weights payments to the C BED groups to take advantage of the
89 present value of money to service the debt load required to build the project. The goal is to be revenue and cost neutral to established utilities and energy cooperatives. Energy standards for public buildings require efficient and/or renewable technology in buildings and are similar to co nstruction design standards, especially in the sense that they should conceivably have a negative effect on energy consumed by the building. In the wind regressions, it appears as a significant policy (accounting for ~300 mwh), but only when demographics are not included. As a result, the predictive power of this policy is somewhat suspect. Generation disclosure is a policy that requires utilities to let consumer know the fuel sources of their electricity. In theory, it would lead consumers to choose ren ewable energy sources for energy generation. In the regressions, however, it is signed negatively (indicating a decline of ~600 mwh in solar and wind energy, and 70mwh in solar.) Again, one interpretation of this is that perhaps generation disclosure is taking place in states that are pulling out the stops to try to increase awareness, and greater data (i.e. more time to analyze) would bear this out. Production incentives are paid (often as tax credits) to utilities that increase the amount of renewable e nergy generated in a state. They are shown to increase wind production by approximately 470 mwh in the states that offer them to the utilities. There does not appear to be any effect on solar generation. Property tax assessments and sales tax refunds are direct financial incentives to encourage the purchase of renewable generating equipment. The sales tax refund is significant and consistently associated with a 3500 4000 mwh increase in the wind a nd s olar + wind regressions, while absent from the solar r egressions. The tax assessment
90 policy exempts the added value of solar heating and power systems from taxes (so a building that has solar panels would be taxed at the same rate as one that does not have panels). As predicted, it is associated with an inc rease in solar power, to the tune of 75mwh. These two groups of polices contribute to the idea that providing credits to consumers can lead to changes in their behaviors / purchasing of renewable energy systems. The final policy that appears only in the wind and solar + wind regressions is that of state level grants and loans. These have helped fund an estimated increase of over 400 mwh in wind power in the states where they are enacted. Demographic and Policy Analysis Conclusions When reviewing all of the policies and demographics as a group, it appears that population is the most consistent predictor of the renewable power generated and that there are characteristics (natural endowment, past government policy, others) that differentiate the states suf ficiently that the vast majority of the variance in the observations can simply be explained by simply looking at populat ion and the individual states. When primarily looking at wind and solar power, the story begins to change far more toward looking at the relative effect of the policies and demographics. Education and conservation efforts are effective in reducing the total amount of energy generated (including renewable energy as a percent of total). In addition, giving individuals the choice to sele ct renewable energy for all or part of their consumption (mandatory options) leads to increases in using renewables, whereas simply telling them about it (generation disclosure) does not seem to have the same effect. Finally, providing financial incentive s (tax assessment exemptions and credits, loans, and feed in tariffs)
91 seems to help move the marginal customer across the decision line to purchasing a renewable energy system. As a result of this fairly exhaustive look at the policies in place along wit h other demographic variable people. If one limits the review to exclusively solar and wind power ( where most of the variance in the last ten years has been), there are policies that seem to have an effect on consumer behaviors, and thus, could be levers used by other states and governments to try to increase the demand for renewable energy. Event Histo ry Analysis The analysis now shifts gears to look at other determinants of policy adoption, outside of the success of similar programs on the adoption of a policy in a state. This section begins with a brief set of comments about the regressions overall, followed by a review of the different effects demographic variables on the policies, and finishes with a summary and general conclusions. There were 40 policies reviewed via this methodology, of which 15 are of primary interest, due to wider adoption and changes in adoption over the study period These widely adopted policies are fairly evenly split between financial incentives and regulatory pressure (7 and 8, respectively) The appearance of the demographic variables in the regressions showed low level s of correlation across the variables. The table below indicates frequency of appearance in the event history analysis.
92 Table 3: Frequency of Significance in Event History Analysis Positive Influence Negative Influence Total Similar policy enacted in n eighboring state (s) 21 3 24 Population 15 3 18 Average hourly wage 21 4 25 Unemployment 4 7 11 Decoupled profits (either gas or electric) 14 2 16 Democratic Legislature in state 12 3 15 Citizens' Initiative 14 7 21 Trend 15 7 22 Table 4: Frequency of Significance in EHA, 15 most popular policies Positive Influence Negative Influence Total Similar policy enacted in neighboring state(s) 6 3 9 Population 7 3 10 Average hourly wage 10 1 11 Unemployment 2 2 4 Decoupled profits (either gas or elec tric) 7 1 8 Democratic Legislature in state 5 2 7 Citizens' Initiative 6 4 10 Trend 5 1 6 The wider range of policies include 17 financial and 19 regulatory policies financial policies provide an incentive for certain behaviors, regulations require performance at a set standard. The received wisdom is that Democratic legislatures would be more likely to enact regulations, and this is borne out in the analysis it was relevant in five regulatory policy enactments and two financial policies. Perhaps a more interesting result is that higher average wages were associated with an increase in the likelihood of adoption in ten regulatory policies and four financial policies. While lks do not want
93 just the option to purchase green energy; they also demand a greater amount of regulation around renewables. The effect of neighboring states is equally split between financia l and regulatory po licies initiative seems to be used the most in creating financial incentives for behaviors, when most of the writing about them in the literature is their use in creating renewable portfolio standards (a regulatory policy) The high negative influence is not surprising, as they can be used in NIMBY efforts. Overall, the trend in time for policies seems to be toward financial incentives and away from regulatory choices, which is congruent with the assumption that government is attempting to use more market based levers and less regulatory power to make changes in behaviors. Table 5: Influence of State Characteristics on Policy Adoption Positive Influence Negative Influence Total Finance Regs. Finance Regs. Finance Regs. Similar policy enacted in neighboring state (s) 9 9 1 1 10 10 Population 4 5 1 0 5 5 Average hourly wage 4 10 2 1 6 11 Unemployment 2 0 2 2 4 2 Decoupled profits (either gas or electric) 6 6 0 1 6 7 Democratic Legislature in state 2 5 1 1 3 6 Citizens Initiative 7 4 4 3 11 7 Trend 6 4 0 4 6 8
94 Table 6: Influence of State Characteristics on Widely Adopted Policies Adoption Positive Influence Negative Influence Total Finance Regs. Finance Regs. Finance Regs. Similar policy enacted in neighboring state(s) 2 4 2 1 4 5 Population 5 2 2 1 7 3 Average hourly wage 3 7 1 0 4 7 Unemployment 2 0 1 1 3 1 Decoupled profits (either gas or electric) 3 4 1 0 4 4 Democratic Legislature in state 2 3 1 1 3 4 Citizens' Initiative 5 1 1 3 6 4 Trend 2 3 0 1 2 4 These findings add to the evidence that legislatures are influenced by what their neighbors are doing. In over half of the policies, its existence nearby significantly increased the probability it would be adopted in another state. For some of these situations, it was perfectly correlated with programs that were found to be effective in the earlier regressions (research outreach, sales tax refunds) and also some other more wonkish policies (construction standards for state buildings, green building in centives). The presence of a similar policy in a neighboring state was a significant contributor to the probability of adoption for 59 % of the policies found to be relevant in the earlier regressions (policies associated with increases in renewable energy .) The policies that were adopted across state borders are equally split between regulatory and financial p olicies they do not seem reflective nor dependent upon the natural endowment of a state. The policy analysis adds further to the idea that green p ower functions as a luxury good. As hourly wages increased, they increased the likelihood that many policies would
95 the grants/loans at the individual level, as well as the tax assessment change for personal property taxes, while tax exemptions and immediate sales tax credits were positively associated with higher wages. This makes sense though, as lower income groups may look for the grants and loans, whereas the hi gher income groups would be able to afford the up front cost, instead pressuring for a tax credit or other such benefit. The next most prevalent characteristic was that of decoupling profits from generation. Based on earlier discussion about the dual le ver nature of making such policies successful, it is unsurprising that its effect is equally distributed across financial and regulatory policies. Looking deeper into the policies, the story begins to tell itself. Decoupling is associated with corporate and state level grants / loans, as well as the sales tax credits. On the regulatory side, there are solar access laws, public education / benefit funds, net metering and interconnection. While n either necessary nor sufficient for a credit market to be cr eated, these policies all contribute toward establishing a distributed energy credit market. Further research into this realm could look at question of chicken and egg part of a holistic approach by the legislature. The final two predictors for policy adoption do not seem to follow any strict pattern population is associated fairly equally with financial and regulatory policies, and the policies it is associated with a s a predictor are not correlated either in theory or adoption. Unemployment follows a fairly inverse course to average hourly wages (negative where average hours was positive), or appearing when average hourly wages does not. Again, there is not a strong correlation to theory or in adoption for why the unemployment appeared in the regressions.
96 CHAPTER V DISCUSSION This section reviews the hypotheses that were asked earlier and provide answers (where possible) and discussion points (when no answer is av ailable). A forewarning is due that the questions were broad in scope, the policies fairly specific, and even with grouping the policies, it is difficult to find a consistent effect. First is a general discussion of the results of the regressions, follo wed by the specific research questions and hypotheses, followed by a summary of the results and further questions. The initial regressions used different dependent variables: total renewable energy generated, the percent of renewable energy relative to tot al energy generated, as well as the change in renewable energy year to year. The latter two sets of regressions did not have enough variance within the observations to be able to draw any conclusions from the results. The first set of regressions include d all of the policies, demographics and states as independent variables. These regressions using total renewable energy generated are the foundation of this paper. While the policies and demographics explain very little of the renewable energy, the very high explanatory power of the states themselves provides strong support to the idea natural endowment is a primary driver of renewable energy. This becomes very relevant when looking at specific policy recommendations, such as renewable portfolio standar ds. If citizen groups are able to enact RPS with minimum production standards (e.g. Colorado requires 30% of energy to be renewable based starting in 2020), this can lead to two sorts of inefficiencies. RPS that require a specific technology may force us age of a non optimal generation technology; the second distortion may come via
97 artificial demand for one good (solar renewable energy) over competing goods (fossil fuels, wind renewable energy, nuclear alternative energy). Combining the natural endowment d iscussion and the results from regressi ng policies against the total renewable energy : while the individual policies do not seem to have an effect in the past decade; t he effect of past federal policies and prioritization shows in hydro power plants and si milar large scale projects The natural endowment of di fferent states also plays a role in defining the ren ewable energy sources exploited. Policy makers will do well to leave open options for choice, rather than mandates for a technology. As mentioned in the beginning of this paper, such natural endowments are a reason to look at state level instead of federal policy; the individual states can use their most abundant natural and/or renewable resources. Historic federal policy in the past has significan tly shaped the profiles of some states (e.g. Washington). State level policies seem to be effective at stimulating modern production of solar and wind power. Despite the difficulties in the full regression these models do establish a baseline for further research. As there have not been other regressions looking at this complete model, there are still some important findings: the last decade of policies have not moved the needle on overall renewable energy generation; future research should look at the d ifferences within the states, rather than policies; the growth in the arena is mostly in the wind area, with some increases in solar. eliminating the states from these models, with inconclusive and sometimes contradictory results. A further restriction of the model to using solar and wind power for the dependent variable opened the door to further insights.
98 Research Questions Discussion The first three research questions can be re viewed and addressed at this stage, usually answering in both the affirmative and negative incentive policies effective at stimulating generation of answered in the negative when looking at total renewable energy. The sheer scale of some of the existing facilities washes out any effect that the marginal consumer may have. This is likely an effect of the path dependency of utilities and the scope of their production it is difficult to move an e ntire industry to a new standard, when the old standard works well enough. When we change the dependent variable to solar and wind power, we find that there are, in fact, some policies that tend to contribute to the growth in generation. Tax incentive s see m to be the most effective at increasing the generation of solar and/or wind power whether offered as grants and loans to corporations or via tax credits and other incentives to individual consumers. The magnitude of the effect of tax credits is well bey ond any other policy (sales tax refunds contributing to a net contribution of over 4000 mwh.) Guaranteeing the rate (as was done in Minnesota) was also effective (net contribution of 2000 mwh). Corporate incentives were also found to be accretive, espec ially for wind power, but at a lower magnitude than the individual policies. One reason for this may be an interaction effect of utility incentives and personal incentives. While this paper has only looked at state level policies, the author is aware tha t many utilities will offer discounts, buy in tariffs, and immediate payment for future renewable energy credits. These programs are often structured differently for corporate consumers than private
99 individuals, which may lead to the interaction effect. This is an interesting area of further research, well beyond the scope of this paper. The flip side of financial gain is protection from loss, which is covered in the productio renewable portfolio standards, as well as consumer protection regulations. Market forces include options for choice and a distributed grid. At the macro level, the answer seems to favor market forces. Both mandatory green options and net metering have a significant impact on the total generation of renewable energy. Many of the most popular regulatory policies are demand side management programs their effect is significant and negativ ely signed. This means they have been focused on reducing overall consumption; the same education and awareness efforts may be used to increase the percentage of renewable consumption. Another popular lever used in government mandates is the renewable po rtfolio standard, setting out a minimum amount of power to be generated from renewable sources. Looking to the future when these policies come into force, it will spark questions around the effect of this technological forcing: did industry respond by cre ating more efficient ways of generating renewable energy, or did the RPSes simply raise power prices? Viewing the regressions with solar and/or wind as the dependent variable the answer again points to a blend of policies being effective. While the finan cial incentives (tax credits and exemptions) are an important part of moving the marginal customer, the most effective regulations include certification and licensing. These regulations reduce the risk of poor quality work / equipment for an individual co nsumer, and lowering
100 search cost barriers to the market consumers are able to ask for a certification instead of performing complete due diligence on a potential contractor. This increases the number of consumers willing to join the market and potential ly increase the number of installed systems. The next question asks if state characteristics or demographics help determine the generation of renewable energy. The regressions point toward a few findings. The first is that population is (unsurprisingly ) a reliable predictor of use of renewable energy. Increases in wages and/or decreases in unemployment are associated with an increase in renewable energy this is consistent with the theory that green power performs in the market similar to a luxury goo d. A question asked and reviewed, but unable to be answered, was if the percentage change in those characteristics were tied to changes in the total renewable energy, or the ratio of renewable energy to total energy generated. However, the data did not h ave significant variation, and the regressions were inconclusive. Decoupling profit from power (particularly electric) generation consistently appears as a way to spur generation of renewable energy. This also speaks to coordinating the use of market and regulatory levers to effect change. Eliminating the cost plus model as the sole source of profit and creating a new market (for renewable energy credits) allows utilities to find new revenues via trading the credits. Utilities can take a long term view i n the market, subsidizing distributed distribution grids (i.e. solar panels on roofs, private wind farms), allowing consumers to focus on immediate need s for power and financial rewards while selling futu re rights to energy credits
101 The variable for Dem ocratic control of the legislature is not significant when looking at renewable energy. The generation data shows a limited effect and this is confirmed in the policy adoption analysis it appears only in 25% of the policies studied, with the majority of those as regulatory policies. This is fairly indicative that renewable energy is a bipartisan issue. W hile running contrary to the stereotype of left leaning environmentalists; it is consistent with the research previously done on conservative viewpoint s. The trend variable (i.e. change of attitudes over time) complements this finding in the full models, as well as in solar and wind models, it indicates increasing generation of renewable energy. Legislative actions can be informed (or directly written ) through citizen initiatives, a powerful tool for the public to express their desires. The research shows it appears that their desires include renewable power. The initiative process was consistently associated with an increase in renewable power; howe ver, that effect disappears when looking only at solar and wind power. When looking at the adoption of various policies, it perfectly predicts the adoption of renewable set asides. This seeming contradiction can be explained by the fact that many of the RSA (and RPS) have not yet come into effect yet, so one expects to see a further ramping up of the renewable power generated as we approach those deadlines. The final research question was if demographics or characteristics help predict policy adoption. W hile the characteristics of a state do inform the estimates for renewable energ y generation there remains a large unexplained gap between the demographics and properties used in this regression and using the state itself as a dummy variable to capture any fixed effects. Despite this, within the limited group of
102 initiative, hourly wage and decoupling profits for utilities all contribute to a large number of policy adopti ons. With the broader research questions answered, we move to looking at the specific hypotheses postulated earlier. As discussed earlier, the results are best seen through a split analysis, separating the total renewable energy and the solar and wind r egressions. With the number of policies that have been investigated, some will have been more effective than others with regard to power generation. The section above detailed the effectiveness of different policies, including those found not to be effec tive at all. Those ineffective policies should not be seen as counterfactuals to the effective policies, rather as failed experiments in policy; they should be considered as something to focus the law makers on successful policies rather than being the gr ound for rejecting a class of policies for future consideration. Hypothesis Analysis and Results Discussion In this section, the research questions from earlier in the paper are rewritten in a or the hypotheses. The first hypothesis ask s if financial incentives were in effective in stimulating production of renewable energy. The full regressions looking at total renewable energy are inconclusive, as all the variance is absorbed at the state lev el Limiting the regression analysis (removing the states from the independent variable list) starts down the path of rejecting this hypothesis, and the solar and wind regressions provide further evidence for rejecting the hypothesis. Policies such as ta x rebates, feed in tariffs and net metering reduce the cost of installing a system or provide surety for the revenues from it, adding to
103 demand for distributed generation systems Financial policies are shown to increase generation of renewable energy. S tate legislators who are interested in increasing generation of solar and wind power would do well to use financial policies to incent their population. The second hypothesis, that government mandates do not increase the production of renewable energy, can be rejected at the limited regression level (no states) as well as when looking at the solar and wind energy policies. This is unsurprising when the government says a an entity must do something, it usually complies. A more important question that ca n not yet be answered is whether or not the government mandates (such where the mandate sets out a goal and the market responds to it, with a goal that the market response will lower costs for the innovative technology sufficiently that it can compete with legacy power generation systems. This is a question that can be revisited over the next decade as utilities and states approach the deadlines enacted in renewable portfolio standar ds. Continuing on in the market realm, the third hypothesis asks if marke t forces are ineffective in generating additional renewable power. While this is tied to financial incentives (in thin RCT frameworks), this hypothesis looks primarily at the effect of choice for renewable energy. There is considerable support to reject this hypothesis. One of the most consistently appearing policies was that of mandatory utility choice allowing consumers to pay a premium to purchase renewable based energy. This is consistent with prior research and remains a powerful tool for research into thick rational choice models. On the supply side, providing alternative s for utility companies to earn a
104 profit through renewable energy credits or other profit decoupling work s to encourage added production of renewables. Overall, the use of market forces is shown to contribute to the generation of renewable energy. Moving from specific types of policies toward states and demographics; the fourth hypothesis is that the states themselves do not have an influence on the generation of renewable energy. This hypothesis can be fairly quickly rejected. As noted multiple times before and forcing an focus on solar and wind power for elements of this dissertation, over 98% of the vari ation in total renewable energy generated can be explained simply by looking at the states themselves. This influence is a heretofore undefined combination of demographics, natural endowment, historic government policies and projects. While this disserta tion does not attempt to define that combination, it does provide evidence that it exists. The theoretical underpinning of the fifth hypothesis is if renewable energy functions as a luxury good and how that informs the thin versus thick RCT perspective It appears that it does function as a luxury good, as income increases and/or decreases in unemployment are associated with a increase in renewable energy generated in that state. While the original expectation was that this would only support a thin RC T view, the increase in population and wages is also associated with passage of more renewable energy policies. This finding is not overwhelming, as average wage is significant in just under half of the regressions, average wage and/or unemployment is sig nificant in 23, and population in only 10 It does lend support to a thick RCT view of increases in population and wages leading to more environmental awareness or willingness to see the
105 policies passed. The mechanism (critical mass of population or wage levels?) has not been investigated, and requires future research before drawing substantive conclusions. This discussion of demographics and policies bridges us to the sixth and final hypothesis: whether state level demographics contribute to the passage of policies. Looking at policy adoption as a whole, only one characteristic was present in over half of voters in states with the initiative process has been explo red; in the renewable energy realm, further research should be conducted to complete the chain. States with the initiative process are more likely to adopt renewable energy policy, but there is still a gap to show that this is a result of legislators resp onding to (or citizens utilizing the process to) the desires of the population to have these policies. Muddying the waters for this hypothesis is the fact that t he single most frequent contributor to policy adoption was the adoption of a similar policy i n a neighboring state that indicates that legislators are informed by what the external influence of what th eir peers are doing across a wide range of policies not just by the internal inputs (including natural endowment) from their own state. As menti oned before, average wage is present in almost half of the regressions, with the next most prevalent variable being the trend in time, another external input. Combining this information, one can say that the state level demographics do play a role in the adoption of the policies, however it is just as important to look at the external inputs to effectively predict adoption of policies. One can apply this dissertation to future public policy in multiple ways. It provides evidence regarding the effectiven ess of various policies can encourage further
106 looking at options for wind and solar energy policy, it appears that a blend of policies some providing financial as sistance for the marginal consumer while regulating the market entrants (either through licensing or certification of equipment) combine to stimulate the market. Studying the effectiveness of policies in other states at a case study level may also be an e ffective way to harness the external inputs for the legislative process. The C BED program in Minnesota stands out as a focused approach to harness community awareness and desire into a renewable energy program. This paper has taken a broad look at mult iple questions, without focusing narrowly on specific policies or individual states. This is done intentionally to fulfill a basic goal of a dissertation identification of a foundation of past research and understanding of the current research state wh ile providing a map of potential papers and research questions What has been found is that a combination of financial, regulatory and market based policies have a place in the future of renewable energy policies no group stands out as more exceptional than the other. In addition, the pressure from external factors (such as neighboring states or evolving opinions over time), as well as internal pressures (initiative process, rising average wages) contribute to policy adoption. These pressures, as measu red in this paper, do not capture all of the variables in a state that influence its renewable energy generation. Reviewing the theoretical implications of this research, it appears that while financial incentives are effective (lending support to thin RC T), it is just as apparent that thick RCT policies are also valuable. The following section looks at various ways to approach these future opportunities.
107 CHAPTER VI FURTHER RESEARCH One oft mentioned item in this paper is that the question of causali ty cannot be addressed effectively the paper was designed to investigate the overall effects of different policies on power generation, as well as what characteristics were associated with policy adoption. To look at causality would require as many pape rs as there are policies, with deep investigation into the theory based frameworks (advocacy coalition frameworks, multiple streams, policy windows). In light of the results (and the lack of overw helming consistency in them), quite a few new areas of inve stigation have become apparent. The level of analysis (states versus local government versus utility level), the amount and duration of data, as well as the policies themselves are all under review in this section. Using states as a level of analysis is a well practiced area of study as they typically form an excellent laboratory, especially in the area of regulation and enforcement of laws. Renewable energy policy definitely shares aspects of this (utilities are highly regulated) and states are a local a choices via the initiative process and direct election of representatives. A measure considered, but not included, is that of stability in the electorate. Are stable electors (i.e. with low threat of being voted out) less likely to innovate / be later adopters of policies? The effect of new political groups (such as the Tea Party) may change the assumed safety of districts. The state and local levels are a good focus, as the shift away from regulatory policy toward incentives means that many more of the policies are directed at utilities (greater generation of renewables, via financial incentive or regulatory mandate) or at consumers
108 rates for dis tributed renewable energy sources). As noted earlier, there were no new state level renewa ble policies recorded in 2009, future research will likely have to compare local entities (perhaps down to city level) or utilities against each other. Looking at th e utility as a measure of analysis would be fecund area, as many of the recent policies have targeted utilities. Renewable Policy Standards often mandate a the uti lity to pick areas that are naturally endowed with wind, solar or hydro opportunities for installing new generation systems. In addition, there are often market mechanisms, such as tradable carbon credits, established by the RPS legislation that allow the utility to offer market incentives (such as installation subsidies or buy in tariffs) to consumers to install solar or wind on their property, creating a distributed electrical grid. Difficulties in working with utilities as the level of analysis i nclu de size, both too big and too small. While major metropolitan areas are covered by public utilities with recognizable names, many of those big names are now consolidated into conglomerates operating in multiple states. As a result, using data from them a dds another layer to the analysis, but can not replace states as a geographic measure for generation. On the other end of the scale, states will often have multiple smaller utilities serving rural or specialized areas for their energy needs. These smalle r generators may be exempt from policies such as RPS or may not offer a full gamut of incentives for consumers. Alternatively, a very progressive small utility may be able to move all consumers to renewable based energy, but still not move the needle when their output is juxtaposed to the rest of the state or large utility players.
109 Add to this the fact that utilities may not be willing to disclose proprietary information. In states where the utilities have been able to decouple profit from generation, t here are multiple paths to profit. The profits (or losses) from trading energy futures, trading energy across state borders or utility boundaries and from generating / selling / stockpiling renewable energy credits may all be seen as trade secrets. Witho ut this information, it is difficult to decompose the profits, and without that, it is difficult to look at utilities through a thin RCT lens. This idea of the profits from renewable energy generated being dwarfed by traditional fuels leads into the next area for further research, one that is time defined. While the growth of renewable energy is continuing, a significant issue is that one is still dealing with rather small numbers. Solar generation capacity doubled between 2010 and 2011, yet still is on ly around .09% of total energy generation (U.S. Energy Information Agency 2012a, 2012b) As a result, a significant change in renewable energy may be obscured by outside generation. Many of the renewable portfolio standards set minimums that need to be achieved at a point in the future; backward looking regressions will onl y capture some of the progress that is being made toward these mandates. Of all of the RPS tracked by DSIRE, only Iowa has a target date before 2014 and it is an absolute generation number, rather than the relative percent of electric generation. Of grea t interest will be the success of these RPS programs (a.) on an absolute scale were the goals accomplished without changing / softening of the amounts required; and (b.) on an excess generation level did the utility stop generating right at the mandate amount, was production scaled up enough
110 to make it a cost effective competitor to traditional fuel sources, did popular demand increase enough that it exceeded the mandated supply? To track this data, one must bide time changes in production relative to established milestones can be seen as an interim marker for success of the program. If utilities are achieving their goals earlier than required in legislation / RPS, then this can inform the planning for public policy in the realm of technological for cing and market level policy tools. A backward look at the effect of RPS may well be a dissertation topic It is important to realize that the RPS number becomes a benchmark that is relative to total energy consu med or generated in an area and the effect of the RPS is set RPS requirement requirement to hopefully level the playing field for comparison of policy effects on energy generation. The results also call for a more in depth look at the states as case studies when 97% of the variance can be explained by simply regressing renewable energy generated against the state, increased to 98% by adding populati on to the mix and the demographic model only explains 33%, it seems that a review of the individual states could be fruitful. could lead to correlations and possibly predictive power for future surveys and planning. While discussing demographics and the internal / external inputs to the policy process, the information transfer mechanism needs to be explored. There is a strong correlation between the citizen initiati ve process and renewable energy policies being adopted; to explain if there is causation, case studies must be completed to look at the
111 (non )adoption of the various policies. Questions to be asked include: how to measure licies (or change in general); if the legislature was responsive to the citizens; if the initiative process was used (or if the threat of it was sufficient); how are resources marshaled within and from outside the state. Enactment of a policy in a neighb oring state has been identified as a predictor for policy adoption for multiple reasons, often centered in a technology transfer framework. A subset of the demographic study completed in this dissertation would allow for investigating the underlying aspec ts of the policies across states. What characteristics of the states, the policies and/or definitions (and achievement?) of success led to adoption in neighboring states? This is a mechanism that has not been fully explored in the renewable energy realm, and could be used across frameworks, both in technology transfer, as well as an argument for thick rational choice decision making. If policies (a) the metrics are citizenry, and (b) the policy success should be better defined. Only further research will be able to explain these questions. S upplement ing this research would be an expansion to follo w environmental legislative histories within states. These case studies can look at the questions above, as well as splitting the states by the initiative process. In the initiative states, the research can investigate if popular initiatives replaced, su pplemented or informed the legislative process. This is already an area of research (Bowler & Donovan, 2002; D. A. Smith & Fridkin, 2008) and applied t o financial policy (Matsusaka, 1995) Individual state case studies can often come to conclusions that are difficult to gener alize to other states, so the
112 focus should be on finding similarities in beliefs and interests that can be seen or recorded across the country. A final area for case study review would be the implementation and specific terms of each of the policies. Some are administered at a local level, some by utilities and some by the state. Some states have high levels of information regarding policies available through various channels (online, media presence, concentration of population near government offices), t his can be tied to the awareness of the population about the policies and their opportunities to participate. Amount of time to qualify, engage in and reap the rewards of the policies could also be measured to analyze the effectiveness of a policy at a st ate by state level. Amount of funding and maximum participation rates could also impact policy success. with (as examples) positive coefficients on financial incentives for increasing production and negative coefficients on incentives / education programs to reduce energy use, there were a number of policies that did not track to theoretical outcomes. A further line of inquiry could look at possible explanations such as the effect of the policy being obscured by state level impacts or an interaction effect with other variables known or unknown. Moving to a more theoretical view, a more holistic understanding of energy policy would review point and non point generation, balanc ing the costs of current solutions (such as generation requirements for electric cars) against other solutions (continued use of fossil fuels), and try to strike a balance for appropriate use of a mix of fuels. As an example, a model could be built using baseline generation using a mix of renewable and traditional fuel sources, with spikes in demand covered by other solutions
113 (battery power, gas turbines) with Pigovian taxation to cover the costs currently allocated to society. While overarching and fraug ht with arguments from partisans on either side of a thick RCT divide, it would definitely inform future policy and goal setting for renewable portfolio standards and generation needs.
114 CHAPTER VII CONCLUSION This dissertation has reviewed the prior lite rature and adds to the academy by making an assessment of the effect of renewable energy policy on generation, as well as what independent variables help predict adoption of renewable energy policies. The primary questions asked were what policies are eff ective in increasing the generation of renewable energy and what influences the adoption of policies are they adopted because they are effective in increasing renewable energy, or are there other reasons. In trying to answer these questions, the researc h showed that the policies have little to no effect on total renewable energy rather, there are characteristics (unmeasured in this paper) in each state that determine the total amount of energy. The state story is mostly one of hydroelectric power and the annual story is one of drought versus rain. However, by looking only at solar and wind power, one can see that a mix of financial, regulatory and market based policies combine to increase the generation from these power sources. There is no one simpl e answer regarding which policies are most effective, as the policies seem complementary to each other, and future research could focus on interaction effects between the policies. The primary determinant of the adoption of renewable energy policy was i f it was adopted in a neighboring state. When looking at solar and wind energy production, the effect of policy success (defined as changing production) in the neighboring states can not be seen as a major contributor to the inform ation process for legisl ators in states considering adoption of the policy. It appears as a significan t indicator in just over half of the policies found to be important in changing solar and wind production. As a result,
115 the author recommends further research into the influenc e that neighboring policies have on the legislative process; this research does add to the literature supporting the influence of bordering states on the actions of state lawmakers. Another interesting contribution is that average wages and / or unemploy ment are seen as important predictors for policy adoption, as well as the overall use of renewable energy. There has long been speculation that renewable energy functions solely as a luxury good, and when times are tough, people will use less energy. Thi s paper has found that while there is this thin rational choice correlation, it also confirms research from Menz & Vachon that mandatory green power options from a utility will also increase renewable energy, which indicates that there may be a shift from simple thin RCT views to a thicker framework. Once the population reaches a certain density and/or level of wealth, then individuals will make choices for renewable energy, even when it costs more. The theoretical basis for renewable energy policy adop tion goes beyond the conversation around environmentalism and adoption lie in thick rational choice modeling rn the switch and the light and goals are, what metrics can be used to measure them, and how they can be used to show legislative responsiveness to the electors.
116 REFERENCES Acemoglu, D., Aghion, P., Bursztyn, L., & Hemous, D. (2010). The Environment and Directed Technical Change [Electronic Version]. SSRN from papers.ssrn.com/sol3/papers.cfm?abstract_id=1668575. Acemoglu, D., & Cao, D. (2010). Innovation by Entrants and Incumbents [Electronic Version]. NBER Working Paper Series Retrieved 10/6/2011 from http://www.nber.org/papers/w16411.pdf Agnolucci, P. (2007). Renewable Energy Policies i n the Netherlands. Renewable Energy, 32 (5), 868 883. Akerlof, G. A. (1970). The Market for 'Lemons': Qualitative Uncertainty and the Market Mechanism. Quarterly Journal of Economics, 84 488 500. Andrews, K. T., & Edwards, B. (2004). Advocacy Organizatio ns in the US Political Process. Annual Review of Sociology, 30 479 506. Arrow, K. J. (1962). The Economic Implications of Learning by Doing. The Review of Economic Studies, 29 (3), 155 173. Arrow, K. J. (1963). Social Choice and Individual Values New Yo rk, NY: Wiley. Arthur, W. B. (1989). Competing Technologies: Increasing Returns and Lock in by Historical Events. The Economic Journal, 99 116 131. Arthur, W. B. (1994). Inductive Reasoning and Bounded Rationality. American Economic Review, 84 (2), 405 4 12. Arthur, W. B. (1999). The End of Certainty in Economics. In D. Aerts, J. Broekaert & E. Mathijs (Eds.), Einstein Meets Magritte Dordrecht, Netherlands: Kluwer Academic Publishers. Astrand, K., & Neij, L. (2006). An assessment of governmental wind po wer programmes in Sweden using a systems approach. Energy Policy, 34 277 296. Atkinson, S. E., & Tietenberg, T. E. (1987). Economic Implications of Emissions Trading Rules for Local and Regional Pollutants. The Canadian Journal of Economics, 20 (2), 370 386. Baumol, W. J. (1972). On Taxation and the Control of Externalities. American Economic REview, 62 (3), 307 322.
117 Belasco, A. (2009). The Cost of Iraq, Afghanistan, and Other Global War on Terror Operations since 9/11 Retrieved 8/16/2009. from http://www.fas.org/sgp/crs/natsec/RL33110.pdf Belkin, P. (2008). The European Union's Energy Security Concerns Retrieved. from http://www. fas.org/sgp/crs/row/RL33636.pdf Berry, F. S., & Berry, W. D. (1999). Innovation and Diffusion Models in Policy Research. In P. A. Sabatier (Ed.), Theories of the Policy Process Boulder, CO: Westview Press. Bertram, G. (1992). Tradable Emission Permits and the Control of Greenhouse Gasses. Journal of Development Studies, 28 (3), 426 446. Besley, T., & Coate, S. (2000). Issue Unbundling via Citizens' Initiatives [Electronic Version]. NBER Working Paper Series 39. Retrieved 4/5/2012 from http://www.nber.org/papers/w8036 Bolinger, M., & Wiser, R. (2006). The Impact of State Clean Energy Fund Support for Utility Scale Renewable Energy Projects [Electronic Version] from http://eetd.lbl.gov/ea/EMS/cases/ Borchers, A. M., Duke, J. M., & Parsons, G. R. (2007). Does willingness to pay for green energy differ by source? Energy Policy, 35 (6), 3327 3334. Bowler, S., & Donovan, T. (2002). Democracy, Institutions and Attitudes about Citizen Influence on Government. British Journal of Political Science, 32 (2), 371 390. Breukers, S., & Wolsink, M. (2007). Wind power implementation in changing instutitional landscapes: An international comparison. Energy Policy, 35 (5 ), 2737 2750. Buckley, J., & Westerland, C. (2004). Duration Dependence, Functional Form and Corrected Standard Errors: Improving EHA Models of State Policy Diffusion. State Politics & Policy Quarterly, 4 (1), 94 113. Buen, J. (2006). Danish and Norwegian wind industry: The relationship between policy instruments, innovaiton and diffusion. Energy Policy, 34 3887 3897. Busenberg, G. J. (2001). Learning in Organizations and Public Policy. Journal of Public Policy, 21 (2), 173 189. Butler, H. N., & Macey, J R. (1996). Using Federalism to Improve Environmental Policy Washington, D.C.: The AEI Press.
118 Calel, R., & Dechezlepretre, A. (2011). Environmental Policy and Directed Technological Change: Evidence from the European Carbon Market. Unpublished Journal a rticle. London School of Economics and Political Science. Carley, S. (2011). The Era of State Energy Policy Innovation: A Review of Policy Instruments. Review of Policy Research 28 (2), 265 294. Carley, S. & Miller, C.J. (2012). Regulatory Stringency a nd Policy Drivers: A Reassessment of Renewable Portfolio Standards. Policy Studies Journal 40 (4),730 756. Carpenter, E. H., & Chester, S. T. (1984). Are Federal Energy Tax Credits Effective? A Western United States Survey. The Energy Journal, 5 (2), 13 9 149. Carrigan, M., & Pelsmacker, P. D. (2009). Will ethical consumers sustain their values in the global credit crunch? International Marketing Review, 26 (6), 674 687. Carson, P., & Mouldon, J. (1991). Green is Gold: Business Talking to Business about the Environmental Revolution Toronto, Canada: HarperBusiness. Center for Climate Change Communication. (2009). Global Warming's Six Americas 2009: An Audience Segmentation Analysis : Yale / George Mason University. Center for Climate Change Communication (2010). Global Warming's Six Americas : Yale / George Mason University. Clayton, S. (2000). Models of Justice in the Environmental Debate. Journal of Social Issues, 56 (3), 459 474. Clinton, B. (1995). Reinventing Environmental Regulation Retrieved 1/31 /2008. from http://govinfo.library.unt.edu/npr/library/rsreport/251a.html Coates, R. H. (1960). The Problem of Social Cost. Journal of Law and Economics, 3 1 44. Darby, S. ( 2006). Social learning and public policy: Lessons from an energy conscious village. Energy Policy, 24 2929 2940. Database of State Incentives for Renewables & Efficiency (DSIRE). (2007). (Publication. Retrieved 12/27/2007: http://www.dsireusa.org David, P. (1985). Clio and tne Economics of QWERTY. American Economic Review, 75 (2), 332 337. deLeon, L. (1996). Ethics and Entrepreneurship. Policy Studies Journal, 24 (3), 495 510.
119 deLeon, P. (1979). Development and Di ffusion of the Nuclear Power Reactor: A Comparative Analysis Cambridge, MA: Ballinger Publishing Company. dePalma, A., Meyers, G., & Papageorgiou, Y. (1994). Rational Choice Under an Imperfect Ability to Choose. American Economic Review, 84 (3), 419 440. DSIRE. (2013). Glossary. Retrieved 02/07/2013, from http://www.dsireusa.org/glossary/ Durham, C. A., Colby, B. G., & Longstreth, M. (1988). The Impact of State Tax Credits and Energy Prices on Adoption of Solar Energy Systems. Land Economics, 64 (4), 347 355. Economist. (2006). The Art of the Impossible. The Economist 381 (8501), SS3 5. Economist. (2012). The dream that failed. The Economist, 47 (10). Economist. (2013a). Business this Week. Business thi s Week Retrieved 3/29/2013, from http://www.economist.com/news/world week/21570765 business week Economist. (2013b). End the ethanol Tax. Babbage: Science and Technology Retrieved 3/29/2013, from http://www.economist.com/blogs/babbage/2013/03/biofuels Economist. (2013c). The future, finally. Shumpeter Retrieved 3/29/2013, 2013, from http://www.economist.com/blogs/schumpeter/2013/02/hydrogen powered cars Eisinger, P. K. (1988). The Rise of the Entrepreneurial State Madison, WI: University of Wisconsin Press. E lster, J. (1994). The Nature and Scope of Rational Choice Explanation. In M. J. Martin & L. C. McIntyre (Eds.), Readings in Philosophy of Social Science Cambridge, MA: MIT Press. Environmental Council of the States. (2008). About ECOS. Retrieved 1/31/2 008, 2008, from http://www.ecos.org Eulau, H., & Karps, P. D. (1977). The Puzzle of Representation: Specifying Components of Responsiveness. Legislative Studies Quarterly, 2 (3), 233 254. European Information Society. (2012). Rogers Innovation / Diffusion Diagram. Exergia Consultants. (2012). Assessment of the implementation of a European alternative fuels strategy and possible supportive proposals Retrieved. from http://ec.europa.eu/transport/themes/urban/studies/doc/2012 08 cts implementation study.pdf
120 Faiman, D., Raviv, D., & Rosenstreich, R. (2007). Using solar energy to arrest the increasing rate of fossi l fuel consumption: The southwestern states of the USA as case studies. Energy Policy, 35 567 576. Fesler, J., & Kettl, D. F. (1991). The Politics of the Administrative Process Chatham, NJ: Chatham House Publishers, Inc. Field, B., & Field, M. (2005). Environmental Economics (4th ed.). New York, NY: McGraw Hill. Forsyth, T., Tombari, C., & Nelson, M. (2006). Lessons Learned from the US Photovoltaic Industry and Implications for Development of Distributed Small Wind Paper presented at the WINDPOWER Con ference. Foster, J. L. (1978). Regionalism and Innovation in the American States. Journal of Politics, 40 179 188. Frondel, M., & Peters, J. (2005). Biodiesel: A New Oildorado? Rheinisch Westfaelisches Institut fuer Wirtschaftsforschung. Gatersleben, B ., Steg, L., & Vlek, C. (2002). Measurement and Determinants of Environmentally Significant Consumer Behavior. Environment and Behavior, 34 (3), 335 362. Gerber, E. R., & Morton, R. B. (1998). Primary Election Systems and Representation. The Journal of Law Economics & Organization, 14 (2), 21. Gillingham, K., Newell, R., & Palmer, K. (2006). Energy Efficiency Policies: A Retrospective Examination. Annual Review of Environmental Resources, 31 161 192. Gouchoe, S., Everette, V., & Haynes, R. (2002). Case S tudies on the Effectiveness of State Financial Incentives for Renewable Energy [Electronic Version] from http://www.nrel.gov Granovskii, M., Dincer, I., & Rosen, M. A. (2006). Economic and environmental comparison of conventional, hybrid, electric and hydrogen fuel cell vehicles. Journal of Power Sources, 159 1186 1193. Gray, V., & Lowery, D. (2000). Where do Policy Ideas Come From? A Study of Minnesota Legislators and Staffers. Journal of Public Administration Resea rch and Theory, 10 (3), 26. Green, D. P., & Shapiro, I. (1994). Pathologies of Rational Choice Theory: a critique of applications in political science New Haven, CT: Yale University Press.
121 Griffith, R., Huergo, E., Mairesse, J., & Peters, B. (2006). Inno vation and Productivity Across Four European Countries [Electronic Version]. NBER Working Paper Series 34. Retrieved 4/3/2012 from http://www.nber.org/papers/w12722 Guber, D. L. (2003). The Grassroots o f a Green Revolution: Polling America on the Environment Cambridge, MA: The MIT Press. Harmelink, M., Voogt, M., & Cremer, C. (2006). Analysing the effectiveness of renewable energy supporting policies in the European Union. Energy Policy, 34 343 351. Hart, G. (2006). The Shield and the Cloak: The Security of the Commons New York, NY: Oxford University Press. Hatch, M. (1995). Nuclear Power and Postindustrial Politics in the West. In J. Byrne & S. Hoffman (Eds.), Governing the Atom: the Politics of Ris k New Brunswick, NJ: Transaction Publishers. Hoffmann, V. (2011). Knowledge and Innovation Management. Module Reader. [Electronic Version]. Retrieved 4/4/12. Holz, L., Kuczera, G., & Kalma, J. (2006). Multiple Criteria Decision Making: Facilitating a L earning Environment. Journal of Environmental Planning and Management, 49 (3), 455 470. Huber, C., Ryan, L., O'Gallachoir, B., Resch, G., Polaski, K., & Bazilian, M. (2006). Economic modeling of price support mechanisms for renewable energy: Case study on Ireland. Energy Policy, 34 1172 1185. Islas, J. (1997). Getting round the lock in in electricity generating systems: the example of the gas turbine. Research Policy, 26 49 66. Jacobsson, S., & Lauber, V. (2006). The politics and policy of energy system transformation explaining the German diffusion of renewable energy technology. Energy Policy, 34 256 276. Johnson, K. C. (2006a). California's greenhouse gas law, Assembly Bill 1493: Deficiencies, alternatives and implications for regulatory climate p olicy. Energy Policy, 34 362 372. Johnson, K. C. (2006b). Feebates: An effective regulatory instrument for cost constrained environmental policy. Energy Policy, 34 3965 3976. Jones, B. D. (2002). Bounded rationality and public policy: Herbert A Simon a nd the decisional foundatoin of collective choice. Policy Sciences, 35 269 284.
122 Julnes, P. d. L. (2000). Evolution of Government Reform: The Center for Accountability and Performance (CAP) and Support of Managing for Results Paper presented at the V Int ernational Congress of CLAD on State and Public Administration Reform. Kirchstetter, T. W., Singer, B. C., Harley, R. A., Kendall, G. R., & Traverse, M. (1999). Impact of California Reformulated Gasoline on Motor Vehicle Emissions. Environmental Science & Technology, 33 (2), 318 328. Klingner, D. E. (2009). Using U.S. Public Administration to Support Global Development. Journal of Regional Studies and Development, 18 (2), 1 30. Klingner, D. E., Nalbandian, J., & Llorens, J. (2010). Public Personnel Managem ent (6th ed.). Englewood Cliffs: Pearson. Knight, F. (1921). Risk, Uncertainty and Profit New York, NY: Harper Torchbooks. Knott, J., & Wildavsky, A. (1980). If dissemination is the solution, what is the problem? Knowledge: Creation, Diffusion, Utlizatio n, 1 (14), 537 578. Kobos, P. H., Erickson, J. D., & Drennen, T. E. (2006). Technology learning and renewable energy costs: implications for US renewable energy policy. Energy Policy, 34 1645 1658. Laird, F. (2001). Solar Energy, Technology Policy and In stitutional Values New York, NY: Cambridge University Press. Landry, R., Lamari, M., & Amara, N. (2003). The Extent and Determinants of the Utilization of University Research in Government Agencies. Public Administration Review, 63 (2), 13. Landy, C. (20 09). Statistical Report News Relase. Retrieved 8/16/2009, from http://www.api.org/statistics/upload/Monthly_Statistical_Report_News_Release_J une_20 09.pdf Lewis, T. R. (1996). Protecting the Environment when Costs and Benefits are Privately known. RAND Journal of Economics, 27 (4), 819 847. Lindblom, C. E. (1959). The Science of "Muddling Through". Public Administration Review, 19 (2), 79 88. Linden A. L., Carlsson Kanyama, A., & Eriksson, B. (2006). Efficient and inefficient aspects of residential energy behaviour: What are the policy instruments for change? Energy Policy, 34 1918 1927. Loughron, D. S., & Kulick, J. (2004). Demand Side Management and Energy Efficiency in the United States. The Energy Journal, 25 (1), 19 43.
123 Lund, P. (2006). Market penetration rates of new energy technologies. Energy Policy, 34 3317 3326. Lund, P. (2007). Effectiveness of policy measures in transforming the energ y system. Energy Policy, 35 627 639. Matsusaka, J. G. (1995). Fiscal Effects of the Voter Initiative: Evidence from the Last 30 Years. The Journal of Political Economy, 103 (3), 587 623. Mayes, F., Gielecki, M., & Poling, J. (2005). Policies to Promote N on hydro Renewable Energy in the United States and Selected Countries [Electronic Version]. Retrieved 5/8/2007 from http://www.eia.doe.gov/fuelrenewable.html Mead, L. M. (2005). Government Matter s: Welfare Reform in Wisconsin (1st ed.). Princeton, NJ: Princeton University Press. Menz, F., & Vachon, S. (2006). The effectiveness of different policy regimes for promoting wind power: Experience from the States. Energy Policy, 34 1786 1796. Mesler, B. (2001). Springtime for Nuclear. The Nation, 273 (4), 16 20. Miller, G. (1992). Management Dilemmas: The Political Economy of Hierarchy New York, NY: Cambridge University Press. Mintrom, M. (1997). Policy Entrepreneurs and the Diffusion of Innovation. American Journal of Political Science, 41 (3), 738 770. Moran, D., & Sherrington, C. (2006). An economic assessment of windfarm power generation in Scotland including externalities. Energy Policy, 35 (5), 2811 2825. Moynihan, D. P., Fernandez, S., Kim, S. LeRoux, K. M., Piotrowski, S. J., Wright, B. E., et al. (2011). Performance Regimes Amidst Governance Complexity. Journal of Public Administration Research and Theory, 21 15. Nadel, S., & Kushler, M. (2000). Public Benefit Funds: A Key Strategy for Adv ancing Energy Efficiency. Electricity Journal, 13 (8), 74 84. National Energy Policy Group. (2001). Reliable, Affordable and Environmentally Sound Energy for America's Future: Report of the National Energy Policy Group Retrieved 5/7/2007. from http://www.whitehouse.gov/energy/National Energy Policy.pdf Nedergaard, P. (2006). Market failures and government failures: A theoretical model of the common agricultural policy. Public Ch oice, 127 393 413.
124 Nonaka, I. (1991). The Knowledge Creating Company. Harvard Business Review, 69 (6), 96 104. Nonaka, I. (1994). A Dynamic Theory of Organizational Knowledge Creation. Organization Science, 5 (1), 25. North, D. C. (1990). Institutions, I nstitutional Change and Economic Performance New York, NY: Cambridge University Press. North, D. C. (2003). Understanding the Process of Economic Change Paper presented at the Forum Series on the Role of Institutions in Promoting Economic Growth. from http://mercatus.org/sites/default/files/publication/Understanding_the_Process_of_ Economic_Change.pdf Olson, M. (1965). The Logic of Colle ctive Action Cambridge, MA: Harvard University Press. Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action New York NY: Cambridge University Press. Parry, I. W. H., & Bento, A. M. R. (1999). Revenue Recycling and the Welfare Effects of Road Pricing Retrieved. from http://library1.nida.ac.th/worldbankf/fulltext/wps02253.pdf Pearce, D. (1998). Cost Benefit Analysis and Environmental Poli cy. Oxford Review of Economic Policy, 14 (4), 16. Peterson, P. (1995). The Price of Federalism Washington, D.C.: The Brookings Institute. Pierson, P. (2000). Increasing Returns, Path Dependency and the Study of Politics. American Political Science Review 94 (3), 251 267. Pitkin, H. F. (1967). The Concept of Representation Berkeley and Los Angeles, CA: University of California Press. Rabe, B. G. (2006). Power to the States: The Promise and Pitfalls of Decentralization. In N. J. Vig & M. E. Kraft (Eds.), Environmental Policy: New Directions for the Twenty First Century (6th ed.). Washington, D.C.: CQ Press. Ratner, M., Belkin, P., Nichol, J., & Woehrel, S. (2012). Europe's Energy Security: Options and Challenges to Natural Gas Supply Diversification Ret rieved. from http://www.fas.org/sgp/crs/row/R42405.pdf Rogers, E. (2003). Diffusion of Innovation (5th ed.). New York: Simon & Schuster.
125 Roques, F., Nuttall, W., Newbery, D., deNeufville, R., & Connors, S. (2006). Nuclear Power: a hedge against uncertain gas and carbon prices? Energy Policy, 27 (4), 1 24. Rowe, G., & Frewer, L. (2000). Public Participation Models: A Framework for Evaluation. Science, Technology & Human Values, 25 (3), 3 29. Sawin J. L. (2004). National Policy Instruments: Policy Lessons for the Advancement & Diffusion of Renewable Energy Technologies Around the World Paper presented at the Internationale Konferenz fuer Erneuerbare Energien. from http://wofuco.inet.de/fileadmin/user_upload/Miguel/Sawin__2004__National_pol icy_instruments.pdf Schmitz, A., Moss, C. B., Schmitz, T. G., Furtan, H. W., & Schmitz, H. C. (201 0). Agricultural Policy, Agribusiness and Rent Seeking Behaviour (2 ed.). Toronto: University of Toronto Press. Schumacher, K., & Sands, R. D. (2006). Innovative energy technologies and climate policy in Germany. Energy Policy, 34 3929 3941. Schumpeter, J. (1950). Capitalism, Socialism and Democracy (3rd ed.). New York, NY: Harper. Schwartz, H. (2000). Down the Wrong Path: Path Dependency, Increasing Returns and Historical Institutionalism Paper presented at the Beyond Markets Retrieved 12/13/2006, fro m http://www.people.virgina.edu/~hms2f/Path.pdf Scott, W. R. (2003). Organizations: Rational, Natural and Open Systems (5th ed.). Upper Saddle River, NJ: Pearson Prentice Hall. Scott, W. R., & Davis, G. F. (2007). Organizations and Organizing: Rational, Natural and Open System Perspectives Upper Saddle River, NJ: Pearson Prentice Hall. Shanks, J. M., & Strand, D. A. (1994). Unresolved Issues in Electoral Decisions: Alternative Perspectives on the Explanation of Individual Choice. In M. K. Jennings & T. E. Mann (Eds.), Elections at Home and Abroad, Essays in Honor of William E. Miller Ann Arbor, MI: University of Michigan Press. Shiller, R. J. (1984). Stock Prices and Social Dynamics. Brook ings Papers on Economic Activity, 2 457 498. Simon, H. (1997, 1947). Administrative Behavior: a study of decision making processes in administrative organizations (4th ed.). New York, NY: Free Press. Smith, A. (1776). An Inquiry into the Nature and Caus es of the Wealth of Nations Indianapolis, IN: Liberty Fund.
126 Smith, D. A., & Fridkin, D. (2008). Delegating Direct Democracy: Interparty Legislative Competition and the Adoption of the Initiative in the American States. American Political Science Review, 102 (3), 333 350. Solomon, B. D., & Banerjee, A. (2006). A global survey of hydrogen energy research, development and policy. Energy Policy, 34 781 792. Sterner, T. (2007). Fuel Taxes: An important instrument for climate policy. Energy Policy, 35 (6), 319 4 3202. Stillman II, R. J. (1999). Preface to Public Administration: A Search for Themes and Direction (2nd ed.). Burke, VA: Chatelaine Press. Thomassen, J. (1994). Empirical Research into Political Representation: Failing Democracy or Failing Models? In M. K. Jennings & T. E. Mann (Eds.), Elections at Home and Abroad, Essays in Honor of Warren E. Miller Ann Arbor, MI: University of Michigan Press. Thompson, J. D. (1967). Organizations in Action New York, NY: McGraw Hill Book Company. Tietenberg, T. H (1980). Transferable Discharge Permits and the control of Stationary Source Air Pollution: A Survey and Synthesis. Land Economics, 56 (4), 391 416. U.S. Energy Information Administration. (2012). Annual Energy Outlook 2013 Retrieved. from http://www.eia.gov/forecasts/aeo/er/executive_summary.cfm U.S. Energy Information Agency. (2012a). Annual Energy Review Retrieved 11/17/2012. from http://www.eia.gov/totalenergy/data/annual/pdf/sec8_4.pdf U.S. Energy Information Agency. (2012b). Utility scale installations lead solar photovoltaic growth Retrieved. from http://www.eia.gov/todayinenergy/detail.cfm?id=8570 van den Hoed, R., & Vergragt, P. J. (2004). Institutional Change in the Automotive Industry, or how fuel cell technology is being institutionalized. Greener Management International, 47 (Autumn 2004), 45 61. van Rooijen, S. N. M., & van Wees, M. T. (2006). Green electricity policies in the Netherlands: an analysis of policy decisions. Energy Policy, 34 60 71. Veblen, T. (2000, 1899). The Theory of the Leisure Class New Brunswick, NJ : Transaction Publishers. Vlasic, B. (2012, August 29, 2012). U.S. Sets Higher Fuel Efficiency Standards New York Times
127 Walker, J. L. (1969). The Diffusion of Innovation among the American States. American Political Science Review, 63 880 899. Walsh, M. J. (1989). Energy tax credits and housing improvements. Energy Economics, 11 (4), 275 284. Wassell Jr, C. S., & Dittmer, T. P. (2006). Are subsidies for biodiesel economically efficient. Energy Policy, 34 3993 4001. Williamson, O. (1987). The Economi c Institutions of Capitalism: firms, markets, relational contracting New York NY: Collier MacMillan Publishers. Yang, K., & Callahan, K. (2007). Citizen Involvement Efforts and Bureaucratic Responsiveness: Participatory Values, Stakeholder Pressuers and Administrative Practicality. Public Administration Review, 67 (2), 249 264. Young, A. A. (1928). Increasing Returns and Economic Progress. The Economic Journal, 28 (152), 527 542.
128 APPENDIX A REGRESSION RESULTS This appendix presents a select number of the regressions, full data and all results are available upon request. These regressions have been selected as they provide a relevant and fairly complete sampling of the models, without being repetitive or unnecessarily extensive. A summary of the regres sions is provided below. For the 7 energy based regressions, all had 450 observations. The dependent variable is measured in megawatt hours, when reviewing the coefficients. The policy regressions follow, with R 2 readings and observation counts include d. Any blanks in the policy adoption regressions indicate perfect correlation between that variable and adoption of the policy. Levels of significance in the charts are marked as follows: 90%/10% = 95%/5% = ** 99%/1% = *** Dependent Variable Indep endent Variables Regression Total Renewable Energy % of Renewable Solar Wind Solar + Wind Policies State Fixed Effect Year Effects Population Demographics Decoupled Profits Decoupled Either R 2 1 X X X X X X X .985 2 X X X X X X X .989 3 X X X X X X X .981 4 X X X X X .298 5 X X X X X X .845 6 X X X X X X .851 7 X X X X X X .857
129 O LS Regression 1: DV: Total Renewable Energy Generated IVs: Demographics, Years States, Trend, Policies Regression 1 tests all of the variables against total renewable energy it casts a wide net to see what variable impact renewable energy generation. Coefficient Std. Error t value P>|t| Significance Alabama 0.00 Alaska 3760.56 7575.40 0.496 0.622 Arizona 9777.07 3059.22 3.196 0.002 *** Arkansas 1937.76 3733.10 0.519 0.606 California 37551.86 52950.74 0.709 0.482 Colorado 8244.76 2002.68 4.117 0.000 *** Connecticut 2399.42 5048.73 0.475 0.637 Delaw are 449.03 7156.11 0.063 0.950 Florida 43886.36 22304.63 1.968 0.055 Georgia 14071.05 7185.06 1.958 0.056 Hawaii 1365.52 6458.66 0.211 0.833 Idaho 4412.18 5831.26 0.757 0.453 Illinois 30364.21 13904.57 2.184 0.034 ** Indiana 17356. 04 3009.22 5.768 0.000 *** Iowa 6652.77 4032.38 1.650 0.105 Kansas 6985.90 3581.46 1.951 0.057 Kentucky 6048.76 1712.87 3.531 0.001 *** Louisiana 8441.03 1361.11 6.202 0.000 *** Maine 5106.01 5890.65 0.867 0.390 Maryland 11314.42 3028 .99 3.735 0.000 *** Massachusetts 11015.08 4976.42 2.213 0.032 ** Michigan 19783.15 8979.75 2.203 0.032 ** Minnesota 11315.75 3120.79 3.626 0.001 *** Mississippi 6958.15 3379.49 2.059 0.045 ** Missouri 12222.57 1802.72 6.780 0.000 *** Mont ana 4190.98 7558.05 0.555 0.582 Nebraska 5477.86 5582.81 0.981 0.331 Nevada 2558.30 5146.53 0.497 0.621 New Hampshire 154.71 6473.34 0.024 0.981 New Jersey 17509.76 7317.74 2.393 0.021 ** New Mexico 2519.95 5617.21 0.449 0.656 New York 18561.40 25471.83 0.729 0.470 North Carolina 17055.90 6872.32 2.482 0.017 ** North Dakota 584.14 7759.75 0.075 0.940 Ohio 27611.71 11042.20 2.501 0.016 ** Oklahoma 5645.43 2537.70 2.225 0.031 **
130 Coefficient Std. Error t value P>|t | Significance Oregon 27631.78 2365.90 11.679 0.000 *** Pennsylvania 26558.70 13232.39 2.007 0.050 Rhode Island 2052.35 6791.94 0.302 0.764 South Carolina 6714.85 1265.70 5.305 0.000 *** South Dakota 2627.14 7618.68 0.345 0.732 Tennesse e 6534.45 2439.31 2.679 0.010 ** Texas 53434.54 30881.11 1.730 0.090 Utah 4334.97 4575.84 0.947 0.348 Vermont 1120.79 7552.82 0.148 0.883 Virginia 15322.45 4859.74 3.153 0.003 *** Washington 61369.54 2946.90 20.825 0.000 *** West Virg inia 3208.59 5852.65 0.548 0.586 Wisconsin 12557.96 2601.94 4.826 0.000 *** Wyoming 1012.36 7775.35 0.130 0.897 Y2001 0.00 Y2002 1652.45 721.74 2.290 0.026 ** Y2003 1731.83 833.96 2.077 0.043 ** Y2004 1620.16 983.12 1.648 0.106 Y2005 1571.34 1231.51 1.276 0.208 Y2006 1444.24 1604.59 0.900 0.372 Y2007 840.09 1934.30 0.434 0.666 Y2008 2129.05 2248.84 0.947 0.348 Y2009 5182.86 2793.62 1.855 0.070 Population 2.72 1.72 1.580 0.121 Average Wage 0.26 0.22 1.152 0.255 De coupled Gas Profits 218.35 547.35 0.399 0.692 Decoupled Electric Profits 661.80 531.32 1.246 0.219 Democratic Legislature 146.17 772.33 0.189 0.851 0.00 Unemployment Rate 756.49 368.62 2.052 0.046 ** Appliance/Equ ipment Efficiency Standards 1637.42 992.44 1.650 0.105 Construction/Design Standard 816.20 1874.09 0.436 0.665 Contractor Licensing 551.71 603.42 0.914 0.365 Corporate Combo 391.19 927.61 0.422 0.675 Corporate Tax Credit 718.88 530.63 1.355 0.182 Energy Standards for Public Buildings 744.17 608.52 1.223 0.227 Equipment Certification 577.51 1045.82 0.552 0.583 Excise Tax Incentive 962.44 664.45 1.448 0.154 Generation Disclosure 433.92 1121.01 0.387 0.700 Green Building Incent ive 371.95 1110.22 0.335 0.739
131 Coefficient Std. Error t value P>|t| Significance Green Power Purchasing/Aggregation 303.91 546.63 0.556 0.581 Industry Recruitment 229.80 475.41 0.483 0.631 Interconnection 617.01 596.36 1.035 0.306 Leasi ng/Lease Purchase 797.10 690.64 1.154 0.254 Line Extension Analysis 0.00 Mandatory Utility Green Power Option 142.45 874.82 0.163 0.871 Net Metering Rules 1169.28 577.45 2.025 0.048 ** Other Incentive 1517.00 781.19 1.942 0.058 Outreach P rogram 322.78 481.98 0.670 0.506 Other Policy 5011.99 1572.66 3.187 0.003 *** Personal Combo 119.95 390.94 0.307 0.760 Production Incentive 230.24 412.12 0.559 0.579 Property Tax Assessment 367.61 617.61 0.595 0.554 Property Tax Exemption 1527 .62 990.28 1.543 0.129 Public Benefits Fund 538.85 657.09 0.820 0.416 Public Education/Assistance 193.09 551.68 0.350 0.728 Renewables DSM/IRP 2046.55 1089.74 1.878 0.066 Renewables Portfolio Standard 758.76 501.41 1.513 0.137 Renewables S et Aside 720.18 448.72 1.605 0.115 Research & Outreach 312.87 1390.41 0.225 0.823 Sales Tax Exemption 564.83 577.71 0.978 0.333 Sales Tax Refund 3072.04 1359.30 2.260 0.028 ** Solar Practitioner Certification/Accreditation 493.39 1107.74 0. 445 0.658 Solar/Wind Permitting Standards 40.21 515.97 0.078 0.938 Special Incentive 1749.07 2584.23 0.677 0.502 State Bond Program 640.08 741.57 0.863 0.392 State Construction Policy 1578.72 991.04 1.593 0.118 State Combo 970.24 418.92 2. 316 0.025 ** Utility Rebate Program 1219.78 1361.05 0.896 0.375 Solar Access Combined 731.99 640.01 1.144 0.258 Constant 10191.65 11221.97 0.908 0.368
132 OLS Regression 2: DV: % Renewable Energy vs. Total Energy IVs: Demographics, Years, States Trend, Policies Regression 2 looks at the change in percent of renewable energy relative to total energy, and what may drive a change in relative rates of renewable energy usage. There is very little variance year to year in the ratio of renewable to total energy Coefficient Std. Error t value P>|t| Significance Alabama 0.00 Alaska 0.09 0.04 2.268 0.028 ** Arizona 0.04 0.02 1.495 0.141 Arkansas 0.02 0.02 1.192 0.239 California 0.44 0.23 1.942 0.058 Colorado 0.06 0.03 2.157 0.036 ** Connecticut 0.03 0.05 0.653 0.517 Delaware 0.12 0.04 2.958 0.005 *** Florida 0.01 0.09 0.142 0.888 Georgia 0.02 0.04 0.405 0.687 Hawaii 0.04 0.03 1.590 0.118 Idaho 0.73 0.03 22.759 0.000 *** Illinois 0.02 0.07 0.351 0.727 Indiana 0.10 0.02 4.006 0.000 *** Iowa 0.13 0.03 4.374 0.000 *** Kansas 0.11 0.02 4.429 0.000 *** Kentucky 0.06 0.02 3.862 0.000 *** Louisiana 0.04 0.01 3.083 0.003 *** Maine 0.24 0.04 6.779 0.000 *** Maryland 0.04 0.04 1.133 0.263 Massachusetts 0.02 0.06 0.367 0.715 Michigan 0.01 0.05 0.302 0.764 Minnesota 0.07 0.04 1.808 0.077 Mississippi 0.07 0.02 3.725 0.001 *** Missouri 0.07 0.02 3.517 0.001 *** Montana 0.19 0.04 5.262 0.000 *** Nebraska 0.10 0.03 3.594 0.001 *** Nevada 0.04 0.03 1.194 0.238 New Hampshire 0.02 0.04 0.631 0.531 New Jersey 0.05 0.06 0.809 0.423 New Mexico 0.09 0.02 3.742 0.000 *** New York 0.23 0.12 1.828 0.074 North Carolina 0.01 0.03 0.155 0.878 North Dakota 0. 07 0.03 2.210 0.032 ** Ohio 0.02 0.06 0.391 0.698 Oklahoma 0.04 0.02 2.143 0.037 ** Oregon 0.58 0.02 24.226 0.000 *** Pennsylvania 0.02 0.07 0.326 0.746 Rhode Island 0.10 0.03 3.007 0.004 *** South Carolina 0.05 0.01 4.719 0.000 *** S outh Dakota 0.36 0.04 9.743 0.000 *** Tennessee 0.01 0.01 0.416 0.679 Texas 0.05 0.14 0.386 0.701
133 Coefficient Std. Error t value P>|t| Significance Utah 0.08 0.02 3.709 0.001 *** Vermont 0.14 0.03 4.316 0.000 *** Virginia 0.05 0.04 1.140 0. 260 Washington 0.63 0.03 18.373 0.000 *** West Virginia 0.10 0.03 3.756 0.000 *** Wisconsin 0.05 0.02 1.955 0.056 Wyoming 0.08 0.03 2.569 0.013 ** Y2001 0.00 Y2002 0.02 0.01 2.316 0.025 ** Y2003 0.02 0.01 1.791 0.079 Y2004 0.01 0.01 0.789 0.434 Y2005 0.01 0.01 0.564 0.575 Y2006 0.01 0.02 0.591 0.557 Y2007 0.00 0.02 0.021 0.984 Y2008 0.01 0.03 0.396 0.694 Y2009 0.04 0.03 1.467 0.149 Population 0.00 0.00 1.090 0.281 Average Wage 0.00 0.00 0.159 0.874 Deco upled Gas Profits 0.00 0.01 0.063 0.950 Decoupled Electric Profits 0.01 0.01 0.614 0.542 Democratic Legislature 0.01 0.01 0.808 0.423 0.00 Unemployment Rate 0.01 0.00 2.384 0.021 ** Appliance/Equipment Efficie ncy Standards 0.01 0.01 0.836 0.407 Construction/Design Standard 0.00 0.01 0.120 0.905 Contractor Licensing 0.00 0.01 0.377 0.708 Corporate Combo 0.01 0.01 0.353 0.725 Corporate Tax Credit 0.01 0.00 1.634 0.109 Energy Standards for Pu blic Buildings 0.01 0.01 1.031 0.308 Equipment Certification 0.01 0.01 0.952 0.346 Excise Tax Incentive 0.07 0.03 2.172 0.035 ** Generation Disclosure 0.01 0.01 1.439 0.157 Green Building Incentive 0.01 0.01 1.469 0.148 Green Power Purchasin g/Aggregation 0.00 0.01 0.167 0.868 Industry Recruitment 0.00 0.01 0.146 0.885 Interconnection 0.00 0.01 0.119 0.906 Leasing/Lease Purchase 0.00 0.01 0.278 0.782 Line Extension Analysis 0.00 Mandatory Utility Green Power Option 0.02 0 .01 1.892 0.064 Net Metering Rules 0.01 0.01 0.797 0.429 Other Incentive 0.01 0.01 1.266 0.211 Outreach Program 0.00 0.01 0.442 0.661 Other Policy 0.06 0.01 5.886 0.000 *** Personal Combo 0.00 0.01 0.459 0.649 Production Incentive 0.01 0.01 1.016 0.315 Property Tax Assessment 0.01 0.01 0.678 0.501 Property Tax Exemption 0.01 0.01 1.390 0.171
134 Coefficient Std. Error t value P>|t| Significance Public Benefits Fund 0.01 0.01 1.706 0.094 Public Education/Assistance 0.01 0.01 1.207 0.233 Renewables DSM/IRP 0.02 0.02 1.271 0.210 Renewables Portfolio Standard 0.00 0.01 0.418 0.678 Renewables Set Aside 0.01 0.01 1.280 0.207 Research & Outreach 0.01 0.01 1.165 0.250 Sales Tax Exemption 0.01 0.01 0.827 0.412 Sa les Tax Refund 0.02 0.02 1.289 0.203 Solar Practitioner Certification/Accreditation 0.01 0.01 1.239 0.221 Solar/Wind Permitting Standards 0.01 0.01 1.360 0.180 Special Incentive 0.00 0.03 0.142 0.888 State Bond Program 0.00 0.01 0.156 0.877 State Construction Policy 0.00 0.01 0.014 0.989 State Combo 0.02 0.01 2.915 0.005 *** Utility Rebate Program 0.01 0.01 1.216 0.230 Solar Access Combined 0.00 0.01 0.450 0.655 Constant 0.15 0.08 1.905 0.063
135 OLS Regression 3 : DV: Total Renew able Energy Generated IVs: States, Population Regression 3 tests the predictive power of just the states themselves and absorbs most of the variance. ( R 2 .9806) Coefficient Std. Error t value P>|t| Significance Alabama 0.00 Alaska 1636.59 5293.27 0.309 0.758 Arizona 9087.85 1903.89 4.773 0.000 *** Arkansas 3373.87 2420.46 1.394 0.170 California 33049.07 42314.94 0.781 0.439 Colorado 10288.89 180.32 57.058 0.000 *** Connecticut 8372.79 1480.00 5.657 0.000 *** Delaware 3405.44 5058.74 0.673 0.504 Florida 39877.97 17703.12 2.253 0.029 ** Georgia 17386.51 6188.18 2.810 0.007 *** Hawaii 3717.97 4488.69 0.828 0.412 Idaho 4772.78 4256.11 1.121 0.268 Illinois 31166.71 11027.59 2.826 0.007 *** Indiana 16 029.24 2303.66 6.958 0.000 *** Iowa 5237.04 2183.36 2.399 0.020 ** Kansas 7218.72 2465.55 2.928 0.005 *** Kentucky 8236.59 515.57 15.976 0.000 *** Louisiana 8612.12 174.53 49.346 0.000 *** Maine 3107.10 4423.50 0.702 0.486 Maryland 12509. 05 1347.43 9.284 0.000 *** Massachusetts 14741.03 2594.54 5.682 0.000 *** Michigan 22159.86 7432.88 2.981 0.004 *** Minnesota 9841.55 748.61 13.146 0.000 *** Mississippi 7152.56 2266.36 3.156 0.003 *** Missouri 14373.66 1691.05 8.500 0.000 *** Montana 5804.98 4926.00 1.178 0.244 Nebraska 4737.61 3819.17 1.240 0.221 Nevada 3947.48 2948.22 1.339 0.187 New Hampshire 2112.69 4437.40 0.476 0.636 New Jersey 21582.27 5485.30 3.935 0.000 *** New Mexico 5020.64 3595.22 1.396 0 .169 New York 20717.13 20024.88 1.035 0.306 North Carolina 16796.06 5668.52 2.963 0.005 *** North Dakota 815.91 5333.47 0.153 0.879 Ohio 28783.42 9376.71 3.070 0.003 *** Oklahoma 6255.09 1371.21 4.562 0.000 *** Oregon 24780.23 1260.34 19.662 0.000 ***
136 Coefficient Std. Error t value P>|t| Significance Pennsylvania 26834.00 10683.79 2.512 0.015 ** Rhode Island 3884.66 4757.98 0.816 0.418 South Carolina 8246.83 378.09 21.812 0.000 *** South Dakota 386.32 5136.33 0.075 0.940 Tennessee 7072.24 1963.50 3.602 0.001 *** Texas 48476.85 24958.32 1.942 0.058 Utah 6760.72 2775.03 2.436 0.019 ** Vermont 1289.84 5359.66 0.241 0.811 Virginia 16163.98 4044.39 3.997 0.000 *** Washington 59251.38 2347.05 25.245 0.000 *** West Virginia 4275.59 3746.64 1.141 0.259 Wisconsin 11795.13 1317.18 8.955 0.000 *** Wyoming 1072.17 5503.21 0.195 0.846 Population 2.47 1.36 1.822 0.075 Constant 1365.85 6198.68 0.220 0.827
137 OLS Regression 4 : DV: Total Renewable Energy IVs: Demographics Regression 4 checks if the predictive power of the demographics captured and measured are equivalent to the states themselves. They are not. Coefficient Std. Error t value P>|t| Significance Y2001 0.00 Y2002 574. 09 493.29 1.16 0.278 Y2003 496.35 575.77 0.86 0.414 Y2004 437.49 380.65 1.15 0.284 Y2005 317.17 275.02 1.15 0.282 Y2006 1073.61 112.36 9.56 0.000 *** Y2007 409.31 244.16 1.68 0.132 Y2008 1156.29 750.55 1.54 0.162 Y2009 3785.14 2295.6 9 1.65 0.138 Population 0.91 0.05 17.29 0.000 *** Average Wage 0.08 0.04 2.22 0.058 Decoupled Gas Profits 2568.40 2043.61 1.26 0.244 Decoupled Electric Profits 5823.63 3422.90 1.70 0.127 Democratic Legislature 5088.77 839.15 6.06 0.000 *** C 6308.93 349.42 18.06 0.000 *** Unemployment Rate 881.24 507.74 1.74 0.121 Constant 10872.95 2830.09 3.84 0.005 **
138 OLS Regression 5 : DV: Total Solar Energy Generated IVs: Demographics, Policies States This regression is the same as Regression 1, only with a different dependent variable (solar) Coefficient Std. Error t value P>|t| Significance Alabama 0.00 Alaska 97.37 158.00 0.616 0.541 Arizona 178.55 157.60 1.133 0.263 Arkansas 55.35 72.40 0.764 0.448 California 504.76 689.66 0.732 0.468 Colorado 317.02 170.32 1.861 0.069 Connecticut 186.39 163.52 1.140 0.260 Delaware 29.51 113.25 0.261 0.795 Florida 106.94 307.20 0.348 0.729 Georgia 87.58 152.19 0.576 0.568 Hawaii 251.10 203.0 1 1.237 0.222 Idaho 202.36 150.56 1.344 0.185 Illinois 277.20 236.23 1.173 0.246 Indiana 56.67 100.11 0.566 0.574 Iowa 824.78 436.41 1.890 0.065 Kansas 71.88 127.73 0.563 0.576 Kentucky 176.37 154.72 1.140 0.260 Louisiana 8.66 7 7.26 0.112 0.911 Maine 169.62 156.32 1.085 0.283 Maryland 40.00 139.10 0.288 0.775 Massachusetts 173.49 189.71 0.914 0.365 Michigan 87.33 151.71 0.576 0.568 Minnesota 691.19 394.67 1.751 0.086 Mississippi 2.54 56.13 0.045 0.964 Missouri 154.70 153.86 1.005 0.320 Montana 714.15 455.64 1.567 0.123 Nebraska 264.28 199.41 1.325 0.191 Nevada 0.05 149.44 0.000 1.000 New Hampshire 30.93 116.73 0.265 0.792 New Jersey 80.54 168.54 0.478 0.635 New Mexico 384.42 378.59 1.015 0.315 New York 112.24 367.87 0.305 0.762 North Carolina 72.24 117.84 0.613 0.543 North Dakota 55.04 169.81 0.324 0.747 Ohio 81.89 221.63 0.369 0.713 Oklahoma 99.64 77.95 1.278 0.207 Oregon 71.82 125.50 0.572 0.570 Pennsylvania 124.04 192.21 0.645 0.522 Rhode Island 0.11 144.17 0.001 0.999 South Carolina 27.58 40.32 0.684 0.497 South Dakota 137.17 113.66 1.207 0.233 Tennessee 47.94 124.03 0.387 0.701 Texas 172.09 438.45 0.393 0.696 Utah 135 .73 135.62 1.001 0.322 Vermont 129.43 103.92 1.245 0.219
139 Coefficient Std. Error t value P>|t| Significance Virginia 9.28 161.99 0.057 0.955 Washington 810.98 410.16 1.977 0.054 West Virginia 215.16 130.31 1.651 0.105 Wisconsin 79. 52 150.15 0.530 0.599 Wyoming 18.85 106.89 0.176 0.861 Y2001 0.00 Y2002 19.46 14.10 1.380 0.174 Y2003 65.72 38.87 1.691 0.097 Y2004 91.06 54.18 1.681 0.099 Y2005 91.95 58.22 1.579 0.121 Y2006 77.59 61.16 1.269 0.211 Y2007 95.01 77.59 1.225 0.227 Y2008 84.57 81.87 1.033 0.307 Y2009 124.00 100.74 1.231 0.224 Population 0.00 0.02 0.207 0.837 Average Wage 0.00 0.01 0.641 0.524 Decoupled Gas Profits 134.82 117.61 1.146 0.257 Decoupled Electric Profi ts 54.85 53.25 1.030 0.308 Democratic Legislature 27.57 18.14 1.520 0.135 0.00 Unemployment Rate 5.76 8.03 0.717 0.477 Appliance/Equipment Efficiency Standards 65.23 49.31 1.323 0.192 Construction/Design Standard 35.66 36.23 0.984 0.330 Contractor Licensing 26.46 23.71 1.116 0.270 Corporate Combo 57.70 40.38 1.429 0.159 Corporate Tax Credit 27.13 25.53 1.063 0.293 Energy Standards for Public Buildings 44.90 30.04 1.495 0.141 Equipment Certif ication 37.88 42.64 0.888 0.379 Excise Tax Incentive 34.87 31.73 1.099 0.277 Generation Disclosure 54.53 43.63 1.250 0.217 Green Building Incentive 57.12 38.81 1.472 0.147 Green Power Purchasing/Aggregation 22.27 18.63 1.195 0.238 Indust ry Recruitment 34.48 35.19 0.980 0.332 Interconnection 37.98 20.98 1.810 0.076 Leasing/Lease Purchase 16.73 20.52 0.815 0.419 Line Extension Analysis 0.00 Mandatory Utility Green Power Option 750.79 498.39 1.506 0.138 Net Metering Ru les 26.06 26.10 0.998 0.323 Other Incentive 51.62 41.75 1.236 0.222 Outreach Program 5.22 11.82 0.441 0.661 Other Policy 81.04 41.08 1.973 0.054 Personal Combo 8.16 26.80 0.305 0.762 Production Incentive 12.75 19.62 0.650 0.519 Pr operty Tax Assessment 136.05 80.23 1.696 0.096 Property Tax Exemption 96.65 58.79 1.644 0.107 Public Benefits Fund 8.43 25.98 0.324 0.747 Public Education/Assistance 4.37 26.25 0.166 0.869
140 Coefficient Std. Error t value P>|t| Significance Renewables DSM/IRP 8.18 47.16 0.173 0.863 Renewables Portfolio Standard 17.42 17.63 0.988 0.328 Renewables Set Aside 50.71 41.99 1.208 0.233 Research & Outreach 7.64 26.42 0.289 0.774 Sales Tax Exemption 52.78 37.32 1.414 0.164 Sales T ax Refund 52.11 59.94 0.869 0.389 Solar Practitioner Certification/Accreditation 30.20 42.99 0.703 0.486 Solar/Wind Permitting Standards 175.66 156.04 1.126 0.266 Special Incentive 26.32 76.72 0.343 0.733 State Bond Program 126.56 56.79 2.229 0.030 ** State Construction Policy 92.31 53.80 1.716 0.093 State Combo 21.74 19.27 1.128 0.265 Utility Rebate Program 835.11 293.54 2.845 0.006 *** Solar Access Combined 134.90 122.51 1.101 0.276 Constant 143.28 248.82 0.576 0.567
141 OLS Regression 6 : DV: Total Wind Energy Generated IVs: Demographics, Policies States This regression is the same as Regression 1, only with a different dependent variable (wind) Coefficient Std. Error t value P>|t| Significance Alabama 0 .00 Alaska 13855.67 5824.86 2.379 0.021 ** Arizona 3876.25 2116.24 1.832 0.073 Arkansas 5810.24 2622.59 2.215 0.031 ** California 88927.20 40917.77 2.173 0.035 ** Colorado 1446.69 804.14 1.799 0.078 Connecticut 5852.64 2459.27 2.380 0 .021 ** Delaware 12573.40 5204.43 2.416 0.019 ** Florida 39644.71 17250.80 2.298 0.026 ** Georgia 11618.44 5494.66 2.114 0.040 ** Hawaii 10916.44 4506.32 2.422 0.019 ** Idaho 9541.62 4211.73 2.265 0.028 ** Illinois 22590.53 10562.13 2.139 0.037 ** Indiana 3604.67 1859.00 1.939 0.058 Iowa 7671.83 2558.31 2.999 0.004 *** Kansas 6245.98 2342.68 2.666 0.010 ** Kentucky 2878.50 1120.86 2.568 0.013 ** Louisiana 730.03 700.18 1.043 0.302 Maine 9907.60 4284.38 2.312 0.025 ** Maryland 1414. 48 1270.36 1.113 0.271 Massachusetts 3917.15 2396.04 1.635 0.108 Michigan 14539.36 6841.28 2.125 0.039 ** Minnesota 939.23 1149.81 0.817 0.418 Mississippi 5763.86 2405.40 2.396 0.020 ** Missouri 1874.49 1191.94 1.573 0.122 Montana 1241 4.54 5114.61 2.427 0.019 ** Nebraska 9345.69 3908.47 2.391 0.021 ** Nevada 8854.65 3512.12 2.521 0.015 ** New Hampshire 11353.78 4652.72 2.440 0.018 ** New Jersey 10635.99 5016.55 2.120 0.039 ** New Mexico 10475.50 3912.60 2.677 0.010 ** New York 42691.06 19290.24 2.213 0.032 ** North Carolina 11580.89 5249.62 2.206 0.032 ** North Dakota 14316.28 5455.42 2.624 0.012 ** Ohio 18498.58 8652.99 2.138 0.038 ** Oklahoma 4752.19 1699.60 2.796 0.007 *** Oregon 5301.47 1634.35 3.244 0.002 *** Pen nsylvania 22015.39 10252.62 2.147 0.037 ** Rhode Island 11960.55 4976.54 2.403 0.020 ** South Carolina 2376.04 791.78 3.001 0.004 *** South Dakota 12348.93 5172.65 2.387 0.021 ** Tennessee 3579.50 1735.69 2.062 0.045 ** Texas 47359.44 24012.20 1 .972 0.054 Utah 7974.51 3330.25 2.395 0.021 ** Vermont 13411.26 5390.14 2.488 0.016 **
142 Coefficient Std. Error t value P>|t| Significance Virginia 7574.33 3723.08 2.034 0.047 ** Washington 1911.67 1892.90 1.010 0.317 West Virginia 10189.07 3 934.74 2.590 0.013 ** Wisconsin 2034.65 1419.78 1.433 0.158 Wyoming 14776.83 5772.37 2.560 0.014 ** Y2001 0.00 Y2002 222.63 151.88 1.466 0.149 Y2003 121.85 245.24 0.497 0.622 Y2004 47.61 313.89 0.152 0.880 Y2005 95.84 380.51 0.25 2 0.802 Y2006 217.01 512.88 0.423 0.674 Y2007 235.51 625.60 0.376 0.708 Y2008 444.96 679.48 0.655 0.516 Y2009 1771.42 902.25 1.963 0.055 Population 3.05 1.32 2.307 0.025 ** Average Wage 0.02 0.07 0.325 0.746 Decoupled Gas Profits 17.42 356.78 0.049 0.961 Decoupled Electric Profits 267.12 217.51 1.228 0.225 Democratic Legislature 263.14 205.50 1.281 0.206 0.00 Unemployment Rate 345.94 137.92 2.508 0.015 ** Appliance/Equipment Efficiency S tandards 182.45 289.47 0.630 0.531 Construction/Design Standard 1341.59 585.66 2.291 0.026 ** Contractor Licensing 176.01 301.30 0.584 0.562 Corporate Combo 369.22 449.12 0.822 0.415 Corporate Tax Credit 585.39 285.19 2.053 0.045 ** Ener gy Standards for Public Buildings 280.64 180.12 1.558 0.126 Equipment Certification 941.45 626.96 1.502 0.140 Excise Tax Incentive 905.88 1019.11 0.889 0.378 Generation Disclosure 363.93 531.92 0.684 0.497 Green Building Incentive 481.16 3 81.83 1.260 0.214 Green Power Purchasing/Aggregation 82.91 204.42 0.406 0.687 Industry Recruitment 138.82 266.52 0.521 0.605 Interconnection 242.48 177.34 1.367 0.178 Leasing/Lease Purchase 91.13 229.44 0.397 0.693 Line Extension Analysis 0.00 Mandatory Utility Green Power Option 145.93 407.70 0.358 0.722 Net Metering Rules 620.83 251.10 2.472 0.017 ** Other Incentive 1445.42 430.12 3.361 0.002 *** Outreach Program 305.77 170.37 1.795 0.079 Other Policy 2750.45 417.68 6.585 0.000 *** Personal Combo 388.24 234.49 1.656 0.104 Production Incentive 246.00 206.89 1.189 0.240 Property Tax Assessment 193.16 245.68 0.786 0.436 Property Tax Exemption 71.28 250.68 0.284 0.777 Public Benefits Fund 200.18 392.27 0.51 0 0.612 Public Education/Assistance 38.75 237.75 0.163 0.871
143 Coefficient Std. Error t value P>|t| Significance Renewables DSM/IRP 575.60 518.18 1.111 0.272 Renewables Portfolio Standard 13.40 183.65 0.073 0.942 Renewables Set Aside 536.48 3 28.80 1.632 0.109 Research & Outreach 1253.20 771.97 1.623 0.111 Sales Tax Exemption 509.15 342.30 1.487 0.143 Sales Tax Refund 2937.69 829.93 3.540 0.001 *** Solar Practitioner Certification/Accreditation 29.31 368.17 0.080 0.937 Solar/Win d Permitting Standards 195.75 309.22 0.633 0.530 Special Incentive 1951.93 1304.57 1.496 0.141 State Bond Program 239.78 385.88 0.621 0.537 State Construction Policy 347.11 343.78 1.010 0.318 State Combo 581.46 261.39 2.224 0.031 ** Utility Rebate Program 273.20 596.16 0.458 0.649 Solar Access Combined 146.97 323.68 0.454 0.652 Constant 12840.76 5669.63 2.265 0.028 **
144 OLS Regression 7 : DV: Total Solar + Wind Energy Generated IVs: Demographics, Policies States This regressi on is the same as Regression 1, only with a different dependent variable (solar and wind) Coefficient Std. Error t value P>|t| Significance Alabama 0.00 Alaska 13953.04 5799.27 2.406 0.020 ** Arizona 3697.70 2102.18 1.759 0.085 Arkansas 5865.58 2601.19 2.255 0.029 ** California 88422.45 40678.28 2.174 0.035 ** Colorado 1763.72 815.52 2.163 0.035 ** Connecticut 5666.24 2462.03 2.301 0.026 ** Delaware 12543.88 5180.18 2.422 0.019 ** Florida 39537.78 17149.59 2.305 0.025 ** Georgia 11530.86 5466.77 2.109 0.040 ** Hawaii 11167.54 4476.17 2.495 0.016 ** Idaho 9743.98 4193.11 2.324 0.024 ** Illinois 22867.74 10502.65 2.177 0.034 ** Indiana 3661.35 1844.78 1.985 0.053 Iowa 6847.05 2506.80 2.731 0.009 *** Kansas 6317.86 233 5.09 2.706 0.009 *** Kentucky 3054.87 1121.94 2.723 0.009 *** Louisiana 721.38 668.73 1.079 0.286 Maine 10077.23 4252.76 2.370 0.022 ** Maryland 1454.48 1278.16 1.138 0.261 Massachusetts 4090.63 2396.35 1.707 0.094 Michigan 14626.69 6803.7 2 2.150 0.037 ** Minnesota 248.05 1262.83 0.196 0.845 Mississippi 5766.40 2394.94 2.408 0.020 ** Missouri 1719.79 1192.95 1.442 0.156 Montana 11700.38 5078.16 2.304 0.026 ** Nebraska 9609.97 3885.02 2.474 0.017 ** Nevada 8854.60 3502.81 2.528 0.015 ** New Hampshire 11322.86 4631.30 2.445 0.018 ** New Jersey 10716.53 4987.01 2.149 0.037 ** New Mexico 10091.09 3893.70 2.592 0.013 ** New York 42803.30 19180.31 2.232 0.030 ** North Carolina 11653.13 5223.79 2.231 0.030 ** North Dakota 1 4261.23 5425.65 2.628 0.011 ** Ohio 18580.46 8603.80 2.160 0.036 ** Oklahoma 4851.82 1685.76 2.878 0.006 *** Oregon 5373.30 1621.79 3.313 0.002 *** Pennsylvania 22139.43 10194.76 2.172 0.035 ** Rhode Island 11960.44 4960.28 2.411 0.020 ** South C arolina 2403.62 788.14 3.050 0.004 *** South Dakota 12211.77 5142.68 2.375 0.022 ** Tennessee 3531.56 1727.18 2.045 0.046 ** Texas 47531.54 23866.73 1.992 0.052 Utah 8110.23 3313.44 2.448 0.018 ** Vermont 13281.82 5359.79 2.478 0.017 **
145 Coeff icient Std. Error t value P>|t| Significance Virginia 7583.62 3703.08 2.048 0.046 ** Washington 2722.65 1904.54 1.430 0.159 West Virginia 9973.92 3910.20 2.551 0.014 ** Wisconsin 1955.12 1413.45 1.383 0.173 Wyoming 14795.68 5736.28 2.579 0.0 13 ** Y2001 0.00 Y2002 203.17 150.09 1.354 0.182 Y2003 56.12 246.65 0.228 0.821 Y2004 43.46 313.19 0.139 0.890 Y2005 187.78 379.47 0.495 0.623 Y2006 294.60 513.79 0.573 0.569 Y2007 330.52 632.41 0.523 0.604 Y2008 360.39 691.06 0.522 0.604 Y2009 1647.42 915.23 1.800 0.078 Population 3.06 1.32 2.324 0.024 ** Average Wage 0.02 0.07 0.249 0.804 Decoupled Gas Profits 117.39 367.21 0.320 0.751 Decoupled Electric Profits 212.27 228.66 0.928 0.358 Democratic L egislature 290.71 210.32 1.382 0.173 0.00 Unemployment Rate 340.17 137.04 2.482 0.017 ** Appliance/Equipment Efficiency Standards 117.21 299.10 0.392 0.697 Construction/Design Standard 1377.25 600.30 2.294 0.026 ** Contractor Licensing 202.47 298.99 0.677 0.501 Corporate Combo 311.51 449.18 0.694 0.491 Corporate Tax Credit 612.53 284.49 2.153 0.036 ** Energy Standards for Public Buildings 325.54 181.54 1.793 0.079 Equipment Certification 903.58 602 .70 1.499 0.140 Excise Tax Incentive 940.75 1004.09 0.937 0.353 Generation Disclosure 309.41 517.53 0.598 0.553 Green Building Incentive 538.28 372.83 1.444 0.155 Green Power Purchasing/Aggregation 105.17 202.01 0.521 0.605 Industry Recr uitment 104.33 265.54 0.393 0.696 Interconnection 280.46 175.71 1.596 0.117 Leasing/Lease Purchase 74.40 231.06 0.322 0.749 Line Extension Analysis 0.00 Mandatory Utility Green Power Option 604.85 435.89 1.388 0.172 Net Metering Rule s 646.89 253.14 2.555 0.014 ** Other Incentive 1393.81 424.11 3.286 0.002 *** Outreach Program 300.55 170.15 1.766 0.084 Other Policy 2669.41 407.21 6.555 0.000 *** Personal Combo 396.40 228.34 1.736 0.089 Production Incentive 258.75 211.13 1.22 6 0.226 Property Tax Assessment 57.12 255.23 0.224 0.824 Property Tax Exemption 167.93 266.08 0.631 0.531 Public Benefits Fund 191.76 399.62 0.480 0.633 Public Education/Assistance 34.38 247.55 0.139 0.890
146 Coefficient Std. Error t value P>|t| Significance Renewables DSM/IRP 583.77 491.23 1.188 0.240 Renewables Portfolio Standard 4.02 186.82 0.022 0.983 Renewables Set Aside 587.19 325.94 1.802 0.078 Research & Outreach 1260.84 764.17 1.650 0.105 Sales Tax Exemption 456.37 346.32 1.318 0.194 Sales Tax Refund 2885.59 806.36 3.579 0.001 *** Solar Practitioner Certification/Accreditation 0.90 367.72 0.002 0.998 Solar/Wind Permitting Standards 20.09 335.53 0.060 0.952 Special Incentive 1925.60 1286.82 1.496 0.141 State Bond Program 113.22 386.15 0.293 0.771 State Construction Policy 254.80 339.90 0.750 0.457 State Combo 559.72 256.33 2.184 0.034 ** Utility Rebate Program 1108.31 608.88 1.820 0.075 Solar Access Combined 281.87 326.15 0.864 0.392 C onstant 12984.03 5645.10 2.300 0.026 **
Policy Determinant Probit regression results: DV in bold, IVs listed with coefficients 2009 (<10%) 147 Appliance Equipment Efficiency/Certification n=450 R 2 =.7095 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 3.395663 0.581183 5.84 0.000 *** Population 0.000012 0.00002 6 0.44 0.657 Average wage 0.000103 0.000035 2.9 0.004 *** Unemployment rate 0.086583 0.092861 0.93 0.351 Decoupled profit 1.243255 0.363185 3.42 0.001 *** Democratic Legislature 0.540275 0.352740 1.53 0.126 Citizens' Initiative 0.574756 0.366542 1.57 0.117 Trend 0.066805 0.121738 0.55 0.583 Constant 7.254288 1.561001 4.65 0.000 *** Construction Design Standards n=423 R 2 =.4093 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 100% Pop ulation 0.000117 0.000035 3.29 0.001 *** Average wage 0.000033 0.000045 0.73 0.467 Unemployment rate 0.094519 0.153972 0.61 0.539 Decoupled profit 0.595488 1.036128 0.57 0.565 Democratic Legislature 1.229456 0.590583 2.08 0.037 ** Citizens' Ini tiative 1.316244 0.840807 1.57 0.117 Trend 0.107435 0.100978 1.06 0.287 Constant 2.269320 1.801974 1.26 0.208 Contractor Licensing n=450 R 2 =.2343 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neigh bor 1.255681 0.303058 4.14 0.000 *** Population 0.000005 0.000014 0.32 0.75 Average wage 0.000031 0.000014 2.22 0.027 ** Unemployment rate 0.051514 0.046458 1.11 0.268 Decoupled profit 0.109354 0.223123 0.49 0.624 Democratic Legislature 0.651563 0.157522 4.14 0.000 *** Citizens' Initiative 1.138016 0.164358 6.92 0.000 *** Trend 0.077497 0.034724 2.23 0.026 ** Constant 2.907773 0.498462 5.83 0.000 ***
Policy Determinant Probit regression results: DV in bold, IVs listed with coefficients 2009 (<10%) 148 Corporate Incentives n=450 R 2 =.1389 Coefficient Std. Error Z val ue P > |Z| Significance Policy Enacted in Neighbor 0.173483 0.420644 0.41 0.68 Population 0.000072 0.000013 5.48 0.000 *** Average wage 0.000028 0.000015 1.84 0.066 Unemployment rate 0.150654 0.053170 2.83 0.005 *** Decoupled profit 0.469853 0.260284 1.81 0.071 Democratic Legislature 0.066176 0.166605 0.4 0.691 Citizens' Initiative 0.626623 0.157613 3.98 0.000 *** Trend 0.129677 0.037419 3.47 0.001 *** Constant 0.478035 0.563263 0.85 0.396 Corporate Tax Credit n= 450 R 2 =.0710 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 0.046126 0.245974 0.19 0.851 Population 0.000024 0.000012 1.93 0.054 Average wage 0.000021 0.000013 1.64 0.102 Unemployment rate 0.130301 0.041022 3.18 0.001 *** Decoupled profit 0.787822 0.199374 3.95 0.000 *** Democratic Legislature 0.031595 0.132188 0.24 0.811 Citizens' Initiative 0.167799 0.127318 1.32 0.188 Trend 0.106051 0.031749 3.34 0.001 *** Constant 0.593019 0.458136 1.29 0.196 Energy Standards for Public Buildings n=450 R 2 =.5586 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 0.902946 0.364614 2.48 0.013 ** Population 0.000037 0.000018 2.1 0.036 ** Average wage 0.000031 0.00001 8 1.69 0.091 Unemployment rate 0.064159 0.058386 1.1 0.272 Decoupled profit 0.038960 0.256845 0.15 0.879 Democratic Legislature 0.161512 0.192625 0.84 0.402 Citizens' Initiative 0.011818 0.188912 0.06 0.95 Trend 0.513880 0.091371 5.62 0.00 0 *** Constant 5.102533 0.798681 6.39 0.000 ***
Policy Determinant Probit regression results: DV in bold, IVs listed with coefficients 2009 (<10%) 14 9 Equipment Certification n=450 R 2 =.1400 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 0.871317 0.259174 3.36 0.001 *** Population 0.000044 0.000013 3.46 0.001 *** Average wage 0.000069 0.000018 3.74 0.000 *** Unemployment rate 0.019842 0.046857 0.42 0.672 Decoupled profit 0.054455 0.226200 0.24 0.81 Democratic Legislature 0.160010 0.157401 1.02 0.309 Citizens' Initiative 0.451190 0.153780 2.93 0.003 ** Trend 0.072510 0.035361 2.05 0.04 ** Constant 0.708108 0.603338 1.17 0.241 Excise Tax Incentives n=337 R 2 =.2385 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 100% Population 0.000099 0.00 0107 0.92 0.357 Average wage 0.000129 0.000066 1.94 0.052 Unemployment rate 0.138825 0.114496 1.21 0.225 Decoupled profit 100 Democratic Legislature 0.630319 0.319437 1.97 0.048 ** Citizens' Initiative 0.141006 0.347298 0.41 0.685 Tre nd 0.293185 0.105244 2.79 0.005 *** Constant 1.944601 1.962647 0.99 0.322 Generation Disclosure n=450 R 2 =.3536 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 1.121446 0.234504 4.78 0.000 *** Popula tion 0.000080 0.000021 3.88 0.000 *** Average wage 0.000162 0.000022 7.53 0.000 *** Unemployment rate 0.060425 0.049922 1.21 0.226 Decoupled profit 0.223880 0.250058 0.9 0.371 Democratic Legislature 0.839223 0.169614 4.95 0.000 *** Citizens' Ini tiative 0.961637 0.170410 5.64 0.000 *** Trend 0.166547 0.037857 4.4 0.000 *** Constant 6.319881 0.697657 9.06 0.000 ***
Policy Determinant Probit regression results: DV in bold, IVs listed with coefficients 2009 (<10%) 150 Green Building Incentives n=108 R 2 =.2499 Coefficient Std. Error Z value P > |Z| Significance Policy E nacted in Neighbor Population 0.000127 0.000087 1.47 0.143 Average wage 0.000099 0.000061 1.63 0.103 Unemployment rate 0.257723 0.174487 1.48 0.14 Decoupled profit 0.208494 0.643113 0.32 0.746 Democratic Legislature Citizens' Initia tive Trend 0.563846 0.255029 2.21 0.027 ** Constant 0.655896 2.125223 0.31 0.758 Green Power Purchasing Aggregation n=450 R 2 =.2465 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 0.466798 0.3254 97 1.43 0.152 Population 0.000022 0.000014 1.6 0.109 Average wage 0.000079 0.000018 4.42 0.000 *** Unemployment rate 0.074655 0.047977 1.56 0.12 Decoupled profit 0.407091 0.214049 1.9 0.057 Democratic Legislature 0.290869 0.158396 1.84 0.066 Citizens' Initiative 0.381663 0.172462 2.21 0.027 ** Trend 0.015917 0.038279 0.42 0.678 Constant 4.453467 0.633554 7.03 0.000 *** Industry Recruitment n=450 R 2 =.1817 Coefficient Std. Error Z value P > |Z| Significance Policy En acted in Neighbor 0.159282 0.381115 0.42 0.676 Population 0.000099 0.000016 6.05 0.000 *** Average wage 0.000006 0.000014 0.45 0.655 Unemployment rate 0.006671 0.046532 0.14 0.886 Decoupled profit 0.048723 0.230476 0.21 0.833 Democratic Legislat ure 0.477847 0.155164 3.08 0.002 *** Citizens' Initiative 0.446910 0.154935 2.88 0.004 *** Trend 0.053399 0.034812 1.53 0.125 Constant 2.282610 0.511914 4.46 0.000 ***
Policy Determinant Probit regression results: DV in bold, IVs listed with coefficients 2009 (<10%) 151 Interconnection n=450 R 2 =.2369 Coefficient Std. Error Z v alue P > |Z| Significance Policy Enacted in Neighbor 0.044836 0.244357 0.18 0.854 Population 0.000024 0.000015 1.65 0.1 Average wage 0.000062 0.000015 4.06 0.000 *** Unemployment rate 0.052476 0.045984 1.14 0.254 Decoupled profit 0.678673 0.2491 94 2.72 0.006 *** Democratic Legislature 0.194848 0.145929 1.34 0.182 Citizens' Initiative 0.520466 0.140489 3.7 0.000 *** Trend 0.158120 0.036640 4.32 0.000 *** Constant 2.422093 0.522962 4.63 0.000 *** Leasing / Purchase agreeme nts n=179 R 2 =.6839 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor Population 0.000328 0.000333 0.98 0.326 Average wage 0.001096 0.000687 1.6 0.111 Unemployment rate 0.305324 0.488860 0.62 0.532 Decoupl ed profit Democratic Legislature 1.223322 1.366212 0.9 0.371 Citizens' Initiative Trend 1.627627 0.931179 1.75 0.08 Constant 19.837830 15.469690 1.28 0.2 Line Extension Analysis n=450 R 2 =.6267 Coefficient Std. Erro r Z value P > |Z| Significance Policy Enacted in Neighbor 12.468410 2.329699 5.35 0.000 *** Population 0.000004 0.000020 0.18 0.857 Average wage 0.000266 0.000063 4.21 0.000 *** Unemployment rate 0.193736 0.114549 1.69 0.091 Decoupled profit 2.0 75536 0.599030 3.46 0.001 *** Democratic Legislature 0.633819 0.344267 1.84 0.066 Citizens' Initiative 0.631144 0.401510 1.57 0.116 Trend 0.276012 0.097296 2.84 0.005 *** Constant 13.70788 2.874558 4.77 0.000 ***
Policy Determinant Probit regression results: DV in bold, IVs listed with coefficients 2009 (<10%) 152 Net Mete ring n=450 R 2 =.1166 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 0.410721 0.245502 1.67 0.094 Population 0.000010 0.000014 0.68 0.494 Average wage 0.000061 0.000017 3.69 0.000 *** Unemployment rate 0.123221 0.043727 2.82 0.005 *** Decoupled profit 0.845688 0.268105 3.15 0.002 *** Democratic Legislature 0.120091 0.146386 0.82 0.412 Citizens' Initiative 0.420716 0.142227 2.96 0.003 *** Trend 0.009057 0.033600 0.27 0.787 Constant 0.979513 0.53923 7 1.82 0.069 Research Outreach Program n=450 R 2 =.6189 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 3.486726 0.404389 8.62 0.000 *** Population 0.000002 0.000028 0.06 0.948 Average wage 0.000009 0.000032 0.29 0.77 Unemployment rate 0.167636 0.124197 1.35 0.177 Decoupled profit 0.916374 0.791451 1.16 0.247 Democratic Legislature 0.502417 0.331541 1.52 0.13 Citizens' Initiative 0.183882 0.295182 0.62 0.533 Trend 0.168833 0.098659 1.71 0.087 Constant 1.874367 1.093133 1.71 0.086 Personal Incentives n=450 R 2 =.0795 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 0.674346 0.228357 2.95 0.003 *** Population 0.000029 0.00001 2 2.34 0.019 ** Average wage 0.000039 0.000013 3.06 0.002 *** Unemployment rate 0.053957 0.040183 1.34 0.179 Decoupled profit 0.300908 0.190779 1.58 0.115 Democratic Legislature 0.536806 0.131744 4.07 0.000 *** Citizens' Initiative 0.260146 0. 126965 2.05 0.04 ** Trend 0.120572 0.030814 3.91 0.000 *** Constant 0.824944 0.448256 1.84 0.066
Policy Determinant Probit regression results: DV in bold, IVs listed with coefficients 2009 (<10%) 153 Production Incentives n=450 R 2 =.1149 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 0.944233 0.260 496 3.62 0.000 *** Population 0.000027 0.000012 2.3 0.022 ** Average wage 0.000006 0.000014 0.4 0.692 Unemployment rate 0.012108 0.042995 0.28 0.778 Decoupled profit 0.376984 0.216317 1.74 0.081 Democratic Legislature 0.088227 0.148086 0.6 0.551 Citizens' Initiative 0.444163 0.142849 3.11 0.002 *** Trend 0.123593 0.036546 3.38 0.001 *** Constant 1.674256 0.504328 3.32 0.001 *** Property Tax Assessment n=450 R 2 =.1361 Coefficient Std. Error Z value P > |Z| Signifi cance Policy Enacted in Neighbor 0.553006 0.524682 1.05 0.292 Population 0.000012 0.000024 0.52 0.606 Average wage 0.000060 0.000022 2.73 0.006 *** Unemployment rate 0.109536 0.057160 1.92 0.055 Decoupled profit 0.451569 0.308925 1.46 0. 144 Democratic Legislature 0.071699 0.197787 0.36 0.717 Citizens' Initiative 0.090225 0.181426 0.5 0.619 Trend 0.196462 0.048250 4.07 0.000 *** Constant 0.602835 0.694738 0.87 0.386 Property Tax Exemption n=450 R 2 =.2081 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 1.712006 0.243440 7.03 0.000 *** Population 0.000044 0.000016 2.69 0.007 *** Average wage 0.000040 0.000015 2.7 0.007 *** Unemployment rate 0.049349 0.043324 1.14 0.255 Decoupled profit 0.794338 0.227653 3.49 0.000 *** Democratic Legislature 0.248701 0.145452 1.71 0.087 Citizens' Initiative 0.472773 0.139289 3.39 0.001 *** Trend 0.028921 0.031668 0.91 0.361 Constant 1.903300 0.511751 3.72 0.000 ***
Policy Determinant Probit regression results: DV in bold, IVs listed with coefficients 2009 (<10%) 154 Public Benefit Funds n=450 R 2 =.2614 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 1.766994 0.263035 6.72 0.000 *** Population 0.000020 0.000014 1.46 0.144 Average wage 0.000053 0.000017 3.07 0.002 *** Unemployment rate 0.051914 0.045998 1.13 0.259 Decoupled profit 0.140498 0.212984 0.66 0.509 Democratic Legislature 0.458430 0.149267 3.07 0.002 *** Citizens' Initiative 0.204509 0.145798 1.4 0.161 Trend 0.066913 0.034502 1.94 0.052 Cons tant 3.325098 0.546233 6.09 0.000 *** Public Education / Assistance n=383 R 2 =.3341 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 0.695929 1.717322 0.41 0.685 Population 0.000043 0.000050 0.86 0.388 Average wage 0.000100 0.000046 2.17 0.03 ** Unemployment rate 0.134570 0.236173 0.57 0.569 Decoupled profit Democratic Legislature 1.014145 0.647432 1.57 0.117 Citizens' Initiative 0.242867 0.474308 0.51 0.609 Trend 0.55005 1 0.211412 2.6 0.009 *** Constant 3.013483 1.660826 1.81 0.07 Renewable Demand Side Management n=273 R 2 =.3482 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 1.258111 1.537634 0.82 0.413 Popu lation 0.000152 0.000031 4.92 0.000 *** Average wage 0.000173 0.000055 3.18 0.001 *** Unemployment rate 0.071633 0.097312 0.74 0.462 Decoupled profit 1.827200 0.374341 4.88 0.000 *** Democratic Legislature Citizens' Initiative 0.360389 0.3137 12 1.15 0.251 Trend 0.044231 0.078536 0.56 0.573 Constant 3.069084 1.552728 1.98 0.048 **
Policy Determinant Probit regression results: DV in bold, IVs listed with coefficients 2009 (<10%) 155 Renewable Portfolio Standard n=450 R 2 =.2822 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 0.820104 0 .252527 3.25 0.001 *** Population 0.000019 0.000015 1.28 0.2 Average wage 0.000111 0.000018 6.32 0.000 *** Unemployment rate 0.047599 0.045610 1.04 0.297 Decoupled profit 0.225149 0.242350 0.93 0.353 Democratic Legislature 0.737923 0.151154 4.88 0.000 *** Citizens' Initiative 0.213447 0.147582 1.45 0.148 Trend 0.000840 0.032340 0.03 0.979 Constant 4.905137 0.597248 8.21 0.000 *** Renewable Set Aside n=234 R 2 =.2704 Coefficient Std. Error Z value P > |Z| Significa nce Policy Enacted in Neighbor 4.508981 0.905455 4.98 0.000 *** Population 0.000048 0.000058 0.83 0.406 Average wage 0.000021 0.000033 0.63 0.532 Unemployment rate 0.260329 0.144616 1.8 0.072 Decoupled profit 0.458186 0.440442 1.04 0.298 Democratic Legislature 0.187047 0.335683 0.56 0.577 Citizens' Initiative Trend 0.030580 0.075482 0.41 0.685 Constant 0.392920 1.259288 0.31 0.755 Research Outreach n=369 R 2 =.1441 Coefficient Std. Error Z value P > |Z | Significance Policy Enacted in Neighbor Population 0.000057 0.000017 3.32 0.001 *** Average wage 0.000037 0.000024 1.53 0.126 Unemployment rate 0.101981 0.085916 1.19 0.235 Decoupled profit 0.061934 0.378534 0.16 0.87 Democratic Le gislature 0.151900 0.212178 0.72 0.474 Citizens' Initiative 0.621981 0.248142 2.51 0.012 ** Trend 0.094964 0.052636 1.8 0.071 Constant 0.726499 0.874484 0.83 0.406
Policy Determinant Probit regression results: DV in bold, IVs listed with coefficients 2009 (<10%) 156 Sales Tax Exemption n=450 R 2 =.1856 Coefficient Std. Err or Z value P > |Z| Significance Policy Enacted in Neighbor 0.464295 0.256740 1.81 0.071 Population 0.000067 0.000013 5.12 0.000 *** Average wage 0.000123 0.000015 8.02 0.000 *** Unemployment rate 0.127411 0.043142 2.95 0.003 *** Decoupled prof it 0.036058 0.216334 0.17 0.868 Democratic Legislature 0.166133 0.143896 1.15 0.248 Citizens' Initiative 0.635439 0.141111 4.5 0.000 *** Trend 0.002357 0.031932 0.07 0.941 Constant 3.947573 0.515788 7.65 0.000 *** Sales Tax Ref und n=197 R 2 =.6574 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor Population 0.000454 0.000317 1.43 0.153 Average wage 0.001277 0.000723 1.77 0.077 Unemployment rate 1.861982 1.085108 1.72 0.086 De coupled profit Democratic Legislature 1.359299 1.357789 1 0.317 Citizens' Initiative Trend 2.867410 1.497239 1.92 0.055 Constant 32.091360 19.567620 1.64 0.101 Solar Practitioner Certification n=450 R 2 =.6546 Coef ficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 0.968614 1.247184 0.78 0.437 Population 0.000170 0.000042 4.07 0.000 *** Average wage 0.000002 0.000051 0.04 0.97 Unemployment rate 0.149130 0.276602 0.54 0.59 Decoupled profit 1.283945 1.298289 0.99 0.323 Democratic Legislature 1.864981 1.249301 1.49 0.135 Citizens' Initiative 0.601982 0.443626 1.36 0.175 Trend 1.133767 0.347777 3.26 0.001 *** Constant 1.736503 1.804612 0.96 0.336
Policy Determinant Probit regression results: DV in bold, IVs listed with coefficients 2009 (<10%) 157 So lar / Wind Permitting Standards n=450 R 2 =.4450 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 0.063315 0.660702 0.1 0.924 Population 0.000001 0.000026 0.03 0.974 Average wage 0.000048 0.000030 1.58 0.115 Unemp loyment rate 0.066667 0.101577 0.66 0.512 Decoupled profit 0.063729 0.386345 0.16 0.869 Democratic Legislature 0.498132 0.374436 1.33 0.183 Citizens' Initiative 0.278245 0.331636 0.84 0.401 Trend 1.034711 0.386246 2.68 0.007 *** Constant 11.83008 3.358082 3.52 0.000 *** State Bond Program n=450 R 2 =.3339 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 0.159526 1.399119 0.11 0.909 Population 0.000027 0.000043 0.64 0.525 Average wag e 0.000141 0.000053 2.67 0.008 *** Unemployment rate 0.251443 0.100654 2.5 0.012 ** Decoupled profit 0.290928 0.374556 0.78 0.437 Democratic Legislature 0.615387 0.359024 1.71 0.087 Citizens' Initiative 0.069669 0.366938 0.19 0.849 Trend 0 .584231 0.147684 3.96 0.000 *** Constant 0.451950 1.590708 0.28 0.776 State Construction Policy n=450 R 2 =.2166 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 1.445202 0.569989 2.54 0.011 ** Populat ion 0.000041 0.000017 2.44 0.015 ** Average wage 0.000038 0.000020 1.88 0.06 Unemployment rate 0.347252 0.107550 3.23 0.001 *** Decoupled profit 0.539387 0.464578 1.16 0.246 Democratic Legislature 0.040433 0.213139 0.19 0.85 Citizens' Initia tive 0.410891 0.205454 2 0.046 ** Trend 0.231989 0.055607 4.17 0.000 *** Constant 0.701754 0.791299 0.89 0.375
Policy Determinant Probit regression results: DV in bold, IVs listed with coefficients 2009 (<10%) 158 S tate Incentives n=383 R 2 =.2268 Coefficient Std. Error Z value P > |Z| Significance Policy Enacted in Neighbor 0.4 42339 0.339281 1.3 0.192 Population 0.000042 0.000022 1.89 0.059 Average wage 0.000148 0.000030 4.99 0.000 *** Unemployment rate 0.157186 0.078037 2.01 0.044 ** Decoupled profit Democratic Legislature 0.065805 0.194506 0.34 0.735 Citizen s' Initiative 0.819862 0.209163 3.92 0.000 *** Trend 0.009184 0.047872 0.19 0.848 Constant 5.562435 1.034175 5.38 0.000 *** Utility Rebate Program n=186 R 2 =.2733 Coefficient Std. Error Z value P > |Z| Significance Policy Enacte d in Neighbor Population 0.000094 0.000100 0.93 0.351 Average wage 0.000082 0.000041 1.99 0.047 ** Unemployment rate 0.160663 0.156273 1.03 0.304 Decoupled profit 0.067632 0.626678 0.11 0.914 Democratic Legislature 0.712471 0.541360 1.3 2 0.188 Citizens' Initiative Trend 0.037622 0.127881 0.29 0.769 Constant 4.384045 1.487731 2.95 0.003 ***
159 APPENDIX B DESCRIPTIVE STATISTICS Included in this section are many of the descriptive statistics around the data set, including num ber of policies enacted by state and changes in number of policies adopted.
160 Number of policies adopted in each state, by year: 2001 2002 2003 2004 2005 2006 2007 2008 2009 Alabama 3 3 3 3 3 5 5 6 6 Alaska 4 5 5 4 4 5 5 7 7 Arizona 19 19 16 15 15 19 19 21 21 Arkansas 9 10 13 12 12 15 15 15 15 California 18 21 24 24 24 24 24 24 24 Colorado 11 11 11 13 13 15 19 20 20 Connecticut 10 10 10 13 13 15 16 16 16 Delaware 4 6 6 5 6 9 9 9 12 Florida 11 12 11 10 10 13 13 14 14 Georgia 3 7 8 7 7 9 9 10 10 Hawaii 10 12 14 13 13 15 15 16 18 Idaho 8 8 9 9 9 11 11 12 12 Illinois 6 6 8 8 8 9 9 13 16 Indiana 7 7 7 8 8 10 11 12 12 Iowa 11 13 14 13 15 20 20 20 20 Kansas 6 10 12 12 12 13 13 13 15 Kentucky 2 2 4 4 4 7 7 7 10 Louisiana 1 1 6 7 7 12 12 12 12 Maine 8 8 12 11 11 15 15 15 16 Maryland 11 13 15 16 16 17 17 17 17 Massachusetts 16 15 18 18 19 23 25 25 26 Michigan 5 7 7 7 8 12 12 13 15 Minnesota 17 17 18 18 18 20 20 20 20 Mississippi 1 2 3 2 2 4 4 4 4 Missouri 4 6 7 8 8 9 10 12 14 Montana 13 15 17 15 16 20 20 20 21 Nebraska 5 6 5 5 5 6 9 9 11 Nevada 8 11 13 12 12 14 14 14 14 New Hampshire 6 7 7 6 6 11 12 13 13 New Jersey 10 10 12 13 13 16 17 18 19 New Mexico 7 9 11 11 13 17 19 19 19 New York 14 15 15 15 15 19 19 20 21 North Carolina 6 7 7 6 7 11 15 17 17 North Dakota 9 9 10 10 10 12 12 12 12 Ohio 11 16 17 16 16 17 19 20 20 Oklahoma 2 3 9 10 10 11 11 12 12 Oregon 13 15 17 16 16 19 20 20 21 Pennsylvania 6 5 7 8 9 11 11 13 15 Rhode Island 11 12 15 15 15 18 18 19 19
161 2001 2002 2003 2004 2005 2006 2007 2008 2009 South Carolina 0 0 1 1 1 7 10 11 11 South Dakota 1 1 4 4 4 5 6 8 11 Tennessee 3 3 6 5 5 7 7 7 9 Texas 11 12 13 12 12 14 14 14 15 Utah 7 10 12 11 11 13 13 14 15 Vermont 3 5 7 6 7 12 12 12 15 Virginia 9 10 10 10 10 11 13 15 15 Washington 9 10 11 10 11 15 15 15 15 West Virginia 3 3 6 6 6 8 9 9 10 Wisconsin 9 11 12 11 11 14 14 14 14 Wyoming 3 2 5 5 5 7 7 8 8
162 Policy adoption count, by year 2001 2002 2003 2004 2005 2006 200 7 2008 2009 Appliance Efficiency Standards 0 0 0 0 0 11 11 11 11 Construction/Design Standard 0 1 3 3 3 0 0 0 0 Contractor Licensing 14 14 13 13 13 13 13 13 13 Corporate Combination 5 5 10 10 10 10 10 10 10 Corporate Tax Credit 14 21 24 24 24 28 28 2 8 30 Energy Standards for Public Buildings 0 0 0 1 1 26 31 40 42 Equipment Certification 13 12 10 10 10 10 11 11 11 Excise Tax Incentive 0 0 2 2 2 2 2 2 2 Generation Disclosure 26 26 27 27 28 29 29 29 29 Green Building Incentive 0 0 0 0 0 1 2 2 2 Gre en Power Purchasing/Aggregation 0 0 9 11 12 13 15 15 15 Industry Recruitment 9 11 11 11 11 12 15 16 20 Interconnection 0 23 28 33 34 36 37 38 40 Leasing/Lease Purchase 0 0 0 0 0 1 1 1 1 Line Extension Analysis 4 4 4 4 4 4 4 4 4 Mandatory Utility Green Power Option 4 4 5 5 5 5 6 6 6 Net Metering Rules 33 35 34 35 36 37 38 41 42 Other Incentive 0 0 0 0 0 1 0 0 0 Outreach Program 0 0 35 0 0 0 0 0 0 Other Policy 0 0 0 0 0 0 1 1 1 Personal Incentive Combo 17 20 22 22 22 25 26 26 29 Production Incentiv e 1 4 12 15 16 16 17 17 17 Property Tax Assessment 4 4 3 3 3 6 8 9 9 Property Tax Exemption 23 22 27 27 27 28 29 30 31 Public Benefits Fund 15 17 16 16 16 19 19 19 20 Public Education/Assistance 2 5 0 0 0 0 0 0 0 Renewables DSM/IRP 6 6 2 2 2 2 2 2 2 Renewables Portfolio Standard 14 15 14 18 19 22 27 31 33 Renewables Set Aside 3 3 2 2 2 2 2 2 2 Research & Outreach 7 7 3 3 3 2 2 2 2 Sales Tax Exemption 14 17 19 19 21 24 26 28 28 Sales Tax Refund 0 0 0 0 0 1 1 1 1 Solar Practitioner Certification/Ac creditation 8 8 0 0 0 0 0 0 0 Solar/Wind Permitting Standards 0 0 0 0 0 0 0 5 13 State Bond Program 0 0 0 0 0 4 4 4 5 State Construction Policy 8 8 9 9 9 0 0 0 0
163 2001 2002 2003 2004 2005 2006 2007 2008 2009 State Combo 33 35 40 42 42 47 47 48 48 Ut ility Rebate Program 0 0 0 1 1 1 1 1 1 Solar Access Combined 34 33 33 33 33 33 34 35 39