Austrian Public Choice: An empirical investigation

Transcription

1 University of Lausanne From the SelectedWorks of Jean-Philippe Bonardi 2012 Austrian Public Choice: An empirical investigation Jean-Philippe Bonardi, University of Lausanne Rick Vanden Bergh, University of Vermont Available at:

2 AUSTRIAN PUBLIC CHOICE: AN EMPIRICAL INVESTIGATION Jean-Philippe Bonardi University of Lausanne Faculty of Business and Economics (HEC) Lausanne CH Switzerland Tel : jean-philippe.bonardi@unil.ch Rick Vanden Bergh University of Vermont School of Business Administration 55 Colchester Avenue, 304 Kalkin Hall Burlington, Vermont Tel: (802) Fax: (802) vandenbergh@bsad.uvm.edu This version: August 2012 JEL categories: B53; D72 1

3 AUSTRIAN PUBLIC CHOICE: AN EMPIRICAL INVESTIGATION ABSTRACT The purpose of this paper is to explore empirically some of the key propositions that can be derived when one combines insights from Public Choice and Austrian economics. We call this approach Austrian Public Choice. These key propositions are (1) that political markets are dynamic processes driven by the action of suppliers and demanders of public policies behaving as entrepreneurs, (2) that these political entrepreneurs are ignorant ex ante of factors that can matter as the regulatory process unfolds, suggesting that they imperfectly anticipate what will ultimately matter for policy decisions, and (3) that these political entrepreneurs learn over time and use local knowledge to discover future opportunities in political markets. Using a dataset of utility initiated rate reviews in the U.S. electricity industry, we find support for these propositions, which suggests that Austrian economics could shed new lights on questions typically raised by Public Choice scholars. JEL categories: B53; D72 2

4 In their introduction to a special issue of the Review of Austrian Economics devoted to Austrian perspectives on Public Choice, Boettke and Lopez wrote: One of the challenges to Austrian economics in general is to establish more and broader empirical relevance, to show that its analytical principles are applicable to standard questions and can provide answers that standard economics cannot. Such accomplishments are necessary to pass the market test. we too emphasize the need for applied-theoretic and empirical work to come from Austrian economists (Boettke & Lopez, 2002: 114). The purpose of this paper is to move in this direction and show the relevance of the Austrian lens for the study of political and regulatory decisions. Following Buchanan and Tullock (1965) or Stigler (1971), Public Choice scholars have showed how private incentives play a considerable role in shaping public policy decisions. Self-interested politicians or bureaucrats participate in political markets and negotiate with various interest groups to set policies, therefore questioning views of public officials as benevolent actors. Public Choice work generally use standard microeconomic tools such as utility maximization, game theory and thus concentrates on solutions at equilibrium. However, DiLorenzo (1988), Holcombe (2002) or Ikeda (1997, 2003) have highlighted how the Austrian focus on dynamic processes rather then equilibrium, knowledge imperfections and entrepreneurship could be used with great benefits to shed new light on the standard Public Choice research program. In this paper, and for simplicity, we call this combination Austrian Public Choice. 1 Apart from the fact that Austrian researchers usually do not see empirical research as a key part of their research program which will not be discussed here another reason for the lack of 1 Compared to Ikeda (1997) or Boettke and Lopez (2002), who use the term Austrian Political Economy to characterize the study of governments decisions and their welfare implications in a broad sense, we adopt a 3

5 empirical studies is certainly related to the data challenge it constitutes. For instance, exploring whether actors behave as entrepreneurs in political markets, a key aspect in an Austrian approach (DiLorenzo, 1988), requires observing when actors engage in political action, which is often quite challenging, especially in a large sample. Another challenge certainly comes from the difficulty to capture the informational/knowledge problems faced by the actors and the surprises generated by political processes, as well as their impact on the final public policy adopted. Last, how can one demonstrate that actors make mistakes or do not make perfect decisions in the context of political processes, due to the fact that they cannot know in advance the behaviors that others (either demanders or suppliers of public policies) will adopt? In this paper, we attempt to address some of these challenges by looking at rate-reviews in the U.S. electricity sector. Our dataset allows us to identify precisely when a political process (the rate-review process) starts, why some actors decide to initiate this process and invest in political activities (utilities initiating the rate-review process and lobbying accordingly), how some other actors react to it (other potential demanders of public policies, especially rivals such as consumers or environmental activists, who lobby as well), and how politicians or bureaucrats, as suppliers of public policies, accommodate these demands. This empirical set-up also provides an opportunity to compare utilities expectations when they engage in political actions with the factors that, in the end, really drive the final policy decision (in that case the variation in the allowed rate of return attributed to the utility). This will allow us to examine empirically the knowledge limitations players face when they first seize political opportunities, and how much of these limitations come from the dynamic nature of political markets. narrower focus here and restrict our analysis to political processes and the dynamics of political markets. Hence we use the term Austrian Public Choice. 4

6 The rest of the paper is composed of four sections. In the first, we outline the theoretical pillars of Austrian Public Choice and identify a few propositions to guide our empirical study. The second presents the empirical context and our strategy to explore Austrian insights in political environments. The third section discusses the methodology and presents the results. The last section discusses these results and concludes. Austrian Public Choice: Key insights The foundations of what we call Austrian Public Choice have been laid out in several pieces, in particular Boettke, Coyne and Leeson (2007), DiLorenzo (1988), Ikeda (1997) and Holcombe (2002). In this section, we review this literature and identify key propositions that will serve as the basis of our empirical investigation. Compared to typical Public Choice models (see Mueller (2003) for a review), an Austrian approach to political economy issues has several characteristics. The first important feature is that policy-making, from an Austrian lens, cannot be modeled as a static equilibrium state, but needs to be considered as a dynamic, competitive and evolutionary process (Benson, 2002). Within this process, each actor involved (both demanders and suppliers of public policies) search for opportunities to take advantage of the process, and by doing so differentiate their actions and prevent the equilibrium described in standard Public Choice models from taking place (Ikeda, 2003). Demanders of public policies might still start this process in certain cases, but their actions are heterogeneous and often hard to predict for their rivals. Suppliers of public policy, similarly, are numerous (politicians, bureaucrats, regulators) and do not passively respond to potential vote support from interest groups (Benson, 2002). They actively seek ways through which to create new opportunities for wealth transfers (Holcombe, 2002). In other words, all players within the policy-making process act as entrepreneurs would in the context of economic markets as defined by Hayek (1945) and 5

7 Kirzner (1973 & 1992): they observe, discover and act upon (political) market opportunities (Rizzo & Driscoll, 1985). Bureaucrats for instance act as entrepreneurs and discover opportunities to expand their resources while applying/enforcing the new regulations put forward by politicians and interest groups. As described in Breton & Wintrobe (1982), this bureaucratic entrepreneurial discovery process involves competition for budgets, power and influence within regulatory systems. Bureaucrats, within the regulatory system, often benefit from a fair amount of discretion, in spite of control by politicians and ultimately by courts. In the short term, maximizing their utility might mean responding to interest groups demands in a way that favors vote maximization for policy-makers (i.e., by giving groups that can provide the highest voting support to politicians policy decisions close to what these groups want). This would be a result compatible with traditional Public Choice insights. However, there might also be cases where bureaucrats/regulators, because they already have ample resources (large budgets, wealth of information, experience, credibility, etc.) might decide to sacrifice short-term obedience to politicians in order to favor longer term objectives such as the building of a reputation that will allow them to stimulate interest group rivalry in the future and thus extract higher rents from policy-making processes in the future. 2 Viewing policy-making as a process generates another important feature of Austrian Public Choice: the knowledge problem and the divergence between intended and actual outcomes (Ikeda, 2003). Because actors have only limited knowledge when they make decisions, errors regularly occur during the policy-making process and unintended outcomes emerge. Bruce Benson summarizes this knowledge problem in political environments when he states: The knowledge problem suggests, among other things, that there are too many uncontrolled margins 2 Rents in that case can be of different kinds, such as extended influence over an industry or even over several opportunities, better career opportunities, etc. 6

8 and unanticipated responses for a rule designer to recognize and anticipate, in part because the changes create a new set of opportunities that have not previously been available (Benson, 2002: 235). In other words, actors in an Austrian approach have to deal with ex ante ignorance of factors that can affect outcomes. Even after conscientiously scanning their environment and planning accordingly, actors end up making mistakes, overestimating certain factors, underestimating other factors, and completely missing others. The implication is that neither demanders of public policies nor suppliers of public policies will, in most cases, obtain exactly what they expect from policy-making processes (Shane, 2000). The assumption of rational ignorance is not, in itself, inconsistent with Austrian political economy for most voters. However, the idea that actors know enough about the relevant probabilities so that, on average, errors are only random and cancel each other out, is inconsistent with Austrian political economy for most voters. An Austrian approach would predict systematic errors made by actors in the policy process because, as all other actors act in an entrepreneurial way, none has accurate information about their own real probability of success. One last important component of an Austrian contribution to the Public Choice research program is that local knowledge is critical to account for the way actors make decisions in political markets. As suggested by Hayek (1945), knowledge differs across time and place, but this does not imply that actors cannot accumulate valuable information and knowledge over time and therefore learn throughout policy processes. Learning, and the local knowledge that an actor has accumulated over time within a specific political market setting, will thus be an important aspect of the future decisions that actor will take. Demanders of public policies will vary not only in their ability to generate voting support to politicians (as in standard Public Choice models) but also in the experience and knowledge they hold about how to proceed and gain an advantage within political processes. 7

9 To summarize, we will try to empirically explore four propositions that define Austrian Public Choice, i.e.: - Both demanders and suppliers of public policies act in an entrepreneurial way, looking for opportunities to change the regulatory status quo to their advantage; - Political entrepreneurs build on knowledge, learnt over time through previous experiences, to discover future opportunities in political markets; - Local knowledge, in particular, is key to generate entrepreneurial discoveries in political processes; - Entrepreneurial behaviors within political processes generate a knowledge/ignorance problem, i.e. an impossibility to know ex ante (even with some probability) all the factors that influence final policy outcomes and how they will do so. This knowledge problem pushes actors (both demanders and suppliers) to make prediction errors, i.e., underestimate, overestimate or even ignore key factors that will ultimately affect the public policy decision. Empirical Setting: Rate Reviews in the U.S. Electric Utility Industry To explore these propositions, we use a dataset of rate-reviews across the fifty states in the U.S. electric utility industry. In the U.S., profit levels of utilities are regulated under a financial rate-of-return regime by state (as opposed to federal) agencies. Utilities are able to improve their financial performance by achieving through political activities higher prices and a higher rate-of-return. State regulatory agencies (Public Utility Commissions, hereafter PUCs ) determine the rate-of-return that a utility is allowed to earn, and hence the final rates charged to consumers, through an administrative process, commonly termed a rate review. Utilities are able to file for rate reviews whenever they wish. Upon initiation of a rate review, a series of public hearings is held where the utility and competing interest groups present arguments and 8

10 information supporting their positions about justifiable rates-of-return and rate levels. At the end of this process, PUC commissioners make a final decision on the rate-of-return for the utility and rates that final consumers pay. Even if they are not supposed to be instrumental in this type of decision, politicians i.e., legislators and governors exercise some monitoring power and control over regulators throughout the process (Hyman, 2000). Figure 1 provides an overview of the different stages of the rate-review process Insert Figure 1 about here This industry context affords a number of advantages for our empirical investigation. Below, we explain how different aspects of an Austrian Public Choice approach can be explored empirically using these data. Identifying political entrepreneurship among demanders of public policies The rate review process is characterized by an intense information exchange between policy-makers, the utility and other interest groups (Hyman, 2000). Since the provision of information regarding policy consequences and alternatives is a central characteristic of actors behaviors during public policy processes, the utility s initiation of a rate review is a rational strategy driven by some calculation or plan by the firms, and can therefore be viewed, from an Austrian perspective, as an entrepreneurial behavior. 3 Because we are able to identify when the utility (a demander of public policy) files a formal regulatory request for rate review, we can identify when it seizes an opportunity to start a policy-making process, i.e. discover an entrepreneurial opportunity in the political market. Note that the Rate of Return (ROR) is firmspecific: each ROR applies to a single utility only even though there are multiple utilities 3 Utilities are in fact likely to engage in activities that complement their regulatory filing with the agency, such as gaining the support of the state governor and legislature (through lobbying, grassroots mobilization, coalition building and financial campaign contributions). For lack of available data on the period under study here, we will 9

11 operating within a given state. This will allow us to identify individual behaviors, and not just collective ones as is often the case for political issues. In the first stage of our empirical analysis, we will therefore account for the decision by utilities to initiate a rate review at a certain point in time. Using a probit model, we will explore the political market conditions which affect the firm s decision (more on this below). We expect to find that this choice will be driven by a variety of political market conditions, which include the rivalry potentially created by competing interest groups (residential consumers, industrial consumers and environmental activists) but also the rivalry faced by legislators and governors for reelection and the amount of resources held by PUCs for reasons mentioned earlier. In effect, the rate review process affords the opportunity for us to explore how a particular Demander (i.e., investor owned utilities) is entrepreneurial in response to characteristics of both the demand- and supply-sides of the political market. Characterizing political entrepreneurship among suppliers of public policy On the supply-side, multiple regulatory and political institutions have a potential role in rate reviews. Final decisions are in the jurisdiction of state PUCs. However, PUCs are overseen by state legislatures that determine their budgets, that can conduct hearings on specific decisions and that can ultimately overturn PUCs through new legislation. PUC commissioners are additionally typically appointed by state governors, giving a further lever to state politicians to exert pressure on PUC decisions. 4 From an Austrian point of view, political opportunities for utilities are thus likely to be shaped by characteristics of both the elected state politicians and the regulatory agency. Capturing uncertainty, or the difference between intended and actual outcomes not explore this dimension, but these behaviors are commonplace during the rate review process and, in our analysis, we will assume they take place (Hyman, 2000). 4 In ten U.S. states PUC commissioners are elected. We control for this variation in our empirical analysis. Below we discuss how PUC selection rules might influence the choice of a utility to initiate and the overall policy outcome. 10

12 Another advantage of using electric utility rate reviews for our empirical setting is that they provide a good measure of the final outcome of the political process. As part of their final rate review rulings, PUCs determine the financial rate-of-return on equity (hereafter ROR ) that the utility may earn. This ROR, in turn, is used to determine the allowed rate levels (i.e., prices a utility can charge to its customers). Since, all else being equal, higher RORs lead to higher profits, utilities are assumed to prefer higher RORs. While PUCs have a statutory duty to set rates that are just and reasonable, in practice they have considerable discretion to set rates and RORs within some implicit range. 5 This will allow us to provide an analysis of the actual outcome of the policy process, compared to the intended outcome that utilities expect to get. A key aspect of our empirical approach will be, therefore, to estimate both the determinants of the rate review initiation with the determinants of the allowed ROR. From a standard Public Choice perspective, since demanders of public policy have extensive knowledge about the process and the factors that will ultimately affect the final decision, one might predict that utilities will initiate when political market conditions are favorable and that nothing else will affect the final outcome. In other words, the decision made by the utility to initiate a rate review will have captured and controlled for the key characteristics of the political market and thus any uncertainty created by the rate review process. On the other hand, from an Austrian point of view, a utility will plan according to its predictions regarding political market conditions, but this utility will turn out to have only imperfectly anticipated what will happen during the political process. Many factors including the entrepreneurial behaviors of rival interest groups or of politicians and regulators will create unanticipated conditions within the political process and will therefore generate differences between what the firm expects and the final outcome. Learning local knowledge and the discovery of future political opportunities 5 Allowed RORs have historically differed significantly across utilities, states and time. For instance, the highest allowed ROR by a state PUC during 1980 was 16.80% while the lowest was 12.50%. 11

13 Rate-reviews are also phenomena which tend to be repeated over time. We observe and analyze utility decisions over a period of 10 years. Thus, we can estimate whether utilities learn from previous periods and whether this learning matters either for their decision to engage in political activities and for the final determination of the allowed ROR by the regulator. We assume that when a utility initiates a rate review and goes through the rate review process the firm is adding to its stock of local knowledge about the political market. We use this to analyze whether local knowledge affects both the entrepreneurial decision to initiate and the regulatory outcomes. Furthermore, using our data, we can identify when a utility observes other utilities initiate rate reviews. When a given utility observes other utilities engaged in the regulatory process, we assume the focal utility is adding to their stock of general knowledge about the political market. Thus, we can compare the effects of learning and accumulating local versus general knowledge on the entrepreneurial actions of the utility and on the final regulatory outcomes. Sample We obtained information on all rate review processes initiated by the population of 190 investor-owned electric utilities during the period 1980 to These utilities represent those operating in all fifty U.S. states except Alaska and Nebraska. While the age of our data may raise concerns, we concentrate on the period since rate reviews were initiated by utilities and not by the PUCs. After 1992, PUCs also initiated rate reviews with the aim of reducing utility rates. Using data after 1992 would have made it difficult to identify whether rate-reviews were initiated by a utility or by the PUC, thus compromising our analysis of entrepreneurial actions by Demanders within the political process. 12

14 The 1980 to 1992 panel of data creates a potential sample of 2470 utility-year observations. After eliminating observations due to missing data, we are left with 1651 utility-year observations. 6 The sample includes 499 rate reviews initiated by utilities. Variables Dependent variables As explained earlier, we will consider various models to isolate the specificities of an Austrian Public Choice approach. To model utilities entrepreneurial actions, we will consider probit models capturing the conditions in which these utilities decide to initiate a rate review. In these models, the dependent variable Firm initiated rate review takes a value of 1 if the utility decides to initiate a rate review in a specific year and 0 if it does not. To measure the outcome of the political process (related to a utility firm s specific action) we calculate the change in the allowed ROR ( Allowed ROR) since the utility s previous rate review. We use the change in ROR rather than the absolute level since this allows us to control for constant firm-level factors that influence the absolute ROR. We obtained the rate review data from a private firm, Regulatory Research Associates, that tracks PUC decisions and crosschecked a sample of rate review results with data available in annual volumes of the National Association of Regulatory Utility Commissions (NARUC) for accuracy. Independent variables Interest group activities: We use three variables to capture interest groups competitive activities and thus important characteristics of the Demanders in the political market. Consumer Advocate is a measure of the degree of residential utility consumer organization per state. In the U.S. utilities sector, 30 states have created consumer advocacy offices to represent residential 6 To measure our dependent variable (change in allowed ROR) we need a baseline measure of allowed ROR. Thus, we eliminate observations on utilities until they initiate their first rate review in the data. We also eliminate observations if we are missing information on the allowed rate of return for a firm since this makes it impossible to calculate the change in allowed ROR. The need for a baseline and the missing data on allowed ROR resulted in a 13

15 utility consumer interests before state regulatory agencies and courts (Holburn and Vanden Bergh, 2006). Consumer advocates, which are publicly funded and which have statutory power to participate in rate review procedures, can provide strong opposition to utility requests for rate increases (Holburn and Spiller, 2002). The variable Consumer Advocate equals one if a consumer advocacy office existed in a given state in a particular year and zero otherwise. Rivalry can also come from industrial consumers who, due to higher average levels of consumption than residential consumers, have stronger incentives to organize. Industrial Consumers, a time-varying variable, is equal to the industrial percentage share of electricity consumption in each state. Data on electricity consumption by consumer sector was obtained from the Energy Information Administration. Finally, we use Environmental activists to capture the extent to which state populations participate in environmental and other non-governmental activist organizations. We use data from the Sierra Club which is the largest environmental NGO in the U.S. Such groups have historically been particularly active against utilities regarding the citing of new power generation plants and the environmental impacts of existing facilities. The variable Environmental activist is equal to the total number of Sierra Club members in the state divided by state population (in thousands). Annual information on state membership was provided directly to us by the Sierra Club. Voting support generated by the utility. To capture the idea that a utility will be more likely to influence policy in a favorable direction when it can provide voting support for policymakers, we also include Market Share for a utility, defined as the total megawatt hours (MWh) of electricity provided by the utility divided the total MWh provided by all utilities in the state. This measure captures both the idea of firm size and its importance as electricity provider within the state, i.e. important political variables for politicians and regulators. The assumption reduction of 311 observations. We also eliminated 475 observations arising from missing data to measure market share. Finally, we have missing data on three utilities resulting in 33 additional observations being eliminated. 14

16 is that if a utility is a major player within a PUC s jurisdiction, then that utility is more likely to provide strong political support to suppliers of public policies. Politicians activities: To attempt to capture conditions under which elected politicians may extend pressure on the PUC, we develop proxies for the degree of rivalry among elected politicians. We construct two dummy variables based on the winning vote margin in the most recent state gubernatorial and legislative elections. For the executive branch (governors), we consider rivalry intense if the margin of electoral victory between the winning and second-place candidates is less than 5%. In this case there is likely to be intense political competition during the next electoral cycle. For the legislative branch, given the importance of party control of the legislature, we consider rivalry intense if the margin of control by the majority party (measured by the number of seats in the combined upper and lower chambers) is less than 5%. Thus, we create dummy variables for Governor rivalry and for Legislature rivalry which are equal to one if rivalry is intense and zero otherwise. We use dummy rather than continuous variables since the underlying distributions of governor vote and legislature party majorities are not normal but highly skewed. We collected this information from annual volumes of The Book of the States. Bureaucrats entrepreneurial activities: As argued earlier in this paper, a dynamic theory of regulators behaviors could suggest that PUCs with greater resources will be more likely to diverge from the pure vote-maximizing approach of politicians to try to build long-term capital such as reputation. Therefore, regulators with greater resources are also more likely to take risks and make decisions against a strong political actor like a utility. Again, we use several measures of resources. Our first, PUC Budget per state capita, is a measure of financial resources. Second, we construct a measure of PUC commissioner experience since experience may partially substitute for financial resources: Tenure PUC is equal to the sum of each commissioner s tenure in years divided by the total number of commissioners on the PUC. We 15

17 expect that more experienced commissioners will have better information/local knowledge regarding utility rate review requests. We obtained annual information on PUC budgets and the identities of PUC commissioners from annual reports of the National Association of Regulatory Utility Commissioners, annual volumes of The Book of the States and the websites of individual PUCs. We also include a variable that may influence the weight that PUCs put on utility versus consumer interests in their ROR decisions. Elected PUC is a dummy variable equal to one in states where PUC commissioners are elected and zero otherwise. PUC commissioners are elected by the voting population in 10 states and are appointed by the governor in other states. Prior research suggests that elected PUCs place greater weight on consumer welfare (Besley and Coate, 2003). Details on commissioner selection were obtained from the Book of the States. Learning local political knowledge: The Austrian Public Choice framework stresses the role of local knowledge in the ability of political entrepreneurs to discover opportunities. We introduce three measures that will help us distinguish local from more general types of knowledge. We create the variable Local knowledge to capture the stock of knowledge that the utilities has potentially learned about the PUC and the local politicians. Local knowledge is equal to the total number of rate reviews the utility has experienced in the past (our sample for this variable started in 1970). To estimate the real impact of this local knowledge on political entrepreneurial discovery, we also compute General knowledge which is a proxy for the utility s stock of general knowledge about the political market. General knowledge (state) and General knowledge (region) are equal to the number of rate reviews for other utilities operating in the same state/region that the focal utility has observed in the past. The knowledge associated with these measures are arguably of a much more general nature since it does not take into account how 16

18 the firm s own characteristics are considered by the local regulators and politicians nor the firm s own past interactions with the political market. Control variables We control for a number of factors that may affect a utility s performance in the rate review process as well as the decision to initiate a rate review. These variables can be interpreted by the fact that regulators and politicians, when they allocate public policies in order to get reelected or reappointed, have to take into account some public interest dimensions and not only the private interests related to interest groups lobbying and highlighted by the Public Choice approach. Interest rates on treasury securities enter into a PUC s decision on the allowed ROR since these are a benchmark to measure the cost of capital. Interest rate change, measured in percentage points, is the difference between the interest rate on ten year Treasury bills at the given time minus the interest rate at the time of the last rate review. Change fuel cost is the percentage change in a utility s average fuel costs (on a per Btu basis) since the last rate review, and is driven mainly by external market forces. Increases in the cost of utilities fuel purchases, as occurred during the early 1980s, directly reduce utility profits, thereby increasing the probability that utilities will initiate rate reviews. In the selection equation, we also control for the absolute level of fuel costs. Since absolute costs are inversely related to profits, we expect a positive relationship between absolute costs and the probability that utilities initiate. We measure Average fuel cost as the average price of fuel per Btu purchased by electric utilities within a state. Fuel cost data is published by the Energy Information Administration. To control for varying economic conditions across the States, we include a measure of the Change per capita income (lagged one year) which is equal to the annual percentage change in 17

19 per capita income in the state. Voter pressure on utility rates may be inversely correlated with recent economic growth trends. We gathered this data from the Bureau of Economic Analysis. A summary of the variables and descriptive statistics can be found in Tables 1 and Insert Tables 1 and 2 about here Table 1 provides statistics for variables included in the full sample of utility-year observations used in the probit models (firms that choose to initiate a rate review) while Table 2 provides statistics for variables included in the Allowed ROR (OLS regression) model. Methodology Empirically, our investigation will move in several stages with the purpose of exploring step by step how the different aspects of an Austrian Public Choice approach improve our understanding of public policy decisions and guide our empirical analysis. First, we will use an ordinary least squares (OLS) regression model accounting for a baseline of what might be seen as a traditional Public Choice approach. Variables capturing the impact of various interest groups and of various policy-makers on the Allowed ROR will be included. We will then explore entrepreneurial political behaviors of utilities by exploring their decision to initiate rate review processes. We will therefore introduce and discuss a set of probit models (Maddala 1983) which allow us to analyze this choice. Within these probit models, and as a third step, we will then add variables considering the role of knowledge, both general and local, in political entrepreneurial behaviors. Last, we will consider a two-stage Heckman model (Heckman, 1979) which allows us to analyze how the utilities entrepreneurial decision to initiate a ratereview in the first stage has an effect on the second stage PUC decision regarding the Allowed ROR. We use this two-stage model to help estimate the total marginal impact of political entrepreneurial behaviors on public policies. Furthermore, this two-stage model will 18

20 allow us to consider which dimensions create the ignorance problem we introduced in the previous section, by comparing the factors a firm considers relevant to the PUC decision when the utility initiates a rate-review with the factors that still affect the ultimate outcome of the policy-making process after the firm initiates. If firms do not face the ignorance problem, then we would expect that the decision to initiate to account entirely for the observed variation in the PUC decision over allowed ROR. RESULTS The first model we consider to start our discussion is a model capturing what might be considered as a straightforward traditional Public Choice analysis, i.e. a model that does not include utilities political entrepreneurship in the Austrian sense. Table 3 presents the results for an OLS regression with changes in the rate-of-return as the dependent variable and variables characterizing the demand (interest group rivalry) and the supply (politicians and regulators) sides of political markets as independent variables. To account for unobserved variations related to firms characteristics, we also include firm fixed effects in the regression Insert Table 3 about here Results indicate that, beyond economic control variables, the variables having a statistically significant impact on public policy decisions are variables related to the supply side of political markets, and in particular the variables characterizing the regulator s behaviors. Elected PUC and Tenure PUC both have a negative and significant impact on the rate-of-return allocated to each utility firm. From a Public Choice perspective, this seems to indicate that regulators that have the most knowledge and consumer connection occupy the best position for creating strong bargaining power vis-à-vis utilities. This fits with arguments indicating that players that truly control the policy-making process are the most likely to extract rent from such a process. 19

21 To consider how an Austrian Public Choice perspective changes the traditional Public Choice perspective, we now introduce utilities behaviors as political entrepreneurs. The OLS analysis above does not take into account that the subset of utilities that engage in rate reviews may have done so due to differential opportunities in the political market. As mentioned earlier, utilities can, but do not have to initiate rate reviews. If utilities are behaving entrepreneurially, they are likely to initiate a rate review when they have identified an opportunity to turn the public policy process to their advantage. We thus consider two probit models with the dependent variable Firm initiated rate review. Results of the probit models can be seen in Table 4. The independent variables in the first model we consider are the same demand and supply dimensions presented in the baseline OLS model presented in Table 3. Column 1 in Table 4 provides the raw coefficient estimates from the probit model, whereas column 2 calculates the marginal effects related to the different independent variables on the probability of initiating. We present the marginal effects separately due to nonlinearities in the probit estimation (Maddalla 1983) Insert Table 4 about here The results presented in Model 1 do provide support for an Austrian-style hypothesis that utilities seek entrepreneurial opportunities in the political market. It appears that the utility is concerned not only with the supply side characteristics, but also with the structure of the demand side of the political market to determine when to initiate a rate review. On the supply side, when we observe an Elected PUC or when the PUC Budget increases we see a reduced likelihood that a utility initiates a rate review by nearly 17% or 3% respectively. These results are consistent with the notion that a utility is less likely to engage with a political actor as the attractiveness of the supply side of the political market declines. 20

22 We see similar effects for the demand side of the market. First, those utilities that possess attractive features for the policy makers (i.e., Market share is larger) are more likely to initiate by about 15.6%. However when the rivalry with other Demanders increases, utilities are less likely to engage with the PUC. An increase in the strength of Environmental Activists or Industrial Consumers the likelihood that a utility will initiate a rate review declines by nearly 2.9% or 31.4% respectively. In Model 2 (columns 3 and 4) of the same Table 4, we present estimated coefficients and marginal effects when we add our knowledge variables. Comparing the statistical significance of the Local knowledge with the General knowledge variables is consistent with an Austrian perspective. When a firm s own knowledge increases, they are more likely to initiate a rate review. However, gaining knowledge by observing others does not appear to have a significant effect on political entrepreneurship. We interpret this result as consistent with an Austrian view of policy-making processes, in which the knowledge that matters for entrepreneurs is mainly local knowledge rather than the general knowledge gained from observing other utilities going through rate-reviews). While the significance of some of the other political market variables changes when we introduce knowledge, there appears to be a robust relationship between characteristics of the supply and demand side of the political market and a utility incentive to act entrepreneurially. The main interpretation is that utilities are more likely to act entrepreneurially when the political market is more attractive. The political market attractiveness improves when PUC budgets decline, PUC commissioners are appointed, and when environmental activists are not as strong. In the last part of our empirical analysis, we turn to a model taking into account both the utilities political entrepreneurial behaviors (initiation of the rate-review process) and the final outcome regarding public policy (the rate-review decision by the PUC). We therefore estimate 21

23 a Heckman two-stage sample selection model which incorporates the utility s first-stage decision to initiate a rate review (Heckman, 1979) into the second-stage decision of the PUC to change allowed ROR. In table 5, we present the estimated coefficients along with the estimated total marginal effects that different characteristics have on the final PUC decision. The first stage results are identical to Model 2 presented in Table 4. The second stage coefficient estimates now take into consideration the entrepreneurial decision of the utility. Given nonlinearities in the estimation of the two-stage model (Heckman, 1979), we estimate the marginal effects separately and present these estimates in the third column of Table 5. The significance of the Inverse Mills Ratio shows the importance of taking into account political entrepreneurial behaviors and their impact on policy-making processes. Failing to take the entrepreneurial behaviors into account would create a bias in the coefficient estimates on the factors that affect the PUC decision. Significance also suggests that we can reject the null hypothesis at the 5% level of confidence that the decision to initiate has no effect on the final policy outcome. This provides support for the Austrian perspective that utilities are entrepreneurial in their decision to engage with the PUC and that this decision is not independent of the PUC rate review process Insert Table 5 about here Turning to our key variables, we find good statistical support for some of the predictions related by the Austrian Public Choice approach. The analysis of the results in Table 5 provides several interesting insights. First, several political market variables appear to be significant in the first stage (selection equation) but not in the second stage (regression): PUC budget, Environmental Activists, and Consumer Advocate. Our interpretation of these joint results is that the factors associated with these variables were correctly predicted by the firms when they 22

24 decided to initiate. In other words, the utilities understood how these factors could affect PUC decisions negatively and managed these in its entrepreneurial choice to engage with the political market. Through an understanding of when to be entrepreneurial, the utilities did not experience any added uncertainty associated with these political market factors. On the other hand, the uncertainty created by longer Tenure PUC and when operating in a state with an Elected PUC seems to have been underappreciated by utilities: the coefficient associated with both Tenure PUC and Elected PUC are significant in the first stage, but also in the second stage (both with a negative sign). Interestingly for an Austrian argument, the variable Legislature Rivalry displays a coefficient estimate that was not significant in the first stage but that became significant in the second stage. Again, our interpretation of these results is that the impact of this dimension, related to the behaviors of regulators and politicians, was not understood ex ante by utilities. In other words, the behaviors of regulators and politicians in the political process created conditions that were largely ignored and not anticipated by utilities. These results related to the supply side of political markets are thus quite consistent with the Austrian view. Now that we control for the entrepreneurial choice of utilities to initiate when political markets are more favorable, we can estimate the overall marginal effect that different political market characteristics have on a PUC decision to change the allowed ROR. The estimated marginal effects are largely consistent with an Austrian Public Choice perspective. When political markets are less favorable, the marginal effect on the change in allowed ROR is negative. We see statistically significant negative marginal effects for PUC Budget, Tenure PUC, Environmental Activists, Consumer Advocate, and Elected PUC. The magnitude of these effects are economically significant as they range from about a 3% to about a 20% effect on the change in allowed ROR relative to the mean of Additionally when a firm increases its 23

25 own knowledge, it has a statistically significant positive marginal effect of close to 10% on the change in allowed ROR (marginal effect of.04 while average allowed ROR = -0.43) CONCLUSION This article sets out to empirically explore some of the key propositions of Austrian Public Choice. Overall, results are supportive and suggest that Austrians have something to bring to the study of how political markets work. Key findings can be summarized by the following points. First, firms involved in public policy processes (in this case, electric utilities) behave as political entrepreneurs in the sense that they scan the competitive environment in the demand side as well as the characteristics of the supply side of the political market, looking for opportunities to change regulations in their favor. This fits with Kirzner s definition of entrepreneurship in economic markets (Kirzner, 1992). Second, our study also uncovers the important role played by knowledge, especially local knowledge, on entrepreneurship in political markets. Compared to general knowledge accumulated by observing other firms going through rate-reviews, the knowledge that firms have learnt through their own experiences with their state regulator is much more likely to impact their decision to engage in political activities and also affect favorably the final policy decision. This is in line with the role of knowledge in markets as characterized by Hayek (1945). Third, other demanders, competing against the initiating firm, also act in an entrepreneurial way, introducing surprises into the process. Our empirical analysis suggests that suppliers of public policies, both regulators and politicians, are the ones that are particularly likely to trigger these surprises and uncertainty within political processes. Other demanders of public policies (residential consumers, activists, industrial consumers) seem to matter much less. This result is 24

26 probably specific to the U.S. electricity sector. However, this result is also broadly compatible with the Austrian view of entrepreneurship, starting with Mises (1949) or Rothbard (1962 & 1985), who stress the importance of property rights in entrepreneurial decisions: entrepreneurial decisions, characterized by profits and losses, matter really for firm owners. In the case of political decisions, a parallel can be made there. Regulators or politicians all have some form of property rights over public policies, whereas consumers or activists do not. In a similar vein, our analysis shows that politicians and regulators behaviors are the ones that create true uncertainty that is hard for utilities to predict and plan for. When considered from a perspective that is out of the equilibrium, suppliers of public policies create more uncertainty than is generally assumed in most Public Choice models. As in any empirical work, this paper has many limitations that should be acknowledged. First, the effort is focused on decisions made by one type of actor (utilities). The decisions and the uncertainty created by other actors entrepreneurial behaviors (consumers, activists, politicians, regulators, etc.) are assumed rather than observed. Second, the firms political activities during the rate-review process are assumed but not measured. For instance, we do not have data on lobbying expenditures or campaign contributions. Adding observations on these data would certainly strengthen the approach presented here. Third, and more importantly, real implications of the Austrian approach are in the nature of the regulation process, and on its implications in terms of welfare and institutional comparisons. As this article only attempts to empirically explore some foundations of Austrian Political Economy, it is limited in its ability to participate in important debates on these questions, such as that of Virginia and the Chicago schools regarding the costs and harm generated by the political market process (Benson, 2002). We leave that for future research. 25