Exploring Changing Patterns of Sponsorship and Cosponsorship in the U.S. House

Transcription

1 / X American Garand, Burke Politics / Sponsorship Research and Cosponsorship in the U.S. House Legislative Activity and the 1994 Republican Takeover Exploring Changing Patterns of Sponsorship and Cosponsorship in the U.S. House James C. Garand Kelly M. Burke Louisiana State University, Baton Rouge American Politics Research Volume 34 Number 2 March Sage Publications / X hosted at Is sponsorship and cosponsorship behavior by U.S. House members responsive to which party controls the chamber? The Republican takeover of the U.S. House in 1995 provides an ideal natural experiment to test whether House members are more or less likely to sponsor and cosponsor legislation as a function of their majority or minority party status. We suggest that changes in control of the House created a set of institutional incentives that shifted sponsorship and cosponsorship behavior by House members. Using data on House members serving from the 102nd Congress to the 105th Congress, we consider whether Democrats and Republicans reconfigured their sponsorship and cosponsorship behavior after the shift in partisan control, controlling for the effects of other variables. We find that majority party status matters, with members of the majority party consistently exhibiting higher levels of sponsorship activity, though our findings on cosponsorship activity are less definitive. Keywords: Congress; legislation; sponsorship; cosponsorship; U.S. House of Representatives; Republican Revolution; majority party The Republican takeover of the U.S. House of Representatives following the 1994 congressional elections heralded in a new era of American politics. Prior to 1994, the Democratic Party had maintained control over the U.S. House for almost 50 years, holding majority party status in every elec- Authors Note: We are indebted to Jim Gimpel and three anonymous reviewers for their extraordinarily constructive comments on an earlier draft of this article. The article has also benefited from constructive comments on this project by Jennifer Victor and Justin Buchler. We also note that both coauthors contributed equally to this project, and the coauthors names are listed in reverse alphabetical order. 159

2 160 American Politics Research tion cycle since the Republican-controlled House during the 80th Congress ( ). When Americans awoke to a Republican Congress in January 1995, the stage was set for a wide array of political and policy changes. The Republicans immediately set out to implement their Contract with America, and the general direction of policy took a decided turn to the right. The period following the Republican takeover was a time of heightened partisanship and conflict, culminating in the Clinton impeachment effort. With Republicans in control of the House, other important changes occurred. Some of these were readily apparent to the mass public, whereas others were noticeable only to careful observers of Congress. At the most obvious level, with the Republicans in the majority the most visible leadership positions of the House shifted from Democratic to Republican control. Committee and subcommittee chairs also shifted completely from the Democrats to the Republicans, leaving the Republicans in a stronger position to control the policy agenda in the House. The distribution of policy preferences in the House moved decidedly to the right as the mean ideological position for the entire membership of the House shifted from the liberal side of the ideological spectrum to the conservative side. Democratic House members became more liberal and cohesive, and the Republican House members became more conservative and cohesive. The result was a more ideologically polarized House membership and more intense interparty conflict. But below the surface other important changes occurred, most notably in terms of the incentive structure governing members behavior. One aspect of legislative behavior that might be affected by partisan control of the House is the level and intensity of legislative activity specifically, the levels of sponsorship and cosponsorship activity by Republican and Democratic House members. When a given party is in the majority, one would expect the members of that party to be active in setting the policy agenda, primarily through their work in chamber, party, and committee leadership positions. One component of setting the agenda is the origination and sponsorship of legislation. On the other hand, members of the minority party are often disadvantaged in their efforts to move legislation through the legislative process, and this may affect their levels of sponsorship and cosponsorship activities beyond their normal propensities to engage in these activities. Did patterns of sponsorship and cosponsorship change when the Republicans took control of the U.S. House following the 1994 elections? If in static models majority party status matters, one would expect changes in majority party status to turn previous patterns of sponsorship and cosponsorship on their head. If majority party members are more likely to sponsor legislation, and if there is a shift in majority party status, then it seems reasonable to expect

3 Garand, Burke / Sponsorship and Cosponsorship in the U.S. House 161 changes in the sponsorship and cosponsorship behavior of the members of the new majority and minority parties. From this perspective, one would expect members of the new Republican majority to have surpassed members of the new Democratic minority in their levels of sponsorship and cosponsorship. On the other hand, the Republican and Democratic parties differ considerably in their policy positions and perceptions of the role of government, so a case can be made that the greater activism by Democrats in sponsoring and cosponsoring legislation prior to 1995 is a natural manifestation of the greater tendency of Democrats to promote government solutions to public policy problems. Under this scenario, the shift in party control starting with the 104th Congress would not be expected to have an effect on patterns of sponsorship and cosponsorship behavior. The observation of a shift in sponsorship and/or cosponsorship activities can be viewed as evidence of institutional effects on member behavior above and beyond what one would expect given legislators individual dispositions as defined by their normal partisan and ideological orientations. The purpose of this article is to take advantage of the natural experiment of the Republican takeover of the U.S. House in 1995 to explore the effects of majority party status on sponsorship and cosponsorship behavior. Using pooled data from the 102nd Congress to the 105th Congress, we develop a model of sponsorship and cosponsorship that considers the effect of majority party status, controlling for the effects of a wide range of variables thought to relate to sponsorship and cosponsorship activity. We speculate that House Democrats would be more active relative to House Republicans in sponsoring and cosponsoring legislation prior to the 104th Congress, controlling for the effects of other variables. If majority party status matters, the patterns of behavior should have shifted following the Republican takeover in early 1995: Republican House members are expected to be more active in sponsoring and cosponsoring legislation relative to Democrats. Hence, one result of the new Republican majority in 1995 may have been changes in the configuration of relative sponsorship and cosponsorship activity by the two major parties in the House. Sponsorship, Cosponsorship, and Majority Party Status In the U.S. House, legislators who introduce a given piece of legislation are classified as primary sponsors. Sponsorship of legislation is not a costless action. Sponsorship of a bill usually carries with it the obligations to oversee the writing of legislation to be introduced and to shepherd the legis-

4 162 American Politics Research lation through the legislative process. On the other hand, legislators can attach their name to a given bill as a cosponsor. Cosponsorship is often seen as relatively low-cost position taking (Kessler & Krehbiel, 1996, p. 555) because legislators can sign on as a cosponsor without necessarily participating in the hard work of moving a bill through the legislative process. Cosponsorship does not necessarily obligate legislators to support a bill throughout all stages of the legislative process (e.g., Krehbiel, 1995), and there has been much writing about how cosponsorship is used by legislators as a mechanism to signal other members of Congress and/or their constituents (Kessler & Krehbiel, 1996; Wilson & Young, 1997). The costs of cosponsorship will be variable across legislators, depending on their level of commitment, their willingness to invest time and energy in a given bill, and their perceptions of benefits and costs of position taking in the chamber and with their constituents. Previous research on sponsorship and cosponsorship has been somewhat limited, though the subject has drawn some scholarly interest of late. A small number of scholars (Burke & Garand, 2005; Campbell, 1982; Schiller, 1995; Swers, 2005) have measured the number of bills sponsored or cosponsored by individual members and then have modeled aggregate sponsorship or cosponsorship behavior. Moreover, there have been several studies of cosponsorship behavior by individual members of Congress on specific bills (Kessler & Krehbiel, 1996; Krehbiel, 1995; Regens, 1989). Krehbiel (1995) considers how legislators waffle from the cosponsorship stage to later stages in the legislative process. Kessler and Krehbiel (1996) use detailed data on the timing of cosponsorship decisions in the 103rd U.S. House of Representatives to explore possible signaling strategies, suggesting that House members are more likely to use cosponsorship to signal other House members rather than outside constituencies. Still other scholars have considered the effects of the number of cosponsoring legislators on the probability that a given bill will pass the legislative chamber (Anderson, Box-Steffensmeier, & Sinclair-Chapman, 2003; Browne, 1985; Wilson & Young, 1997). To date, although some scholars have discussed the possible effects of majority status on sponsorship and cosponsorship behavior in the U.S. House (Anderson et al., 2003; Wawro, 2000) and U.S. Senate (Schiller, 1995), there has been no study of which we are aware that has explicitly considered the effects of majority party status and, in particular, the Republican takeover of the U.S. House in 1995 on patterns of sponsorship and cosponsorship among U.S. House members.

5 Garand, Burke / Sponsorship and Cosponsorship in the U.S. House 163 Theoretical Considerations The theoretical orientation for our research has been discussed in more detail elsewhere (Burke & Garand, 2005), but we describe it briefly here. First, we suggest that in building a theory of sponsorship and cosponsorship activity, it is important to consider the motivations and goals of legislators. What motivates legislators to act in the legislative arena? What are legislators goals, and what outcomes do they prefer? Moreover, because legislators do not exist in a vacuum, they must consider how their goal-seeking behavior interacts with their environment to increase or decrease the prospects that they can achieve their goals. We suggest that any successful theory of legislative sponsorship and cosponsorship should consider both legislators internal motivations and the external environments within which they must strategically operate to achieve their goals. Motivation and Goals As a starting point, what motivates legislators to be active in sponsoring or cosponsoring legislation? We begin with the assumption that legislators are purposive, boundedly rational actors who take actions that they perceive will, on balance, increase the probability that they will be able to achieve their goals. By purposive we mean thatlegislatorsaregenerally awareof theoutcomes that they prefer, and they intentionally take actions designed to increase the likelihood that those outcomes can be achieved. Moreover, we suggest that legislators seek to increase the benefits and reduce the costs associated with their actions. Sponsoring or cosponsoring legislation may bring to the legislator expected benefits and costs, and legislators will participate in these actions insofar as the actions increase their expected benefits relative to costs. If sponsoring or cosponsoring legislation will increase the probability that they will reap the benefits associated with the achievement of their goals, they will increase their participation in these activities. If, on the other hand, legislators perceive sponsorship or cosponsorship activity as decreasing the likelihood of achieving their goals, or if legislators perceive that the costs of sponsorship and cosponsorship activity exceed the expected benefits of these actions, they will exhibit lower levels of sponsorship and cosponsorship activity (cf. Schiller, 1995). To be sure, legislators reside in limited information environments, and hence they will have some uncertainty about the degree to which sponsorship and cosponsorship behavior will help them achieve their goals. It is not always predictable how sponsorship or cosponsorship activity will be received by other actors (e.g., constituents or fellow legislators), so estimates of benefits

6 164 American Politics Research and/or costs associated with these activities will be imprecise. However, sponsorship and cosponsorship activity will be expected, subject to some level of uncertainty, to affect the achievement of goals and preferred outcomes. Legislators will select the mix of sponsorship and cosponsorship activities that they believe will provide them with the best chance of meeting their goals. Like Fenno (1973) and Mayhew (1974), we recognize that legislators have a variety of goals, including reelection, influence within the legislature, making good public policy, a career beyond the legislature, and personal gain. Legislators will vary in how they rank these goals, though there is a fairly wide consensus that the primacy of reelection to the achievement of other goals gives the reelection goal a high priority for most legislators. But other goals will vary in importance, and legislators who are relatively unconcerned about their reelection prospects either because they intend to leave the legislature or because they represent very safe districts may give a high priority to other goals, particularly legislative influence and good public policy. The motivations and goals of legislators are not directly observable. However, there are likely to be patterns of sponsorship and cosponsorship behavior across legislators and across time that are consistent with these various goals and motivations and that will increase (or decrease) the prospects of achieving those goals. First, some (and perhaps many or most) legislators will be driven by electoral considerations. Although electoral pressures may lead legislators to sponsor or cosponsor specific legislation and remain mum on other legislation, it is unlikely that vulnerable legislators will be generally active in sponsoring or cosponsoring in legislation. Because constituency interests do not overlap completely, over multiple bills the act of sponsorship and cosponsorship can create a coalition of minorities in legislators districts, much in the same way that presidential actions create a coalition of minorities that erodes presidential approval. For legislators who are active in sponsorship or cosponsorship, the result may be an erosion of electoral support. For electorally secure legislators, the erosion of support is unlikely to be a problem. Other legislators will structure their sponsorship and cosponsorship activities around the goals of influence in the legislature or making good public policy. For these legislators there is reason to expect a higher level of sponsorship and cosponsorship activity. For instance, one way in which legislators can make a name for themselves and become influential in the legislature is to be actively involved in the day-to-day work of the policy process. Sponsoring and cosponsoring legislation gives legislators an opportunity to

7 Garand, Burke / Sponsorship and Cosponsorship in the U.S. House 165 develop and hone policy proposals and to show that they can contribute to the hard work of putting together voting coalitions to support legislation. Sponsorship can be a highly visible activity, particularly for successful legislation. Hence, we would expect that legislators who are intent on achieving positions of power in the legislature (or who are already in leadership positions) to take an active role, particularly in the more highly visible sponsorship activities. The goal of making good public policy is also likely to move legislators in the direction of more extensive sponsorship and cosponsorship activity. Making good public policy often involves the development and promotion of policy proposals part of what Wawro (2000) refers to as legislative entrepreneurship and sponsorship (and, possibly, cosponsorship) activities are at the forefront of these efforts. Moreover, much has been written of late about the use of cosponsorship as a signaling mechanism (Wilson & Young, 1997). Legislators may sign on as cosponsors as a means of signaling other like-minded legislators, hoping to create a bandwagon effect in support of the legislation. Members of the legislature who have strong policy preferences and who are intent on promoting good public policy may be active in cosponsorship as a means of contributing to the coalition-building process on a wide range of bills before the body. External Environment Ultimately, we suggest that legislators are boundedly rational, purposive, and goal oriented, and they become active in sponsorship and cosponsorship (or not) as a means of achieving those goals. However, legislators do not exist in a vacuum, and they must tailor their activities based on the environments within which they reside. What this means is that legislators must be somewhat strategic, insofar as they cannot pursue their interests blindly and without consideration of other actors. Rather, strategic legislators will link their activities and goals but only as mediated through their external environments. Legislators must consider how other actors (e.g., fellow legislators, campaign contributors, lobbyists, the presidential administration, and constituents) will respond to their actions and how legislators pursue their interests in this external environment will determine the degree to which they are able to achieve their goals. How do external environments work to shape sponsorship and cosponsorship behavior? First, legislators may vary in their sponsorship and/or cosponsorship activities based on the attributes of their fellow legislators. For instance, legislators who are members of the majority party may be called on to be more active in setting the policy agenda. Legislators of both parties may

8 166 American Politics Research also be encouraged to participate in signaling games whereby they cosponsor legislation as a means of taking a public position on a bill before it is up for consideration and of encouraging like-minded legislators to support the legislation, creating a bandwagon effect (Wilson & Young, 1997). Moreover, legislators may become active in sponsoring and cosponsoring bills as a result of pressure from like-minded members who are pushing a policy agenda. Second, external actors, such as interest groups and campaign contributors, may shape the degree to which legislators are active in sponsorship and cosponsorship. The degree to which campaign contributions have an effect on the roll-call behavior on final passage votes is subject to some disagreement in the scholarly literature (Smith, 1995; Wright, 1985). The problem is endogeneity between campaign contributions and legislative behavior; it is difficult to conclude that legislators systematically shift their voting behavior in response to campaign contributions because those very campaign contributions are often made to individuals who already agree with the goals of the contributors. However, there is evidence that campaign contributions shape levels of legislative activity, particularly behind the scenes at the committee and subcommittee stages (Hall, 1987; Hall & Wayman, 1990). We speculate that bill sponsorship and cosponsorship activities will be similarly important for campaign contributors, and hence sponsorship and cosponsorship may be shaped by pressure from campaign contributors for greater legislative activism. Third, legislators sponsorship and cosponsorship activity may be either enhanced or constrained by the president and his administration. Members of the president s party may be asked to carry the president s water on important policy matters being pushed by the administration. Certainly members of the president s party will be encouraged to sponsor the president s legislative agenda, and there may also be strong incentives for members of the president s party to lend their support to administration-sponsored legislation through cosponsorship designed to signal other legislators to follow suit. Finally, constituents serve as a major external force that shapes legislators sponsorship and cosponsorship activity. Some constituencies expect high levels of policy activism from their legislative representatives. Other constituencies are content to let their legislators ignore the policy process and focus on the provision of constituency services. In one sense, there is a market for varying levels of activism in sponsorship and cosponsorship. For some districts, legislators who are inactive in sponsorship and cosponsorship will not survive regular election cycles. For other districts, legislators who are active in sponsorship and cosponsorship may face the prospect of electoral defeat.

9 Garand, Burke / Sponsorship and Cosponsorship in the U.S. House 167 Toward a Model of the Effects of Majority Party Status The proceeding discussion lays out a general theoretical framework that guides the development of our operational model of sponsorship and cosponsorship activity. We do not presume to propose a direct test of this theory, because some of the important components of this theoretical framework are not directly observable (e.g., legislator motivation, legislator goals). However, our discussion of the effects of party control of the U.S. House, and the selection of other independent variables in our model, is guided by a general theoretical framework that views legislative sponsorship and cosponsorship behavior as a result of self-interested legislators seeking to achieve their goals in a context that often requires them to think strategically. Most of the literature on sponsorship and cosponsorship has been focused on time periods of fixed partisan control of the U.S. House. Scholars have explored the effects of partisanship on sponsorship and cosponsorship behavior (cf. Campbell, 1982; Krehbiel, 1995; Schiller, 1995; Swers, 2005), but these efforts have been directed primarily at distinguishing Democrats and Republicans in terms of their propensity to sponsor or cosponsor legislation. During periods of Democratic control of the U.S. House, it is impossible to distinguish the effect of being a Democrat (Republican) from being a member of the majority (minority) party. Because majority party control is something that changes only over time, the only way of distinguishing the effect of partisanship and majority party status is to conduct research over multiple years that vary in terms of which party is in the majority. There are a number of reasons to presuppose that majority party status matters for sponsorship activity. Bill sponsorship is a resource-consuming activity that often costs the sponsor allocations of time, staff, and political capital. Given these often considerable costs, legislators are unlikely to introduce legislation and invest the necessary resources unless they believe that there is a good chance that there will be a pay-off at the end of the legislative process. 1 We suggest that majority party status carries with it a set of incentives and disincentives for behavior that should both encourage members of the majority to sponsor legislation and depress sponsorship activity by members of the minority, all else being equal. What are these incentives associated with majority and minority party status? First, when a given party is in the majority, the role of its rank-and-file members in helping to shape and promote the legislative agenda of the majority party takes center stage. Majority party members are expected to contribute to the legislative agenda, and legislation sponsorship is one way of achieving this goal. Second, the majority party is able to select committee chairs and subcommittee chairs, and a majority of members on each House

10 168 American Politics Research committee come from the majority party. Both committee members and rank-and-file members from the majority party know that the bills that they introduce are more likely to face a favorable hearing from committees led by leaders from their party and composed of a majority of fellow party members. Moreover, members of the majority party control the committee staff slots that are used to generate policy proposals and legislation. Third, should a bill sponsored by a member of the majority party clear the committee hurdle, it should be more likely to face favorable treatment by the Rules Committee (which is controlled by the majority party) and in scheduling by the majority party leadership. A majority party member who sponsors a given piece of legislation is less likely to face onerous rules governing amendments and length of debate and is more likely to receive favorable treatment from the majority leadership. This creates an incentive for members of the majority party to sponsor legislation above and beyond the normal propensity of party members to sponsor legislation in the first place. Besides these institutional advantages accruing to the majority party, there may also be less formal incentives for activism by majority party members in comparison to their minority party brethren. Organized interest groups are very likely to have greater contact with members of the majority party because they control much of the policy apparatus in the House. When interest groups are interested in promoting policies requiring legislation, they are likely to target members of the majority party for their efforts. This propensity to target the majority may result in greater contributions from political action committees (PACs) to encourage greater policy and legislative activism by the majority party. The result is that these informal forces may encourage members of the majority to exhibit higher levels of sponsorship and cosponsorship behavior. On the other hand, members of the minority party face a much steeper climb when they introduce legislation. Minority members who sponsor legislation are more likely to face a hostile reception in committee, and even if their bills get through committee, they are likely to receive less favorable treatment by the Rules Committee, particularly in terms of amendability and the scheduling of floor time for consideration. Minority party members are also likely to be relatively disadvantaged in terms of their contact with interest groups and with PAC contributions, and this may depress levels of sponsorship and cosponsorship activity. Simply, members of the minority party are less likely than are members of the majority party to receive a pay-off for their investment of time and staff necessary to introduce legislation. What about cosponsorship? Although sponsorship often implies a level of commitment to a given piece of legislation, cosponsorship is a much less

11 Garand, Burke / Sponsorship and Cosponsorship in the U.S. House 169 costly activity that can be undertaken without a great deal of commitment. House members can cosponsor legislation simply by attaching their name to a given bill, and it is not unheard of for House members to withdraw their cosponsorship or to vote against a cosponsored bill if the political situation demands it (Krehbiel, 1995). Cosponsorship can be done for a wide variety of reasons, but ultimately cosponsors are not required to invest considerable resources to serve in that capacity. Given the low cost and low pay-off nature of cosponsorship, the incentive structure for cosponsorship is much more ambiguous. On one hand, cosponsorship can be viewed as another piece in the puzzle of legislative policy entrepreneurship, with members of the majority party exhibiting higher levels of cosponsorship activity relative to their minority party counterparts. Wawro (2000) defines legislative entrepreneurship as a set of activities that a legislator engages in, which involves working to form coalitions of other members for the purpose of passing legislation by combining various legislative inputs and issues in order to affect legislative outcomes (p. 4). When placed in this context, sponsorship is clearly an entrepreneurial activity, and in many (or at least some) cases, cosponsorship may be as well. Members of the majority party are more likely to engage in entrepreneurial activity, at least during the period of Democratic control through the 103rd Congress (Wawro, 2000), and if cosponsorship is part of entrepreneurial activity, there are theoretical reasons to expect members of the majority to be more active in cosponsorship activities. On the other hand, it is also possible that members of the majority party will be no more likely (or perhaps even less likely) to cosponsor legislation than similarly situated members of the minority party. Members of the majority party are expected to prioritize their legislative activities on the sponsorship side of the ledger because it is in sponsorship activity that the direction and content of policy proposals is most clearly defined. Although cosponsorship activity is relatively cost free, cosponsorship is not completely cost free, and so majority party members may prioritize their efforts and focus their attention on sponsorship activities. This does not mean that majority members will eschew cosponsorship activities altogether but rather that they will adjust their behavior to give greater weight to sponsorship activities. This could result in moderately lower levels of cosponsorship activity than predicted based on House members normal propensities toward legislative activism. Hence, an alternative perspective is that majority party status will be positively related to sponsorship activity but will be either unrelated or negatively related to cosponsorship activity.

12 170 American Politics Research Estimating the Effects of Majority Party Status To estimate the effects of majority party status on sponsorship and cosponsorship behavior among U.S. House members, we develop a model in which we depict the count of sponsorships or cosponsorships for each member as a function of partisanship and majority party status, along with controls for other variables thought to be related to the level of sponsorship and cosponsorship activity. At the outset, it should be noted that the dependent variables in this study are the count of sponsored bills and the count of cosponsored bills for each member of the U.S. House of Representatives in each Congress from the 102nd Congress ( ) to the 105th Congress ( ), though we supplement parts of our discussion with data that include the 93rd Congress ( ) to the 107th Congress ( ). These sponsorship and cosponsorship data are available from Thomas, the Library of Congress s Web site ( There are two primary independent variables in our model. First, we include a simple dichotomous measure of member partisanship, coded 1 for Republican members and 0 for Democratic members. At the simplest level, we would expect Democratic House members to be more likely to envision a strong role of government in solving the problems of the country and hence to develop an active policy agenda that includes a heavy dose of sponsorship and cosponsorship activities. Republicans, on the other hand, are typically more reticent to propose public sector solutions to policy problems and may often pursue their policy priorities by opposing proposals for new programs rather than sponsoring or cosponsoring new programs themselves; hence, Republicans may exhibit somewhat diminished levels of sponsorship and cosponsorship activity (Schiller, 1995). Given this, we hypothesize that the coefficient for this variable will be negative, controlling for the effects of other variables. As noted, we contend that the effect of party is not so simple. Rather, the effect of party will be conditional, depending on which party holds the majority of the House. Majority party status is a contextual condition that will have an effect on how House members see their roles and how party affiliation is translated into levels of sponsorship and cosponsorship activity. Following the arguments made above, we suggest that majority party members will be more active in sponsoring legislation, all else being equal. The effect of majority party status on cosponsorship is less definitive because the competing theoretical arguments about the direction of the majority party effect are in conflict. We measure this variable as a dichotomous variable, coded 1 for members of the majority party and 0 for members of the minority party. If

13 Garand, Burke / Sponsorship and Cosponsorship in the U.S. House 171 our hypothesis is supported, we expect the coefficient for the majority party variable to be positive for the sponsorship model, although the expected effect for the cosponsorship model is less clear. We include a number of control variables in our model. For the sake of brevity, we do not discuss the full contours of our model here; the theoretical rationale for including these variables in the model is discussed in more detail elsewhere (Burke & Garand, 2005). Briefly, we include the following control variables. First, we include two spatial variables, represented by Poole and Rosenthal s (2005) DW-Nominate scores. We capture the effects of ideological conservatism by including the raw DW-Nominate score in the model. Like Schiller (1995), we speculate that conservative House members will be more likely to see a limited role of government and hence less likely to promote government activity through sponsorship and cosponsorship. Because DW-Nominate scores are coded so that high values represent the more conservative policy positions, we hypothesize that the coefficient for DW-Nominate scores will be negative in both the sponsorship and cosponsorship models. On the other hand, ideological intensity and strong policy preferences should promote greater sponsorship and cosponsorship activity. We create a folded DW-Nominate score around the central 0 value by taking the absolute value of the raw DW-Nominate scores. The result is a measure ranging from 0 (pure moderate) to more than 1 (intense liberal or conservative). We hypothesize that the coefficient for this variable will be positive in both models. Second, we consider the effects of various kinds of legislative and political experience. We include a series of dichotomous variables representing whether House members had served as a former state legislator, former congressional staff member, or former state and/or local official. These variables are measured using biographical data collected from various issues of the Almanac of American Politics. We speculate that House members with political and/or legislative experience prior to arriving in Congress will have developed substantial expertise in the legislative process that can be translated into higher sponsorship and cosponsorship activity. Hence, the coefficients for these variables should be positive. Moreover, we consider the effects of seniority, which is a presumptive surrogate for legislative experience in the House. Seniority should facilitate sponsorship by giving House members the experience to move comfortably through the legislative process, and hence we hypothesize that the coefficient for seniority will be positive in the seniority model. However, we suggest that the positive effect of seniority may be limited to sponsorship. On one hand, senior members of the House should be the most likely to be legislative entrepreneurs, and one might speculate that senior members will hence be most

14 172 American Politics Research likely to be active in a wide variety of activities designed to promote public policies, including cosponsorship activities. On the other hand, senior members have many demands on their scarce resources, and they are more likely to prioritize their efforts and focus attention on sponsorship activities. Even though cosponsorship is a relatively low cost action, it is not entirely cost free, and we suggest that cosponsorship will not be a high priority for more senior members. Hence, we speculate that senior members will opt for a mix of sponsorship and cosponsorship that is weighted toward the former and away from the latter. Given this, the coefficient for seniority in our cosponsorship model should be negative. Third, the institutional positions that House members hold are expected to have an effect on sponsorship and cosponsorship activity. Committee and subcommittee chairs and ranking minority members should have a special propensity to sponsor and cosponsor legislation because they are leading committees and subcommittees from which legislation is reported. Committee and subcommittee chairs and ranking minority members are likely to focus their energies more on higher profile sponsorship activities, so the relationship between chair status and sponsorship should be particularly strong, though we do hypothesize that committee and subcommittee leaders will also be more likely to cosponsor legislation, all else being equal. In addition, we speculate that party leaders (i.e., the Speaker of the House, Majority and Minority Leaders, and Majority and Minority Whips) will take a bit of a hands-off approach on sponsorship and cosponsorship activities. Party leaders are likely to focus their policy activities on agenda setting and lobbying rank-and-file members to support party positions. Moreover, party leaders should hoard their sponsorship and cosponsorship activities so that they sponsor or cosponsor selected legislation likely to be important to the party leadership. Given this, we hypothesize that party leaders will tend to eschew sponsorship and cosponsorship for other policy-related activities. Fourth, we consider the effects of being a member of an underrepresented group. For women and Blacks, membership in the underrepresented group may bring special obligations to represent other members of society who are party to the underrepresented group, particularly on issues of importance to the group. We include binary variables for gender (coded 1 for women, 0 for men) and Blacks (coded 1 for Black House members, 0 for all others). We hypothesize that the coefficients for these variables will be positive. Finally, we contend that House members representing safe districts will be more active in sponsoring and cosponsoring legislation. There is a degree of risk associated with sponsorship and cosponsorship because these activities can generate higher levels of opposition among coalitions of minorities

15 Garand, Burke / Sponsorship and Cosponsorship in the U.S. House 173 in House members districts. Members with high levels of electoral vulnerability must be particularly cautious in their legislative activity, whereas members from safe districts can sponsor and cosponsor bills without fear of electoral retribution. Given this, we hypothesize that electoral safety, defined as the percentage of the vote received by the incumbent in the last election, will be positively related to sponsorship and cosponsorship. Data and Model Estimation To estimate our models, we have collected data on all members of the U.S. House who served during the 102nd, 103rd, 104th, and 105th Congresses. This represents two Congresses before and two Congresses after the Republican takeover following the 1994 elections. This data set is part of a largescale project that includes data collected from the 93rd Congress ( ) to the 107th Congress ( ), and we supplement part of our descriptive analysis with data from the larger data set. A summary description of the variables used in this analysis can be found in the appendix. The models that we propose to estimate raise a number of thorny methodological issues that render ordinary least squares (OLS) regression inappropriate. First, each of the dependent variables in our models (i.e., the number of bills sponsored and bills cosponsored by each House member) is best described as an events count, and the distribution of events count data often means that OLS regression is an inappropriate estimation tool. The most commonly used approach for estimating models with events count data is the Poisson model, which assumes that the dependent variable follows a Poisson distribution and which uses a log transformation to correct for the possibility of negative predicted values and inconsistent standard errors (King, 1989; Long, 1997). Furthermore, the Poisson model assumes that the mean and variance of the dependent variable are relatively equal (referred to as equal dispersion), and violations of this assumption where the variance exceeds the mean (referred to as overdispersion) represent a deviation from the Poisson distribution. In such cases, one can substitute negative binomial regression, which explicitly accounts for the overdispersion in the dependent variable. An examination of the means and variances of the sponsorship and cosponsorship variables reveals substantial overdispersion, and this means that the negative binomial model is the best choice for these data. In addition, the data in our study are measured both in the cross-section and over time. The use of pooled, cross-sectional, time-series (i.e., panel) data often results in violations of model assumptions, including autocorrelated errors and heteroskedasticity. To correct for these violations of

16 174 American Politics Research Figure 1 Mean Number of Sponsored Bills, by Party, 93rd Congress to 107th Congress Number of Sponsored Bills Congress Republican members Democratic members Note: The vertical line is at the midpoint between the 103rd and 104th Congresses and represents the shift in partisan control of the U.S. House from the Democrats to the Republicans following the 1994 congressional elections. assumptions, we estimate our negative binomial regression model using a population-averaged approach with semirobust standard errors that is appropriate for use with panel data. Sponsorship and Cosponsorship Over Time As a starting point, we describe patterns of sponsorship and cosponsorship by Democratic and Republican House members over time. In Figures 1 and 2, we place these patterns of mean sponsorship and cosponsorship rates for Democrats and Republicans in the context of longer term trends from the 93rd Congress to the 107th Congress. In Table 1, we report mean sponsorship and cosponsorship activity for all House members, and for Democrats and Republicans, for the 102nd to 105th Congresses, which is the period that is the focus of this study.

17 Garand, Burke / Sponsorship and Cosponsorship in the U.S. House 175 Figure 2 Mean Number of Cosponsored Bills, by Party, 93rd Congress to 107th Congress Number of Cosponsored Bills Congress Republican members Democratic members Note: The vertical line is at the midpoint between the 103rd and 104th Congresses and represents the shift in partisan control of the U.S. House from the Democrats to the Republicans following the 1994 congressional elections. The first thing that emerges from an examination of Figure 1 is that Democratic House members typically have higher rates of sponsorship than do Republicans, at least until the 104th Congress. With the exception of the 95th and 96th Congresses, from the 93rd to the 103rd Congresses Democrats sponsor several bills more per Congress than Republicans. This is not a huge difference, but it is clear that the typical Democrat is more active in bill sponsorship than is the typical Republican. When one aggregates all Democratic sponsorships and all Republican sponsorships, it is equally clear that Democrats collectively sponsor a great many more bills than Republicans because (a) the Democratic average is higher than the Republican average and (b) until the 104th Congress, there are more Democrats than Republicans. Up until the 104th Congress, it appears that Democrats are more active than are Republicans in sponsoring legislation. The picture changes substantially with the 104th Congress, during which the Republicans gained control of the U.S. House for the first time in almost

18 176 American Politics Research Table 1 Mean Number of Bills Sponsored and Cosponsored, 102nd to 105th Congresses M Party Difference Congress Party Control Overall Republican Democratic M t Sponsorship 102 D ** 103 D ** 104 R ** 105 R *** Cosponsorship 102 D *** 103 D R R *** Note: Party difference is calculated by subtracting the Democratic mean from the Republican mean. The reported t statistic is from the test of the difference of mean sponsorship and cosponsorship for Republicans and Democrats. **p <.05. ***p < years. The pattern of higher Democratic sponsorship is reversed in scissor-like fashion, and for the first time Republican House members have higher sponsorship rates than do their Democratic counterparts. These higher levels of Republican sponsorship persist through the 107th Congress. The shift toward greater Republican sponsorship occurs less because of any major increases in Republican sponsorship per se; rather, the shift is primarily because of a decline in Democratic sponsorship, particularly following the 103rd Congress, accompanied by a modest upward shift in Republican sponsorship. Of course this pattern is reinforced by the results reported in Table 1. During the 102nd and 103rd Congresses, the relationship between party and sponsorship is negative; Republican House members have significantly lower levels of sponsorship than do Democrats. However, after the Republicans took control of the House following the 1994 elections, the relationship between party and sponsorship is reversed, with Republicans significantly more active in sponsorship than Democrats. The differences are not huge; Democrats sponsored about 3 more bills during the 102nd and 103rd Congresses, whereas Republicans sponsored 2.17 and 4.54 more bills after they gained control of the House in the 104th and 105th Congresses, respectively. But the shift toward greater Republican sponsorship following their takeover of the House is consistent, and coupled with Democratic sponsorship advan-

19 Garand, Burke / Sponsorship and Cosponsorship in the U.S. House 177 tage prior to the 1994 elections, it suggests that majority party status has a nontrivial effect on sponsorship. Turning to cosponsorship, the pattern is somewhat less clear. In Figure 2, we show that patterns of cosponsorship were on the rise for both parties from the 93rd Congress to the 101st Congress, though Democrats typically had an advantage. Following the 101st Congress, there was a sharp decline in cosponsorship activity for both parties through the 104th Congress, after which Democratic cosponsorship increased dramatically to levels similar to those at the peak of Democratic cosponsorship activity during the 101st Congress. Republican cosponsorship increased as well, but by much more modest amounts. The bottom line is that it appears that cosponsorship is characterized more by a Democratic effect than a majority party effect. Simply, Democrats are typically more active in cosponsorship than are Republicans, though admittedly the partisan gap reached unprecedented proportions in the 106th and 107th Congresses. Moreover, these ambiguous patterns are mirrored in the results from Table 1, which reveal nonsignificant party differences during the 103rd and 104th Congresses the former the last Congress in which the Democrats controlled the House and the latter the first Republican-controlled House but significant differences during the 102nd and 105th Congresses that favored the Democrats. A Multivariate Test of Majority Party Effects Thus far we have examined the simple relationship between majority party status and patterns of sponsorship and cosponsorship activity among U.S. House members. These initial results reveal a discernible majority party effect on sponsorship activity, but these results suggest an apparent Democratic Party effect on cosponsorship. But does this pattern stand up when we subject it to an empirical test using a full multivariate model? It is possible that the simple patterns observed in Figures 1 and 2 and in Table 1 either overstate or understate the effects of the Republican takeover on sponsorship and cosponsorship. We suggest that the most appropriate test is one that includes both party and majority party variables in the same model and that is conducted for a period in which the two do not coincide perfectly. Sponsorship Models In Table 2, we present the results for our pooled negative binomial regression model of sponsorship activity for the 102nd to 105th Congresses. In addition to the coefficients for our independent variables described above, we also include coefficients for fixed effects for the 103rd, 104th, and 105th