POWER PLAY: GUBERNATORIAL USE OF EXECUTIV E ORDERS IN THE AMERICAN STATES By MITCHELL DYLAN SELLERS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FO R THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2016
2016 Mitchell Dylan Sellers
To my family
4 ACKNOWLEDGMENTS I thank my committee chair, Daniel A. Smith, who took an active interest in helping shape my research agenda and theoretical approach to studying state politics, even before I started classes. Throughout my tenure at the University of Florida, Dr. Smith challenged me to better theorize and consider the policymaking process beyond simple explanations. I also want to thank the other members of my committee, David M. Hedge, Michael D. Martinez, Kenneth D. Wald and Lawrence W. Kenny for their suggestions and feedback on my dissertation and my other projects. I would like to thank my p rior mentor Patricia Shields from Texas State University. Her strong emphasis on pragmatism and research design helped me manage my research agenda, as well use my time efficiently. Many of my projects benefited from both her notebook method and advice on framing research questions. Donald P. Haider Markel and Jami K. Taylor were also very influential in helping me create my research agenda and mentoring me throughout graduate school. I thank my parents and brother for their support throughout my time in gr aduate school. Their listening to summaries of my projects and future plans helped me throughout my time at the University of Florida. My niece and cuteness especially helped to encourage me to finish and move back to Texas. I would also like to t hank my grandma for hosting me for each break and would like her to know that I finally finished that paper for Finally, I would like to thank my friends, A. Diana Forster, Lou Weaver and Nicole Zambrana for listening to me talk about statis tics, lacrosse and research far too much.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 12 Unilateral Action in Action ................................ ................................ ....................... 12 Why Study Governors? ................................ ................................ ........................... 19 Executive Power in the United States ................................ ................................ ..... 21 Governors as Policymakers ................................ ................................ .................... 23 Dissertation Outline ................................ ................................ ................................ 28 2 EXECUTIVE POWER AND BEHAVIOR IN THE UNITED STATES ....................... 32 Summary ................................ ................................ ................................ ................ 32 Considering Governors in the American Context ................................ .................... 32 Gubernatorial Power ................................ ................................ ............................... 34 Insti tutional Power ................................ ................................ ................................ .. 37 Presidential Use of Executive Orders ................................ ................................ ..... 39 Gubernatorial Use of Executive Orders across the States ................................ ...... 43 Executive Legislative Relations ................................ ................................ .............. 49 Levels of Analysis ................................ ................................ ................................ ... 53 3 DATA COLLECTION AND S UMMARY ................................ ................................ ... 56 Summary ................................ ................................ ................................ ................ 56 Introduction ................................ ................................ ................................ ............. 56 Partisanship of Governors ................................ ................................ ...................... 57 Divided Government ................................ ................................ ............................... 58 Data Collection ................................ ................................ ................................ ....... 59 Coding Executive Orders ................................ ................................ .................. 62 Executive Order Availability ................................ ................................ .............. 63 Measuring Power ................................ ................................ ............................. 64 Use of Executive Orders by Region ................................ ................................ ........ 67 Discussion ................................ ................................ ................................ .............. 69 4 AGGREGATE USE OF EXECUTIVE ORDERS ................................ ..................... 88
6 Summary ................................ ................................ ................................ ................ 88 Modern Concerns of State Government ................................ ................................ 88 Executive Order Usage across the States ................................ .............................. 90 Macro level Gubernatorial Use of Executive Orders ................................ ............... 95 What Causes Governors to Issue Executive Orders? ................................ ............. 98 Data and Expectations ................................ ................................ ............................ 99 Estimation Methods ................................ ................................ .............................. 104 Unconditional Three Level Model ................................ ................................ ......... 108 Multilevel Negative Binomial Models ................................ ................................ .... 108 Discussion ................................ ................................ ................................ ............ 112 5 FUNCTIONS OF EXECUTIVE ORDERS: A LOOK AT GUBER NATORIAL AND PRESIDENTIAL ACTIVITY WITHIN POLICY AREAS ................................ .......... 121 Summary ................................ ................................ ................................ .............. 121 Shaping Policy Areas ................................ ................................ ............................ 121 Meso Level Analysis ................................ ................................ ............................. 123 Design and Case Selection ................................ ................................ ................... 124 Coding Executive Orders ................................ ................................ ...................... 127 Functions of Executive Orders ................................ ................................ .............. 129 Health and Human Services ................................ ................................ ................. 130 Presidential Use of Ex ecutive Orders ................................ ............................. 130 Gubernatorial Use of Executive Orders ................................ .......................... 131 Terrorism Preparedness ................................ ................................ ....................... 133 Presidential Use of Executive Orders ................................ ............................. 133 Gubernatorial Use of Executive Orders ................................ .......................... 133 Discussion ................................ ................................ ................................ ............ 135 6 DRIVING FACTORS OF SPECIFIC POLICY ADOPTION ................................ .... 141 Summary ................................ ................................ ................................ .............. 141 Methods of Estab lishing Employment Protections ................................ ................ 142 Gubernatorial and Presidential Use of Executive Orders ................................ ...... 144 Statute Adoption and Executive Order s ................................ ................................ 146 Data and Expectations ................................ ................................ .......................... 150 Estimation ................................ ................................ ................................ ............. 155 Results ................................ ................................ ................................ .................. 157 Determinants of Executive Orders ................................ ................................ .. 157 Determinants of Statute Adoption ................................ ................................ ... 160 D iscussion ................................ ................................ ................................ ............ 161 7 CONCLUSIONS ................................ ................................ ................................ ... 171 Review of Findings ................................ ................................ ............................... 171 Future Resea rch ................................ ................................ ................................ ... 172 Executive Legislative Relations ................................ ................................ ............ 173 Executive Policymaking ................................ ................................ ........................ 177
7 Implications for the States ................................ ................................ ..................... 178 Final Thoughts ................................ ................................ ................................ ...... 179 APPENDIX A: DESCRIPTIVE SUMMARY OF CONTROL VARIABLES ................................ ..... 181 B: CODING SCHEME FOR GUBERNATORIAL POWER ................................ ......... 184 LIST OF REFERENCES ................................ ................................ ............................. 187 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 194
8 LIST OF TABLES Table page 3 1 Summary of power within across states from 1975 to 2013. .............................. 74 3 2 Summar y of power by state from 1975 to 2013. ................................ ................. 75 3 3 Summary of power by state in 1980, 1990, 2000 and 2010. .............................. 77 4 1 Summary of executive orders used in analysis. ................................ ................ 118 4 2 Results of the unconditional three level model. ................................ ................ 119 4 3 Multilevel ne gative binomial model estimating executive orders issued in a given year/state. ................................ ................................ ............................... 120 5 1 Function of executive orders issued relating to health and human services fro m 2004 to 2014. ................................ ................................ ........................... 139 5 2 Function of executive orders issued relating to terrorism preparedness from 2004 to 2014. ................................ ................................ ................................ .... 140 6 1 Factors that Drive the Issuance of LGBT Inclusive Executive Orders. ............. 169 6 2 Adoption of LGBT Inclusive Statutes. ................................ ............................... 170
9 LIST OF FIGURES Figu re page 3 1 Partisanship of governors in office, 1980 2014 ................................ .................. 71 3 2 Percentage of governors facing divided government by partisanshi p from 1980 201 4 ................................ ................................ ................................ .......... 72 3 3 Average of state indicators of power from 1980 2014. ................................ ....... 73 3 4 Northeast: New England data summary ................................ ............................ 79 3 5 Northeast: Middle Atlantic data summary ................................ .......................... 80 3 6 Midwest: East North Central data summary ................................ ....................... 81 3 7 Midwest: West North Central data summary ................................ ...................... 82 3 8 South: South Atlantic data summary ................................ ................................ .. 8 3 3 9 South: East South Central data summary ................................ .......................... 84 3 10 South: West South Central data summary ................................ ......................... 85 3 11 West: Mountain data summary ................................ ................................ .......... 86 3 12 West: Pacific data summar y . ................................ ................................ ............. 87 4 1 Annual total executive orders issued ................................ ............................... 117 6 1 Adoption of LGBT Protections from 1975 2014. ................................ ............... 167 6 2 Predicted Probability of a Governor Issuing an Executive Order to Protect Sexual Orientation in Employment in 201 3. ................................ ...................... 168
11 The relative power of the governor helps to explain the use of unilateral action. Relatively stronger governors are less likely to use executive orders as an aggregate strategy because they have more options available to them to pursue their policy goals. However, weaker governors issue more orders in a given year because they have a narrower range of options to pursue their policy objectives Institutional power is less influential on behavior when exploring how gover nors shape public policy within policy areas but it becomes important once more when analyzing the adoption of controversial specific executive orders. Weaker governors are more likely to issue executive orders overall, but these weaker governors are less likely to do so when the executive order is owned by a particular party, or if the issue is controversial. The pattern found at the aggregate level is not retained on the micro level due to the lack of power Weaker governors issue more executive orders a s a general policymaking strategy to shape their administration, but weaker governors turn to executive orders less when pursuing controversial policies because the legislature can more easily mobilize and overturn executive action
12 CHAPTER 1 INTRODUCTION Unilateral Action in Action Executive power in the American states is understudied. Yet, the states establish a wide array of public polic ies that influence many policy areas, such as emergency management, social and environmental policy. As with presiden ts, governors have the power to influence policy within the executive branch, particularly when legislation delegates authority to the executive branch to promulgate and implement statutes Governors can, and have, establish policy unilaterally via executi ve orders, signing statements and memorandums (Ferguson and Bowling 2008). For example, Virginia Governor Terry McAuliffe issued an executive order to protect transgender employees at the start of his administration in January 2014 Some governors provide signing statements when signing newly adopted legislation into law This follows the presidential trend that recent presidents, particularly President George W. Bush, use as a means of directing agencies or voicing opposition to legislation Many governors take apply the concept of an administrative presidency to running their administrations (Nathan 1983 ). However, appointment powers vary across the states, so this is more effective in states that give governors more discretion. Unilateral action is partic ularly appealing for governors who confront a hostile legislature or one that is resistant to the ir policy agenda because it allows for them to signal policy preferences and/or direct agencies regarding their preferred policy implementation. I argue that governors use executive orders as a power play to advance their policy agenda and long term interests. Governors who are institutionally weaker relative to other states and time periods are more likely to turn to executive orders as an
13 aggregate strategic plan to pursue their policy agenda through executive orders. 1 The reason for this is that weaker governors have fewer options available to them to pursue their interests. Governors who have more power also tend to have greater support in the legislature, more discretion in making political appoints of executive heads, and more leverage in the legislative arena because of budgetary and veto power T herefore, stronger governors are able to use other institutional tools to push for their policy agenda rather than relying on unilateral policymaking, such as executive orders, that could later hinder their efforts for negotiation on desired legislation For weaker governors, the ir negotiating strategy must differ from relatively stronger governors. They have limi ted alternatives that allow them to avoid working with the legislature and that restriction places them on a different footing in terms of pursuing their policy objectives than institutionally stronger governors and, thus, have more leverage when confron ting the legislature Therefore, weaker governors are more likely to forgo pursuing some policy objects within the legislature, where they are relatively weak, and issue executive orders instead. I focus on gubernatorial use of executive orders to underst and when and why governors issue executive orders. I conduct analysis by conceptualizing motivations to use executive orders to differ by different levels. O n the macro leve l, I analyze factors that drive at the aggregate counts of executive orders issued This is similar to conventional presidential studies of executive order usage. I also consider executive orders on the micro level By this, I mean I look at factors that lead to explicit executive 1 2007 ) formal power index that is calculated for each year under analysi s. This measure was originally created in 1960, but adjusted several times, the most recent time by Beyle in 2007.
14 orders that establish the same policy throughout the stat es. I also consider the purpose of these orders. I look at the meso level, which looks at the functions of executive orders within particular policy areas It is essential that analysis considers all three levels in order to provide a better understand ing of gubernatorial behavior. The only study of gubernatorial use of executive orders focuses on the aggregate count of executive orders issued ( Ferguson and Bowling 2008; Ferguson and Shor 2014). However, as several recent publications that focus on the Pres idency demonstrate, not all executive orders are equal in their intent or influence (Ouyan and Waterman 2015); additionally, the strategy of executives shift once confronting divided government versus a unified government (Fine and Warber 2012) Partisan m akeup and divided government are important, but they are by no means the only factor that explain behavior. I focus on power and gubernatorial characteristics to develop a more nuanced understanding of when and why governors issue executive orders Looking at these three tiers allows the analysis to look at the big picture to understand the quantity, quality and strategy behind gubernatorial use of executive orders. I argue that governors are more likely to issue executive orders during divided government a nd when the executive is facing resistance to his or her new policy goals. Governors use executive orders as a power play when they have fewer options available to them; in other words, I expect institutionally weaker governors and governors facing a resis tant legislature to issue more executive orders. For instance, Democratic governors in red states, or in office when the Republicans control the legislature, are more likely to issue executive orders than pursue policy objectives through the legislature, p articularly to enact policies that are more liberal While they
15 may not do so every time they desire new policy, over the course of a year, there is a notable difference between strong and weak governors This was seen in several states to expand state wid e employment nondiscrimination policies to include transgender individuals. A new controversy provides another example of how executives can influence spouses of gay and bisex ual servicemembers can receive the same marriage benefits as their heterosexual counterparts (Department of Defense 2013). But, governors from ten states declared that marriage benefits will not be offered to married servicemembers in same sex relationship governors issued executive orders and directives that instructed personnel not to grant benefits to members of the National Guard because state statutes or constitutions restricted the definition of m arriage to one man and one woman They argued that any services provided to same sex couples violated state law or constitutions, and should not be given Georgia, Louisiana and Texas refuse d to process applications from servicemembers that were in same se x relationships and require d federal employees to process applications During this conflict, states also confronted the question of the 14 th amendment that guarantees equal protection As a result, some states including Oklahoma and South Carolina, cease d processing all marriage benefit requests within state sex marriage. These actions were later ruled invalid and removed, but they effectively established policy within the state for a time period.
16 In th ese scenario s the polic ies were wholly in the domain of the executive branch and the outcome s were decided by the governor s since they are the Commander s in Chief of the National Guard within their state This is rarely the case. Rather it is more common for a combination of the legislature, judicial branch and the executive branch to shape public policy within their state. For example, the executive and the legislative branches are responsible for altering electoral rules. Hicks et al. (2013) find that there are different factors that lead to the proposal versus adoption of voter identification laws. The authors find that greater Republican control of the legislature is associated with stricter voter identification legislation, but Demo cratic governors in office reduce the probability that these bills will actually be enacted. The governor can threaten to veto, or they can use other means to negotiate on bills in the developmental phase of the policymaking process. The y find that the res ulting voter identification policies are a reflection of the legislature and executive action. The legal framework, thus, can be created by any branch of government, along with bureaucratic procedure, because they overlap in domain. State politics scholars focus largely on how legislatures or direct democracy influences policy outcomes, and Rarely is gubernatorial action considered in large n studies Particularly within the p olicy diffusion literature, research focuses on policymaking as primarily a legislative responsibility and do not account for actions taken by governors or the courts. 2 I focus on the role of governors in policymaking by looking specifically at gubernatori al use of executive orders. This is not the only alternative to making policy outside of the legislature and 2 For more on policy diffusion, see: Berry and Berry (1990); Hicks et al. (2015); Karch (2007); Mooney and Lee (1995); Sharp (2005); Shipan and Volden (2008).
17 there are more avenues available to governors to pursue policymaking, but I focus on executive orders because this is a common approach to policyma king Additionally, this is a power available to all governors, and is frequently used by governors Since 1980, governors issued over 35,000 executive orders. Some governors issued executive orders frequently (issuing hundreds or even thousands of executi ve orders while in office), while others did so sparsely while in office (issuing only a few to dozens of executive orders). I contend that governors, as well as executive power more broadly, matter and can influence policy outcomes over a range of issues These executive orders do not establish policies that are as stable as legislation because future governors can alter or remove the policies in place and they require reauthorization after a period of time in some stat es However, they enable governors t o pursue their policy interests while in office without having to negotiate with other political actors, such as legislators interest groups or other high ranking political figures and they allow governors to win political points by signaling their polic y preferences to the public (Ferguson 2006). I assert that governors utilize executive orders to establish policy unilaterally. This executive action is akin to how presidents circumvent Congress by pursuing an administrative presidency approach (Nathan 19 83); otherwise known as the institutional presidency (Burke 2000) or managerial presidency (Pfiffner 1999), in order to quickly achieve policy goals and adjust implementation of statutes already on the books and regulated by state agencies. This administra tive executive approach allows governors to quickly gain control of the relevant state agencies (Nelson 1982; Waterman 1989), as well as signal policy preference to the public and win political points with voters.
18 Consequently, these executive orders are a form of policymaking and a political tool to further future electoral goals so executive orders allow governors to pursue their political agenda and further their viability in future elections simultaneously Likewise, governors use executive orders for multiple purposes and there are varying conditions under which executives issue orders. While the federal and state governments differ considerably in regard to institutional variation and procedures in place, there are common traditions within, as well a s across states, and policy learning occurs for executives across the two levels of government, as well as across the states, which helps to further develop theory that explains state level activity. Executives consider factors that are circumstantial to t heir particular term in office. For instance, there are emergencies or economic concerns that governors must consider that are specific to their own states However, across time, institutional features change within states The relative power of executives also changes gradually as states alter their constitutions or powers given to governors through legislation (Abney and Lauth 1983). Another way in which gubernatorial power changes is through legal challenges in the courts or policy learning from other st ates that cause states to reevaluate their current distribution of power or their practices within government. Governors do not hold the same powers across the states; however, all states offer the governors powers that allow the executive to check the leg islature or act without legislative approval. Every governor can issue executive orders ( Council of State Governments 2014 ). Most governors derive the power to issue execut ive orders from the state constitution or statutes. For some governors, the state co urts ruled that the power to issue executive orders is an implied power. The source of authority varies
19 by state, and some governors derive their power to issue executive orders from a combination of the constitution, statutes and judicial interpretation. Consequently, some governors can use executive orders for more purposes than others, and the strength of executive orders differs across the states. Some governors can only use executive orders for emergency purposes, while other states give governors cons iderable leeway in terms of what executive orders can do to the ir administration and the executive branch. For instance, some states allow for executive reorganization and alterations to executive policies; whereas, other states narrowly define the ability of governors to alter policies through executive orders. For instance, governors can issue executive orders, veto legislation, give State of the State Addresses annually to citizens, as well as a number of other powers that give governors political clout not afforded to the legislature, or any one legislator (Ferguson 2003). However, all governors can issue executive orders that can influence their administration and direct personnel, which makes executive orders an ideal place to begin analysis of guberna torial policymaking through unilateral action. Why Study Governors? Focusing on executive action in the states opens up a large array of methodological approaches, and allows for to test hypotheses of the role of institutions that can only be treated as f ixed when studying presidents. Analysis of factors that drive the use of executive orders concentrates on presidential action. This is problematic in multiple regards: 1) from a statistical standpoint, not enough observations on presidents exist to conduct a medium or large n study and time series analysis with adequate observations may be misleading if motivation to issue orders changed ; 2) analysis
20 cannot directly compare presidents to other cases in the same time period ; and 3) due to the first two probl ems, it is hard to establish causality and to distinguish contextual from institutional constraints Analysis cannot turn any comparable cases for insight on the effects of different approaches that executives use, or how different political actors influen ce presidential behavior Counterfactual research is an option, but all presidential scholars must cope with the N of 1 problem At any given time, o ne administration is in office, which severely reduces the statist ical methods open to analysis. Time serie s analyses require at least 50 observations to create models of stochastic processes and reliable findings. This greatly inhibits presidential scholars from using many quantitative methodological approaches that state scholars utilize in analysis. One way for presidential scholars to get around limited observations is to include a longer time period, but this assumes that factors that drive executive behavior remain constant throughout time. Studying gubernatorial behavior does not have this challenge becau se each year potentially brings an additional 50 cases in the analysis This allows for longitudinal studies or even research on a short time period for relatively new phenomenon The second problem discussed notes an even bigger problem that restricts pr In contrast, governors confront 50 varying legislatures, as well as institutional features, each year, which allows researchers to explore a broader range of politica l contexts. I develop a conceptual framework to explain under which conditions governors Republicans and Democrats alike elect to issue executive orders. I contend that governors are most likely to issue executive orders when political payoff is highes t, i.e.
21 when the legislature is less likely to adopt statutes (during divided government or quasi divided government), or to fulfill campaign promises by rewarding their partisan base when initially entering office. State politics allows this kind of analy sis because there are regularly Democratic and Republican governors in office who are constantly facing different challenges in the legislature and various institutional constraints. Executive Power in the United States The states and federal government s hare a common structure across the country. Separation of powers disperses control of government over the three branches of government and each branch of government is able to check, or to some extent challenge, or prevent abuses of power. It is a common m isconception that state government is modeled after federal government. The reverse is actually the case. After the demise of the Articles of the Confederation, federal government was reconceived (Maddex 2006). The thirteen colonies had fairly stable syste ms of government in place that were distinct from the British structure, with all of the perceived corruption and potential threat of tyranny. As a result, parts of the Constitution were derived from different elements of the early state governments. Exe cutives on the state and federal governments gained power throughout the start of the government. The state governments originally crafted legislatures that were deliberately stronger than the executive branch (Ferguson 2006). Presidents gained power throu gh times of crisis and through opportunistic presidents who challenged ambiguities in the constitution or statutes (Marshall 2008 ). Additionally, presidents gained power by judicial interpretation and Congress acquiescing and granting greater power to the President. Presidents gained part of their power through Congressional
22 action that specifically grant ed power to the president in order to handle a crisis or take charge in implementing new policy. Yet, presidents also gained power through Congressional in action i.e. Congress not acting to prevent executive action or expansion of power The courts ruled that inaction was a means for Congress to confer additional rights and powers to the President under the assumption that Congress would act if it disapprov Likewise, the power of the governors grew in the past 200 years However, much of the changes happened across the country throughout the 19 th century through Constitutional Conventions and state legislatures redefining execu tive legislative relations (Tarr 1998). Unlike the federal government, it is not uncommon for the states to hold Constitutional Conventions to restructure their governments. Some states held up to as many as six Constitutional Conventions to reinvent or to adopt current ideals of what the best structure and purpose of government should uphold (May 2003). This made state government structures to look more and more similar, as well as reflective of federal government. Many states originally had governors serv e shorter terms in the 1800s, governors served one, two and three year terms across the states. For practical purposes, governors term in office lengthened. These constitutional changes also added or removed term limits for governors Despite all of thes e changes, variation still exists in tenure potential. A few states allow governors to run indefinitely for two year terms, while governors in most states serve four years. Additionally, states vary in how their term limits function, if any are even in pla ce for governors. Over the course of the 19 th century the balance between the state legislature and governors changed, with governors gaining power, but the balance is not entirely equal
23 in most states (Dometrius 1999). Some states grant more power to the state legislatures; whereas, other states favor the executive branch O verall the trend in recent decades is towards stronger governors with powers that emulate the American President. Over time, the majority of the states adopted a model of the executive branch that resembles federal government. The most common structure is for governors to serve a four year term that has a term limit. However, term limits occasionally differ from presidential term limits. For instance, 28 s tates currently place limits on the number of consecutive terms or years allowed, while eight other states establish a lifetime term limit. Governors as Policymakers Governors have three official responsibilities chief executive officer crisis managers and the bureaucracy ( Donovan et al. 2015 ). These roles place the executive as the leader of a broad public organization that protects, serves and responds to constituents within their state. Executives are frequently spoken of as the legislature in chief. D ue to their high profile posit ion and me dia attention, governors have considerable p olitical clout that allow them to pressure legislators to act on their agenda. Weaker governors are more likely to pursue their policy objectives unilaterally. Stronger governors typically have veto pow ers that are harder to override which provides them with more leverage to negotiate with the legislature. Stronger governors can provide exchange for continuing part icularized benefits they can use informal powers to sway political actors opposed to their political agenda and stronger governors can pressure their cabinet to act on concerns of legislators (Rosenthal 1990). Yet, weak governors
24 cannot make these threat s or quid pro quo negotiations to the same extent Therefore, weak governors cannot offer the same incentives to legislators to negotiate and cannot pressure agencies to the same degree as presidents or stronger governors. These weaker institutional powers provide greater incentive to governors to pursue their policy objectives through executive orders, which could hurt their political interests if negotiations fail. Governors pursue their policy agenda strategically as well as to maintain administrative functions Fine and Warber (2012) demonstrate on the federal level that presidents are more likely to issue meaningful executive orders that make considerable changes to policy during divided government, but are more likely to issue symbolic executive orde rs during unified government. Presidents issue fewer orders during divided government, but the ones issued make meaningful changes. The reason behind this is that presidents elect to pursue their policy objective s through the legislature when they are more confident that their desired legislation will pass. These symbolic orders cause little change to the current policy equilibrium, but they signal to agencies and to the public that the president is in support of the current policy framework. Presidents win political points with these symbolic orders. However, during divided government, presidents make changes through executive orders, but pursue a less ambitious plan to making policy changes. Changes are made through executive orders, but presidents do not issue executive orders as frequently as unified so that Congress will be less likely to oppose and undo the purpose of the executive order. Governors are similar when confronting state legislatures. Although I cannot analyze all of the 35,000 executive o rders individually, I find support in Chapter 5 that
25 the purpose of the executive orders change for governors facing divided government versus unified government. Kousser and Phillips (2012) also fin d evidence that governors act strategically when confront ing the legislature. Governors are always disadvantaged when facing the legislature for policy oriented concerns. The only instance where governors have the upper hand in the legislative arena is when determining the budget. A budget is required to pass by law, but legislation altering policy is not required. In the scenario laid out by Kousser and Phillips, governors confronting a professional legislature are weaker than governors confronting one that is less professional. Therefore, g overnors have the bes t negotiating positions when they work with the legislature on budgetary concerns. This is because governors typically establish the initial budget proposal and their primary job is to serve as governor ; whereas, legislators in citizen legislature hold a p rimary job other than being a career politician Budgetary delays punish the legislature more than the governor, particularly in states with citizen legislatures. Whether the governor faces a hostile legislature and the degree of professionalization in the legislature mater in terms of how successful governors are in the legislative arena. Governors serve in their position all year and do not have other tasks to return to, unlike many legislators, particularly in less professional state legislatures. When l egislatures are less professional, governors can simply slow down or delay negotiations on the budget to pressure legislatures to negotiate. As Kousser and Phillips (2012) show, this forces legislators with other primary jobs to negotiate more over time in order to return to their main job and home. Yet, this does not happen for policy oriented legislation. There is no required outcome for these policies. I n other
26 words, the legislature does not have to negotiate with the governor. If the legislature fails to reach an agreeable settlement the session ends without any new legislation or legislation that is counter to what the governor lobbied to adopt This puts the governor at a disadvantage throughout the legislation making process. When governors want to make policy changes, they must evaluate the best method to pursue their policy objectives. One common strategy used within government and the business sector is to pursue a principled negotiation within the legislature or through state agencies. This nego tiation approach has the two actors involved in negotiations, in this case the policymaking process, determine their Best Alternative to a Negotiated Agreement (BATNA) (Fisher and Ury 1991). The BATNA is the best case scenario for each political actor if t hey are unable to reach favorable negotiations. Governors and legislatures plan to further their policy interests with their best alternative if they cannot create a mutually beneficial policy in the legislative arena. This is essentially the fallback plan that represents the worst case scenario for the respective political actors if negotiations fail I argue that institutionally weaker governors are more likely to turn to executive orders in an effort to pursue their policy objectives. These governors hav e fewer options of how to pursue their policy objectives because they are weaker when confronting the legislature. If negotiations fail, they are less able to alter state agency heads and veto legislation they are opposed to, so BATNA is le ss desirable than stronger governors BATNA is to turn to unilateral action which further irritates an already opposed legislature. In light of these difficulties for governors in the legislative arena, there are times when governors will elect to turn immediately to
27 executive orders. Rather than facing a resistant legislature to begin with, they pursue unilateral action. Governors that are weak have little incentive to enter and remain in the legislative arena to pursue the more important elements of their political agenda. In these instances, I expect governors to turn to unilateral action in order to further their interests for the purposes of this dissertation, I assess executive orders specifically. These weaker governors issue execut ive orders more frequently than stronger governors or governors in more amicable political conditions. The lack of success in the legislative arena was explained by Kousser and Phillips (2012) They find that governors are drastically more successful in th e legislative arena when working on budgetary concerns as opposed to policy oriented legislation. Governors have an advantage when working on the budget because legislatures must complete a budget every year, or biannually in a few states Governors in sta tes that have citizen legislatures are at greater advantage because these legislators have other careers that they must return to once the session is over. Legislators face punishment if they do not complete the budget and time, in that they have to contin ue the session and remain away from their primary occupation, if they do not complete the budget. This gives a large incentive for the legislature to negotiate and oriented legislation. For policies on the governors agenda, the governor must use his or hers informal powers to encourage to the legislature to negotiate. Unlike the budget, the legislature does not have to create new legislation The governor can call special ses sions for the purposes of giving the legislature more time to create legislation, but the legislature is not required to adopt legislation that the governor pushes Therefore,
28 the legislature has the upper hand on policy since a stalemate results in none o f the policy changes that the governor desires Dissertation Outline The remainder of this dissertation aims to explore when and why governors issue executive orders. In Chapter 2 I overview the current literature on executive action, executive legislativ e relations, and governors. Herein, I also distinguish points where literature on the presidency should differ from governors due to institutional differences, as well as differing political context and traditions. I then set out to provide a comprehensive theory to explain when and why governors turn to executive orders overall more frequently. In this chapter, I introduce the possible levels of analysis of gubernatorial use of executive orders and my overarching expectations throughout analysis. On each level, when and why governors issue executive orders differ, but I argue that power influences governors in systematic ways. In the aggregate, or looking at the macro level, executive orders are explained as a political tactic to promote a general approach to handling an administration; whereas on the micro level, governors push controversial policies of special importance. In the macro analysis, I expect weaker governors to issue more executive orders because there are fewer options available for them to p ursue policy. On the micro analysis, I expect strong governors to push through controversial policies with executive orders because they are stronger and have more discretion on how to create policy. However, on the meso level, governors use executive orde rs to shift the administration priorities, but are also meant to start new initiatives and maintain current policies.
29 I analyze all three levels in order to fully understand why executive orders are issued. Analyses that disregard other possible levels of analysis would fall prey to phenomenon (Converse 1990; Erikson, MacKuen and Stimson 2002 ) This is a mathematical occurrence that highlights the potential for ecological fallacy. Aggregate trends may differ from individual level trends. This is a concern for other areas of Political Science. For instance, complications occur when estimating marcropartisanship because uninformed voters introducing random noise relat ionships found when analyzing aggregate patterns of change do not necessarily hold when analyzing individual partisanship I argue that executive orders function as a power play to help governors accomplish their goals in office. This is not the only optio n available for governors to influence policymaking, but it is a useful tool for governors with weaker powers. These governors have fewer options, therefore, they are more likely to turn to executive orders to accomplish their goals. I find that this is th e case when analyzing the annual use and the issuance of explicit executive orders, i.e. the macro and micro levels. However, executive orders serve practical purposes and governors issue executive orders for a number of purposes that are administrative, pragmatic and symbolic. Analysis of the meso level or within policy areas, shows that governors use executive orders to establish new policies and to maintain ongoing operations. Governors may use executive orders strategically to create new policies that the legislature or agencies are unlikely to craft on their own, this is not the only purpose of executive orders. The remainder of the dissertation analyzes each level of executive order. I explain the methodology and data used throughout the dissertatio n in Chapter 3. I
30 create an original dataset composed of executive orders issued from 1980 to 2014 from all 50 states. The observations are in state/year/administration format but the different chapters utilize this information differently Due to accessi bility problems, not all years are available for all states, but I adjust for this limitation by using multilevel modeling and mixed methods to triangulate findings. In Chapter 3 analysis begins by looking at raw data to first develop an understanding of use of executive orders. This provides an overview of how frequently governors issue executive orders across the states. I find a wide degree of variation in executive order usage but with the exception of three southern states, most states use executive orders sparingly with less than 100 executive orders issued in almost all years After I detail the theory and data used, the dissertation turns to quantitative analysis. Chapters 4 and 6 investigate when and why governors issue executive orders whereas Chapter 5 analyzes the function of the orders themselves to understand the intended purpose of the orders. Chapter 4 analyzes gubernatorial use of executive orders on the macro level. Here I find that governors use executive order strategically to pursue and establish policy objectives outlined in their administration. Governors are most likely to issue these orders upon first entering office, but then become likely to issue executive orders once more towards the end of their tenure. Governors who are inst itutionally weaker are particularly more likely to issue executive orders. In Chapter 5 I analyze the meso level by looking at executive orders issued with in two policy fields: health and human services, and emergency preparedness With this, I assess the broader goals of governors by looking at all executive orders issued within various policy areas to broadly explain how governors use executive orders to shift the policy
31 framework and implementation of policies within several diverse policy areas. In Cha pter 6, I analyze what leads to the issuance of specific controversial executive orders. I draw conclusions and summarize the main findings in Chapter 7 I discuss the implications of gubernatorial use of unilateral powers to establish policy as compared t o other more democratic means of enacting public policy, such as through the legislature or through direct democracy. I also discuss the implications and concerns of using executive orders to establish public policy. Finally, I highlight how gubernatorial action within bureaucracy, via executive orders and other institutional powers, influences policymaking, as well as alters intergovernmental relations and public policy on all levels of government.
32 CHAPTER 2 EXECUTIVE POWER AND BEHAVIOR IN THE UNITED STAT ES Summary Presidential scholars offer explanations regarding how presidents use unilateral powers. However, institutional differences exist across the states and that may cause governors to behave differently from presidents I apply presidential literat ure to state government to develop a theory for understanding factors that drive gubernatorial use of executive orders Not only does political context change by state, but the powers given to the executive branch differ as well. Drawing particularly from the presidency literature, I offer an original theory for understanding when and why governors issue executive orders. I point to when presidents and governors face similar conditions, but also detail when I expect governors to behave differently than pres idents. I approach the question by looking at annual use, orders aimed at a particular policy area and specific executive orders. I argue that the makeup of the legislature matters, but the relative strength of the governor is another important that explai ns tactics. In the aggregate, stronger governors use executive orders less frequently because they have more negotiating options available to them than weaker governors; however, I expect these same stronger governors to issue controversial executive order s because they are given more range in how to use their powers. Weaker governors use executive orders more over the course of a year, but they are less likely to issue orders when the legislature or courts might work to overturn their actions. Considering Governors in the American Context I seek to understand executive behavior and unilateral action broadly within the United States context. I argue that these findings are applicable to understanding
33 gubernatorial and presidential power, and to a lesser ex tent, executives on the local level. Yet, there is one key element that presidential scholars can only speculate regarding, which is the role of power on influencing executive behavior. Presidents in recent decades are considerably stronger than their pred ecessors (Marshall 2008 ). A number of factors influence executive power. The constitution and statutes grant components of formal power, such as veto powers or ability to propose the budget, to executives to act unilaterally (Bernick 1979; Beyle 1968) E le ments of informal power are specific to the executive, which gives rise to the strong emphasis of presidential scholars focusing on specific characteristics of presidents to explain their success at policymaking. While president s have their own strength s and approach to governing, there is very little quantifiable differences in terms of power in recent administrations. Additionally, it is hard to attribute presidential success to power rather than different time periods or political contexts. Scholars can not analyze the role of power appropriate with only one president at a time. The N of 1 problem limits how much the political context is taken into account in accounting for the historical events throughout analysis. I argue that this limitation and unders tanding of power hinders understanding gubernatorial use of executive orders. Schlesinger (1965 ) created a n enduring measure of power the Formal Power Index that is commonly used to assess gubernatorial power across the states that looks at the powers given to executives and the ability of governors to remain in office indefinitely Thad Beyle recreated this data throughout the years I assert that governors turn to executive orders early on in their tenure, during divided government when there are more legislators from oppositional political parties,
34 in response to citizen interests and, most importantly, as a power play when they are relatively weak as compared to their counterparts. Governors use executive orders to establish policy and they turn to t his approach more whenever they have fe wer options at their disposal. As I show, g overnors use executive orders as a power play, but this is not the only time in which they turn to executive orders. I argue that governors broadly use more executive order s as an overall strategy to establish their policy preferences. This is best seen at the aggregate and micro level when explaining the annual use of executive orders and when governors are likely to issue policy specific orders that their ever, when looking at the intent of executive orders issued by governors within specific policy fields, the executive orders are less political and are aimed at establishing policy or responding to current threats. Therefore, the level of analysis, in addi tion to annual counts or specific orders issued, is important to fully explain gubernatorial use of executive orders. Gubernatorial Power As discussed in Chapter 1 executive power is similar across all levels of government. The executive functions as the head of state and is given greater power to force through policies that they favor The founding fathers structured the federal government after the early state systems in an effort to distinguish their system of government from Britain ( Donovan et al. 201 5 ). These early governors were not like modern day governors. The constitution themselves gave greater power to the legislature. In some instances, the legislatures selected the governor. In most states, the citizens selected governors, along with other st atewide elected officials and
35 significantly less power was given to the executive branch. The role of state government in society was much smaller, and the powers granted to the governor were weaker. Governors faced a disadvantage t hroughout most of the t ime period because they were less equipped to face off with the legislature. Part of this was due to the fact that the constitution and the political environment simply favored the legislative branch, but also because unprepared or inexperienced governors served as leaders of the executive office. Recent decades saw a rise in the education and experience of the elected governors. A new breed of professional politicians repla ce the once prominent goodtime Charlies that served in office in the decades before (Sabato 1978). This contrasts considerably with the current governors that are relatively strong, and almost all governors in the past three decades have at least a college degree, with approximately a third of all governors having an advanced degree commonly a law degree. Governors have three forms of power for mal, informal and implied (Donovan et al. 2015 ). Formal powers are powers that are clearly defined by the state constitution or statutes (Dommetrus 1988). These powers are the most easy to define and identify across the states, with most states having similar powers, but procedural differences. Informal powers, on the other hand, are harder to parameterize Informa l powers are not explicitly granted to executives R ather they are tools that the governor has to influence political actors and the public through the ability to persuade and utilize networking to advance their policy agenda (Dometrius 1979; 1987). Cl assical approaches to assessing presidential behavior most prominently relied
36 influenced by the power of persuasion, first put forth by Neustadt (1960). Presidential sc holars rely heavily on this conceptualization of executive power as an informal power to explain presidential power. This is likely because presidential power remains fairly constant within an administration. Marshall ( 2008 ) shows that executive power has grown substantially throughout the past century, yet this has been largely gradual, with the biggest gains in power during times of crisis, such as the great depression and immediately following the 9/11 attacks. One major limitation of the Presidency lit power on policymaking and executive and legislative relations is that the formal power of the president changes gradually and there is only one case within each time period and political context. Presidents claim po wer throughout time through times of crisis and court rulings (Marshall 2008 ). Yet, there are very few amendments to formal powers of but also due to lack of variatio n. In addition to institutional power not varying much by administration, there are no cases to compare the president to during the same time period. Governors are in a similar situation, but presidents differ on too many dimensions, and are responsible fo r considerably different tasks, to compare governors to presidents. Conclusions drawn about presidential power is limited because scholars are unable to determine what accomplishments of the president is due to formal powers of the president or due to indi vidual powers that vary by the individual president. Since there is little institutional variation, presidential successes and failures are often Studying governors allo ws analysis to control for institutional features not found in
37 federal government, and control for problems that all governments faced in certain time periods. This allows research to consider context and the role of institutional features that is not feas ible when studying presidents. Institutional Power Power is commonly discussed in Political Science, but the definition and operationalization is much debated. Scholars such as Neustadt (1960) Eshbaugh Soha (2006) and Kernell (2006) argue that informal p owers are key to understanding the success of executives in their overall policy objectives. S cholars focus on informal powers of executives, particularly the ability to communicate with individuals within Congress, and the public broadly, to send signals to legislators and individuals within key agencies that are responsible for implementation of existing policies. Informal powers, such as staff, public approval and greater media coverage, help governors maintain leverage over other policymakers. The infor mal powers vary by individual and their importance varies over time based on national trends. I model executive orders by state and administration to account for variation caused by the governor specifically, as well as the state. I also control for formal powers adjusted measure of institutional power. Beyle revised and made accessible a measure of institutional gubernatorial power or formal power, which quantifies several elements of institutional power that differ for governors across the United States. This measure is commonly used by gubernatorial scholars. It is composed of five different dimensions that are frequently associated with the power that an executive holds while in office tenure potential, appointment powers, how
38 many other executive heads are elected positions, veto power, and the partisan makeup of members within the legislature. 1 Controlling for the formal powers of the governor is vital to understanding gubernatorial use of executive orders. In man y cases, executive orders are in and of itself a formal power However in many instances executive orders are an implied power or granted through statutes. I control for institutional power of the governor as a means to explain the relative power of gover nors across the states and time. The time period under analysis includes 35 years from 1980 2014. All of the states change institutional power through this time, with some creating a fairly strong governor, and others creating a moderately strong governo r who can assume the greater responsibilities of state government in recent decades. The partisan makeup of legislatures is the most obvious element of the measure of power that will vary throughout time, but many states have altered their statutes and con stitutions to favor the executive branch more than a century ago ( Council of State Governments 2014 ). For instance, elections were altered so that there are fewer statewide elected executive positions, terms for governors went from two years to four in the mid 1900s and now governors run with lieutenant governors in several states as compared to prior decades ( CSG 2014 ). All of these elements are controlled for and considered a component of gubernatorial power across the states. 1 2007 ) measure of formal power. A few of the categories were redefined by Beyle throughout the past several decades. This measure was not created by Beyle every year and not all values are compara ble throughout the available dates because the definition of values were adjusted as variation across the states reduced. I regenerated power based on the most recent ndix for more details
39 Presidential Use of Executiv e Orders P residential scholars have begun to study why presidents use executive orders and their role in establishing policy. Therefore, I draw from the presidential literature to fill in theoretical gaps in understanding executive legislative relations a nd the implications of utilizing executive orders. Presidential use of executive orders was also a relatively unexplored phenomenon within political science until the late 1990s. While in recent years numerous studies have sought to understand determinants of executive orders (Krause and Cohen 1997) there remains a paucity of literature dedicated to completely transferable, I use presidential literature to guide theory regardi ng gubernatorial behavior. requires or authorizes some action within the executive branch." To broaden this definition, I define executive orders as directives that apply to the executive branch and instruct personnel in formulating policy, procedures or actions. This is akin to states as others (Beyle 1999). Yet, all states allow gove rnors to issue executive orders for anything from administrative activities to policy setting purposes (Ferguson and Bowling 2008). Since the federal government shares many features of the states, the evolution s helpful to understanding states and past than in earlier time periods (Muller 1985; Bowman and Kearney 1986). This is due
40 to legislature s deffering to the executive, as well as the executive claiming p ower. Hebe (1972) notes that executive orders are not mentioned within the US Constitution, but presidents derive power through interpretation of the Constitution and through Congressional delegation of power. Nevertheless, every president has used executive orders, or decrees, to establish policy (Hebe 1972). He argues that the expansion of power via executive orders grew during times of crisis and through strong presidents extending their power and role as president. At th e start, most of the orders were related change when Lincoln took drastic action during wartime to charge wartime criminals and to free slaves. He notes that the Supr eme Court upheld the validity of federal executive orders in 1915 in United States v. Midwest Oil Co. by finding that presidents frequent legislature. Executives in the Uni ted States can, and have, expand their power when the legislature is slow to act. I expect this same pattern to occur across states. The presidential literature argues that presidents issue executive orders strategically to develop policy unilaterally. The Strategic Model, as named by Deering and Maltzman (1999, p. 768), holds that "executive orders are instruments used by rationale is that presidents rely on executive order s more when the legislature is control by the oppositional political party or when agencies are uncooperative (Deering and Maltzman 1999, 768). Deering and Maltzman (1999) find several conditions under which presidents are more likely to issue executive or ders to circumvent confronting Congress. First, they note that this route is taken as the likelihood of the legislature
41 overturning executive action decreases. For example, presidents are more likely to issue executive orders when Congress is unlikely to p ass legislation to invalidate their unilateral action They also find that presidents are also more likely to issue executive orders when their approval ratings decline, and when the ideology of the president and the median member of both houses grow. Howe ver, as a whole the Strategic model is not borne out by subsequent research. The above studies do not find differences in the frequency of executive orders during unified and divided government. These scholars looked purely at the count of the policies a nd not their content or their subsequent effect. Fine and Warber (2012, p. 272) argue that this approach may be why there are null findings; further, they assert that on the federal level executives tend to issue more symbolic and routine executive orders during unified government and when Congress is ideological ly closer to the presiden t E x ecutives are more likely to make major po licy changes when the le gislative and e xecutive branches have different policy preferences. Th erefore, I expect executive orders that are symbolic and less likely to bring substantial change to policy implementation during unified government. I also anticipate governors to issue executive orders that have major policy implications during divided go vernment. Fine and Warber (2012) also stress that the ideological distance between the executive and play a vital role in explaining when presidents use executive orders T his suggests that of the extremes than the legislature, and that governors utilize executive orders strategically to implement their policy goals devoid of legislativ e action.
42 Consistent with this notion, Krause and Cohen (1997) suggest that executive orders function similar to vetoes for the purposes of bargaining, at least on the federal level of government. They argue that both of these political tools can be used as a strategy, the end game is then shifted to having mild to moderate change favoring the executive versus no change, which is likely preferable to the legislature. Referring back to Kousser and Phillips (2012), this shifts relations in favor of the executive because the effective end game is no longer the status quo it is a policy that i s closer to the The lack of support for the Strategic Model is somewhat moot in terms of understanding the use of executive orders in any one policy field. It suggests that executives do not immediately or overwhelmingl y rely on executive orders to accomplish policy goals when confronting an adverse legislature. Yet, this may be the mode of choice in altering policy on key issues. Krousser and Philips (2012) argue that governors alter strategies to approaching legislatur popularity as well as the popularity of their policy agenda. I assert that executive orders are an additional option for governors to adopt policies. They may seek this route when they are unpopular or when the policy is further to the one of the extremes than the state legislature. In both these instances, the governor is less likely to win significant support from legislators, so unilateral action offers greater rewards than pursuing policy objectives through the legislative are na.
43 Executive orders are tools to circumvent the legislature, but they are also a source of political points. Pledging to or publicizing the use of executive orders allows executives to position take and credit claim (Mayhew 1974). Mayer (1999, p. 446) a sserts that "Executive orders even enter public discourse from time to time, and presidents use them as a symbol of their intention to act decisively" in order to institute policies quickly and gain immediate popularity (Mayer 1999, 446). Presidential cand idates promise to use executive orders to quickly fulfill campaign promises or reward groups that supported their election because "actions that lack broad impact can still be extraordinarily important to particular interest groups or constituencies" (Maye r 1999, 446). This is also seen on the state level. Therefore, I expect governors to utilize executive orders when they believe the potential voters are likely to support the executive order or to reward their groups that were supportive to their election campaign, such as their partisan base or particular interest groups. Governors are also likely to issue executive orders when similar policies are unlikely to pass in the legislative arena particularly when a governor wins a mandate election, but faces an oppositional legislature. This allow s governor s to increase their popularity, which would also give them more political clout to pursue other policies. Gubernatorial Use of Executive Orders across the States Little research exists on gubernatorial use of executive orders. The premier work done by Ferguson and Bowling (2008) describes the use of executive orders in 2004 and 2005. It found that executive orders accomplished a range of purposes with some orders satisfying administrative tasks, while others established policy. Their results showed that executive orders are frequently used in the American states, and though
44 they do not highlight this, it shows that there is considerable variation in the total number of executive orders issued annually. For in a total of seven executive orders over the same time period. Ferguson and Bowling (2008) provide a needed description of the purpose of ex ecutive orders across the states. Yet, there are several concerns that arise from drawing firm conclusions regarding state policymaking from this study alone. The limited timeframe under analysis raises concerns regarding why such variation is seen across the states. Is the phenomenon random for all years of their administration? Only analyzing two years presents a limited picture of the patterns that states exhibit. Was the use of executive orders a result of differing contexts and approaches to policy dev elopment across administrations, or is it a result of differences across the states? Given the differing powers granted to governors and the different political systems created by the constitution, it is reasonable to assume that the states inherently use executive orders for different purposes and at different rates. While these studies provide important insight, my dissertation addresses the missing components of Ferguson and Bowling and develops a framework to understand gubernatorial behavior. Governors themselves vary on many dimensions and approaches to running or the powers given to the executive branch vis vis the ability of governors to pursue their policy agenda with executive orders. Statement about managerial presidency/executive. Ferguson and Bowling (2008) find evidence that suggests the frequency at which governors use executive orders is similar across years. States also
45 differ in how they use execut ive orders. For instance, in recent years, Oklahoma governors used executive orders primarily to order flags to half staff to honor various individuals or events (Oklahoma Secretary of State 2014). Governors in Arizona have been more proactive, issuing ord ers to establish commissions for regulation or to support their initiatives (Arizona Memory Project 2014). Neither state prohibits using executive orders for other purposes; both can make administrative changes. The source of authority for executive orders also varies across the states. Some states give the governor this power directly in the constitution, some do so through statutes, while the courts ruled in other states that executive orders are an implied power (Council of State Governments 2014). Howev er, unlike the federal level, most states do not require statutory authorization to issue orders. Executive orders are unilateral tools that all governors can use to promote their agenda or to direct agencies on how to carry out executive functions (Mayer 1999). This allows governors to institute their desired policies immediately and without having to negotiate with political actors that have different priorities. It removes negotiations and makes the primary focus of the governor on implementation rather than policy development. Since voters hold executives responsible for policy outcomes, governors have motivation to institute policy unilaterally to circumvent opposition in the legislative arena (Krause and Cohen 1997). Governor can win political points by simply instituting their desired, or promised, policy agenda and ensuring that it is fully implemented. Executives must be strategic in how they approach the legislative arena, as well as administrative changes made in office, or risk suffering at the p olls for perceived effects of policies. Yet, governors can quickly win support by making administrative changes
46 signal overall policy stances (Partin 2001). P residential scholars offer a framework to explain executive use of unilateral powers. Moe and Howell (1999) argue that due to the ambiguity of the Constitution in defining executive powers, presidents gain power by challenging the courts and xecutive power. Additionally, federal legislation occasionally, whether deliberate or not, leaves ambiguity in the legislation, which allows the executive branch to implement the resulting policy in the manner that they feel is best. State constitutions an d legislation share this same ambiguity. Governors help to shape public policy by directing agencies and issuing directives to signal their policy preference while in office. In fact, the source of authority for executive orders in most states is either an implied power or derived from the constitution (CSG 2014), which allows governors even greater ability to influence policy implementation than the president. Use of executive orders can also serve as a symbol of new administrative goals. Mayer (1999, p. 4 46) notes that presidents use executive orders "as a symbol of their intention to act decisively" on an issue devoid of Congressional approval or action. This allows executives to pursue their interests regardless of the preferences of the legislature, whi ch is particularly useful when the policy preferences differ. Due to the decline in straight ticket voting, many governors in recent years face divided government and opposition in the legislative arena. Mayer and P rice (2002) argue that presidents use ex ecutive orders on significant issues in order to direct agencies and signal to other political actors that their policy preferences differ from their predecessor. My evidence suggests that this is true for governors since
47 many of the governors under analys is issued protections for lesbian, gay, bisexual and transgender state employees within the first month of entering office. Presidential research initially suggested that use of executive orders can be explained primarily by the rise of the institutional presidency (King and Ragsdale 1988) or divided government (Deering and Maltzman 1999), but later research suggests that motivations to issue executive orders are more complex. One factor does not explain the strategy that executives take to managing the ex ecutive branch. Krause and Cohen (1997) argue that presidents use executive orders based on differing managerial styles, as well as popularity and legislative conditions. In addition, Fine and Warber (2012, p. 272) argue that the intent of executive orders differs based on unified or divided expect executives to issue more policy oriented or ders when they view their objectives as unlikely to pass in the legislature, so governors issue executive orders in response to the makeup of the legislature, as well as the degree of change that the executive wants to pursue while in office. I expect g ove rnors to both create new policies and to respond to other policymaking institutions federal government, direct democracy, and the state legislature. When the federal government and state legislature is active, I expect g overnors to deploy executive order s more frequently in a reactionary manner either to qualify legislation or to add direction to regulatory agencies. I argue that executive orders are a form of policymaking for governors. They strategically use executive orders to pursue their agenda. Pr evious work finds that divided government actually reduces
48 the number of executive orders issued. However, this is taking in aggregate approach to explaining executive orders. Throughout this dissertation I hope to explain how executives collectively use e xecutive orders in the macro sense, within policy areas and for specific orders. By looking at all three policy levels, I am able to explain more successfully what governors aim to accomplish with their executive orders. I expect governors to use executive orders as a power play on the macro and micro levels. Over the course of a year, or in the macro sense, weaker governors use executive orders to push through policies that they are less able to advance in the legislature or pressure agencies to implement as they want. On specific controversial executive orders, or I the micro level, power is also important in influencing gubernatorial behavior. However, unlike on the aggregate analysis, it is stronger governors that use executive orders as power plays. St ronger governors can put forth executive orders that are high priority to their administration or are otherwise controversial and unlikely to pass through the legislature. In this instance, strong governors turn to executive orders because they are given s o much institutional power that they can push their agenda without procedural concerns. Weak governors issue executive orders more to advance their interests, but they are not inherently controversial; whereas, stronger governors have more bargaining lever age on all policy, but may turn to executive orders when other avenues of policymaking seem unlikely to succeed. Unlike in the aggregate or explicit controversial executive order cases, g ubernatorial use of executive orders within policy areas, or on the meso level, is less of a power move. Governors use executive orders collectively within a policy area for
49 administrative purposes, start new initiatives and to complement their overall administration Rather than the governor using executive orders for sym bolic purposes to issue policy stance s or as a calculated coping mechanism to circumvent oppositional legislatures as weak governors do throughout the course of a year I expect governors to strategically and purposefully use executive orders to establish a policy framework on the meso level that shapes healthcare, emergency management and education. Executive orders should be more policy oriented in order to shape the fields as they had proposed prior to entering office. It is at this level that executive orders are less about power and politics, but more about shaping the policy within their state. Executive Legislative Relations Bernick (1979) surveyed state senators to understand how legislators view the formal and informal powers of the governor. Execu tive orders were not listed, but they about, which was the formal power that the majority of senators identified as the most important. Senators from states with strong governors viewed formal powers as more vital to their strength than senators from states that have relatively weaker governors. Senators in states that had stronger governors saw the same governors use the formal powers to their advantage; whereas, this i s not the case in states with weak governors. This suggests that governors deploy different strategies to approaching the legislature based on the tools given to them and the relative strength of their formal powers. The dynamics of governors and state le gislatures are similar to the federal government, since one was modeled after the other and followed similar patterns of the executive branch gaining power throughout time. The strength of governors is partly
50 dependent on personal characteristics (Barth an d Ferguson 2002), but also who they are facing in the legislature (Bowling and Ferguson 2001; Ferguson 2003; Kousser and Phillips 2012). Governors who have weaker powers may approach policy objectives with the mentality that the ability to persuade is powe r; 2 whereas, governors who assume office that provide stronger powers to the executive are more likely to take a headstrong approach to confronting the legislature. Analogous to how presidents utilize unilateral powers differently based on confronting a un ified government and divided government, I expect governors who enter office with relatively weaker institutional powers to adapt their policymaking powers to account for the fewer options available to them. Features of the legislature can make the govern or relatively stronger or weaker. Kousser and Phillips (2012) argue that in some cases the governor has an advantage over the legislature to setting policy. This happens when legislators are citizens that are not professional politicians. Kousser and Phili ps find governors face harder negotiations when confronting professional legislatures because factors that characterize professional legislatures, such as more time in session and more attention to lawmaking, reduces the urgency for the legislature to quic kly reach an agreement. Legislators in a professional legislature can draw out negotiations with fewer consequences than legislators serving in a citizen legislatures, which enables professional legislatures to develop policy outcomes that favor their inte rests. Citizen legislatures must make concessions to the governor in order to continue negotiations 2 This is to borrow from Neustadt (196 0 objectives as highly dependent on their power to persuade legislators and legislatures to adopt policies in line with their policy agenda.
51 This luxury is not afforded to elected officials that are working in amateur legislatures. This suggests that there are times when governors will favor unil ateral action over purs u ing objectives in the legislative arena. I will show in subsequent chapters that the aggregate number of executive orders issued varies based on characteristics that are specific to the governor and the states. It is also influenc ed by the political context wherein the governor is confronting, which includes divided government, presidents from oppositional political parties and institutional constraints. I will also show that executive orders on specific factors are driven by the p olitical context in which governor face. Yet, this is not as pertinent when explaining policymaking through executive orders within general policy areas. My analysis bridges the gap in this chapter between the micro and macro explanations of policy adoptio n and aggregate use of executive orders by focusing on two policy areas. Governors typically enter the legislative arena disadvantaged when it comes to policymaking. Kousser and Phillips (2012) distinguish between two types of legislation: 1) the budget; a nd 2) other policy. They argue that the governor is stronger in the legislative arena when working on the budget than when pursuing other policy. A budget must be passed and the cost of longer negations is more for legislators than it is for the governor, so legislators must work with the governor to develop legislation that is acceptable to both political actors. However, the condition does not exist when the governor pursues other policies. If legislators chose to not participate or they do not support th e policy, they do not have to negotiate with the governor or pass any version of the bill. They argue that governors must use bargaining chips, such as small favors, the honeymoon period of the administration, and popularity, to push their policies
52 through the legislature, particularly when the legislature is resistant to passing legislation or unmotivated to address the issue. highly dependent on the political context that the executive inherits (Ferguson 2003; Kousser and Phillips 2012). I argue that governors use executive orders strategically when the policy is more extreme than the ideology of the state government and unlikely to pass in the legislature. 3 I argue that the go vernor will attempt to change policy within the legislature in this situation because it is easier to get the legislature to agree to the conditions and the negotiation required between the executive and legislature is less tenable. The governor may pursue either unilateral action or working in the legislative arena in this case, likely with the governor attempting in the legislature first. I argue that only when the policy is ideologically closer to the legislature, or in the middle of the executive and le gislative branches, is the governor unlikely to issue an executive order, and to pursue policy in the legislative arena. Therefore, I expect governors to pursue legislation when the policy is consistent with the ideological preference of the legislature. A likely to respond to the objections of in state constituencies because they must face running for reelection (Karch 2010, p. 152). The governor and legislators confront different constituencies, so mobilization against legislation can make it harder for it to pass. Winston (2002) found that more controversial aspects of welfare reform were removed when constituents voiced opposition to them. This is an 3 By extreme, I mean further to the right or left of the legislature in ideology.
53 additional reason why I expect governors to issue executive orders on more extreme policies the governor does not have to compromise in order to fulfill policy objectives. Levels of Analysis I focus on gubernatorial use of executive orders to understa nd when and why governors issue executive orders. I conduct analysis on three different levels. First, I look at the macro level which considers factors that drive the aggregate counts of executive orders issued in a given state/administration/year. Secon d, I consider the meso level, which analyzes the functions of executive orders within three particular policy areas This allows analysis to explore collective goals of executive orders and their effects on public policy. Third, I assess the micro level, w hich looks specifically at what leads governors to issue specific controversial executive orders. Current research on the state level focuses on the aggregate count of executive orders issued (Ferguson and Shor 2014). Looking at these three tiers allows th e analysis to look at the big picture to understand the quantity, quality and strategy behind gubernatorial use of executive orders. As the study of macro partisanship and partisan identity shows, looks can be deceiving. Aggregate patterns do not necessari ly explain individual behavior. For this reason, I look for differing trends on multiple ways of analyzing executive orders. The about gubernatorial use of executive orders, but it is important to analyze executive orders across time and administrations to understand what leads to greater use of executive orders throughout time and administrations I argue that power plays an important role primarily on the macro and micro l evel s of analysis. In other words, I
54 expect weaker governors to issue more executive orders within a given year and administration strategically as a method of coping for the fewer powers granted to them from the constitution and statutes. On the meso lev el, power is still important, but less so than analysis at other levels of government because governors issue executive orders to influence the entire policy area to shape the policy more so than as a power play. I expect governors to issue executive order s to maintain programs that already have support, either as indicated during polling on the campaign trail or by approval of existing programs. Executive orders within a policy area create or maintain current policies should be less political because colle ctively, these orders are intended to maintain and refine the current policy framework in place. Therefore, most executive orders are functional in that they serve an administrative role or create new policy but as a whole, they are not inherently counter to the state legislature or controversial. I expect power to once ag ain come into play on the micro level to explain why governors push through specific executive orders. I think governors will particularly turn to this tactic on policy areas that of pri mary concern to the governor or are controversial and unlikely to pass in the state legislature. However, I also expect partisanship of the governor and legislature playing a much larger role. I run a survival analysis to estimate the likelihood of governo rs issuing executive orders to add employment protections, and a subsequent analysis to explore executive order usage with adoption of similar legislation. Presidency scholars acknowledge the flaw in looking at solely annual counts and have taken to studyi ng the significance, as well as purpose, of executive orders issued
55 by presidents. I approach the study of gubernatorial use of executive orders by analyzing more than just the aggregate use of executive orders. Since analyzing all executive orders individ ually is not feasible, I take a three level approach to assess when and why governors issue executive orders, in addition to the influence these executive orders have on the framework of public policy. This allows me to consider gubernatorial behavior from multiple perspectives and it allows me to triangulate my findings because all three approaches explain part of the story of gubernatorial behavior. Additionally, this multilevel approach avoids ecological fallacies and false characterizations of gubernato rial behavior that can occur by only considering the aggregate use of executive orders.
56 CHAPTER 3 DATA COLLECTION AND SUMMARY Summary This chapter provides an overview of the data used throughout the dissertation. Each chapter uses different portions o f the dataset, but the sources and collection methods are the same for every chapter. I collected data from 1980 2014 for all years when the data were available by the state website or archives Chapter 3 starts by overviewing collection methods, the data collected, and then discusses what prima facie patterns that exist across regions. I also provide a discussion of the key variable that measures power. This factor ( 2007 ) measure of power using his most latest coding scheme provided in 2007. I create an annual estimate of power for all years under analysis. These descriptive overviews of executive orders available and how power changes throughout the time under analysis provides a basis to help guide later analysis throughout the disser tation. Introduction Context matters. Though Ferguson and Bowling (2008) on gubernatorial use of executive orders provide a thorough description of the function of executive orders no literature explains motivations for gubernatorial use of executive ord ers. L iterature explaining gubernatorial and presidential use of any unilateral powers highlight s that the conditions under which the executive faces are key to explaining the strategy that they chose Overall, I argue that the institutional powers given t o governors differentiate the approach that governors pursue when issuing executive orders, but the partisan makeup of the executive and legislative branches are also important to understanding why governors elect certain strategies. For instance, if Democ ratic governors only face
57 unified government, then analysis would suggest that Democrats are prone to using executive orders systematically different than Republicans or Independents, when the findings may merely be an artifact of strategies used during un ified government. Similar concerns could exist for governors who are members of the opposing party of the governor. This chapter provides a descriptive overview of the context that governors face while in office throughout the time period under analysis I provide an overview of the dependent variable that I use throughout this dissertation. I discuss the data collection, as well as the patterns by state and region in order to establish a better understanding of the dependent variable. Subsequent chapter s utilize this understanding of patterns of partisanship in office and any anomalies in the dependent variable to justify the analysis methods selected. Though Democratic governors were more abundant in earlier years and tended to face unified government con siderably more than their Republican counterparts, by 1995 both of these advantages faded. Additionally, I identify three outliers (Florida, Georgia and Kentucky) in terms of how states and governors use executive orders. Partisanship of Governors State political parties craft their own platform but state party platforms share fundamental principles that the national platform espouses. Additionally, political parties share many of the same benefits across the states due to guidance and funding from nati onal parties Therefore, Democrats and Republicans may approach unilateral powers differently based on advice and official stances of national party members regardless of the conditions in the legislature or the party in government at the federal
58 level. F igure 3 2 shows that the United States experienced a competitive two party system across the states from 1980 to 2014. Third party candidates occasionally won office, but at no point were more than two Independent governors in office at the same time, so t he norm for almost all states across the entire time period was for a governor in office that was either Democrat or Republican. Partisans from both main political parties served in office throughout the entire time period. Democrats had the most governor s in office at one time in 1984 with control of the legislature in 35 states. The greatest control Republicans held was in 2001 with 34 Republican governors in office. Throughout 1980 to the early 1990s, Democrats were the most common governors in office. However, along with the Republican Revolution in Congress, Republican governors swept into office in the mid 1990s and remained prevalent for the next decade. Governors from both political parties faced divided and unified government throughout the time p eriod under analysis. T here should be minimal concerns regarding the number of observations of Democratic and Republican governors throughout the time period. However, many states are red or blue and are prone to elected particular types of governors over others, so analysis still must account for state. Divided Government As discussed in the previous chapter, the literature suggests that divided government versus unified government matters in terms of explaining gubernatorial use of executive order usage. Deering and Maltzman (1999) suggest that ex ecutives will use executive orders to further pursue their policy objectives. However, subsequent studies find that governors do not use more executive orders during divided
59 government. Presidents are more likely to make major policy changes through executive orders during divided government, even if the aggregate use of executive orders is lower during divided government This provides some support of the strategic model, but also suggests that any analysis of ex ecutive orders should also consider the purposes in conjunction with the overall use. Continuing along the same line of literature that argues that partisanship helps to explain gubernatorial use of executive orders, the makeup of the legislature and execu tive branch influence gubernatorial behavior overall strategy changes to conform to the fact that there are fewer receptive members within the legislature, so their policy objectives seek to pursue objectives that are more achievable before the next election. Divided government is common throughout the time period under analysis. Republican governors frequently faced divided government prior to 1995. At its peak in 1992, 17 of the 20 Republican governors confronted divided government. Democrat ic governors fared better in the 1980s by mostly working with unified government but their advantage faded throughout the next two decades. Evidence of partisan sorting as well as party switching in the South, is seen starting in the 2000s .U nified govern ment is increasingly common for governors from both parties. Data Collection The executive orders used to create this dataset come from a plethora of sources on state government. Since only limited analyses of gubernatorial use of executive orders were p reviously undertaken and prior research on governors largely ignores personal factors in explaining behavior, I collect and code much of the data. The
60 website, registrar, legisl ature, online archive or a combination of the aforementioned sources. Controls come from either coding information available online or from Klarner (2012). Throughout analysis, three types of independent variables exist for analysis of governors: 1) instit utional power; 2) factors specific to the governor; and 3) controls regarding political conditions that the governor confronts. Institutional power is a composite measure of formal power that was adjusted by Beyle ( 2007 ) created by considering six dimensio ns of power It codes elements of ranging from 0 to 5, with the strongest governors having higher institutional power values. The six dimension are: 1) the number of separately elected executive positions; 2) tenure potential; 3) ability to appoint high le vel executive positions; 4) veto power; 5) budgetary power; and 6 ) the strongest governors can run repeatedly without constraint, are solely responsible for appointing exe cutives, as well as have the ability to influence the legislature through veto powers on all bills that require supermajorities to overturn and by having more members of their same party holding office within the legislature. While this variable wa s origin ally constructed in 1965 Beyle adjusted the coding scheme throughout the decades and measures were not constructed every year. I recreated the variable throughout the time period and cross years that the me asures are available. 1 1 See the appendix for in depth discussion of the coding scales.
61 For the factors specific to the governor, much of the data needed for analysis was collected and coded by accessing state websites and the National Governors Association (NGA). I obtained personal elements about each governor, such as gender, biography and personal history. Public transparency laws across the states encouraged states to retain public records and the advancement of technology made it easie r for governments to provide information to citizens via the internet. As a result, many states, through state websites, agencies or state archives, provide links to archival information or digital copies of executive orders so that people can readily view orders issued by governors in recent years. Supplemental data on direct democracy usage, legislative activity and partisanship came from the National Conference of State Legislatures ) provides political conditions, such as partisanship and divided government. I collected the total executive orders issued from each state. In order to do so, I searched several websites from each state in order to collect the data. For all states, I sear ched orders issued from 1980 to 2014 However, many states do not provide executive orders for the entire time period. I contacted stat departments for additional records when the states did not provide all years going back until 1980 The resulting dataset created is highly extensive time series cross sectional analysis that structures data in a state/administration/year format Table A 1 provides a matrix that summarizes the data collected from each state It shows that almost all states have at least 10 years of data included in the dataset The states vary by what years they include, but any missing
62 years for a state are assumed to be missing completely at random, i.e. there is no systematic reason a state would not provide orders for a given year other than lack of information or resource limitations. Most of the states do not have any gaps throughout the time period that was obtained. However, Maryland, New York and North Dakot a are missing some observations. Table A 1 summarizes the number of observations collected in each year and by state. Over half (2 6 ) of the states have observations throughout the entire time period. The total executive orders issued for the past 10 years are included for 48 states, and the total executive orders from 3 3 states are available for the past 20 years. I collect ed e xecutive orders for 1, 232 observations out of the possible 1,7 5 0 for all 50 states from 1980 to 2014 Coding Executive Orders The o utcome of interest varies by chapter. However, they all relate to gubernatorial use of executive orders. Analysis of the macro level analyzes the total number of executive orders issued in a given state/ administration/ year. The count of executive orders is sued is recorded for each year that is available. However, I adjust this value to avoid introducing measurement error or additional random noise, into the dependent variable. Governors commonly have to make amendments to executive orders to correct for te chnical problems or to revise sections of the order. In order to avoid overinflating the total executive orders issued in a given year, I do not count amendments issued in the same year as additional executive orders. These orders served the same purpose, so including them would introduce error by inflating the number of executive orders issued. However, I included executive orders that rescind a previous order or amend orders from previous years because they are establishing a new policy in that year.
63 Exe cutive Order Availability I searched for executive orders online for all 50 states for the entire 1980 to 2014 time period under review. This information is not available for many states throughout the entire time period. In some instances, I was unable t o obtain executive orders even after contacting state resources through online inquiry, or the aide of archivists, historical groups and government agency officials. In other instances the data could only be obtained through archival work. Additionally, in a few states prior to 2000, the records were poorly maintained and no longer existed. For budgetary and time constraints, analysis focuses on data that is available through the state government websites, archives or historical groups. I contacted the sour ce, or sources, that provided executive orders for a particular state in all instances that executive orders could not be obtained for the entire time period under analysis. In some cases, executive orders for several additional years were available, and i n a few states the source could only provide the count of executive orders for earlier years. This approach to data collection yielded an elaborate data set that spans 35 years. Unfortunately, the dataset is unbalanced However, I collect ed the actual ex ecutive orders for 41 states for the past 8 years (2007 to 2014) which allowed for the analysis in Chapter 5 that analyzes the purposes of executive orders within policy areas Table 3 1 shows the data collected for all states for all years. All states pr ovide executive orders for the current administration; 2 however, the number of years back that states provide executive orders varies by state. Most commonly states provide 2 Kentucky provides executive orders from most recent two administrations, but errors occur i n the website before most of the orders can be downloaded in any given year.
64 executive orders online for the entire time period though only four states provid e all executive orders issued by governors throughout the 1900s. Recent pushes for transparency and e government in the past two decades are apparent when analyzing the accessibility of executive orders. Executive orders are available online for all curr ent administrations. Further, executive orders are available going back until 2005 for 47 states, with another state (Nevada) provides the count of executive orders issued in those particular years 30 states provide executive orders going back until at le ast 1995, with an additional 7 states providing the count of executive orders issued in a year. Though transparency concerns are not a new phenomenon, with pushes in 1980s and 1990s, the rise of e government led to much greater access and some states retro actively posting their executive orders to encourage citizen engagement with government. State archives and historical papers. 3 The dataset has 22 states with complete timespans, and an additional 5 states with the counts for all of the later years available. Measuring Power A large benefit to studying the states is that there is institutional variation within the dataset. The states differ from one another. Additiona lly, some states altered their term limits, appointment powers and other powers over the three decades included in analysis This analysis spans from 19 75 to 201 4 Over this time period states changed powers given to their executives with some making gover nors relatively stronger, while 3 These administration papers include a number of artifacts, such as State of the State Addresses, executive orders, proclamations and any personal records that governors used wh ile in office and donate once their administration ends.
65 others made governors weaker. Overall, governors became stronger throughout time. Using the most current coding scheme created by Thad Beyle for 2007, I created annual estimates of institutional power for each state from 197 5 201 4 4 This is a six point composite scale that measures the total statewide elected positions tenure potential of governors, appointment powers, budgetary power, veto power and the level re. These six elements capture several dimensions of institutional power. Lieutenant governors and vice presidents serve similar roles in government. However, unlike vice presidents, lieutenant governors do not always run on the ticket with governors. The governor and lieutenant governor may even be members of different political parties. Also, most states allow citizens to elect other high ranking executive positions. Some states also reduce appoint power by requiring governors to share appointment power with the legislature, or by fully granting appointing appointment power to the legislature or other organizations. Additionally, the ability to set the budget influences gubernatorial ability to pressure state agencies. The legislature is an important component of gubernatorial power. The veto power explains how much opposition legislators need to overcome gubernatorial opposition. Additionally, partisan control indicates how much same party support governors have in the legislature. These two elements combine to indicate how strong the governor is in comparison to the legislature. 4 This coding scheme was adjusted several times since it was first conceived and prior estimates were not recreated once the coding was adjusted. I recreated the measure, but unlike Beyle, I make all components of the measure continuous in an effort to get the most precise estimation of institutional power. I used the Book of States to code the data and cross referenced measures against the available estimates provided by Beyle (r > 0.9 for all measures).
66 I start by considering the overall patterns of power across time and states. Figure 3 1 shows that overall the average state power did not change drastically throughout the years under analysis. In fact, states changed in the early 1980s and 1990s to adjust some procedural adjustments to executive power. Rarely were there drastic changes to executive power ; rat her, these changes were procedural changes slowly adjusted the power of governors within their own states. Table 3 1 shows the level of power that existed within states and between states throughout time. Unlike some of the data on executive orders, the d ataset includes all years under analysis. All 50 states are available for the 34 years of analysis, which accounts for 1,700 observations. Most of the variation exists between states, but roughly a quarter of the variation in power exists within the states themselves, so states have changed modestly throughout the past three decades. I provide descriptive summaries of institutional power throughout the years under analysis state by state. Table 3 2 provides the summary statics throughout the time under ana lysis. Most states do not change drastically throughout the time period; however, the average standard deviation is 0.47. Additionally, the lowest value of power is 2.0 and the maximum is 4.80. Table 3 3 compares the institutional provisions in 1980, 1990, 2000 and 2010. While some of the states decreased the powers granted to governors, most of the governors gained power through longer tenures or fewer executive positions elected into office. Overall, governors held the most power in the 1990s, with a shar p increase from 1980 to 1990. Power declined modestly from 1990 to 2000, but remained fairly stable for the next decade.
67 Use of Executive Orders by Region The remainder of the chapter overviews the data available for each state regarding executive orders issued from 1980 to 2014 and patterns found across the states. I discuss these states by United States Census regions for organizational purposes. Though I discuss broad patterns found within regions and sub regions, this is primarily for exploratory purpo ses. Overall use of executive orders and the purposes are similar along some neighboring states, but not necessarily all neighboring states or states within the same region. These states share similar traditions in terms of historical experiences, shared p olitical pressures and broad political cultures (Elazar 1972 ; 1984 ), but there is considerable differences in political cultures within government and among citizens (Lieske 2010). Northern states have considerable variation in regards to how frequently governors issue executive orders Figure 3 4 and Figure 3 5 display the total use of executive orders from 1980 to 2014 for states in the Northeast: New England and Northeast: Middle Atlantic A couple of states in the north issued between 40 and 60 orders across the time period, but most governors issued fewer than 20 in any given year. Executive orders were less frequent in the New England area, but fewer states had complete datasets in the North, so conclusions about the 1980s cannot be formed. There i s less variance in terms of how often governors issued executive orders in the Midwest. Figure 3 6 and Figure 3 7 show executive order usage in the East North Central and West North Central states, respectively. These states typically issued between 10 and 40 executive orders throughout the time under analysis. Additionally, a larger portion of the states in the Midwest made data available going through the 1980s.
68 After initial review, no general trend regarding use of executive orders can be found. The cou nt varies by year. Additionally, many of the executive orders are in response to the current environmental conditions that the state faces. Governors use executive orders in large part based on emergencies or other conditions specific to that year. Southe rn states use executive orders differently than other regions The south includes: Southern Atlantic (Figure 3 8) East South Central (Figure 3 9) and West South Central (Figure 3 10) These states exhibit the most variation and include a large proportion of states in the United States. Florida, Georgia and Kentucky deviate from other southern states, as well as all other states in the United States. These states frequently issue hundreds of executive orders in a given year and thousands within an administr ation. Florida averages between 300s and 400s throughout 1980 to present time. Georgia consistently issued over 400 executive orders in a given year since the 2000s (when data first became accessible). Kentucky is the largest outlier. Kentucky issued 1,000 s of executive orders in a given year. Part of the reason for these anomalies is that all three states issue executive orders to hire personnel within state government, as well as appoint other officials to office. Kentucky also uses executive orders for o ther purposes, such as issuing pardons and removing individuals from office. Looking beyond these three outliers, southern states still frequently use executive orders more than the Northern states and more than most Midwestern states. I now turn to patt erns of gubernatorial use of executive orders for Western states. These states are most like southern states, but considerable variation also exist across these states. The executive order use for West: Mountain is shown in Figure 3 11 and Figure 3 12 disp lays usage in West: Pacific Both the Southern and Western states tend
69 to issue executive orders more than the Northern and Midwest states. There are no comparable outliers found in the West that revival Florida, Georgia and Kentucky, but many of the state s use executive orders frequently in any given year. Many of these executive orders respond to emergencies, such as, fires and other disasters, but many governors issue executive orders to establish policies. Discussion While no concrete conclusions can be drawn from the raw data analyzing the aggregate use of executive orders, it is clear that overall certain regions turn to executive orders more. It appears that states with longer histories as American states tend to use executive orders less frequently There considerable variation throughout time in terms of who holds the executive branch and whether or not divided government is in place that is not controlled for when reviewing the raw counts. Subsequent chapters take partisan control into considerati on within analysis However, review of the partisan control of the executive branch and the total governors facing divided government suggests that there are an adequate number of governors from both parties that face divided and unified government through out the time period Power looks relatively constant throughout the time period under analysis when analyzing the average power granted to governors across all 50 states. I estimated est variation over time when considering the average of all states. However, power changes within and across states throughout the whole time period under analysis. As Table 3 1 shows, there is variation across the dataset within and across the states. Mor e variation is seen across states, but governors within the same state vary modestly across time
70 Gubernatorial power strengthened towards the end of the 20 th century, but term limits and minor procedural changes in the early 1990s reduced some of this pow er. Overall the measures of the state power changed to a modest existent. The range itself varied more within states, but still very few states experienced large sweeping changes to institutional power over the three decades under analysis. Additionally, the partisan component of institutional power causes power to change each year; yet, partisanship plays less of a role of power variability when the party in government remains fairly stable. (19 84 ) coding scheme, suggests that traditionalis tic states use executive orders most frequently, but even this is not consistently supported. On average, states in the North and Midwest use executive orders less frequently than the South and West. However, once analysis removes the three main outliers, none of these patterns are large enough to warrant overall controls by region.
71 Figure 3 1 Partisanship of governors in office, 1980 2014.
72 Figure 3 2 Percentage of governors facing divided g overnment by partisanship from 1980 2014.
73 Figure 3 3 Average of state indicators of power from 1980 2014.
74 Table 3 1 Summary of power within across states from 1975 to 2013. Variable Mean Std. D ev. Min Max Obs. Power Overall 3.57 0.49 1.93 4.83 Total = 1,950 Between 0.40 2.72 4.23 States = 50 Within 0.29 2.61 4.59 Years = 39
75 Table 3 2 Sum mary of power by state from 1975 to 2013. State Obs. Mean Median Std. Dev. Max Min Alaska 39 3.89 3.89 0.23 3.50 4.28 Alabama 39 3.09 3.03 0.51 2.42 4.03 Arkansas 39 3.40 3.36 0.43 2.86 4.06 Arizona 39 3.49 3.53 0.30 2.89 4.13 California 39 3.48 3.39 0.47 3.00 4.50 Colorado 39 3.71 3.75 0.30 3.17 4.30 Connecticut 39 4 .04 3.92 0.39 3.42 4.70 Delaware 39 3.63 3.56 0.26 3.33 4.40 Florida 39 3.63 3.64 0.30 3.03 4.17 Georgia 39 3.32 3.07 0.38 2.86 4.07 Hawaii 39 4.10 4.17 0.32 3.50 4.50 Iowa 39 3.88 3.83 0.26 3.47 4.60 Idaho 39 3.50 3.47 0.12 3.17 3.70 Illinois 39 3. 94 3.92 0.18 3.47 4.50 Indiana 39 3.24 3.25 0.25 2.75 3.63 Kansas 39 3.62 3.61 0.27 3.17 4.10 Kentucky 39 3.57 3.56 0.19 3.06 3.90 Louisiana 39 3.38 3.33 0.37 2.94 4.13 Massachusetts 39 4.04 4.22 0.53 3.08 4.83 Maryland 39 4.23 4.28 0.17 3.78 4.40 M aine 39 3.43 3.47 0.30 2.97 3.89 Michigan 39 3.82 3.75 0.14 3.58 4.08 Minnesota 39 4.02 3.92 0.36 3.50 4.67 Missouri 39 3.48 3.50 0.22 3.14 4.00 Mississippi 39 3.03 3.00 0.26 2.58 3.69 Montana 39 3.52 3.53 0.17 3.23 3.86 North Carolina 39 2.72 2.83 0 .28 2.22 3.20 North Dakota 39 3.88 3.89 0.13 3.61 4.28 Nebraska 39 4.13 4.14 0.13 3.83 4.37 New Hampshire 39 3.13 3.14 0.25 2.72 3.56 New Jersey 39 4.10 4.08 0.19 3.75 4.42 New Mexico 39 3.51 3.53 0.15 3.25 3.78 Nevada 39 3.05 2.97 0.21 2.78 3.53 Ne w York 39 4.17 4.22 0.12 3.92 4.33 Ohio 39 3.85 3.92 0.18 3.42 4.08 Oklahoma 39 3.15 3.06 0.42 2.63 4.03 Oregon 39 3.39 3.33 0.26 3.08 4.03 Pennsylvania 39 3.97 4.00 0.16 3.67 4.22 Rhode Island 39 3.89 3.89 0.23 3.50 4.28
76 Table 3 2 Continued. State Obs. Mean Median Std. Dev. Max Min South Carolina 39 2.72 2.81 0.38 1.93 3.39 South Dakota 39 3.75 3.80 0.29 3.22 4.20 Tennessee 39 3.90 3.94 0.23 3.58 4.28 Texas 39 2.94 2.97 0.36 2.00 3.33 Utah 39 3.98 4.00 0.21 3.58 4.33 Virginia 39 3.35 3.33 0.2 5 3.06 3.92 Vermont 39 2.84 2.78 0.32 2.31 3.37 Washington 39 3.39 3.33 0.24 2.89 3.75 Wisconsin 39 3.63 3.56 0.20 3.39 4.14 West Virginia 39 3.90 4.00 0.28 3.17 4.30 Wyoming 39 3.57 3.64 0.28 3.08 4.13 Total 1,950 3.57 3.58 0.49 1.93 4.83
77 Table 3 3 Summary of power by state in 1980, 1990, 2000 and 2010. State 1980 1990 2000 2010 Alaska 4.06 4.28 3.75 4.03 Alabama 3.53 2.75 3.14 2.69 Arkansas 3.58 4.03 2.86 3.36 Arizona 3.53 3.39 3.61 3.56 California 3.92 3.47 3.42 3 .03 Colorado 3.92 3.83 3.75 3.75 Connecticut 4.42 4.42 3.67 3.72 Delaware 3.83 3.67 3.42 3.67 Florida 3.89 3.44 3.47 3.64 Georgia 3.56 3.44 3.04 3.03 Hawaii 4.40 4.44 4.19 3.50 Iowa 4.33 3.92 3.58 4.00 Idaho 3.39 3.64 3.50 3.39 Illinois 3.92 4.00 3.81 3.92 Indiana 3.53 3.19 2.92 3.19 Kansas 3.58 4.00 3.67 3.17 Kentucky 3.67 3.67 3.56 3.50 Louisiana 2.94 3.58 2.97 3.06 Massachusetts 4.39 4.83 3.56 4.22 Maryland 4.33 4.28 4.28 4.19 Maine 3.64 3.50 3.11 3.69 Michigan 3.75 3.92 3.75 3.75 Minne sota 4.14 4.33 3.50 3.86 Missouri 3.75 3.50 3.56 3.22 Mississippi 2.69 3.64 3.08 3.00 Montana 3.44 3.78 3.58 3.36 North Carolina 2.83 2.58 3.03 2.92 North Dakota 3.75 4.03 3.89 3.89 Nebraska 4.14 4.23 4.00 4.17 New Hampshire 3.31 3.53 2.89 3.14 New Jersey 4.42 4.42 4.08 3.75 New Mexico 3.50 3.28 3.31 3.78 Nevada 2.78 3.28 2.97 2.81 New York 4.25 4.23 3.92 4.25 Ohio 3.75 3.83 3.94 3.64 Oklahoma 3.44 2.97 2.63 2.72 Oregon 3.14 3.83 3.17 3.47 Pennsylvania 4.00 4.20 4.00 3.83 Rhode Island 3.39 2 .65 2.69 2.67
78 Table 3 3 Summary of power by state in 1980, 1990, 2000 and 2010. State 1980 1990 2000 2010 South Carolina 2.44 2.81 2.92 3.17 South Dakota 4.00 3.97 3.86 3.22 Tennessee 3.94 4.28 3.58 3.61 Texas 2.08 2.67 2.97 3.28 Utah 3.75 4.19 3.8 8 4.00 Virginia 3.06 3.92 3.47 3.33 Vermont 3.22 3.06 2.83 2.36 Washington 3.14 3.75 3.33 3.67 Wisconsin 3.39 3.72 3.56 3.72 West Virginia 3.92 4.02 3.70 4.08 Wyoming 3.69 3.61 3.58 3.17 Total 3.63 3.72 3.46 3.48
79 A B C D E F Figure 3 4 Northeast: New England data summary. A) Connecticut, B) Maine, C) Massachusetts, D) New Hampshire, E) Rhode Island, F) Vermont.
80 A B C Figure 3 5 Northeast: Middle Atlantic data summary. A) New Jersey, B) New York, Pennsylvania.
81 Figure 3 6 Midwest: East North Central data summary. A) Indiana, B) Illinois, C) Michigan, D) Ohio, E) Wisconsin.
82 A B C D E F G Figure 3 7 Midwest: West North Central data summary. A) Iowa, B) Kansas, C) Minnesota, D) Missouri, E) Nebraska, F) North Dakota, G) South Dakota.
83 A B C D E F G H Figure 3 8 South: South Atlantic data summary. A) Delaware, B) Florida, C) Georgia, D) Maryland, E) North Carolina, F) South Carol ina, G) Virginia, H) West Virginia.
84 Figure 3 9 South: East South Central data summary. A) Alabama, B) Mississippi, C) Tennessee.
85 AB CD Figure 3 10 South: West South Centra l data summary. A) Arkansas, B) Louisiana, C) Oklahoma, D) Texas.
86 Figure 3 11 West: Mountain data summary. A) Arizona, B) Colorado, C) Idaho, D) Montana, E) Nevada, F) New Mexico, G) Utah, H) Wyoming.
87 Figure 3 12 West: Pacific data summary. A) Alaska, B) California, C) Hawaii, D) Oregon, E) Washington
88 CHAPTER 4 AGGREGATE USE OF EXECUTIVE ORDERS Summary Chapter 4 analyzes the determinants of aggregate use of executive orders. I analyze t he macro level use of executive orders by governors from 1980 to 2013. I first assess where the variation is across the states. I find that most of the variation is seen across states, but considerable variation is seen within administrations. In other wor ds, different traditions across the states cannot fully explain gubernatorial use of executive orders, nor can a purely politician centered argument. Differences across the years and states are partly due to the individual governor as well as the state it self. I find that once analysis controls for relevant factors, such as legislative activity, year s in office served by the governor and the makeup of the legislature executive orders are more common for weaker governors, governors with prior experience in the state legislature and when governors have more same partisans in office. Additionally, time in office is found significant, consistent with the presidential literature, governors issue executive orders most at the start of their tenure. This generally continues throughout their time in office, but as their time in office winds down, governors are more likely to issue executive orders once more. This indicates that governors respond to their political context and the makeup of the legislature but facto rs specific to the governor themselves also explain use of executive orders Modern Concerns of State Government All executives on the state and federal level can issue executive orders to accomplish a multitude of functions. Though the constitutionality o f some of the orders
89 are is in question, President Obama issued executive orders to establish policies that Congress is not responding to or will not act to change. For example, he used executive orders to expand protections to lesbian, gay, bisexual and t ransgender employees, as well as halt deportations. All governors also have the ability to issue executive orders, but the legal source of authority, and the purposes for which orders can be used, vary across the states (Council of State Governments 2014). Some governors have the power to issue executive orders to respond to emergencies, implement federal statutes and handle administrative protocol, as well as other agency tasks; whereas, in other states, governors can only use executive orders pursuant to state statutes or to respond to administrative tasks. Still, in all scenarios there is leeway for governors to use executive orders for policymaking. Understanding gubernatorial patterns of issuing executive orders helps to explain executive power and exec utive legislative relations on the state level. Looking at aggregate use of executive orders helps to understand broad patterns. Figure 4 1 shows that over the past 10 years, executive orders were used fairly frequently across all of the states. The total orders issued falls between 3,000 and 3,500 for all of the years. This suggests that there is not an aggregate trend upward or downward. Yet, Figure 4 1 shows that there is variation by total year s served in office for specific administrations. Using the 7 6 governors that have at least 8 years in office reveals that governors use executive orders more frequently at the start of their tenure, and this use quickly declines in subsequent years. Use of executive orders also differ by state Figure 4 1 shows tha t Arizona, Florida, Indiana, Kansas, Nevada and Washington vary considerably in terms of how frequently governors issue executive orders.
90 across states and administrations. This question has been partially addressed by presidential scholars, but has largely been ignored by State Politics literature. The primary research purpose of this chapter is to explain the frequency in which governors use executive orders. I do so by ana lyzing how the frequency of executive orders varies across state, across administrations, and by year, as well as by analyzing factors that influence gubernatorial use of executive orders. The overall use of executive orders influences how states create p olicies through the executive branch Furthermore, these findings can help more broadly to understand executive power on the federal level in the United States. Research on presidents is severely limited by the N of 1 problem with few observations availab le across time. President Obama is increasingly turning to executive orders as a means to circumvent a deadlocked Congress. This chapter provides insight into whether this is a route used by a more tactical executive or if this is something that occurs fre quently in the face of a gridlock or divided government. Understanding when and why governors issue orders, in addition to how this can vary by executive, provides insight to executive power for all levels of government. Executive Order Usage across the S tates their frequent use in state government. The premier work done by Ferguson and Bowling (2008) shows that there is huge variation between states in terms of how frequently governors utilize executive orders and for what purposes. It found that executive orders accomplished a range of purposes with some orders satisfying
91 administrative tasks, while others established policy. Their analysis reveals that from 2004 through 20 05 hundreds of executive orders issued across the country that establish governmental policies, clarify org anizational goals or reorganize agencies. Ferguson and Bowling show that governors in the American states frequently use executive orders, and though they do not highlight this, it shows that there is considerable variation in the total number of executive orders issued annually The frequency and purpose of executive orders vary considerably by state and ued 898 orders in these two years, The limited timeframe under analysis raises concerns regarding why such variation is seen across the states. Is the phenomenon random for all years? Is it a result of differing contexts and approaches to policy development across administrations, or is it a result of differences across the states? States and governors differ in how executives use this unilateral too. For example, Oklahoma g overnors used executive orders primarily to order flags to half staff to honor various individuals or events (Oklahoma Secretary of State 2014). Governors in Arizona were more proactive, issuing orders to establish commissions for regulation or to support their initiatives (Arizona Memory Project 2014). Neither state prohibits using executive orders for a number of other purposes; both can make administrative changes. I argue that the executive matters in establishing policy on the state level and that gov ernors use executive orders as a power play. Governors are more likely to take advantage of executive orders in two situations specific to their political context: 1) when the governor is weak relative to other executives; and 2) when the legislature is le ss
92 likely to respond or block actions. Governors that have weak institutional powers do not have as many alternatives in terms of what tools are available for them to utilize to influence policies. Additionally, governors that are weak, also frequently fac e shorter term lengths and cannot appoint as many supporters to higher executive positions to carry out desired agendas (Beyle 1999). In the former situation, governors may be forced to campaign during their second year in office, or their time in office i s limited to a single four year term. In both of these cases, governors must act quickly to establish protocol and to transition from the previous administration. Executive orders spur the transition process. The other consequence of weaker institutional powers is that a governor is limited by what options are available to them to influence policy. Appointment powers Waterman 1989; Black et al. 200 7 administration in the White House proves that carefull y selected appointees can be a very effective way of changing policies and informal behavior of government agencies (Waterman 1989). By carefully selecting appointees that held strikingly similar polit ical views and policy objectives, he was able to influe nce policies in a variety of policy areas, such as the environmental regulation and healthcare. 1 Other powers that varies across the states, and that many relatively stronger governors can use, are the ability to design the executive budget and veto powers In all states, governors help to establish the executive budget, but this influence ranges from total control, with no outside input from the legislature, to moderate influence, with the 1 For more information, see: Waterman (1989); Layzer (2012): Chapter 13.
93 governor act ing as one of many actors that establish the budget (CSG 2014). The power to veto is not in itself a policymaking power, but it does give the governor the ability to prevent the legislature from changing the status quo. All governors have the ability to ve to legislation, but not all governors can line item veto legislation, and the required vote from the legislature to override vetoes ranges from simple majority to special majority. I expect governors to rely on executive orders more to pursue their agenda when they are less likely to turn to other institutional powers The alternative situation that governors use executive orders is when the legislature is less likely to respond with legislation that overturns the ir executive orders the premise of the S trategic Model. I expect g overnors to issue executive orders more during unified government because there are fewer legislators that will act to oppose executive action. Governors may choose to enact policies through executive orders because opposition is unlikely to mobilize a majority within the legislature to oppose policies made unilaterally. Unified government, or simply more control within the legislature by members of the same party as the governor, removes a large degree of the threat that the legis policymaking tools. While I expect governors to issue orders based on their political context, executive orders are political tools that governors can utilize regardless of the partisan or ideological makeup of the legislature, so at times it is appealing, or the only option, for the governor to enact policies that were promised to supporters while campaigning. I expect g overnors to issue executive orders more early on in their administratio n in order to reward or quickly establish policies. G overnors use this as a means to appeal to their
94 partisan base or prospective constituency to create or adjust policies (Bishin 2000), as well as to adjust policy unilaterally devoid of ideological compro mise. This allows governors to position take and provide citizens with new information regarding their policy stances in order to appeal to their potential supporters. Governors act not only in short term to establish policy, but also in the long term to e nsure their own viability in future elections. I expect governors to turn back to using executive orders in later years once their political clout fades I also expect g overnors to respond to other policymaking institutions direct democracy, and the stat e legislature. I expect g overnors to use executive orders less frequently when citizens place more concerns on ballot, specifically when citizen legislation originates from the citizens initiatives and popular referendums. Statewide constituencies elect governors and approve of ballot measures. Therefore, I expect the statewide electorate to select governors in tune with the state as a whole. For this reason, I expect g overnors to issue fewer executive orders when citizens are more active. There is less o f a need for the governor to act because the citizens are enacting policies through direct democracy. However, the districts within the state elect members of the legislature Legislators reflect their constituents and lobby for regional interests, which m ay place the interests of the legislature at odds with the governor. I expect the reverse relationship with legislative activity Different constituencies e lect s tate legislators so the legislature may enact policies that benefit the majority within the l egislature, but not necessarily the state as a whole. L egislation may be at odds with the I expect
95 governors to use executive orders more when the legislature is more active in order to either qualify legislation or to add direction to regulatory agencies. Macro level Gubernatorial Use of Executive Orders Presidential scholars originally chose to focus on the macro level of executive orders. By this, I mean that scholars analyz e what leads to the aggregate count of executive orders issued in a given year and state. Scholars that approach executive order use by focusing on the total use of executive orders tend to focus more on partisan explanations, divided government and length of time in office. These scholars find that presidents are more likely to issue executive orders at the start of their tenure, as well as towards their time in office, and the makeup of the legislature explains how executives approach their use of unilate ral action. Yet, presidential literature cannot assess whether presidents turn back to executive orders later in their term due to impending term limits forcing them to act before leaving office or due to declining prospectus of success in the legislature as time passes. This is one of the questions that I address in this chapter. An important theory discussed on the macro level is the Strategic Model, as termed by Deering and Maltzman (1999). This model suggests that governors are more likely to issue ex political agenda. In other words, presidents are more likely to issue executive orders when the legislature is controlled by the oppositional part. However, scholars do not find much s upport for this theory on the federal level (Mayer 1999; Howell 2005), and it has not been tested on the state level. While not the primary goal, I test this theory with my analysis. I argue that relative power is the largest factor to explain gubernatoria l
96 behavior, but governors are more likely to issue executive orders when they confront a hostile legislature Bolton and Thrower (2015) find that presidents Congress constrains the president better when the legislature is composed of more ideologically sim ilar members; yet, when there is greater divergence within the legislature, the president is less restricted in terms of executive orders they can issue without the legislature mobilizing against the president Power is partially correlated with the makeup of the legislature, but the analysis does not find significant multicollinearity concerns. I define macro level for my dissertation to mean the aggregate use of executive orders in a given state, any given administration. In other words, the analysis of t he macro level will look at the counts of the executive orders issued in a given year in a given state and within a particular administration. This is one observation in the dataset. Previous studies did not control for the administration itself. In most i nstances, this was not possible. Presidential scholars do not use multilevel models and the few studies that considered gubernatorial use of executive orders either did not conduct advanced analysis or did not control beyond the state and year. However, if as scholars on administrative control and management styles exert, the specific executive is important to understanding the informal power of governors or presidents, then there are powers that cannot be directly measured or controlled for by simply contr olling for the state and year. There are differences by the individual executive that must be considered. Controlling for administration in time series cross sectional analyses provides a rough the state. I argue that over the course of a year weaker governors issue more executive orders Unlike federal government, research focused on the state level can assess the
97 role of the degree of formal power granted to the executive branch as a possible reason of differing approaches to use of executive orders. As outlined in the theory section, I expect weaker governors to rely on executive orders more because they have fewer options outside of the executive borders. Governors with weaker institutional power are more likely to turn to executive orders when fewer options are available to them to influence their administration or make headway in the legislative arena. I also expect governors to use executive orders differently based on how long they have served in office Governors frequently issue executive orders upon entering office. This is because in some states executive orders terminate after 30 days in office. This is also because governors want to reward partisan supporters and signal administrati ve positions (Sellers 2014). The newly entering administration can issue executive orders to create new policies. This allows the newly elected governor to quickly establish policies and objectives of the administration. Once the governor creates his or he r desired policies or directs agencies to alter the initiatives there is less of a need to continue issuing executive orders, so the aggregate use of executive orders declines. Governors continue using executive orders throughout their tenure, but not as f requently since they have already acted unilaterally on concerns they feel were most pressing. Presidential scholars find a similar pattern of decline after the start of an administration with an eventual upswing in use of executive orders once the presi dents nears the end of their administration. However, governors in 14 states never face term limits, so governors in these states can serve indefinitely. This analysis distinguish es between lame duck executives and governors who are later in their tenure. I expect
98 governors that are about to leave office, i.e. lame ducks to issue executive orders more I model the relationship of years served in office as nonlinear to test if governors use executive orders more at the start of their administration followed by an uptick in the later years of administration s I account for governors who are lame ducks. I expect to governors to use executive orders as their time in office closes. I also expect governo rs to issue more executive orders as the number of years served in office increases. I also expect the partisan makeup of the legislature to influence over behavior. I expect governors confront a divided government to use executive orders more often, to fo llow the Strategic Model However, I expect this analysis to differ from presidential while governors alter their strategy based on the partisan makeup, they still turn t o executive orders more when it seems infeasible to pursue policy objectives in the legislature. What Causes Governors to Issue Executive Orders? Governors use executive orders at different rates and for different purposes. Therefore, it is undoubtable t hat executive orders usage should change by year to year and administration to administration. However, use of executive orders is not completely explained by the specific governor and the political conditions they confront. Another important element to co nsider is the state in which a governor serves. The state is key to explaining the frequency and purpose that a governor issues an executive order. For instance, Florida, Georgia and Kentucky all require governors to issue executive orders in order to hire new employees to state government and to appoint numerous positions.
99 This naturally inflates the overall number of executive orders issued across the years. Similarly, in some emergency declaration and changes to state protocol in response to emergencies must be made in order for the state to continue to function. States have their own traditions and patterns of executive order usage that must be taken into account when discussing gubernatorial behavior. Data and Expectations The dataset created for this dissertation ranges from 1980 2014, which includes 35 years The dependent variable is the count of how many executive orders issued in a given state/administration/year. Figure A 1 shows the distribution of counts for all states and years included in anal ysis, with the exception of the three outlier states Florida, Georgia and Kentucky The independent variables included can be grouped into four classes of variables gubernatorial factors, partisan factors, institutional factors and controls. One large component of the overall use of executive orders are covariates specific to the governor In addition to controlling for state and administration, I expect the length in office to influence how frequently governors use executive orders I expect g overnors to issue more orders to establish policies or reward supporters early on in the administration. I include Years Served in Office and expect it to have a negative relationship with total executive orders issued. However, the presidential literature asserts that this relationship is curvilinear. I allow for this possibility as well so that results are comparable to presidential studies and to test whether time served in office or term limits better explain gubernatorial behavior Governors use executive orde rs most often at the start of their tenure, followed by a steady decline, and finally ending
100 with an upward tick as their time in office comes to a close As gubernatorial popularity declines and political capital is spent by the governor, governors will b egin to use executive orders towards the end of their tenure. Years Served in Office 2 tests this relationship This provides the basis for the initial hypotheses: Hypothesis 1: Governors issue more executive orders at the start of their t enure all else equal Hypothesis 2: Governors turn back to using more executive orders late in their tenure all else equal One large premise of this dissertation is that governors act differently based on the insti tutional power available to them. Governors with l ess institutional Gubernatorial Power use executive orders more as an overall strategy to shape their administration in lieu of alternate executive tools Governors with more power have more options to influence agencies and the policymaking process, but w eaker governors must select from within their limited options. Therefore, I expect weaker governors to turn to ( 2007 ) Institutional Power I ndex. 2 This is a time variant, continuous composi te measure that ranges from one to five and varies annually. It takes into account six different dimensions of gubernatorial power separately elected executive offices, tenure 2 This was originally created in 1960 for the National Governors Association, but reappeared in multiple editions of Politics in the American States : A Comparative Analysis with the coding scheme was further developed with later editions, but Beyle did not generate estimates for every year. I recreated the measure based on the provided coding scheme. The coding used herein is the amalgamation of all of the coding schemes outlined in the text, and may deviate from reported values in the text from earlier years due to minor changes to the coding scheme See Table B 1 for the coding scheme.
101 potential, appointment powers, veto power, budgetary influence and partisan co ntrol of the legislature. My hypothesis for institutional power is: Hypothesis 3: Weaker governors use more governors in a given year, all else equal. I expect characteristics specific to the governor to also influence their use of executive orders. Gov ernors with experience I the legislature have a better understanding of how the legislature function, so I expect executives with this experience to deploy executive orders differently than those with no experience in government, in statewide elected posit ions or in other governmental roles I expect g overnors who have experience in the state legislature to use executive orders less due to previous experiences as a legislator because they know how to anticipate the behavior of the state legislature better t han executives without this experience This leads to the following hypothesis: Hypothesis 4: Governors with state legislative experience issue fewer executive orders in a given year than all other governors all else equal The political party in gov ernment also plays an important role in understanding how governors use their power. The makeup of the legislature, along with the partisanship of the governor is important to understanding executive legislative relations. The partisanship of the governor is controlled with the dichotomous variables Democratic Governor and Independent Governor Yet, partisanship of the governor
102 alone should not influence their use of executive orders ; rather, I expect gov ernors to behave differently based on how much of the legislature is composed of members that share their party affiliation Presidential literature finds that presidents use executive orders less during divided government (Mayer 1999; Howell 2003; Fine and Warber 2012). Therefore, I expect g overnors to use executive orders more when a legislative check cannot, or is unlikely, to occur. Therefore, governors should use executive orders more as the percentage of the legislature is composed of more legislators that are members of the same political party as the governor. The Same Party Control of Legislature is measured as the average percentage of the legislature that is composed and should be associated with governors using executive orders more frequentl y The following hypothesis asserts: Hypothesis 5: Governors issue fewer executive orders when a higher proportion of legislators from the same political party as the governor serve in the legislature all else equal. Other policymaking institutions ca n also influence gubernatorial activity. As stated previously, when citizens are more active with direct democracy, governors do not have to take as active of a role in policy creation so I expect governors to use executive orders less often as Citizen Ac tivity increases This measure of direct democracy takes into account all citizen legislation that generated by the public initiatives and popular referendum, and accounts for their proportioned passed. Unlike direct democracy, I expect the reverse relati onship of Legislative Activity with executive
103 orders. Governors use executive orders to further stipulate policies or protocol so more active legislatures should prompt executive to issue more directives regarding the way that the executive branch impleme nts legislation Legislative Activity takes into account total bills introduced in the legislature and the proportion enacted. I create estimates of Citizen and Legislative Activity with the following formula: In both of these cases, the total policies proposed have a positive, but diminishing, effect on activity. The v alue is then adjusted by the proportion passed to punish, or account for, states that propose more citizen legislation or bills, but do not pass most of them, such as California for direct democracy or New York for state legislation. Both of these measures incorporate moving averages by including the current year, in addition Citizen Activity accounts for the current year activity, along with three previous elections Figure A 2 provides the distribution of non zero values of Citizen Activity I create Legislative Activity by using the total bills propose d and enacted in the current year and prior year 3 The final hypotheses assert: Hypothesis 6: Governors issue fewer executive orders when citizens are more engaged in creating legislation through direct democracy, all else equal. 3 By including two years, accounts for biennial state legislatures to better measure the activity of each legislature. Figure A 3 provides the potential values of Legislative Activity. It shows how Legislative Activity changes as proportion of bills passed increases and where most observations in the dataset fall. Using annual estimates, it plots the average bills proposed, as well as the upper and lower 5 percentile of bills proposed to demonstrate how Legislative Activity changes.
104 Hypothesis 7: Governors issue more executive orders when the legislature is more active in proposing and enacting legislation, all else equal. I control for temporal relationships in the models s ince my analysis spans several decades First I include a lag of the total execu tive orders issued. This helps to account for autocorrelation or the correlation in errors that occurs over years This has the benefit of accounting for correlating errors in the estimation that can shrink p values and inflate statistical significance. S econd, I control for the year with a linear estimation of time 4 Estimation Methods I argue that the state and governor matter to explaining gubernatorial use of executive orders. If this is true, then any analysis must correctly model the variance that exists within both of these levels of analysis. This argument suggests that a three level model is most appropriate to estimate the mechanisms that drive gubernatorial use of executive orders This analysis includes o ver 200 administrations across all 50 s tates so it is not feasible to include a dummy for each state and administration. Fortunately, the administrations are nested within states. Multilevel analysis take s into account that some of the variation is a result of a governor working in a particula r state and administration. The analysis is a mixed effects model that uses fixed effects of the independent variables and random effects to account for the variance of the state and/or administration. 4 After analysis, a linear estimation is the best estimation of the temporal relationship. I also modeled time as a quadratic, cubic and quartic function; however, all measures of fit indicated that a linear function provides the best estimation of the temporal relationship.
105 For this chapter, the dataset is structured so that t here are three levels, which allows analysis to explore where variation in the use of executive orders exists Level 1 is the count of the total number of executive orders issued in that state for that year. Level 2 is the administration throughout the gov time in office which typically includes multiple years. Level 3 is the state. Table 4 1 presents t he key characteristics of the executive orders. The analysis includes 220 administrations from 47 states. 5 T he average length of administrations is 5 .22 years which indicates that the average administration lasted just over one term The interclass correlation (ICC) shows that about 19 percent of the variation within the dataset occurs across administrations. Of the 47 states included, the average num ber of years observed is 24.45 for each state. T he ICC shows that 53 percent of the variance is seen across states which indicates that most variance occurs across the states However, considerable variations occurs on all levels of analysis. For this ch apter, the entire timeframe is analyzed using multilevel models and Time Series Cros s Sectional (TSCS) analysis First, I estimate a Three Level Unconditional Multilevel Model to determine the distribution in variance. This analysis is key to understanding why differences exist across the states and governors. It also provides insight into how to properly specify models throughout analysis in this dissertation. For instance, if most of the variance exists across the states, then factors that differ by state such as power or legislative professionalism, should be controlled for 5 As noted earlier, Florida, Georgia and Kentucky are ext reme outliers. These states issue thousands more executive orders across the time period under analysis, which heavily skews results to reflect trends found within their states. This chapter omits them from analysis to estimate use of executive orders more accurately.
106 or considered in analysis. By using the proportion of variance in level 2, I can explain how much variation is explained between administrations By using the proportion of variance i n level 3, I can explain how much variation is explained between states. The significance of the variation is tested with likelihood ratio tests. The initial findings from the unconditional model helps to develop the model construction throughout the remai nder of the dissertation. Since this chapter is concerned with how the counts of executive orders vary across administrations and states, I begin analysis with an unconditional three level model to help guide specification of subsequent models. This model assumes that the state and administration have their own random effect on the count of executive orders, unconditional on other variables. Robust maximum likelihood estimates t he models and a log likelihood test evaluates fit. The following equation estima tes the unconditional three level model : Where Executive Orders iat is the count of the executive orders issued in a given 000 is the mean total of executive orders issued, 00t is th e random effect of the state, r 0at iat is the random effect of the year. The remaining analysis utilizes Time Series Cross Sectional (TSCS) analysis, specifically with Multilevel Negative Binomial Models, to determine what influences gubernatorial behavior. This allows for the models to take into consideration the
107 temporal and political context of the specific governor. They all assume overdispersion and if the dispersion equals zero, the model reduces to a m ultilevel Poisson (Long 1997). An additional advantage to TSCS analysis is that I can include more observations because the analysis accounts for the fact that the dataset is not balanced, but still provides reliable estimates. To develop an understanding of how statistical significance changes once the different levels of analysis is controlled, I estimate t hree count models: the first only controlling for observational level variance; the second controlling for the state level heterogeneity; and the final controlling for the state and administration heterogeneity. The final model is estimated with the following equation: In this model I indexes the state, a indexes the administration and t indexes the year. is a matrix of th e independent variables. is a matrix of the control variables. T is a time counter variable. 6 are observation level, administration level and state level variance terms, respectively. The simplest mo del controls for only observation level variance, the next model controls only for state level variance, and the final model controls for all three variance components. Each additional variance component controls for heterogeneity cause by that particular level (Raudenbush and Bryk 2002). The heterogeneity occurring in the most basic model is most likely a result 6 Time was specified in multiple ways linear, squared, cubic and quadratic the linear specification yielded the best fit for all of the models.
108 of state level and administration level heterogeneity, meaning that heterogeneity should decline significantly once the models allow for random ef fects to capture the variance that exists as a result of the state and administration. T he negative binomial model accounts for any remaining overdispersion with an observational level variance component (Long 1997). Unconditional Three Level Model Before advanced analysis can be done, an unconditional model tests what levels contain the most variance and should be controlled for in later analysis. Thi s is done by first running an unconditional three level model. This analyzes the outcome, the total execut ive order, without regard to any other factors than the three levels. Table 4 2 presents t he results of the unconditional three level model. All three levels are statistically significant. This indicates that a statistically significant degree of variance is seen within and across administrations, as well as across states. As the ICC indicate s approximately 50 percent of the variance is seen between states, and roughly 20 percent of the variance occurs within states. Multilevel Negative Binomial Models T he unconditional model indicates that each level of analysis contains statistically significant degree of variation. This indicates that a three level multilevel model allows for the best estimation of the use of executive orders. I now turn to analysis to determinants of gubernatorial use of executive orders. Table 4 3 shows t he results of the Negative B inomial models The findings are fairly consistent across the models However, the statistical significance changes for a couple of the terms The partisa nship of the governor has little bearing on how governors use executive orders. In the second
1 09 model, that controls for states, the results find that Democratic Governors are positive and statistically significant (p<0. 10 ), but this result only just reaches statistical significance T he incidence rate ratio (IRR) for Democratic governors indicates that the substantive effect of partisanship is negligible (1.06). Still, this suggests that Democratic governors are more likely to issue executive orders than Rep ublicans, and suggests that partisanship should be further evaluated as a predictor. However, once analysis controls for the variance on the administration, this relationship loses statistical significance. The term Independent Governor is not statisticall y significant in any of the models. As a whole, these covariates indicate that partisanship does not distinguish gubernatorial use of executive orders. Despite partisanship not playing a significant role, other gubernatorial factors help to explain use of executive orders. State Legislative Experience is statistically significant throughout the models and suggests that previous experience in the state legislature is linked to a marginal increase (a factor of 1.10) in use of executive orders. Governors use executive order more if they have experience in the legislature. Unlike hypothesized, this experience leads to greater use of executive orders. For example, if governors without legislative experience issued 50 executive orders, then if all else were equal we would expect governors with legislative experience to issue approximately 55 orders. Years Served is robustly significant throughout the models. Both terms estimating time in office show that years served in office retain high statistical significant in all of the models. The relationship shown indicates that gubernatorial use of executive orders is curvilinear i.e. governors begin their administration by using more executive orders, and then their use quickly declines; yet,
110 later in their administrat ion, they begin to use more executive orders once again (with everything else held constant). This is consistent with the proposed hypothesis and with findings that analyze presidential use of executive orders. Gubernatorial Power is also robustly signific ant All of the models show a negative and highly statistical significance of Gubernatorial Power with aggregate use of executive orders. This supports my theory that weaker governors issue executive orders more than stronger governors when focusing on the macro level of analysis In the final model that controls for random effects of the state and administration the IRR is 0.72, which indicates that for a one unit increase in Gubernatorial Power, we expect the total executive orders issued to decrease by approximately a factor of 1.4. To put it in concrete terms in a state where the governor issues 50 executive orders a year were to increase gubernatorial power by one unit, we would expect approximately 36 executive orders, all else equal. Additionally, i f a 2.5 unit change in Gubernatorial Power (of a composite index that ranges from 1 to 5) occurred, we would expect a decline in total executive orders by a factor of 2.25. Partisanship play s an indirect role in gubernatorial use of executive orders. The partisan dynamics between the executive and legislative branches of government helps to explain when governors turn to executive orders. I find that the % Same Party Legis lators is positive and statistically significant. This indicates that the higher the percentage of legislators that are members of the same party as the governor, we can expect the governor to issue more executive orders. The IRR is fairly low for a one un it change (1.004), but the IRR increases to 1.02 for a 5 percentage change, 1.04 for 10
111 percentage change and 1.06 for a 15 percentage change. This supports the proposed hypothesis and remains consistent with presidential literature that finds that divided government and fewer same party members in the legislature decreases use of executive orders. Institutional actors appear to also play a role in executive orders. Citizen Activity is negative and statistically significant in the first model While it no longer reaches conventional statistical significance once controlling for the state and administration, the coefficient remains negative throughout the models. This means that a statement can be made about the states, but not specific governors. Governors in states that use direct democracy more are less likely to issue executive orders. The IRR indicates that for a one unit increase in Citizen Activity, e xecutive orders decrease by a factor of 1.35 (an IRR of 0 .72). This supports the proposed hypothesis th at governors use fewer executive order when citizens engage in making citizen legislation more. As the state as a whole is more involved in policymaking through direct democracy, governors use executive orders more. The statistical significance decreases o nce we control for the effects of states and administrations. Legislative Activity does not have a consistent influence on executive orders. I expected fewer executive orders associated with more active legislatures. The coefficient is consistent with this expectation once the state and administration analysis controls for the effects of the state, and administration; yet, none of the models reach statistical significance. While I expect the legislature to heavily influence the executive branch, their actua l productivity in producing legislation may not influence the governor as much as their partisan makeup or the previous experiences governors have when working with the legislature.
112 I control for two temporal concerns of gubernatorial use of executive ord ers. To control for autocorrelation and correlation in the errors commonly associated with time series analysis, I include a simple time counter for the year, and I control for the Ti me is significant across the models but this term loses statistical significance when controlling for the state and administration Since the dataset is unbalanced, no meaningful interpretation can be made regarding time, and descriptive analysis reveals that this varies by state It is likely that the finding in the standard Negative Binomial model is spurious and biased due to states with longer time spans influencing the results more than states with shorter timespans available Additionally, the findin gs show that governors use executive orders more frequently as the number of executive orders issued in the previous year is higher. In other words, there is evidence that states that use executive orders more frequently in a given year are more likely to issue executive orders in the subsequent year, which is indicative of traditions within administration or states from year to year. There is a strong autocorrelation, so including the lag of the previous year helps to remove correlation in the error term. The AIC of all of the models indicate that they outperform the unconditional model. Each additional variance component shows significant improvement over the previous models, and the AIC for the final model shows a significantly better fit compared to all of the other models. Discussion I show that the number of executive orders varies greatly across all levels states, administrations and across years. Over half of the variation is seen across states, but approximately 19 percent of the variation is acro ss administration level
113 (level 2). This has important implications because it suggests that political context matters. Variance is not isolated to a single level the state and administration play a role on how frequently governors issue executive orders. In other words, while the particular state in which a governor works is important, it is not the single most important factor to explaining gubernatorial use of executive orders. Governors use executive orders with varying frequencies, and particular even ts in time can cause a spike in total number of executive orders issued in a given year. Presidential scholarship offers several reasons for why executives resort to unilateral policymaking divided government, disasters, public approval ratings, executi ve charisma etc. These possibilities suggest that the varying use of unilateral powers is a result of political context. While quantitative analysis cannot pinpoint the cause, it reveals patterns that cannot be seen on the federal level due to data restric tions and limited institutional variation. Significant variation occurs across administrations, as well as time. I also find support for the Strategic Model proposed by presidential scholars. Executives are more likely to use executive orders with when the that the relative strength of governors plays a role in the frequency in which governors issue executive orders. Relatively weak g overnors are more likely to turn to ex ecutive orders than their counterparts in other states. Additionally, governors are more likely to use executive orders at the start and nearing the end of their administration. Unfortunately, analysis could not include all years and states. Analysis exc ludes the Florida, Georgia and Kentucky because they are extreme outliers. Additionally, accessibility of executive orders to the public differs across the states Analysis does not
114 include all years for all 50 states. In statistical terms, the dataset is not balanced This leaves the potential flaw that the Negative Binomial models provides biased results that better explain states with more observations This is a concern for the models that do not control for the variance within the state and administrat ion levels, but analysis in the final model mediates this concern for two reasons First, the time period selected restricts the years to range from 1980 to 2013. All states, with the exception of Arkansas, include at least 11 observations and two administ rations The longest time period is 35 years. The difference in years included for the states with the most observations compared to the states with the fewest years does not raise concern that estimates favor states with more observations Analysis accoun ts for over 200 administrations in 47 states. The resulting models are extremely powerful by this, I mean that we are highly confident that the findings are in fact true relationships and are not spurious. 7 Second, multilevel modeling accounts for concer ns with unbalanced data. The only models that the estimates may be a concern are the negative binomial models that do not control for state or administration heterogeneity. Even for the model that does not account for these levels, it accounts for overdisp ersion with an additional term for observational level variance. An alternative concern is that executive orders accomplish different tasks, as shown in Chapter 4. The quantity and use of executive orders varies considerably by state. Some orders establish policy; whereas, others are more administrative or 7 For more information on measuring power of multilevel models see: Konstantopoulos (2008, 2009 ), Schochet (2005).
115 of policy oriented executive orders. Again due to time restrictions, this could not be accomplished, since it requir es coding over 32,000 orders based on content. However, since governors within states are fairly consistent in terms of the patterns and purposes upon which governors use executive orders, as shown in the previous chapter, this limitation is not as problem atic as it suggests. Symbolic or policy oriented executive orders should be issued at a fairly consistent rate throughout the time period; therefore, future research sh ould study this more in detail. Despite these limitations, these findings provide critical insight regarding Ferguson and Bowling (2008), analyzed solely 2004 and 20 05. This study analyzed orders with descriptive statistics. While their study offers great depth in understanding use of executive orders over time or administrations. T his chapter explores that gap in knowledge. I show that context matters the state and administration help to explain gubernatorial use of executive orders. The state and governor under analysis matter greatly. States vary in their use of executive orders owing to their different constitutions traditions and powers granted to governors, but the amount of variation across administrations (~20%) is sizable and not addressed by previous research. However, resident, in this case executives, matter. Along similar lines, I argue that there are characteristics about governors that explain their use of executive orders and other unilateral tactics that they deploy throughout their tenure, beyond partisanship.
116 These results also reveal that partisan and state institutional factors matter. Governors are more likely to issue executive orders when members of their same party comprise a greater proportion of the legislature. This also means that governors are less l ikely to issue executive orders during divided government, which suggests that governors are more likely to make use of this unilateral policymaking tool when the legislature is unable, or unlikely, to mobilize a coalition in opposition to executive action Additionally, governors respond to other policymaking institutions. Governors do not issue executive orders as frequently when citizens are more active with direct democracy. These findings open up several areas of inquiry. First, research should study the implications of executive orders. On average, these 47 states issued 27.28 executive orders a year. Research should address how these executive orders influenced statute adoption across the states. Second, research should further look to determine how the content in executive orders influences executive legislative relations. This would help to understand factors leading to more executive orders, as well as the implications of these orders.
117 Figure 4 1 Annual total execu tive orders issued. A) Arizona, B) Florida, C) Indiana, D) Kansas, E) Nevada, F) Washington.
118 Table 4 1 Summary of e xecutive order s used in analysis Administration Level Total Administrations 220 Average Administration Si ze 5.22 ICC for Administration level 0.19 State Level Total States 47 Average State Size 24.45 ICC for State level 0.53 Total Observations 1,149 Total Executive Orders Issued 32,550 Annual Average of Executive Orders Issued 27.28 Note: This is excluding FL, GA and KY
119 Table 4 2 Results of the unconditional three level model. Component Estimate S.E. Within Administration Variance 57.55** 13.89 Between Administration Variance 37.35** 8.08 Between State Variance 106.03** 26.05 Average Executive Orders Issued 1 18.77** 1.59 Total Observations 1,149 AIC 8,338.6 BIC 8,358.8 Adjusted BIC 8,346.1 Note: 1 The dependent variable in a given state/administration/year is coded as the count of orders issued. The first column of each cell entry provides the coefficient estimates. The second column of each model provides the standard errors in parentheses. 1) This denotes the average of the entire included population, not the annual average. p < 0.10. *p < 0.05. ** p<0.01.
120 Table 4 3 Multilevel negative binomial model estimating executive orders issued in a given year/state. Covariates Estimate I.R.R. Estimate I.R.R. Estimate I.R.R. Democratic Governor 0.03 1.03 0.06 1.06 0.09 1.09 (0.05) (0.04) (0.06) Independent Governor 0.12 0.89 0.02 1.02 0.04 1.04 (0.20) (0.16) (0.19) Gubernatorial Power 0.26** 0.77 0.39** 0.68 0.33** 0.72 (0.05) (0.09) (0.10) State Legislative Experience 0.08 1.08 0.09* 1.09 0.10 1.11 (0.04) (0.04) (0.06) Years Served by Governor 0.10** 0.90 0.12** 0.89 0.12** 0.89 (0.02) (0.02) (0.01) Years Served by Governor 2 0.01** 1.01 0.01** 1.01 0.01** 1.01 (0.00) (0.00) (0.00) % of Legislators from the s Party (log) 0.01* 1.00 0.01** 1.01 0.01 1.00 (0.00) (0.00) (0.00) Citizen Activity 0.30** 0.74 0.18 0.84 0.12 0.89 (0.11) (0.13) (0.14) Legislative Activity 0.05 1.05 0.11 0.90 0.08 0.92 (0.05) (0.08) (0.08) Time 0.01** 1.01 0.01 1.00 0.01 1.00 (0.00) (0.00) (0.00) Previous Yea r 's Total Orders 0.01** 1.01 0.01** 1.01 0.01** 1.00 (0.00) (0.00) (0.00) Constant 8.94 0.67 2.07 (4.61) (3.79) (5.60) 0.94** 1.91** 2.45** (0.05) (0.0 7) (0.09) 0.32** 0.40** (0.07) (0.10) 0.091** (0.016) Total Observations 1,008 1,008 1,008 2 1,174.3** 254.4** 146.3** AIC 7,574.7 6,992.3 6,885.6 Note: The dependent variable is the total number of ex ecutive orders issued in a given year/administration/state. p < 0.10. *p < 0.05. **p<0.01.
121 CHAPTER 5 FUNCTIONS OF EXECUTIVE ORDERS: A LOOK AT GUBERNATORIAL AND PRESIDENTIAL ACTIVITY WITHIN POLICY AREAS Summary Chapter 5 analyzes what executive orders actually do. This chapter looks at the intent of executive orders issued from 2004 to 2014. I conduct content analyzes of executive orders to determine why these executive orders were issued throughout the time under analysis for two distinct policy areas health and human services, and emergency preparedness I find that most executive orders issued in these policy fields either establish new policies or to make administrative adjustments to policies already in place. The majority of executive orders alte r current policies that were originally created by earlier executive orders in previous years This provides findings counter to what presidential literature concludes, most executive orders in fact influence the current policy framework and governors issu e these orders to make meaningful changes rather than to send signals to bureaucrats or the public. Shaping Policy Areas Thus far, I argue that motivations to issue executive orders differs based on the level that analysis By this, I mean that the aggrega te number of executive orders is explained by different covariates than when studying when governors issue specific executive orders. In this chapter, I explain gubernatorial use of executive orders at the meso level or in other words, I analyze what exec utive orders do within policy areas I show in chapters 4 and 6 that many factors influence orders Presidential scholars focus on control of the legislature, but I assert that factors specific to the governor also influence use of executive orders For instance, governors behave differently at the start of their tenure as compared to later years and weaker
122 governors are more likely to turn to executive orders owing to their lack of options. I argue that governors do this to esta blish their initial policies Motivation to issue new orders dissipates as governor s finish transitioning into office. However, governors eventually turn back to executive orders later in their tenure to launch new policies that seem unlikely to pass in th e legislature prior to the end of their term. I also showed that features specific to the governor, such as partisanship and tenure, help to explain when policy area. Addi tionally, explaining explicit policy adoption of the executive orders is explained largely by the political context in which the governor is facing. In this chapter, I analyze executive orders issued in two different policy areas in 22 states from 2004 to 2014. First, I look at executive orders influencing health and human services from governors (and to a lesser extent presidents). Interestingly, governors frequently turn to executive orders to start initiatives, programs or councils to influence the publi c health and safety. Presidents do not use executive orders in the same manner. Executive orders that presidents issue for health and human serves were related to specific legislation and did not approach the issues as proactively as most governors. Most o f the executive orders were also administrative to amend already established health programs. Next, I look at another current hot topic terrorism, specifically terrorism prevention and preparedness for emergencies. Here we find the reverse pattern of act ivity. Presidents were not as active in public health and governors were very active. However, in terrorism prevention presidents are the more active executives. Governors were also active, but much of their activity was in response to either federal legis lation,
123 presidents or the state legislatures to prepare to possible threats from terrorist organizations. This shows a clear distinction from health and human services where governors were very proactive. Yet, governors were not proactive, or even relative ly as active, for terrorism preparedness as health and human services. This may be due to a number of factors, such as the federal government taking more initiatives and starting more programs that required a response from the states. Regardless of whether executive orders are proactive or reactionary i t is clear that executive orders used in state government focus much more on policy when dealing with modern pressing issues. Presidents tended to issue executive orders that dealt with executive reorganizat ion and amending previous policies, but the majority of executive orders issued by governors were focused on new initiatives, councils or implementation of policies. These findings are actually counter to what presidential literature finds recent literat ure suggests that executive orders do not serve much of a functional purpose. This discrepancy in findings is reasonable considering that this chapter finds that governors and presidents use executive orders for different purposes. Governors issue many adm inistrative and executive orders with other intents, but many of their executive orders focused on health and terrorism prevention are plainly meant to alter state level policies. Meso Level Analysis The meso level of analysis is the intermediate level bet ween the micro and macro level It examines the role of executive orders in shaping public policy within specific policy areas. The mico level analysis considers factors that drive the adoption of individual executive orders, but this level includes many executive orders that
124 collectively shape a policy area The theory driving the macro an d micro level of analysis focuses on politically motivated reasons to issue executive orders, and argue that governors systematically use executive orders to pursue the ir policy objectives. In large part, governors confront the same political conditions throughout the year, so governors use systematic political strategies that can be modeled analyzing the aggregate use of executive orders Political factors also influenc e the deployment of specific controversial, or partisan agenda or spearheaded initiatives Political factors drive behavior when dealing with c ontroversial or partisan owned topics because governors have g reater motivation to act unilaterally to create or maintain the se specific policies, but governors cannot indefinitely turn to executive orders to pursue policy objectives without sparking public outcry or invoking retribution from the state legislature in the form of legislative vetoes or coalitions forming to prevent The meso level bridge s the gap and explaining ca pability of the micro and macro levels of analysis. Up until this point, I have discussed specific policies and discussed broad patterns of gubernatorial strategies to use executive orders, but this chapter gives little insight into the actual uses of the executive orders. The meso level looks within policy areas broadly and analyzes the purpose of all the executive orders issued within the time period under analysis Design and Case Selection I focus on two policy areas to develop a deeper understanding of how executive orders shape policy areas within the United States I focus on healthcare and
125 emerg ency preparedness owing to the activity of the states and federal government in recent decades. The media consistently discusses a ll of these policies areas These are important issues to the public, so calls to improve current policy frequently appear on the political agenda for state and national governments The recent healthcare reform from the Obama Administration sparked current interest in healthcare, but reforms were made throughout the 1900s. Additionally, all governments became particularly active and cognitive of emergency response in the past decade and a half since the 9/11 attacks Y et, the 9/11 attacks did not initiate interest in emergency response. This was a common concern throughout current decades due to weather disasters, mass shootings and a lack of coordination among the levels of law enforcement. The federal government and state governments are very active in all of these policy fields. This study focuses on two policy areas that governors are consistently active in throughout recent d ecades: healthcare and emergency preparedness Recent events, such as changes to social insurance programs in the 1990s and the adoption of Obamacare, as well as the 9/11 attacks, kept these policy fields in constant flux. Presidents and governors directed their agencies and related organizations on how to institute their desired policies, all with different levels of government and other branches of government influencing the public policy framework. Health and human services have always been a concern of state governments. As you would expect, this is still true today. The modern concerns of public health relate to disease outbreaks, communicable diseases and general threats to health or safety. Governors actually have a large sway in influencing public he alth and safety initiatives. Some g overnors can redirect money to respond to outbreaks or fund research on
126 diseases. This is typically done by declaring a public emergency and redirecting emergency funds to respond to the present health threat Additionall y, governors start their own initiatives in many states to address problems within their state or national trends that the National Governors Association is attempting to address The federal government activity in public health has typically been started by the Public Health Service and other agencies that deal with public health rather than the president himself. Presidents have issued executive orders to influence public health, but they have been less active than governors. Government activity regarding terrorism has followed a different path. For health and human services, the states have taken a more active role by establishing proactive policies through executive orders Federal government mostly focuses their work through agencies rather than Congres s or the president once major reforms are passed. However, this is not the case for counterterrorism. Presidents, Congress and agencies have all been very active in shaping the landsc ape of terrorism preparedness. State legislatures participated in this tr end as well. Coordination increased considerably after the 9/11 attacks. Part of this was due to Congress and the president mandating that states must be prepared for terrorist attacks with tested plans, but also states got involved because 9/11 raised awa reness that few regions were prepared for a major terrorist attack or a catastrophe. The need for preparedness was most evident in the states and localities that had major populations. In other words, terrorist organizations are most likely to select citie s that were among the biggest in the nation, such as Houston or New York City due to their high population and ability to cause the most chaos and gain attention For instance, the terrorists that planned the Boston Marathon
127 Bombing selected the day and l ocation due to the high concentration of people in town for the event. All levels of government constantly revise their emergency response plans. This is in part due to lessons learned from current events, problems discovered during practice exercises or p olicy learning from other governments. However, the response even by the time of the Boston Marathon Bombing, emergency preparedness plans improved drastically since the 9/11 attacks I conduct a comprehensive examination on these two policy areas rather t han focusing on a single specific executive order, as was done in the previous chapter. This approach takes a much more wide ranging take on executive orders that influence health and preparation of terrorist attacks. This allows me to analyze collectively what governors attempt to do with executive orders and the purpose of executive orders, beyond just simply describing what type of order is in place. I analyze the purpose of these executive orders in the hopes that we can find out what is causing some go vernors to issue more orders than others, as well as to determine if these executive orders effect state governments Coding Executive Orders This analysis focuses on intent of the executive orders issued by governors within each of the selected policy ar eas Due to data limitations and time constraints, I cannot include all states in this analysis. I drew a random sample of 25 states to closely review the purposes that executive orders serve across the United States Data limitations restricted the use of three of the states selected in the random draw (Arkansas, Kentucky a nd Nevada) I code the remaining 22 states in this analysis. I
128 review all executive orders issued between 2004 and 2014 for these selected states to identify all executive orders that re late to the two selected policy areas After preliminary review, I created a dataset composed of all executive order s that address health and human services, and disaster preparedness. This includes over 400 executive orders issued by governors and 108 or ders issued by presidents in this time period. This makes up approximately 7 percent of the almost 6,500 state executive orders issued in this time within the 22 states. I classified the executive orders as either administrative, budgetary, policy oriented or symbolic I then take a more nuanced approach to understanding what the goals of governors and the presidents were with their executive orders when describing the findings. While some executive orders achieve multiple goals, I code the primary purpose of the executive order. The preamble typically explains the motivation for issuing the executive order, which helps to clarify the primary objective of the order The criteria used to code these orders is outlined in the following paragraph To ensure codi ng reliability, I and another coder coded the entire dataset and find that the intercoder reliability is high. I coded the executive orders once I identified all of the relevant executive orders. I reviewed each of the relevant orders to identify the prim ary reason that governors issued the executive order. The categories are: 1) administrative; 2) budgetary; 3) policy; and 4) symbolic. Administrative and budgetary concerns overlap to some degree in that governors issue both forms of executive orders in or der to enable agencies to operate. However, the key distinction between the two is that governors issue budgetary executive orders for the sole purpose of directing extra funds to agencies so that they can maintain functions or respond to a crisis Governo rs frequently issue these orders
129 during an emergency or epidemic. I code executive order as policy when the executive order establishes new councils, directs agencies on how to implement policy or alters interpretation of previous policy. In short, these e xecutive orders changes how agencies understand or implement policy. I code executive orders as symbolic when the governor issues an executive order with the primary intent of recognizing accomplishments or individuals. Governors issue these orders primari ly to bring attention to a person or particular issue, but the executive order does not establish new policies, direct agencies or redirect funds. Functions of Executive Orders I first describe the general purpose of executive orders thr oughout this time period. Table 5 1 and Table 5 2 show the executive orders coded based on their purpose by policy area. As you would expect, given the distribution of the counts from state to state, there is considerable variation across the states. Interesting ly some sta tes that rarely issue executive orders, such as Illinois or Vermont, issued a fairly high proportion of executive orders explicitly on these two policy areas. The states were much more active in health and human services than in terrorism preparedness. 1 Ad ditionally, there are similar patterns across the states in terms of how they use executive orders. In both cases, the purpose of the executive orders was more policy orientated. Governors attempted to institute new policies or initiatives to address both policy areas. In some instances, executive orders removed policies, councils or initiatives. This was 1 This is partly due to the current time period under analysis. Once the dataset is expanded to 2000, I expect much more activity for terrorism preparedness in immediate response to the 9/11 attacks.
130 included because it aimed at changing the status quo. The next most common form of executive order was administrative. These orders would continue counci ls, initiatives or direct agencies to amend their plans. This was due to the fact that many executive orders only last for a set time period and some executive orders were designed to temporarily make new policy. While policy oriented and administrative ex ecutive orders were the most common, some governors issued orders that were budgetary or symbolic. These were among the least common. The executive orders for budgetary concerns were issued because in some states governors must issue executive orders to tr ansfer funds in times of emergencies. The symbolic executive orders were least common ; however, governors use these orders to signal their policy stance to the public. Health and Human Services Presidential Use of Executive Orders To establish an understa nding of how executive orders from governors differ from presidents, it would be helpful to first know how presidents use executive orders to influence policymaking and administrative behavior from their executive branch. Therefore, this analysis also look s at presidential use of executive orders to provide a comparison to governors. This is performed for both policy areas. Presidents issued a number of executive orders to influence health and human services in this time period. Almost all of the executive orders focused on health or improving the health of employees or citizens by encouraging activity. The majority of the executive orders issued were administrative. Rather than starting new initiatives, presidents made amendments to previous orders that car ried out the same tasks. Unlike many states, presidential executive orders remain in place indefinite ly. Frequently presidents issue
131 subsequent executive orders to adjust the policy, but presidents do not have to issue executive orders to main them as is t he case for many governors. Additionally, presidents were less proactive than the states. Governors aimed to create new programs to address health concerns or to provide better services to citizens. Yet, presidential executive orders took a narrower approa ch to changing the tasks of health agencies, what is defined as a current health threat to the country and im proving activities in schools. Gubernatorial Use of Executive Orders Unlike presidents, governors were very active throughout this time period on a range of issues. Governors issued orders that created research and regulatory councils, transferred funds to encourage research on current health concerns and to counteract public threats. Unlike the president, most of the executive orders issued by gove rnors were not administrative. Rather, they were focused on policymaking through either creating new policies, new initiatives or creating councils to oversee a new points of emphasis for governors. There were several popular initiatives that were created throughout this time period, such as initiatives to counteract homelessness or serve the homeless, health coordination and adjustments to Medicaid. The majority of these executive orders issued were focused on policy. However, other governors used executi ve orders to maintain these initiatives by either extending the program or adjusting the intent of an initiative, or the council that seeks to improve it. In the case of more administrative executive orders issued in a given state, they were intended to in fluence already existent initiatives. Many of the administrative orders were meant to adjust the policy, but some were issued as a result of how executive orders must be issued in that
132 particular state. For instance, some executive orders only lasts a numb er of years. In other states, a new executive order must be issued within a set number of days when a new governor takes office in order to retain that executive order and its intended purpose. Most of the executive orders issued by governors in this time period were initiatives pushed for on the state level or by the governor. It was not until the Affordable Care Act (ACA) that the governors start to respond to the federal government with their executive orders that deal with health and human services, sp ecifically health exchanges and analyzing the affordability of healthcare within their state. States became very active in healthcare in response to the ACA. Yet, this is really the only instance where governors consistently respond to the federal governme nt in regards to health and human services. While many of the executive orders that were administrative were for maintenance of previously issued executive orders, there were other executive orders that were in performing an administrative function. In som e cases, governors wanted to make adjustments to the previously created councils, so they issued executive orders to add or remove certain positions or seats on various councils to promote health or human services. Additionally, there were many administrat ive orders that were focused on the budget. Governors issue d an executive order to transfer funds so that the agency, typically confront ing some sort of health emergency, could you continue to function and serve their purpose. There were a few instances wh ere executive orders were issued for primarily symbolic reason; however, this was pretty uncommon and
133 made up only 1.7 percent of all of the executive orders and in most states no symbolic executive order was issued. Terrorism Preparedness Presidential Use of Executive Orders Presidents and governors acted differently in regards to terrorism. In this case, presidents were much more proactive than governors. Governors primarily issued executive orders in response to the president or federal or state legislat ion. Again, presidents made a considerable number of administrative executive orders, but there were a lot of policy oriented executive orders as well. Presidents used executive orders to make alterations to policies overseas, but some changes were made at home as well. These orders related to coordination with other levels of government and orders that involved the military. Some of the more administrative tasks involved adjusting the order of succession in times of emergency, as well as amending formal po licies relating to homeland security and the military. Gubernatorial Use of Executive Orders Governors were active in issuing executive orders regarding terrorism prevention, but not to the same extent as with health and human services. Governors issued a lmost 100 orders throughout the time under analysis. A number of executive orders dealing with counterterrorism are in direct response to Congress the president or federal agencies The executive orders issued by governors are reactionary They issue exec utive orders in response to political actors in federal government. This differs from executive orders addressing health and human services. Governors issue orders relating to health and human services largely based on their own policy agenda or in respons e to public safety crises within their own state. Governors respond to federal
134 government for emergency preparedness, but to state factors for health and human services. Executive orders were once more policy oriented for terrorism preparedness. A lot of these orders worked to create policies or systems to allow for better coordination with local and federal governments for emergencies. In some states, this involved loc al officials collect and transfer breaking news to local, state and federal officials in order to allow for coordination and to allow all governments to respond quickly to new threats. This originally was meant for terrorism preparedness, but it also allow s for use for any emergency. In fact, many of these centers and the infrastructure created in preparation for terrorist attacks can double as disaster or catastrophe response, which is why some of the orders incorporate both of these issues within the same executive orders in some cases. Governors also issue executive orders for the same reason as health and human services. In many instances, governors must reauthorize executive orders after a specified elapsed time or upon entering office Additionally, a s time went on, presidents and governors amended their initial orders. Technology improved, which enabled the coordination efforts to adapt and also improve. Furthermore, in the late 2000s, some of the concerns were regarding technology preparedness. This meant that not only was terrorism prevention in the form of direct attacks, but also bioterrorism attacks and cybersecurity concerns. The issue itself evolved and executive orders were adjusted to account for and respond to the new threats.
135 As with execu tive orders on health and human services, the bulk of the orders were policy oriented or administrative. There were not many symbolic or budgetary orders that relate to terrorism prevention. This was even less common than with health and human services. On ly one executive order was symbolic and one was budgetary, which shows that these orders were deployed for very explicit purposes. The executive orders were meant to create new policies or councils and to maintain programs that were created by the governor Discussion This chapter analyzed how governors issue executive orders in two policy areas. The goal was to understand how governors broadly use executive orders to influence current concerns in their administration. I focused on two hot topics in Americ an politics health and human services, as well as terrorism prevention. These are policy areas that all governments are constantly working on and striving to improve. There was significant activity throughout the time under analysis by both the federal g overnment and state governments. The analysis shows that governors were also quite active in shap ing the policy area with executive orders. Governors issued over 400 executive orders (approximately 300 involving health and human services, and roughly 100 on terrorism preparedness) to influence the two policy areas. The majority of the orders were focused on policy either creating new policies, plans or creating councils or departments to oversee and regulate a given concern. The other frequent purpose of executive orders were administrative orders that maintained previous policies, direct agencies to adjust policies or to make technical changes to the way the state handled the two policy areas.
136 There is a noticeable difference between how governors use ex ecutive orders for health and human services versus how they do so regarding terrorism preparedness. This discrepancy might stem from the activity of the federal government during the time period. The federal government was active with health and human ser vices in terms of having the US Public Health Service work to help lower income citizens with health responses, as well as to enact legislation to address public concerns. Yet, there was no large impetus for the federal government to create new legislation that influence the health policy until President Obama entered office and pushed for healthcare reform. That is likely why governors were not very responsive to the federal government action prior to 2009. Governors did not respond to the federal governme nt simply because the federal government was not very active During this time period, some states were active in changing their healthcare, so this would help to explain gubernatorial behavior. It was not until the ACA was discussed and eventually adopted by Congress that we see governors responding to federal legislation or presidential executive orders. prevention. Healthcare in most states did not see any shock to their policy equilibrium until the ACA was adopted. Yet, at the start of the century we had the 9/11 terrorist attacks, which led to a significant amount of activity on the federal level in the early ing formal plans for terrorism and catastrophe preparedness across the country. In response to 9/11, Congress and the president acted quickly and on rather wide ranging areas in order to prevent any other possible terrorist attacks. This is significant for two reasons. First, this meant that there were much more activity for the governor to respond to. Second, it set
137 an imperative for the governors needed to react to in order to assure citizens that they are safe. These findings show that governors use execut ive orders as a form of policymaking. Most of the orders issued by governors either create policy or make administrative changes Future research should more states in analysis, but my analysis indicates that the remaining states follow similar patterns in regards to the purpose that executive orders accomplish Only three of the twenty two states analyzed issued more executive orders that were more administrative than policy oriented (no states issued primarily budgetary or symbolic orders). Consistent with previous chapters, governors use executive orders to set the agenda and to institute policies in these two broad fields. While this chapter does not isolate the factors that lead to why governors issue policy oriented versus administrative executive order s, it helps to show that many of the executive orders issued by governors are more than simply a symbolic gesture or political posturing, it is a deliberate attempt to shape policy and current policy areas. Several recent presidential analyses show that, w hile presidents issue many executive orders throughout their time in office, these executive orders are more symbolic and have little effect in terms of the way that the executive branch institutes public policies. 2 My findings are counter to presidential scholars in that I show that the main function of executive orders within policy areas is either administrative or policy oriented. In the former case, government would be forced to cease existing policies or wait for legislative approval; whereas, in the latter case, I find that many governors pursue their 2 See: Fine and Warber (2012)
138 policy objectives or work to make changes in executive branch government by issuing executive orders.
139 Table 5 1 Function of executive orders issued relating to health and hu man services from 2004 to 2014. State Administrative Budget Policy Symbolic Total Alaska 1 0 0 0 1 Arizona 20 19 17 1 57 Georgia 4 3 11 0 18 Illinois 5 0 6 1 12 Indiana 4 0 7 0 11 Iowa 1 0 0 0 1 Kansas 1 0 5 0 6 Louisiana 8 2 10 0 20 Maryland 5 0 20 2 27 Michigan 10 0 47 0 57 Minnesota 4 0 6 0 10 Missouri 9 0 8 0 17 North Carolina 4 0 6 0 10 Oregon 3 0 6 0 9 Rhode Island 13 0 13 1 27 South Dakota 1 0 12 0 13 Tennessee 1 0 3 0 4 Texas 1 0 1 0 2 Utah 0 0 1 0 1 Vermont 3 0 16 0 19 Washing ton 2 0 6 1 9 West Virginia 3 1 10 0 14 Total 103 25 211 6 345
140 Table 5 2 Function of executive orders issued relating to terrorism preparedness from 2004 to 2014. State Administrative Budget Policy Symbolic Total Alask a 0 0 0 0 0 Arizona 6 1 11 2 20 Georgia 6 0 2 0 8 Illinois 2 1 1 1 4 Indiana 4 0 5 0 9 Iowa 1 0 0 0 1 Kansas 2 0 1 0 3 Louisiana 0 0 3 0 3 Maryland 0 0 8 0 8 Michigan 2 0 5 0 7 Minnesota 0 0 0 0 0 Missouri 4 0 1 0 5 North Carolina 0 0 0 0 0 Or egon 0 0 1 0 1 Rhode Island 0 0 2 1 3 South Dakota 0 0 5 0 5 Tennessee 1 0 0 0 1 Texas 1 0 1 0 2 Utah 0 0 0 0 0 Vermont 2 0 2 0 4 Washington 0 0 1 0 1 West Virginia 1 0 5 0 6 Total 32 1 54 4 91
141 CHAPTER 6 DRIVING FACTORS OF SPECIFIC POLICY ADOPT ION Summary I analyze gubernatorial use of executive orders at the micro level and highlight the limitations of viewing policy creation as being driven solely by the legislature. Chapter 6 analyzes policymaking on the state level using lesbian, gay, bisex ual and transgender (LGBT) employment protections as an illustrative case. First, I analyze factors that cause governors to issue controversial executive orders. I find that governors are more likely to issue these executive orders when first entering offi ce and when they are institutionally stronger. Findings also suggest that partisan makeup of the legislature may play a role in helping to explain executive order usage. Governors were less likely to issue executive orders when the legislature was made up of more members of the same party as the governor, as well as when government is unified. I next turn attention to the role that executive orders play on adoption of legislation that parallels these same executive orders. I argue that executive action, or inaction, influences the adoption of LGBT inclusive legislation. I find that there is no statistical difference in the likelihood of adoption of legislation between states with and without executive orders in place, but states with executive orders are mo re likely to adopt legislation as time goes on. Counter to executive order findings, states with weaker governors are more likely to adopt legislation, which suggests differing strategies to policy adoption. Governors tend to issue executive orders at the start of their tenure, but adoption of stronger, more enduring policy does not happen for years later, in part, due to differing incentives for governors and the legislature. In the case of LGBT protections,
142 partisanship is strongly linked to both gubernat orial use of executive orders and adoption of legislation. Democratic governors issued approximately 80% of the LGBT inclusive executive orders. Counter intuitively, states with executive orders in place experience a longer wait time for the legislature to adopt similar legislation Additionally, states with executive orders in place are not inherently more likely to adopt similar legislation but states with executive orders in place are more likely to adopt legislation as time moves on In neither case of sexual orientation or gender identity protections do executive orders preclude the adoption of legislation, but legislatures are less likely to adopt similar legislation in the near future These results show that policy adoption cannot be understood by s imply considering legislative action. Methods of Establishing Employment Protections Presidential scholars have long emphasized the role of the executive branch in federal policymaking. Presidents develop policies formally through unilateral action, but they also pursue their objectives in the legislative arena. Governors fill an analogous role within their states. They manage the bureaucracy and they help set the policy agenda through speeches, calling special sessions or by taking unilateral action. I a nalyze factors that explain gubernatorial use of executive orders, and I assess how these same executive orders influence statute adoption of similar legislation, using lesbian, gay, bisexual and transgender (LGBT) employment protections as an illustrative case. This case is particularly well suited to test the role of executive action on legislative policy adoption. Since the 1970s, dozens of governors issued executive orders to establish protections for the LGBT community by prohibiting discrimination bas ed on sexual orientation and/or gender identity. As Figure 6 1 shows, executive
143 orders created many of the first LGBT protections. Only one governor issued an executive order after legislation was already in place (Massachusetts Governor Mitt Romney in 2003). In all other cases these executive orders created new policies These early protections only protected sexual orientation in public employment. Only one governor in histo ry has been openly out as bisexual or gay (Kate Brown Democratic Governor of Oregon) while in office and no governor has been out as transgender. Therefore, this is not a story of descriptive representation. Years, or even decades, lapse between when a governor issued an executive order to the adoption of legislation, if adopted at all. This variation in policy creation allows for analysis of gubernatorial motivation to issue executive orders, while also providing an opportunity to consider the role of executive action on legislative statute adoption. legislation. Commonly states are the unit of analysis (Berry and Berry 1990; Mooney and Lee 1995), but many studies analyze determinants of policy ad option at the local level (Sharp 2005; Shipan and Volden 2008). Many recent scholars try to explain diffusion beyond defining policy adoption as a single stage process. They attempt to understand the content complexity of the adopted policies (Karch 2007) and the mechanism that leads to their adoption (Hicks et al. 2015; Shipan and Volden 2008). There is a key limitation to the majority of these studies they rely almost entirely on policy adoption by the legislative branch of government. However, this is not the only manner in which governments create policies. 1 Indeed, studies show that other political 1 The courts and various agencies also create their own form of p olicy. However, I focus primarily on executive orders and their influence on statute adoption for this chapter
144 actors can greatly influence the policy landscape whether through interest groups challenging the legality of policies in the courts or acting as policy ad vocates to provide information to politicians (Godwin and Schroedel 2000), or even by executives acting as policy entrepreneurs to pursue their political agenda (Spill, Larci and Ray 2001). I look solely at employment protections for LGBT individuals to de monstrate that the executive orders can play a role in adoption of legislation. Gubernatorial and Presidential Use of Executive Orders Governors themselves vary on many dimensions and approaches to running ditions or the powers given to the executive branch vis vis the ability of governors to pursue their policy agenda with executive orders Gubernatorial power diverges along several lines, such as ability to veto legislation, set the budget and tenure pot ential, which puts governors on a different footing from one another when comparing relative strength particularly when pursuing their agenda in the legislature (Kousser and Phillips 2012) Executive orders are unilateral tools that all governors can use to promote their agenda or to direct agencies on how to carry out executive functions (Mayer 1999). Since voters hold executives responsible for policy outcomes, governors have motivation to institute policy unilaterally to circumvent opposition in the leg islative arena (Krause and Cohen 1997). Governors can quickly win support by making administrative changes that align policy stances (Partin 2001). Use of executive o rders can also serve as a symbol of new administrative goals. Mayer (1999, p. 446) notes that presidents use executive orders "as a symbol of their
145 intention to act decisively" on an issue devoid of Congressional approval or action. Presidential literature also suggests that executives are more prone to issue executive orders to alter policy. Mayer and Price (2002) argue that presidents use executive orders on significant issues in order to direct agencies and signal to other political actors that their pol icy preferences differ from their predecessor. My evidence suggests that this is true for governors, since the majority of the governors under analysis issued protections for LGBT members within the first month of entering office. Early presidential studi es suggests that use of executive orders can be explained primarily by the rise of the institutional presidency (Burke 1992) or divided government (Deering and Maltzman 1999), but later research suggests that motivations to issue executive orders are more complex. Fine and Warber (2012, p. 272) argue that the intent of executive orders differs based on unified or divided government, and that into reality. This suggests a refined approach to the strategic model. Executives should issue more policy oriented orders when they view their objectives as unlikely to pass in the legislature. Executive orders are not equivalent to legislation, and should not be conceptualized as su ch, but these orders emulate many of the qualities of legislation and are often seen as one possible step in the incremental progression to adoption of legislation. Executive orders can define a protected class, 2 as well as indicate the licy preferences. In the best case scenario, states view executive orders as equivalent to the law. However, executive orders are much less stable than 2 For more information on employment protections see: Colvin (2007); Sellers and Colvin (2014).
146 legislation because executive orders are not necessarily permanent. Seven states had protections created by executive order that were subsequently removed or invalidated. For instance, the Iowa Supreme Court nullified protections when it ruled that the governor went beyond his constitutional power to issue the executive order (HRC 2014). Oregon citizens also successfully passed a ballot initiative in 1988 that temporarily invalidated sexual orientation protections. 3 The remaining five states had protections taken away once a new governor took office, and in all cases, this occurred when party control of the e xecutive switched from Democrat to Republican. Several state courts ruled that transgender discrimination is a form of sex discrimination under state statutes (Transgender Law Center 2010), but often these cases remain tied up in court for years and can st il l be challenged in appeals or state supreme courts. Statute Adoption and Executive Orders Policy diffusion literature in recent years is critical of assessing the adoption of multiple variations of policy in the same analysis because the determinants o f some components of LGBT policy may differ from others, which can cause misleading conclusions (Boehmke 2009). The same is found when assessing the factors that lead to adoption of different kinds of protections for the LGBT community (Taylor et al. 2012) For this reason, my analysis is broken down into two parts: 1) factors that lead to executive order use; and 2) determinants of adoption of legislation establishing LGBT employment protections. I argue that executive action and adoption of legislation in fluence one another ; further, executive orders can delay the adoption of legislation. 3 The Oregon Supreme Court eventually reinstated t he protections when it the ruled the ballot initiative unconstitutional, but the initiative caused protections to briefly be taken away.
147 Howell (2005) posits that executives have two broad choices to pursue policy objectives they can either pursue unilateral action or submit proposals to the legislature. He asserts that the role of the executive in policymaking, and his or her ability to influence political outcomes, varies depending on whether the chief executive creates policies unilaterally or the legislature crafts legislation. I argue that governors with stronger institutional powers are more likely to establish policy unilaterally, but these executive orders make legislatures less likely to adopt similar pro LGBT legislation by reducing legislature (Krause and Cohen 1997; Wigton 1996). State politics and policy diffusion literature assert that there are internal and external factors that lead to adoption of legislation (Berry and Berry 199 0 ). External factors include innovations from o ther states and diffusion. Findings indicate that regional pressures and innovations made by neighboring states influence the adoption of similar policies; though, scholars in recent years question the power of interstate factors to influence policy adopti ons due to the rise in technology and national advocacy groups (Haider Markel 2001). Mooney and Lee (1995) find that significant value conflicts occur for morality politics, which LGBT policies fall within, that make intrastate factors pivotal to explainin g the adoption of policies. Social and political conditions within a state are particularly important when analyzing the adoption of morality politics (Gray 1994). Social conditions, such as values or citizen ideology, influence the adoption of legislation by altering the individuals elected to office and the overall support for creating change. Lax and Phillips (2009) found that, overall, liberal states are more likely to adopt pro LGBT policies; further, states with larger proportions of Evangelicals
148 were less likely to adopt these pro LGBT policies (Barclay and Fisher 2003; Haider Markel 1999; Taylor et al. 2012). Previous studies also find that greater Democratic control of the legislature increases the probability that states will adopt pro LGBT legisla tion (Haider Markel 2001; Taylor et al. 2012). These studies did not control for governors, but based on shifting alliances of political parties in the 1990s, Democratic governors should issue pro LGBT executive orders more. Thus, the first two hypotheses are: Hypothesis 1: Democratic Governors are more likely to issue executive orders. Hypothesis 2: The likelihood of adoption of LGBT employment protections increases as the percent of Democratic legislators increases, all else equal. Presidential scholar s find that governors issue executive orders strategically. Howell (2003) shows that presidents use unilateral tools, particularly executive orders, in order to further their policy objectives. In a similar regard, I argue that stronger governors establish protections upon entering office through executive orders because they are given greater leverage to act unilaterally and there is strategic incentive for the governor to do so, but this action may run counter to the incentives of the legislature and over all long term goals of the governor (Bolton and Thrower 2015). As Howell asserts, the legislature is often unable to respond with legislation, even when a majority is opposed to this new policy, because building a successful coalition to overturn the execu tive order is challenging. While Howell focuses on executive orders that run
149 coalition building problems when considering the adoption of LGBT inclusive statutes that stre ngthens protections created by the executive order. This leads to following hypotheses: Hypothesis 3: Governors are more likely to issue executive orders at the start of their tenure. Hypothesis 4: Stronger governors are more likely to issue executive o rders, all else equal. Governors may utilize executive orders when they view the legislature as unlikely to pass inclusive legislation. On the other hand, it is possible that part of the delay in the adoption of legislation may be due to the governor add ing protections through executive orders. Among the governors who issued executive orders to protect sexual orientation, 14 of 24 faced divided government and an additional 5 confronted legislatures with smaller Democratic majorities (55% or less). A simil ar story is seen with gender identity. Of the 11 governors who added protections, 7 faced divided government and an additional governor had a smaller partisan advantage in the legislature. Ultimately, however, short term gains for the LGBT community actual ly lead to long term difficulty in achieving more comprehensive protections. 4 Executive orders do not prevent statute adoption, but they impede the development of a successful coalition and executives turn to executive orders more when the legislature is less capable of mobilizing around the 4 Presidents increasingly use executive memos and signing statements. The Chief of Staff for Governor Jack Dalrymple of North Dakota sent an executive memo protecting sexual orientations for employees in 2015, but no other state issued a similar memo throughout the time period.
150 issue (Bolton and Thrower 2015). As a result, I expect that LGBT inclusive executi ve orders to hinder the spread of similar statutes, but not altogether prevent their adoption. Hypothesis 5: Legislatures are more like ly to adopt p ro LG B T legislation the longer it has been since a governor issues an inclusive executive order. Data and Expectations To assess factors that explain policy adoption, I conceptualize policy adoption in two ways: 1) as created by executive o rders and 2) as created by legislation. I created a of statutes that establish sexual orientation and trans inclusive employment protections. I identified executive orde rs by first searching within the National Gay and Lesbian Task Force (2014 ) a similar searches withi The Human Rights Campaign (2014) provided employment nondiscrimination statute adoptions. The resulting dataset spans from 1975 the first year that any form of protection was in place to 2013 for all fifty states. I argue that factors specific to individual governors lead to their issuance of executive orders. Klarner (2012) provided the executive political features. The term Democratic Governor is a dichotomous variable coded 1 for Democ ratic governors and 0 otherwise. I expect Democratic governors to be more likely to issue executive orders since on average they appeal to a more liberal voting base and have greater motivation
151 to do so than their Republican counterp arts (Brewer 2007 ). Whi le not all Democrats are proponents of LGBT rights, they have more electoral incentives to push for nondiscrimination policies. Because the Christian Right Movement (and social conservatives more generally) comprise a strong and highly active component of the Republican Party, creating LGBT inclusive policies could harm reelection bids of Republican governor s (Wilcox and Robinson 2010 ). Therefore, Republican governors have little incentive to issue an executive order to protect LGBT employees. A similar con cern confronts Democrats in more conservative states. These governors have less incentive to issue an executive order than a governor from a more liberal state because of potential backlash in subsequent elections. If governors work to appease their base a t the expense of swing voters, they endanger holding together a plurality coalition to win reelection, so this is not simply a story of partisan control of government; intrastate factors must be considered as well to explain gubernatorial behavior. The ne xt explanatory variables consider the how time in office influences how governors issue executive orders I argue that governors who are more liberal or governors who confront more liberal constituencies want to gain political points by issuing protections Issuing executive orders can quickly establish new policy without losing much political capital in liberal states that are amenable to these protections. This allows governors to establish policy and avoid confrontation with a legislature that is either controlled by the opposition party or resistant to the policy, but early unilateral action may thwart future negotiations by worsening ties with the legislature. Since this does not require negotiations with other political actors, governors can achieve th eir objectives upon first entering office without losing much political capital. However,
152 governors may pursue their policies through legislative or administrative means first if it appears that there are receptive members in the legislature or agencies. I f the later in their tenure. The variable Years Served by Governor is the number of years that a governor has served in office as recorded by Klarner (2012). I expect govern ors to issue orders most upon first entering office and towards the end of their tenure as they lose political clout. Therefore, I also include Years Served by Governor 2 to account for the nonlinear relationship. Unlike presidents, not all governors face t erm limits Knowing that you cannot run for reelection could encourage governors to add protections in their last year in office, once governors realize that the legislature is unlikely to act prior to the end of their term. Therefore, I expect lame duck g overnors to be more likely to issue an executive order. I include Term Limited to account for governors serving their last year in office. I argue that stronger governors are more likely to issue executive orders prior to pursuing legislation in the legisl ative arena. Weaker governors may not have the ability to issue these protections, so they take a more conciliatory approach to policymaking. I account for the relative strength of an administration by controlling for annual estimates of institutional powe r of the governor based off of Beyle ( 2007 ) This cannot get directly at capturing managerial style, but it creates a proxy for the potential ability of governors to manage their administration. The resulting measure, Gubernatorial Power can range from 0 to 5 with values increasing for relatively stronger governors. I expect stronger governors to issue executive orders, while I expect weaker governors to be more
153 successful in legislation adoption because they have less discretion to establish policy unilat erally. policy adoption. Consistent with policy diffusion literature, I model these variables as a function of the proportion of neighboring states that adopted similar polic ies. 5 I include these variables to account for regional influences and practices. Executive Order Diffusion and Statute Diffusion measure the proportion of bordering states that have an executive order in place, or passed legislation, respectively. The mod els require two distinct measures for both forms of protections because analysis looks at the spread of protections through two separate mechanisms 1) through unilateral executive action; and 2) through legislative action. 6 I create these measures for se xual orientation and gender identity, so the models use a total of four terms t o e stimate the results in Table 6 1 and Table 6 2. In keeping with previous literature, I expect policies to diffuse regionally, but I do not expect the diffusion variable to be as influential as when it was first conceived, owing to the rise in technology and national organizations that prior scholars raised as limitations on the diffusion variable. I expect the strategy deployed by governors on whether or not to issue executi ve orders to influence the adoption of legislation. I expect governors to issue executive orders when legislation appears unlikely and, therefore, associated with delayed enactment of similar legislation in the near future. Two variables control for this. I expect the term Executive Order to make legislatures less likely to adopt legislation. This 5 See Mooney and Lee (1995); Haider Markel (2001). 6 However, as a robustness check, the alterna tive models included a composite measure that accounted for the presence of either protection, in addition to including both terms in the model.
154 explores the role of executive orders on the adoption of legislation. This variable is coded 1 when a state has an executive order in place in the given year, and 0 otherwise. Additionally, I include Years since E.O. Issued to account for the increasing probability that a state will adopt legislation in later years. The term is transformed using the square root function to account for the increasing probability of states adopting legislation, but with diminishing returns. Partisan control of government helps to explain both models, but partisanship is controlled differently based on the guiding theory. Presidential literature suggests that the proportion of the leg islature that is controlled by the same party as the governor should explain executive behavior. Therefore, in the model analyzing the issuance of executive orders, I control for the in order to capture the that similar legislation may pass. I expect governors to issue orders when there are fewer same party members in the legislature. The policy diffusion and LGBT literature, however, finds that the adoption of pro LGBT policies is highly dependent on the partisan makeup of the legislature, with higher proportions of Democrats linked to the adoption of LGBT protections ( Haider Markel (2001); Soule and Earl 2001; Taylor, et al. 2012 ) I model partisan composition of the legislature with the term Democratic Legislators (logged %) The models account for b oth terms as the log of the percent age to account for the fact that there is diminishing returns on the likelihood of policy adoption as a single party takes over greater control of the legislature. 7 I also include Divided Government to 7 This modeling approach was selected to account for the fact that not all changes to the partisan makeup of legislatures have the same effect on altering behavior. For instances, the probability that a legislature adopts legislation will increase much more by picking up 1 new seat to move from 55% to 56%, as compared to when the new seat shifts partisan control from 70% to 71%.
155 account for partisan tension that occurs with divided or quasi divided government In keeping with previous literature (Mayer 1999; Howell 2003; Fine and Warber 2012), I expect governors to be less likely to issue executive orders and legislatures to be less likely to adopt legislation during divided gover nment. I account for several additional controls. Liberal Citizen Ideology which models series from Berry et al. (1998). 8 I expect governors and state legislatures to b e more likely to add protections when their constituents are more liberal. Evangelical Rate is the percentage of Evangelical Protestants within the states as measured by the 1980, 1990, 2000 and 2010 Religious Congregations and Membership surveys S ocially conservative constituents tend to favor policies that reflect their personal beliefs (Sharp 2005), and I expect that states are less receptive to both forms of policies when there are higher proportions of Evangelical within the population Finally, the co ntrol Time is a simple time counter. Estimation Using Multilevel Event History Analysis, with the state/year as the unit of analysis, I evaluate: 1. The probability that a governor i will issue an executive order protecting LGBT employees in time t given that no executive order is in place. 2. The probability that state legislature i will adopt an LGBT inclusive employment nondiscrimination statute in time t given that it has not already done so. 8 The provided da ta ranges from 1974 2010. I extrapolate values for 2011 2013 to provide estimates for these years.
156 Analysis runs from 1975 to 2013 for all of the models. 9 As d iscussed previously, states vary in tradition and use of executive orders. Multilevel modeling accounts for these differences and within state patterns of adoption seen throughout the years. Two separate analyses estimate factors that lead to the adoption of sexual orientation and gender identity protections This is because the determinants that lead to their adoption differ from one another. In fact, Taylor et al. (2012) found that the effect of determinants that lead to successful statute adoption of LG BT protections share common elements, but differ based on type of protections added sexual orientation versus gender identity. Therefore, I estimate the likelihood of policy adoption separately for sexual orientation and gender identity. The models estim ate state level heterogeneity with a level 2 variance component; additionally, year level heterogeneity is accounted for with a level 1 variance component (see Rabe Hesketh and Skrondal 2008). The dependent variables are all binary variables coded one for policy adoptions the state adds protections. The analysis defines an event differently based on the type of policy under analysis because the two forms of adoption occur independent from one another In many states the both legislation and an executive order was issued, but it did not occur in the same year (Massachusetts is the only exception) For executive orders, an event occu rs when either a governor issues an executive order or the 9 The year 1975 was selected because it was the first year that an LGBT inclusive nondiscrimination policy was in place in this case, by executive order to protect sexual orientation. However, I tested three alternative start years (1981, 1993 and 1999) as a robustness check. State legislatures adopted sexual orientation and gender identity legislation in 1981 and 1993 (respectively). Additionally, the first year a transgender inclusive executive order was in 1999. No start year yielded substant ively different results, but the variance explained was higher as more years were included in analysis.
157 legislature adopts a statute. 10 For statutes, an event only occurs when the legislature adopts a statute. An advantage of using Multilevel Event History Analysis is that it geneity as a random effect, [therefore,] coefficients for level variance component partly accounts for the different traditions, and frailties, across the states. Results Determinants of E xecutive Orders First, I examine the factors that lead to governors issuing executive orders. The first model presented in Table 6 1 considers factors that lead to protections for sexual orientation and the second model analyzes determinants that influenc e governors to issue executive orders protecting gender identity. The dependent variable is whether a governor issues an executive order in a given state/year, given that no executive order is in place and the legislature has not already adopted similar le gislation. Both models present the results of a Multilevel Event History Analysis. Time is centered at 2013. The covariates Gubernatorial Power, Executive Order Diffusion, Liberal Citizen Ideology and Evangelical Rate are all grand mean centered, which doe s not alter the estimates, but aids in interpretation, as the intercept is now meaningful. The findings are similar for gubernatorial factors across both models. As theory and previous suggests Democratic governors are much more likely to issue executive 10 While ideally states would re enter the dataset if the state removes an executive order, this raises several concerns. First, this approach introduces multiple observations for the same state, so the probability is high that there is statistical dependence between the observations and errors. This would cause p values and standard errors to be lower than in actuality. This can be adjusted for through ro bust estimation methods; yet, a larger concern lies with the question of re entry. It is unclear when the state becomes at risk again. Depending on modeling choice, this could introduce considerably more observations, but could also introduce considerably more error if states reenter at the wrong time.
158 orders. Governors issue executive orders at the start of their tenure. In the model focused on sexual orientation protections the square term is statistically significant, which is consistent with the presidential literature and suggests that governors tu rn to executive orders at the start and end of their tenure. However, the term is not statistically significant for transgender protections. The differing findings make it unclear whether the relationship is linear or nonlinear, but demonstrate that govern ors are most likely to issue protections at the start of their administration. Governors in their last year in office are not statistically more likely to issue protections. 11 The political conditions appear to play a lesser role in explaining gubernatori al actions, but no clear relationship can be drawn. In both models, the term is statistically significant, but it is positive for gender identity and negative for sexual orientation. This is unclear why and will be further discussed in the conclusion. Citizen ideology is not statistically significant for either of the models. Divided government is shown to increase the likelihood that a governor will issue an executive or der by a factor of 2.8 for sexual governors are less likely to issue orders when their party controls more of the legislature. This gives some credence to the notion that go vernors issue executive orders when they do not expect legislation to pass; however, this does not hold true for transgender protections. The Evangelical population is negative and statistical significant for sexual orientation protections, but also does n ot reach conventional 11 However, it should be noted that there is a weak correlation (r = 0.15) of Years Served by Governor 2 with the last year in office, which could account for a decline in statistical significance.
159 statistical significance for the transgender model. Both models show that governors are more likely to issue protections later into the time frame and the variance across the state is significant. es to their 2013 mean. I estimate the predicted probability that a governor will issue an executive order to protect sexual orientat ion in their first eight years in office, assuming they experienced unified government throughout their tenure and the governor cannot be term limited. 12 Figure 6 2 plots the results separated by partisanship of the governor. Clearly, partisanship matters. Democrats are much more likely to issue executive orders to establish LGBT protections. At no point in time do the confidence intervals of Democrats and Republicans overlap. Figure 6 2 shows the relationship of time served in office with the declining like lihood of issuing an executive order. There is an upshot in likelihood approximately five years into office, but this incline is modest. Governors from both parties are most likely to issue protections upon first entering office. The model predicts that in 2013 there is a very high likelihood that Democratic governors will add protections upon entering office; whereas, the predicted probability for incoming Republican governors is roughly 0.4. Only 1 Democratic governor has taken office in a state that with out these LGBT protections in place since 2013 (Virginia 2014). In that particular case, Governor Terry McAuliffe added protections for sexual orientation and 12 Figure 6 2 uses the model estimating the likelihood that a governor will add sexual orientation protections since there are more cases of governors issuing executive orders. The results are more robust.
160 gender identity in his first month in office. No Republican governors added protections after 201 Determinants of Statute Adoption The analysis so far shows that gubernatorial characteristics and the political conditions that the administration confronts help to explain when gov ernors issue executive orders to protect the LGBT community. The analysis now turns to considering the role of executive action by examining factors that lead to statute adoption of LGBT inclusive employment protections. Once more, Time is centered at 2013 while the covariates Gubernatorial Power, Executive Order Diffusion, Liberal Citizen Ideology and Evangelical Rate are all grand mean centered. The model provided in Table 6 2 estimates the likelihood that a state legislature will adopt legislation that provides LGBT protections given that it has not already done so. Executive order is not statistically significant for either of the models, nor was the partisanship of the governor. However, legislatures are more likely to adopt legislation the longer it h as been since an executive order was issued. Executive orders do not inherently prevent the adoption of the laws, but this reveals that there is a relationship between the two. Either governors unintentionally delay the adoption of legislation by adding ex ecutive orders or governors issue executive orders because legislation is unlikely. The findings on Gubernatorial Power in Table 6 1 and Table 6 2 indicate that both of these explanations may be at play. The remainder of the findings is consistent with ex pectations and previous literature. Statute Diffusion is consistent in both of the models, which suggests that neighboring states may influence the adoption of legislation, but this finding may also be related to neighboring states being homogenous in term s of ideology and constituents.
161 Legislatures are more likely to adopt legislation in states where the citizens are more liberal. The Evangelical population suggests a similar conclusion. States are less likely to adopt legislation protecting sexual orienta tion when the Evangelical population is higher, but this does not hold true for trans protections. Additionally, states with more Democrats in office are more likely to adopt protections. Consistent with previous policy diffusion research more Democrats in office leads to a higher probability of states adopting pro LGBT legislation. Divided government is not statistically significant, which may be a reflection of the partisanship of the governor also not being a statistically significant determinant on the adoption of legislation, but may also speak to the overwhelming importance of the partisan composition in the legislature. As with executive orders, states are more likely to adopt statutes throughout the time period. Discussion Scholars have largely negl ected why governors issue executive orders and the role that executive action plays on influencing the adoption of legislation. Analyzing executive orders systematically, and in conjunction with legislation enacted, reveals a very different picture of the extent of LGBT protections than when looking solely at legislation. As of 2015, 24 states have some degree of trans inclusive nondiscrimination employment protections, but only 18 states have statutes in place preventing employment discrimination. Likewise 28 states have statewide sexual orientation protections, but only 21 of these states established protections through statutes. Employment protections for sexual orientation and gender identity are an interesting area of analysis for gubernatorial behavi or. Typically, policy diffusion looks at the adoption of legislation, but this analysis shows that this is not the only way that
162 states added protections. Further, it is interesting to consider the adoption of controversial, Democrat owned, policy area bec ause it allows us to explore behavior when the governor and the state often times differ starkly in policy preferences. To date, no similar analysis of LGBT protections across the states exists, and this analysis simply cannot be conducted quantitatively o n the federal level. LGBT protections was restricted to exploring legislation (see Haider Markel 2001; Taylor et al. 2012), the courts (see Mooney 2000 ; Sharp 2001), and frequently excluded gender identity protections. While one work, Sellers (2014), analy zed the factors that led to gubernatorial use of executive orders, this analysis was restricted to gender identity only, and did not account for the effects of diffusion or the potential relationship of executive order use with adoption of legislation. Fut ure research should consider other less controversial areas or Republican owned po licy areas to consider if the same relationships hold across multiple policy areas. Sexual orientation and gender identity protections followed considerably different adopti on patterns. In the 1970s through 1990s states added sexual orientation protections mostly through executive orders. Whereas, gender identity protections were not common until the 2000s, and many of the states that added protections did so through statutes Despite these differences in policy development and the time period in which states created these policies, the governor specific factors that cause governors to issue executive orders were similar across the sexual orientation and gender identity models In both cases, governors were more likely to issue protections at the start of their administration. This is nonlinear for sexual orientation protections and linear for gender identity, but both models indicate that governors are more likely to issue
163 exe cutive orders upon entering office. The difference between the two might be due to the fact that many sexual orientation protections were added during changing alignment of the Democratic Party with the LGBT Movement, while transgender protections were add ed after this shift occurred. Governors knew whether they wanted to add protection s immediately upon entering office. Both of the models in Table 6 2 show that the presence of executive orders are important to statute adoption with states with executive or ders in place associated with delayed statute adoption. The presence of executive orders is not statistically significant, which indicates that states with executive orders in place are no more likely to adopt pro LGBT legislation, all else equal. Yet, the time since the executive order was issued is statically significant and positive, which indicates that state legislatures are unlikely to adopt legislation in the near future when a governor issues an executive order. Institutional power has an interestin g effect on the adoption of both of the policies that may suggest differing tactics by strong and weak governors. Stronger governors are more likely to issue executive orders. Yet, it is the legislatures confronting relatively weaker governors that end up adopting legislation. Coupled with the finding that legislatures are more likely to adopt legislation after more time lapses since an executive order is issued, this suggests that executive orders influence legislation adoption and legislation may influenc e executive order use. Stronger governors act at the start of their administration, but governors whose powers are weaker chose to approach policymaking in the legislative arena or the legislature establishes the policy agenda. These early acts of unilater al action create institutional conflict and alters the motivations of legislators to adopt legislation.
164 Political and social factors also influenced policy adoption. Social factors explained gubernatorial behavior only to a modest extent, but played a vit al role to explaining adoption of legislation. Consistent with presidential literature, governors were less likely to act unilaterally to establish sexual orientation protections when the legislature is composed of more legislators from the same party as t he governor. However, this finding did not hold for the gender identity analysis. Social and political factors were much more important when modeling the adoption of legislation. States with more liberal citizens and fewer Evangelicals in the population we re more likely to adopt LGBT protections. Consistent with previous research, states were more likely to adopt pro LGBT policies the greater control Democrats had over government. Diffusion also was found influential in all of the models, but in an inconsis tent manner Sexual orientation protections was the earliest form of protections, so later innovations, such as adding gender identity or passing legislation, may have replaced this policy and led to this anomaly in findings It is also possible that the f act that so few states were at risk in later years could lead to this finding No previous diffusion literature analyzed the spread of executive orders, so diffusion may be driven by other factors than what is commonly found in policy diffusion literature that focuses on legislation Many of the states that start off with executive orders go on to adjust or reinvent their policies; some adopt legislation to protect private employment as well, and others with only sexual orientation protected eventually add gender identity. However, considerable time passes in between. The average wait time from when a governor issues an executive order for sexual orientation to when the legislature adopts a transgender inclusive statute is 14.3 years. Many states have yet t o expand their
165 policies. In fact, the first state to add sexual orientation protections (Pennsylvania in 1975) has not adopted legislation to expand protections to private employment, and it did not add gender identity until 2003, which suggests that gover nors may inadvertently delay their overall policy objective by issuing executive orders. It should be noted that there may be endogeneity at play. Governors may issue executive orders because they perceive legislation is unlikely to pass. Controlling for i nstitutional power helps to mediate some of this concern by suggesting that governors pursue different strategies based on their relative strength, but analysis is unable to control for the expectations of governors. Future research should consider the ove rall managerial style of these governors who issued executive orders as well as their success on other policies within the legislative arena. Executive orders appear to influence statute adoption in ways not previously considered. The results suggest that the success of statute adoption may in fact be delayed by the presence of existing executive orders rather than the legislatures being less open to the statutes. It should be noted, however, that executive orders cannot influence every policy field equall y. In some cases, the governor may have nominal claim to influence policy outcomes because the governor does not have constitutional or statutory rights to dictate policy, so unilateral efforts to influence policy outcomes in these fields will be unsuccess ful. Governors are more likely to influence the policy area in this analysis employment nondiscrimination protections than in other areas, such as modifying the penal code. Future research should consider assessing multiple policy areas, or perhaps cond ucting a pooled event history analysis of several orders within a specific policy area to help circumvent this generalizability concern.
166 Despite this limitation, my analysis reveals that governors establish their own form of LGBT protections and that these policies influence the adoption of legislation. orders, and these orders they can influence the adoption of legislation. Stronger governors are more likely to issue ex ecutive orders, but legislatures do not adopt similar legislation until years after governors issue executive orders. I gnoring the role of executive orders on statute adoption leads to misleading conclusions regarding the current state of nondiscrimination protections by overlooking many policies in place across the country I show that protections for LGBT public employees has progressed further than is usually asserted. Yet, this was done through executive orders, which are neither as stable nor as compre hensive as their statutory counterparts. Nevertheless, scholars should expand their conception of policy development to consider the importance of executive action.
167 Figure 6 1 Adoption of LGBT Protections from 1975 2014.
168 Note: Estimates provided are for a governor who is not term limited and is working under unified government. All other covariate values used the averages of the states still at risk in 2013. Figure 6 2 Predicted Probability o f a Governor Issuing an Executive Order to Protect Sexual Orientation in Employment in 2013.
169 Table 6 1 Factors that Drive the Issuance of LGBT Inclusive Executive Orders. Covariate Sexual Orientation Transgender Democratic Governor 2.425** 3.497** (0.344) (1.030) Gubernatorial Power 1.226** 2.749* (0.446) (1.357) Years Served by Governor 0.442** 0.781 (0.120) (0.469) Years Served by Governor 2 0.045** 0.048 (0.011) (0.078) Term Limited (last year in office) 0.09 2.148 (0.592) (1.927) Executive Order Diffusion 7.599** 15.281** (0.934) (3.022) Liberal Citizen Ideology 0.010 0.003 (0.014) (0.033) Divided Government 1.034** 0.289 (0.334) (0.919) 0 .999** 0.138 (0.279) (1.916) Evangelical Rate 0.038 0.004 (0.022) (0.056) Time 0.522** 0.501** (0.042) (0.119) Constant 0.650 10.497** (0.463) (2.374) State Level Variance Constant 22.435** 33.813** (3.576) (12.027) N 1,823 1,823 2 16 5.60** 36.32** AIC 536.13 176.58 BIC 607.74 248.18 Note: Dependent variable is coded 1 if in a given state, an executive order is issued in year t, 0 otherwise (1975 2013). Gubernatorial Power, Executive Order Diffusion, Liberal Citizen Ideology and Eva ngelical Rate are grand mean centered. Time is centered at 2013. p < 0.10. *p < 0.05. **p<0.01.
170 Table 6 2 Adoption of LGBT Inclusive Statutes. Covariate Sexual Orientation Transgender Executive Order 1.341 1.832 (0. 987) (2.135) Years since E.O. Issued (sqrt) 1.200** 1.655 (0.317) (0.989) Democratic Governor 0.143 1.186 (0.708) (2.847) Gubernatorial Power 2.453** 3.806* (0.702) (1.784) Statute Diffusion 5.917** 7.018* (1.253) (3.719) Liberal Citizen Ideology 0.218** 0.358** (0.043) (0.103) Divided Government 0.769 5.253 (0.698) (3.212) % Democratic Legislators (log) 13.101** 51.001** (2.185) (12.865) Evangelical Rate 0.349** 0.056 (0.067) (0.090) Time 0.977** 2.572** (0.134) (0.589) Constant 52.754** 203.104** (8.839) (52.003) State Level Variance Constant 57.380** 175.163* (15.451) (80.629) N 1,833 1,833 2 58.13** 19.61* AIC 250.24 155.92 BIC 316.41 222.08 Note: Dependent variable is coded 1 if in a given state, an i nclusive statute is adopted in year t, 0 otherwise (1975 2013). Gubernatorial Power, Statute Diffusion, Evangelical Rate and Liberal Citizen Ideology are grand mean centered. p < 0.10. *p < 0.05. **p<0.01.
171 CHAPTER 7 CONCLUSIONS Review of Findings My ana lysis took a comprehensive approach to explain ing gubernatorial use of executive orders. I set out a theory that approaches explaining when and why governors issue executive orders from multiple levels of analysis. Each level of analysis shows that governo rs strategically utilize executive orders to pursue their political agenda. My findings indicate that, while some executive orders are for political reasons, other times governors use executive orders to amend policies, create new initiatives or maintain e xisting programs by making administrative changes. My analysis shows that executive orders on the state level are part of an overall approach to governing that governors utilize. I find that the relative power of a governor helps to explain when and why g overnors issue executive orders. However, this is not as simply as it sounds. Power influences behavior, but expectations differ based on whether analysis focuses on aggregate use of executive orders or analyzing individual orders. Weaker governors issue e xecutive orders more frequently throughout a given year/administration, owing to fewer relative options. These g overnors use more executive orders systematically to institute their own policy agenda. However, strong governors are more likely to institute c ontroversial policies via executive orders when they view the policy as unlikely to pass in the legislature or instituted by bureaucrats Governors with greater support in the legislature perceive that their policy o bjectives were successful, so they are less likely to act unilaterally to pursue their objectives. Likewise, weaker governors use executive orders more frequently in a given
172 year and administration because there are fewer options for them to make policy. S tronger governors are given more powers that allow them to negotiate and challenge the legislature; however, this route is less successful for weaker governors because they are not given as many tools in their toolbox. Governors attempt to establish a cons istent policy framework or new initiatives by issuing executive orders. Future Research My analysis focused on the causes and purposes of executive orders. The results indicate that governors issue executive orders to make meaningful changes to the way i n which administrations implement legislation and create new programs. Future research should delve further into how governors pursue their policy objectives. Executive orders are unilateral policymaking tools that all state governments give to their gover nors. Yet, governors can also use appointment powers, vetoes and members of their cabinet to advance policy agenda. One line of investigation that should be considered is when governors elect to influence policymaking in the legislative arena versus the r ulemaking process. Altering the formation process allows governors to shape the intent and wording of the legislation, which in the long run influences the ability of government to effectively regulate or shape public policy. However, the makeup of the leg islature, the powers given to governors and the receptiveness of legislators to executive influence limits the ability of governors to pursue their policy agenda in the legislative arena. Therefore, governors may choose to pursue a strategy akin to the adm inistrative presidency. By appointing key positions and deploying unilateral policymaking tools, governors can affect the implementation of public policy. Inquiry analyzing the governor in the
173 legislative arena or administrative governorship should assess how governors chose pursue their policy agenda, and gauge how effective governors are at instituting public policy. Future research should also consider the effects of governors on the implementation of policy a long with assessing how governors influence the development of policy in the legislative arena and within the executive branch. Executive Legislative Relations My dissertation shows that governors are in fact using their unilateral powers as a power play to systematically pursue their policy objecti ves. On the macro level, governors consistently utilized executive orders throughout their administration when they were relatively weaker, both in terms of institutional power of the governor, but also in terms of the partisan makeup of the legislature. T his suggests that governors with fewer options consistently turn to the few unilateral powers available to them. While looking within policy areas, what I term the meso level, showed a less partisan motivated tactic in policy areas that were recently in th e national and state attention, governors still turned to executive orders to create, maintain or alter initiatives of their administration. Power once again drives behavior as we look at factors that explain the issuance of controversial executive orders that establish employment protections, what I referred to as the micro level. While I analyze gubernatorial use of executive orders on a controversial policy area, my analysis reveals that power is once again important. Strong governors issue executive ord ers; whereas, legislatures pass statutes to add protections when the governor is weak. This clearly shows that strong governors make use of their powers when they want to assure that their policy objectives are successful. This could be a story of weak gov ernors electing to pursue policy through the legislature
174 to assure stronger protections, or it could be a reflection of weak governors being unable to block legislation. The relationship between legislation and gubernatorial power should be further explore d. The relationship of power to gubernatorial use of orders still raises the question of what does this do to executive legislative relations? Unilateral action strains relations between the executive and legislative branches particularly when the govern or or president does so to side step a resistant legislature. As Chapter 6 shows, there is a delay between statute adoptions when there a similar executive order in place exists. It is possible that there is a degree of endogeneity at play between executiv e action and legislative adoption of statutes. However, if executive action irritates relations, then governors strategically utilizing executive orders to establish policy devoid of legislative participation or consent could lead to continued strain in po licymaking processes across the states. We frequently see divided government across the states for the same reasons that it occurs within federal government. The decline of straight ticket voting, rise in severely gerrymandered C ongressional districts and a general negative response to the current party in government led to the rise of split ticket voting, and the subsequent pattern of divided government. My results indicate that the partisan makeup of the legislature and the relative power of the executive in office explains in large part when and why governors chose to issue executive orders. This a ffects their initial time in office, but also influences their behavior throughout their administration. If executives consistently turn to unilateral action to pursue their policy devoid of legislative approval or participation, then this may help to explain the conflict seen in recent years between a Republican Congress versus a Democratic president common gridlock and of the
175 inability of the president to pre ss for change in major policy areas that are commonly owned by one party or the other Y et, this also leads to stalemate on other issues that the public in general agrees needs to change, such as gun control in regards to preventing dangerous individuals f rom gaining access. Unfortunately a similar analysis cannot be done on the federal level, but does provide insight into behavior of presidents. One component of institutional power, as 2007 ) measure of power used throughout this analysi s, is the President Obama was criticized for turning to unilateral action, particularly executive orders. However, as my analysis of macro and micro level use of executive orders shows, governors turn to executive orders more as their time in office nears an end and when they confront a hostile legislature. Governors who face term limits and an oppositional legislature should turn to executive orders more. President Obama fa ced a legislature controlled by the oppositional party for most of his tenure For federal government, p arty politics plays an even bigger role and media coverage is more attentive to government activity so not all of the elements of executive behavior is comparable. However, President Obama is relatively strong compared to most governors and faces divided government, where at one point the leader of the Senate, accomplishment up until that point (Kessler 2012). President Obama is precisely on par with what my analysis predict governors will behave do when the legislature is unlikely to act on the executiv There is no counter factual to compare
176 to in terms of overall use of executive orders issued, but President Obama used executive orders to establish controversial executive orders, such as expanding background requirements for gun sales or extending employment protections to LGBT individuals. Analysis of gubernatorial use of executive orders, as well as these anecdotal examples of how President Obama and recent Congresses behave, provide some interesting questions regarding how politics will continue to play out during extreme polarization within Congress These questions will remain important over the next decade as the states continue to face divided government and the ideological gap between the parties remains large Future research s hould consider how tactics of executives explicitly changed throughout the late 20 th century. Additionally, studies should consider how presidents issue controversial compared to neutral policy throughout recent decades. Qualitative research or mixed metho ds could broach this topic particularly well given that large n analysis cannot be conducted with only one president at a time, but there are several morality politics areas or social policy areas that might allow for a pooled analysis of presidential beh avior or for an enlightening case study Future research should be conducted to see if partisanship of the legislature and time in office plays a role in gubernatorial use of executive orders and also consider whether those covariates explain statute and overall policy adoption on policies that are not controversial While power cannot be controlled effectively by looking simply at federal government governors and presidents share many parallels. This would contribute to the overall debate regarding pres idential use of executive orders for symbolic reasons as compared to executive orders issued for policymaking
177 reasons, and while it does not directly provide insight, it provides an alternate theoretical framework for considering how executives, in this ca se presidents, utilize unilateral powers Executive Policymaking These results show that gubernatorial use of unilateral powers is extremely common Governors use executive orders most commonly to make administrative changes, but executive orders frequent ly institute new policies, or innovations, within government This raises several normative questions regarding democracy in the United States along with the role of the executive branch across the states. The separation of power s within American states e stablishes state legislatures as the primary policymakers and delegates power to the executive to im plement any developed policies. State legislatures and constitutions grant governors unilateral powers to allow them to act decisively during times of cris is within the state. However, when governors use executive orders and other unilateral powers consistently to circumvent the legislature, this oversteps the intent of the separation of powers. Governors are creating policies rather than implementing it. Wh ile in the case of lesbian, gay, bisexual and transgender (LGBT) protections, these orders expanded protections and governors issued executive orders to protect employees, the findings suggest that this is counter to what policies constituents want This r aises questions regarding what other policies do governors create that the public does not widely support. I find that governors use executive orders for primarily administrative and policymaking processes. Chapter 4 shows that executive orders are more c ommon from weaker governors and hints that governors use executive orders as an overall strategy
178 to allow their administration to function, but this finding does not mean that weaker governors who issue more executive orders issue more policy oriented exec utive order. Looking on the within policy areas and specifically at controversial executive orders gave greater insight in to how governors use executive orders throughout their time in office. In fact, findings suggest that weaker governors issue more orde rs overall in a given year and administration, but it is stronger governors who issue controversial executive orders Policy diffusion and state politics literature should study the adoption of specific executive orders across more policy areas before dra wing any concrete conclusions, but analysis of LGBT employment protections shows that stronger governors are more likely to issue individual executive orders. I look at the use of executive orders within policy areas, but further research should delve deep er to determine what causes differing strategies among governors. It may be the case that strong governors only are more likely to issue controversial executive orders, or it may be the case that strong governors are more likely to issue policy changing ex ecutive orders. Future research should consider the role of governors on policy adoption through unilateral action and statute adoption, but findings indicate that gubernatorial power plays a role in policy adoption through both branches of government. Imp lications for the States Frequent use of executive orders raises some concerns regarding the role of unilateral action and ideals of democracy. Although Chapter 6 shows analyzes an instance where executive orders enact anti majoritarian policies, these pol icies provide protections to a subpopulation that often cannot effectively mobilize and drive change.
179 On one hand, these policies establish protections in states for an often marginalized population. On the other, these new policies are anti majoritarian. Presently, support of the lesbian, gay and bisexual community is at an all time high. Support of protections for the transgender population is growing with high visibility of famous transgender individuals, such as Laverne Cox, Chaz Bono and Buck Angel. Y et, the public did not overwhelmingly support sexual orientation or gender identity protections for most of the time period under analysis so governors issued these executive orders for reasons other than government responsiveness. This demonstrates that gubernatorial use of executive orders could establish policies counter to public opinion, as seen in Colorado with Governor John Hickenlooper issuing an executive order to support hydraulic fracturing. While voters approve of him overall, they oppose speci fic policies of the administration. To a limited extent, executive orders allow governors to adopt policies without public or legislative approval, which means that current state governments allow governors to act however they see fit. This places the role of governors as not only responsible for implementing legislation, but also as largely unchecked policymakers. Final Thoughts With the rise of the states in the last half of the 20 th century and the growth of bureaucracy throughout the past 100 years, s tate politics is increasingly important. It is clear that governors work with the legislature to pursue their policy agenda, as well as bring about litigation to pursue their interests. What my findings show is that governors also pursue their policy objec tives through executive orders. While some governors issued executive orders that are symbolic, most executive orders are intended to either
180 start new initiatives or amend current programs. Executive orders are not the only form of unilateral action that a re available to governors, but they are a tool that all governors turn to within their administration to help manage their administrations. This analysis shows that governors frequently turn to executive orders as a way of enacting goals of their administr ation, which helps to set the landscape of state public policy, along with creating conflict between the executive and legislative branches.
181 APPENDIX A DESCRIPTIVE SUMMARY OF CONTROL VARIABLES Table A 1 Distribution of Executive Orders.
182 Table A 2 Distribution of citizen activity for states that have citizen activity.
183 Table A 3 Potential values of legislative activity.
184 APPENDIX B CODING SCHEME FOR GUBERNATORIAL POWER Component 1: Separately Elected Power (SEP) Scale Coding 5.0 The only elected executive positions are Governor and Lieutenant Governor (on a single ticket) 4.5 The elected executi ve positions are Governor and Lieutenant Governor (on a single ticket) with 1 other elected official 4.0 The elected executive positions are Governor and Lieutenant Governor (on a single ticket) with some "process officials" (Attorney General, Secretary o f State, Treasurer, Auditor, Comptroller) elected 3.0 The elected executive positions are Governor and Lieutenant Governor (on a single ticket) with "process officials" and major/minor policy officials elected 2.5 The Governor is elected (separately from the Lieutenant Governor) and fewer than 7 other executive positions are elected. Among those other positions, none are major policy officials 2.0 The Governor is elected (separately from the Lieutenant Governor) and fewer than 7 other executive positions are elected. Among those other positions, 2 or fewer are policy officials 1.5 The Governor is elected (separately from the Lieutenant Governor) and fewer than 7 other executive positions are elected. Among those other positions, 2 or more are policy offi cials 1.0 The Governor is elected (separately from the Lieutenant Governor) and 7 or more other executive positions are elected. Among those other positions, several are policy officials Component 2: Tenure Potential (TP) Scale Coding 5.0 Governors are elected to 4 year terms, with no constraints on how many terms can be served 4.5 Governors are elected to 4 year terms, and can serve for a maximum of 3 terms*, eventually reach a total limit consecutive only limit 4.0 Governors are elected to 4 yea r terms, and can serve for a maximum of 2 consecutive terms and 3 total (my revision) 3.5 Governors are elected to 4 year terms, and can serve for a maximum of 2 total terms 3.0 Governors are elected to 4 year terms, and cannot serve consecutive terms in office 2.0 Governors are elected to 2 year terms, with no constraints on how many terms can be served 1.0 Governors are elected to 2 year terms, and can serve for a maximum of 2 terms I recoded this so that: 5) 4 year terms and no limits; 4.5) 4 year terms, some limits applied, but could run again indefinitely; 4) restricted to a strict term limit; 3) elected to
185 a 1 4 year term, cannot run again immediately; 2) 2 year term, no limits; 1) 2 year terms, hard limit Component 3: Appointment Power (AP) Scale Coding 5.0 The governor appoints and no other approval is required 4.0 The governor appoin ts but a board, council or legislature must approve appointments 3.0 Another entity appoints, the governor approves; or the governor shares appointment powe rs with another entity 2.0 Another entity appoints, the governor, in conjunction with others, approve 1.0 This is a statewide elected office; or appointments are made without gubernatorial approval Component 4: Budgetary Power (BP) Scale Coding 5.0 The Governor has full responsibility, and the legislature may nor increase the executive budget 4.0 The Governor has full responsibility, but the legislature can increase the executive budget by special majority vote or subject to line item veto 3.0 The Governor has full responsibility, but the legislature has unlimited power to change the executive budget 2.0 The Governor shares responsibility, and the legislature has unlimited power to change the executive budget 1.0 The Governor shares responsibility with other elected officials, and the legislature has unlimited power to change the executive budget Component 5: Veto Power (VP) Scale Coding 5.0 The governor has the power to item veto. In order for the legislature to override vetoes, a special m ajority vote is required (3/5 of all elected members or 2/3 of legislators present) 4.0 The governor has the power to item veto. In order for the legislature to override vetoes, a majority vote of all elected members is required 3.0 The governor has the power to item veto. In order for the legislature to override vetoes, a majority vote of all members present is required 2.5 The governor has the power to item veto appropriation bills only. In order for the legislature to override vetoes, a special majori ty vote is required (3/5 of all elected members or 2/3 of legislators present) 2.0 The governor has the power to item veto appropriation bills only. In order for the legislature to override vetoes, a majority vote of all members present is required
186 1.5 T he governor does not have the power to item veto any bill; However, they can veto legislation. In order for the legislature to override vetoes, a special majority vote is required (3/5 of all elected members or 2/3 of legislators present) 1.0 The governor does not have the power to item veto any bill; However, they can veto legislation. In order for the legislature to override vetoes, a majority vote of all members present is required Component 6: Partisan Control (PC) Scale Coding 5.0 More than 75% of both houses are controlled by same party as the governor 4.5 More than 75% of one house is controlled by the same party as the governor; the party control of the remaining house is substantial, but less significant (control>65%) 4.0 More than 50% of b oth houses are controlled by same party as the governor 3.5 1 house is controlled by the governor's party; the remaining house is evenly split 3.0 For both houses, 50% of both houses are controlled by same party as the governor, or 1 house is controlled by the governor's party and the remaining house is controlled by an oppositional party 2.5 1 house is not controlled by the governor's party; the remaining house is evenly split 2.0 Less than 50% of both houses are controlled by same party as the governo r 1.5 Less than 25% of one house is controlled by the same party as the governor; the party control of the remaining house is substantial, but less significant (control<35%) 1.0 Less than 25% of both houses are controlled by same party as the governor Note: this is usually qualified by saying within a 12 or 16 year time period. AP is a composite measure that takes the average value of gubernatorial appointment power from 6 major governmental functional areas. The functional areas included are: Correct ions, Education (K 12 specifically), Health, Highways/Transportation (preference given to Transportation), Public Utilities Regulation and Welfare. These are the fields originally selected by Thad Beyle.
187 LIST OF REFERENCES Abney, G., & Lauth, T. P. (1983). The governor as chief administrator. Public Administration Review 43 (1), 40 49. A rizona Memory Project. ( 2014 ) e xecutive o http://a zmemory.azlibrary.gov/cdm/search/collection/execorders (accessed October 14, 2014). s tates and the d iffering i mpetus for d ivergent p aths on s ame s ex m arriage, 1990 The Policy Studies Journal 31:331 52. Barth J., & Ferguson, M. R. (2002). American governors and their constituents: The relationship between gubernatorial personality and public approval. State Politics & Policy Quarterly 2 (3), 268 282. Bernick, E. L. (1979). Gubernatorial tools: Formal vs. inf ormal. The Journal of Politics 41 (02), 656 664. Berry, F. S., & Berry, W. D. (1990). State lottery adoptions as policy innovations: An event history analysis. American political science review 84 (02), 395 415. Berry, W. D., Ringquist, E. J., Fording, R c itizen and g overnment i deology in the American s tates, 1960 American Journal of Political Science 42: 327 48. chair. Public Admi nistration Review 28 540 545. Beyle, T L (2007). The governors. In Politics in the American States: A Comparative Analysis by Virginia Gray and Russell L Hanson, 191 231. Washington, DC: CQ Press, 1999. Beyle, T. L. (2007). Gubernatorial Power: The I nstitutional Power Ratings for the 50 Governors of the United States University of North Carolina, at Chapel Hill. http://www.unc.edu/~beyle/gubnewpwr.html Bishin, B. G. (2000). Constituency i nfluence in Congress: Does s ubconstituency m atter? Legislativ e Studies Quarterly 389 415. Black, R. C., Madonna, A. J., Owens, R. J., & Lynch, M. S. (2007). Adding r ecess a ppointments to the p t ool c u nilateral p owers. Political Research Quarterly 60 (4), 645 654. hes to m odeling the a doption and d iffusion of p olicies with m ultiple c State Politics & Policy Quarterly 9:229 52.
188 Bolton, A., & Thrower, S. (2015). Legislative c apacity and e xecutive unilateralism. American Journal of Political Science Bowlin g, C. J., & Ferguson, M. R. (2001). Divided government, interest representation, and policy differences: Competing explanations of gridlock in the fifty states. Journal of Politics 63 (1), 182 206. Bowman, A. O. M., & Kearney, R. C. (1986). The resurgence of the states Prentice Hall. Brewer, P. R. (2007). Value war: Public opinion and the politics of gay rights Rowman & Littlefield Publishers. Burke, J. P. (1992). The institutional presidency Baltimore: John Hopkins University Press. Burke, J. P. (20 00). The institutional presidency: Organizing and managing the White House from FDR to Clinton Johns Hopkins University Press. Colvin, R. A. (2007). The rise of transgender inclusive laws how well are municipalities implementing supportive nondiscrimina tion public employment policies?. Review of Public Personnel Administration 27 (4), 336 360. Conlan, T. J. (1991). And the beat goes on: Intergovernmental mandates and preemption in an era of deregulation. Publius: The Journal of Federalism 21 (3), 43 57. Converse, P. E. (1990). Popular representation and the distribution of information. In Information and democratic processes ed. Ferejohn, J. A., and Kuklinski, J. H. Chicago: University of Illnois Press. Council of States. "Governors' Powers and Authori ty." http://www.nga.org/cms/home/management resources/governors powers and authority.html (accessed on October 12, 2014). Council of State Governments. ( 2014 ) Book of States http://knowledgecenter.csg.org/kc/category/content type/bos 2014 (a ccessed October 10, 2014). Deering, C. J., & Maltzman, F. (1999). The politics of executive orders: Legislative constraints on presidential power. Political Research Quarterly 52 (4), 767 783. Department of Defense. (2013). DOD Announces Same Sex Spouse B enefits. Retrieved from https://owl.english.purdue.edu/owl/resource/560/10/ Dometrius, N. C. (1979). Measuring gubernatorial power. Journal of Politics 41 589 610.
189 Dometrius, N. C. (1987). Changing gubernatorial power: The measure vs. reality. Western Political Quarterly 40 319 28. Dometrius, N. C. (1999). Governors: Their heritage and future. American State and Local Politics: Directions for the 21st Century, eds. Ronald E. Weber and Paul Brace. New York: Chatham House 38 70. Donovan, T., Smith, D A., Osborn, T., & Mooney, C. Z. (2015) State & local politics: institutions and reform 4 th ed. Stamford, CT: Cengage Learning Elazar, D. J. (1972). American federalism: A view from the states Crowell. Elazar, D. J. (19 84 ). American federalism: A vie w from the states 3 rd ed. New York: Harper and Row, Chapter 5 Erikson, R. S., MacKuen, M. B, and Stimson, J. A. (2002). The macro polity New York: Cambridge University Press. Eshbaugh Soha, M. (2006). The President's Speeches: Beyond" going Public" B oulder, CO: Lynne Rienner Publishers. Ferguson, M. R. (2003). Chief executive success in the legislative arena. State Politics & Policy Quarterly 3 (2), 158 182. Ferguson, M. R. (Ed.). (2006). The executive branch of state government: people, process, a nd politics. ABC CLIO. Ferguson, M. R., & Bowling, C. J. (2008). Executive orders and administrative control. Public Administration Review 68 (s1), S20 S28. Ferguson, M., & Shor, B. (2014). Unilateral l awmaking in the American g Working pap er. Fine, J. A., & Warber, A. L. (2012). Circumventing adversity: Executive orders and divided government. Presidential Studies Quarterly 42 (2), 256 274. Fisher, U., & Ury, W. (1991). Patton. Getting to YES 111 113. Godwin, M. L., & Schroedel, J. R. (20 00). Policy diffusion and strategies for promoting policy change: Evidence from California local gun control ordinances. Policy Studies Journal 28 (4), 760 776. Haider Markel, D. P. (2000). Lesbian and gay politics in the states: Interest groups, elector al politics, and public policy. In The Politics of Gay Rights by Craig Rimmerman, Kenneth Wald and Clyde Wilcox, 290 346: Chicago: University of Chicago Press.
190 Haider Markel, D. P. (2001). Policy diffusion as a geographical expansion of the scope of pol itical conflict: Same sex marriage bans in the 1990s. State Politics and Policy Quarterly, 1 (1): 5 26. Hebe, W. (1972). Executive orders and the development of presidential power. Vill. L. Rev. 17 688. Hicks, W. D., McKee, S. C., Sellers, M. D., & Smith D. A. (2014). A principle or a strategy? Voter identification laws and partisan competition in the American States. Political Research Quarterly 68 (1), 18 33. Howell, W. G. (2003). Power without persuasion: The politics of direct presidential action Princeton: Princeton University Press. Howell, W. G. (2005). Unilateral powers: A brief overview. Presidential Studies Quarterly 35 (3), 417 439. Human Rights Campaign. (2014). http://www.hrc.org/ (accessed November 12, 2014). Karch, A. (2007). Democrati c laboratories: Policy diffusion among the American states University of Michigan Press. Karch, A. (2010). Policy feedback and preschool funding in the American states. Policy Studies Journal 38 (2), 217 234. Kernell, S. (2006). Going p ublic: New s trate gies of p residential l eadership: New s trategies of p residential l eadership Washington, D.C.: CQ Press. Washington Post Retrieved from https://www.washingtonp ost.com/blogs/fact checker/post/when did mcconnell say he wanted to make obama a one term president/2012/09/24/79fd5cd8 0696 11e2 afff d6c7f20a83bf_blog.html King, G., & Ragsdale, L. (1988). The elusive executive: Discovering statistical patterns in the presidency. Washington, DC: Congressional Quarterly Press. Klarner, Carl. (2012). http://www.indstate.edu/polisci/klarnerpolitics.htm (accessed December 22, 2012). Kousser, T., & Phillips, J. H. (2012). The power of American governors: Winning on budgets and losing on policy Cambridge University Press. Krause, G. A., & Cohen, D. B. (1997). Presidential use of executive orders, 1953 1994. American Politics Quarterly 25 (4), 458 481. Lax, J. R., & Phillips, J. H. (2009). Gay rights in the states: Public o pinion and policy responsiveness. American Political Science Review 103 (03), 367 386.
191 Layzer, J. A. (2015). The environmental case: Translating values into policy CQ Press. Long, J. S. (1997). Regression models for categorical and limited dependent var iables. Thousand Oaks: Sage Publications. Maddex, Robert. (2006). State c onstitutions in the United States (2nd ed.) Washington, DC: CQ Press. Marshall, W. P. (2008). Eleven r easons w hy p residential p ower i nevitably e xpands and w hy i t m atters. Boston Uni versity Law Review 88 505 22. May, J. C. (2003). of the States 2003. Mayer, K. R. (1999). Executive orders and presidential power. Journal of Politics 61 (2) 445 466. Mayer, K. R., & Price, K. (2002). Unilateral Presidential Powers: Significant Executive Orders, 1949 99. Presidential Studies Quarterly 32 (2), 367 386. Mayhew, D. R. (1974). Congress: The electoral connection Yale University Press. Mills, M. (2011). Introducing survival and event history analysis Thousand Oaks: SAGE Publications. Moe, T. M. (1985). C ontrol and feedback in economic regulation: The case of the NLRB. American Political Science Review 79 (04), 1094 1116. Moe, T. M., & Howell, W. G. (1999). Unilateral action and presidential power: A theory. Presidential Studies Quarterly 29 (4), 850 873 Mooney, C. Z., & Lee, M. H. (1995). Legislative morality in the American states: The case of pre Roe abortion regulation reform. American Journal of Political Science 39 (3), 599 627. Mueller, K. J. (1985). Explaining variation and change in gubernatori al powers, 1960 1982. The Western Political Quarterly 424 431. Nathan, R. P. (1983). The administrative presidency John Wiley & Sons. Nathan R. P., & Doolittle F. C. (1987). Reagan and the s tates. Princeton: Princeton University Press. National Confere nce of State Legislatures. NCSL's Ballt Measures Database. ( 2014 ) http://www.ncsl.org/research/elections and campaigns/ballot measures database.aspx (accessed September 22, 2014). National Gay and Lesbian Task Force. (201 4 ). Home p age http://www.thetas kforce.org/ (accessed November 6, 2012).
192 Nelson, M. (1982). A short, ironic history of American national bureaucracy. The Journal of Politics 44 (03), 747 778. Neustadt, R. E. (1960). Presidential p ower and the m odern p residents: The p olitics of l eadersh ip from Roosevelt to Reagan. New York: Maxwell Macmillan International. Oklahoma Secretary of State. ( 2014 ) o https://www.sos.ok.gov/gov/execorders.aspx (accessed November 8, 20 14). Ouyang, Y., & Waterman, R. W. (2015). How l egislative ( i n) a ctivity, i deological d ivergence, and d ivided g overnment i mpact e xecutive u nilateralism: A t est of t hree t heories. In Congress & the Presidency (Vol. 42, No. 3, pp. 317 341). Routledge. Parti n, Randall W. ( 2001 ) "Campaign i ntensity and v oter i nformation: A l ook at g ubernatorial c ontests." American Politics Research 29 (2): 115 140. Pfiffner, J. P. (1999). The managerial presidency (No. 4). College Station: Texas A&M University Press. Rabe H esketh, S., & Skrondal, A. (2008). Multilevel and longitudinal modeling using Stata STATA Press. Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods 2 nd edition. Thousand Oaks: Sage Publications. Rosenthal, A. (1990). Governors and legislatures: Contending powers Washington D.C: CQ Press. Sabato, L (1978). Goodbye to good time Charlie. Lexington: Lexington Books. K. N. (eds.). Politics in the American States p. 207 38. Boston: Little Brown. Sellers, M. D. (2014). Discrimination and the t ransgender p opulation a nalysis of the f unctionality of l ocal g overnment p olicies t hat p rotect g ender identity. Administratio n & Society 46 (1), 70 86. Sellers, M. D., & Colvin, R. (2014). Policy learning, language, and implementation by local governments with transgender inclusive nondiscrimination policies. In Transgender Rights and Politics: Groups, Issue Framing, and Policy Adoption ; Jami K Taylor and Donald P. Haider Markel (eds): University of Michigan Press. Sharp, E. B. (2005). Morality politics in American cities Lawrence: University of Kansas Press.
193 Shipan, C. R., & Volden, C. (2008). The mechanisms of policy diffus ion. American journal of political science 52 (4), 840 857. Soule, S. A., & Earl, J. (2001). The enactment of state level hate crime law in the United States: Intrastate and interstate factors. Sociological Perspectives 44 (3), 281 305. Spill, R. L., Lic ari, M. J., & Ray, L. (2001). Taking on tobacco: policy entrepreneurship and the tobacco litigation. Political Research Quarterly 54 (3), 605 622. Tarr, G. A. (1998). Understanding s tate c onstitutions Princeton: Princeton University Press. Taylor, J. K. Lewis, D. C., Jacobsmeier, M. L., & DiSarro, B. (2012). Content and complexity in policy reinvention and diffusion gay and transgender inclusive laws against discrimination. State politics & policy quarterly 12 (1), 75 98. Transgender Law Center. (2010) Litigation: Case l aw. http://www.transgenderlaw.org/cases/ (accessed May 18, 2014). Walker, J. L. (1969). The diffusion of innovations among the American states. American P olitical S cience R eview 63 (03), 880 899. Waterman, R. W. (1989). Presidential influence and the administrative state Memphis: University of Tennessee Press. Wigton, R. C. (1996). Recent presidential experience with executive orders. Presidential S tudies Q uarterly 26 (2), 473 4 84. Wilcox, C., & Robinson, C. (2010). Onward Christian soldiers?: the religious right in American politics Boulder: Westview Press. Winston, P. (2002). Welfare policymaking in the states: The devil in devolution Washington, D.C.: Georgetown University Press.
19 4 BIOGRAPHICAL SKETCH Mitchell Dylan Sellers received his Ph.D. from the University of Florida in the summer of 2016. His substantive research focuses on state politics, executive legislative relations and public policy. His methodological specia lties include multilevel modeling, survival analysis and maximum likelihood estimation. Prior to entering the University of Florida, he graduated from Texas A&M University with a B.S. in p olitical s cience, with a minor in u rban and r egional p lanning, and r eceived an M.P.A. from Texas State University, with a minor in u rban and e nvironmental p lanning. His research on policy diffusion and transgender rights has appeared in several scholarly outlets, including Political Research Quarterly Administration & Soc iety and an edited volume on transgender rights and politics from the University of Michigan Press.