The Speaker s Discretion: Conference Committee Appointments from the 97 th -106 th Congress

Transcription

1 The Speaker s Discretion: Conference Committee Appointments from the 97 th -106 th Congress Jeff Lazarus Department of Political Science University of California, San Diego jlazarus@weber.ucsd.edu Nathan W. Monroe Department of Political Science University of California, San Diego nmonroe@ucsd.edu Abstract: When choosing members to serve on a conference committee for a particular bill, the House Speaker selects at least some members of the bill s standing committee(s) of origin greater than 99% of the time. This has lead scholars to conclude that the Speaker defers to committee actors, who ultimately control the selection of conferees. We break with this conventional wisdom, arguing that when the Speaker anticipates that the conferees suggested by committee actors would produce an unfavorable conference report he uses his discretion to appoint other members to the conference. These other members do not replace conferees from the bill s originating committee; rather, they go to conference with the standing committee members. The intention of this packing by the Speaker, we argue, is to affect the ideological bent of the conference committee in a manner favorable to the majority party. We derive and test hypotheses regarding when this is likely to occur and what conferences look like when it does occur. We find that, contrary to the prevailing wisdom in the conference committee literature, the Speaker exercises significant influence over conference appointments, by way of conference committee packing. Prepared for presentation at the 2003 Annual Meeting of the American Political Science Association, Philadelphia, Pennsylvania, August 28-31, The authors would like to thank Rob Hennig and Mat McCubbins for their assistance in obtaining data. We would also like to thank Jamie Carson and panel participants at the 2003 meeting of the Midwest Political Science Association for their helpful comments.

2 The official rules of the House give sole authority to make conference committee appointments to the Speaker of the House. When choosing members to serve on a conference committee, however, the Speaker selects at least some members from the bill s standing committee(s) of origin more than 99% of the time. The regularity of this practice has lead many scholars to argue that the Speaker is somehow obligated to include members of the committees of origin on conference delegations, or that committee leaders such as the chairman or ranking minority member are actually in control of the selection process. Indeed, this has been the academic view dating back at least to Rogers (1922). This conventional wisdom suggests that the Speaker does not use his institutional prerogative to determine the make-up of House delegations to conference committees for any purpose of his own design - be it self or party driven - but rather abdicates the responsibility to other actors. We contend that in narrowly focusing on conferees from the committee(s) of origin, this scholarship has overlooked a central feature of the Speaker s ability to affect the make-up of conference committees: packing. We expect that the members from the committee of origin go to conference so often 1) because these members are most likely to have a constituency interest at stake in bills originating from their committee, and 2) because the appointees from most standing committees are usually representative of the majority party, since most committees are representative of the majority party, thus satisfying the Speaker s need to protect the majority party s label. However, in some cases, the Speaker may wish to improve the majority party s position on conference by augmenting the slate of conferees from the committee of origin. In these cases, we argue, it is beneficial for the Speaker to appoint members outside of the committee of origin in addition to, rather than instead of, members from the slate proposed by the committee of origin. Though this argument is observationally equivalent to the

3 conventional wisdom in many cases, it yields predictions as to when and how the Speaker is likely to use his discretion in appointing conferees, and what the make-up of conference delegations should look like under those conditions. Our hypotheses suggest, and empirical tests confirm, that the Speaker adds members who are not on the bill s committee of origin when doing so is in his or rather, his party s best interests. This is most likely when adding outside members produces a conference delegation which is more likely to issue a conference report in line with what the majority party wants. Since each conferee receives one vote on sending conference reports back to the chambers, appointing a number of conferees who all substantially agree with one another as well as the Speaker can significantly pull the conference in the Speaker s and thus the party s direction. In this way, we argue, the speaker does impose significant influence on the appointment of conferees. The remainder of the paper proceeds as follows. Section I describes the traditional rules of conferee assignment in more detail, and discusses the Speakers incentives to adhere to, and deviate from, regular appointment procedures. Section II presents a one-dimensional spatial model of conference appointment which generates predictions about the conditions under which a Speaker has incentives to impose his influence on conference committee appointments. Section III uses an original dataset to test those predictions, and section IV concludes. Section I Conference delegation appointments occur in a very regular pattern. The Speaker makes the formal selections, but only after the chairman and ranking minority member from the bill s committee(s)-of-origin have given the Speaker a list of recommended conferees. The Speaker is

4 supposed by many to rarely select members who are not on the list. The conference is typically dominated by members of the committee(s) of origin, leans toward high-seniority members rather than low-seniority members, and almost always includes the committee s chairman and ranking minority member. Finally, the conference delegation typically contains majority- and minority-party members in the same ratio as they appear on the committee of origin. This pattern has been long noted in the literature, 1 and throughout the paper we refer to this pattern of appointment as traditional selection procedures. We also use the term traditional conferee to refer to any conferee who is a member of the bill s committee of origin. 2 Traditional conferees appear on nearly all conference committees, which has lead scholars to conclude that those who submit lists to the Speaker, rather than the Speaker himself, have final say in determining the makeup of conference committees. For example, Oleszek (1996) notes that the Speaker becomes directly and aggressively involved in naming conferees [only] on issues of fundamental importance (279). Shepsle and Weingast (1987) similarly observe that, when it comes to conference assignments [T]he views of the committee chairpersons are dominant (95), and this has been the case for more than a century (97). 3 This conclusion lies in stark contrast to the House rules, which give the Speaker sole authority to appoint conference committee members, and a great a great deal of autonomy in doing so. Rule I, Section 11 states: The Speaker shall appoint all conference committees ordered by the House, and requires the Speaker to appoint a majority who generally supported the House position, as determined by the Speaker. Though, the Speaker is admonished to 1 See, for example, Froman (1967), Smith (1989), Longley and Oleszek (1989), Smith and Deering (1991), Bach (1991), Clapp (1963), Pressman (1966), Volger (1971), Ferejohn, (1974), Shepsle and Weingast (1987), Oleszek (1996), and Brown (1996). 2 Although this criterion does not coincide with all traditional appointment rules, it is inclusive of all conferees who are appointed from this procedure. 3 Others who argue from this point of view include Sinclair 1997, Clapp 1963, Smith 1988, Rogers 1922, Pressman 1966, Volger 1971, and Ferejohn, 1974.

5 include members who supported the bill on final passage, it is left up to the Speaker to determine who those supporters are and what qualifies as support. Finally, the Speaker is to name those who are primarily responsible for the legislation, and shall, to the fullest extent feasible, include the principal proponents of the major provisions of the bill or resolution passed or adopted by the House. 4 This contrast between the traditional appointment procedure and the seemingly dominant position of the Speaker outlined in the House rules gives rise to several questions. If the Speaker has sole authority to make conference committee appointments, why does he so often follow the lists given to him and appoint traditional conferees? If the Speaker needs to, can he violate tradition? If he can, how and under what conditions would he choose to do so? We argue that traditional conferees are so predominant because their presence on conference committees almost always benefits both the traditional appointees and the Speaker. Members often request, and receive, appointment to specific standing committees because those committees give the members opportunities to ingratiate themselves to constituents. These opportunities might come in the form of work on legislation (Rhode and Shepsle 1973), or credit-claiming and position-taking opportunities (Mayhew 1974, pp 95-97). A conference appointment is an extension of this opportunity structure. The appointment gives a member a chance to claim credit for, or to simply take strong public positions with regard to, the bill at hand. 5 Additionally, at times a member may seek out a conference committee appointment to protect provisions that materially benefit his constituents. A conference committee appointment is therefore electorally valuable to members of the bill s committee of origin. The Speaker, who 4 In some previous Congresses, identical language was located in Rule X. 5 Credit claiming for and position taking on conference is made possible by the fact that conference proceedings have been public record since 1974; Longley and Oleszek 1989, p. 51

6 is a selected agent of the majority party 6, must pay attention to the electoral needs of those who select him. If he does not, he risks losing his Speakership via either de-selection by disgruntled members of his party, or his party s loss of majority status. Speakers, therefore, also have career-based incentives to appoint members of the standing committee(s) of origin to conference committee. Despite these incentives, sometimes appointing traditional conferees gets in the way of another of the Speaker s goals as a selected agent of the majority party. That goal is to engineer the passage of legislation which is beneficial to (a majority of) the majority party. Conference delegations made up of strictly traditional conferees may under certain conditions (which we fully discuss in Section II) interfere with that goal. For example, we might see a conference full of members writing narrow, personalistic provisions into a conference report which other party members consider wasteful. Once appointed, such a conference could inflict real damage on the party s label with respect to the bill at hand, because conference committees act autonomously in writing a conference report, and can submit any report they like. 7 Further, once reported, conference reports are privileged on the House floor and always considered under a closed rule, making them difficult to ignore and impossible to change. As such, the Speaker has no recourse to fix the damage done by a conference committee short of killing the bill in its entirety. A preferable remedy is to prevent its occurrence. If he has reason to believe that a traditional conference will produce party label damaging conference report, then the Speaker can use his appointment power to send a different slate to conference. He can do so in one of two ways. The first is to simply ignore traditional procedures all together when selecting conferees, and appoint only conferees that will produce a 6 Technically, the entire floor selects a Speaker. In practice, however, this is always a straight party line vote. The actual selection takes place in the majority party caucus, by a majority vote of the majority party members. 7 The floor of either chamber can vote to instruct their conferees, but these instructions are nonbonding.

7 conference report to the Speaker s liking. We call this option replacement. The second option is to appoint the traditional conferees as well as others who will work with the Speaker to produce a favorable conference report; we call this packing. 8 In theory, both options work equally well in terms of securing a favorable conference report. The assent of only a majority of conferees is needed for a conference to send a report back to the chambers. If a delegation is packed heavily enough, so-called outsider conferees form the necessary majority (either by themselves or with a faction of traditional conferees). The difference between the two strategies is that replacement is more costly; it denies traditional conferees opportunities in conference to take positions and defend provisions. By packing, the Speaker gives traditional conferees the opportunity to defend their constituents interests and take public positions with reelection in mind. They may not sign the conference report, but they can fight against it for position-taking s sake. Ultimately, their opposition does little damage, since packing makes their signatures unnecessary for the majority party friendly conference report to pass. Thus, we contend that the Speaker prefers packing to replacement, since it achieves the same objective with fewer political costs. As a result, packing should occur much more frequently than replacement, and it does. In the 97 th -106 th Congresses, traditional and outsider conferees jointly appeared on 21% of conference delegations (143 of 679 cases). On the other hand, all traditional conferees were left off of the conference delegation less than 1% of the time (6 of 679 cases). 9 8 Though the speaker could pack a conference with members from the committee of origin that were not on the proposed slate, we expect that the primary draws from other committees when packing a conference. 9 There are two other strategies available to the Speaker which we do not investigate. In cases of multiple referral, the Speaker may leave one or more of the committees of origin off of conference committee, while including the others. In fact, this occurs in almost half of multiply referred bills which go to conference (49 out of 103). However, this is qualitatively different from replacement because in these cases, because all appointed conferees are traditional. Additionally, the Speaker might appoint conferees from the bill s committee(s) of origin, but go off of the suggested slate which was provided by the committee chairman and ranking minority member. This strategy is

8 We do not claim that manipulation of the conference committee s preferences is the only reason the Speaker appoints outsider conferees. This paper identifies conference-packing which pursues the passage of legislation that is favorable to the majority party. However, the Speaker might also appoint a member with specific policy expertise applicable to the bill at hand, even if that member isn t on the bill s committee of origin. Additionally, he might appoint a conferee to give that member a chance to protect a personal interest in a bill. We do not pursue these possibilities here, since the aim of this paper is not to present a theory of nontraditional conferee placement. Rather, we present a theory of the Speaker s control of conference appointments. We contend that the Speaker employs the strategy of packing conference committees to pursue the task of passing legislation that is favorable to his party, and investigate the use of that strategy. Section II This section discusses the specific conditions under which traditional conferees contrast with the interests of the majority party at large, thus giving the Speaker incentives to add additional conferees. We develop a one-dimensional spatial model of floor consideration of bills in the House of Representatives, represented in Figure F represents the floor median, M the majority party median, and C 1 and C 2 the medians of the majority party contingent on two types of outlier committees: C 1 is more moderate than the majority party median (i.e., closer to F than M), and C 2 is more extreme. Together with committees in which the majority party contingent is nearly impossible to investigate because the lists submitted by committee chairmen are not public knowledge, so we cannot compare them to Speakers conference selections. 10 This simplifying analogy between policy choice and one-dimensional space follows the lead of many congressional scholars (Downs 1957, Krehbiel 1991, 1998, Cox and McCubbins 1993, Poole and Rosenthal 1997). Note that Figure 1 represents a Republican congress, but is generalizable (by simply producing the mirror image) to Democratic congresses.

9 representative of the majority floor median, these committee types are exhaustive of all possible committees in a single dimension. We assume that at final passage, all bills B are in the F-M interval, in accordance with theories on both sides of the debate over congressional decision-making. 11 Assume a committee proposes a bill at its majority-contingent median s ideal point, such as C 1 or C 2, and that the floor considers the bill under an open rule. 12 In this case, the floor median would propose to amend the bill to her ideal point, and the amendment passes in accordance with the median voter theorem. Alternatively, recent theories of congressional organization (Sinclair 2002; Dion and Huber 1996) suggest that the Rules Committee uses its ability to limit amendment and debate time to pull bills to the majority party side of the floor median. As such, a bill may inhabit any point within the F-M interval, though we do not require that bills stray from F. 13 We further assume that after final passage, the bill is considered by a conference committee which moves the proposal at will, then reports the bill back to the floor of the House. This movement is determined by some monotonic function of the conference median s ideal point, although we are agnostic as to what this function looks like, with two exceptions. Here, we mean to be as general as possible to accommodate a range of possible outcomes of conference committee negotiations. Though we assume conference reports to be the weighted average of the medians of the House and Senate confernece delegations, these weights may be of 11 Krehbiel (1991) suggests, and Cox and McCubbins (1993, 2002) assume that in the House, all bills are amended to the floor median s ideal point before passing. Krehbiel (1998) is less clear on the subject. Though the floor may amend bills to F, it might also reflect the inflection points of the House equivalents of the filibuster and veto pivots in the Senate. If the latter is true, it changes our theory where M is between the floor median and one of the pivots. We do not consider this special case. 12 Cox and McCubbins (2002) argue that the majority party asserts most of its influence in determining which bills (or more precisely which status quo policies) make it to the floor. One implication of their theory is that there is a block out zone between M and F where bills addressing certain status quo policies should not make it to the floor. Our assumption is not in conflict with this prediction, however, as their block out zone refers to the location of the status quo policy, not the location of the bill on final passage. 13 Our predictions are theoretically robust if we relax this assumption to allow bills to end up on the opposite side of F from the M. However, this is empirically unlikely.

10 any strength. The first exception, monotonicity, requires that as the House conference committee median (which is the relevant actor, as conference committees operate under an open rule) moves left, the conference report moves left as well, and vice versa. We impose a second exception, that the weights cannot result in the report being determined entirely by the Senate. 14 Our third assumption is that the Speaker has full discretion in appointing members to conference, with regard to both size and composition. This follows directly from Rule I of the House. The Speaker can thus solely determine the House delegation s median, and by extension, where the conference report lies. This raises the question of what the Speaker s ideal point is; we assume that the Speaker s ideal point is M. This is an assumption that the Speaker acts like the majority party median because he is an agent of the party, not an empirical assumption about the actual location of the Speaker s ideal point. Rather, we take into account the Speaker s incentives to minimize the distance between the majority party median and the conference report and thus maximize the payoff of the bill to a majority of the majority party. The Speaker can allow the conference median to be set by traditional modes of appointment, or he can move it toward M. If he sticks with tradition, the conference committee s majority-party median is that of the committee. 15 If he chooses to move it he can use replacement or packing. Our final assumption, based on the arguments of the previous section, is that packing a conference committee is less costly than replacing members of the proposed slate from the committee of origin. 14 Conferences are supposed to be restricted by a set of conditions collectively known as The Interval Rule. In brief, these state that if the Senate proposes X, and the House proposes Y, the conference report must be within the (X, Y) interval. However, the interval rule is unenforceable within the processes of the conference committee itself, as a result we do not consider it in our model. It s inclusion, however, should not substantively alter the model. 15 We make no claim about the Speaker s ability or willingness to manipulate the appointment of minority party members, although there is no procedural impediment to his doing so.

11 We can now generate hypotheses regarding conference committee appointments. Hypothesis 1 is that the Speaker is more likely to augment conference delegations when a bill comes from an unrepresentative committee than if the bill comes from a representative committee. Hypothesis 2 reflects our final assumption that packing is less costly than replacement: Conference delegations augmented by the Speaker contain more conferees than those which are not. Hypothesis 3 reflects that the Speaker s motive in augmenting a conference delegations is to move the conference median away from the unrepresentative committee s median ideal point and toward that of the majority party. When the Speaker augments a conference, outsider conferees are closer to the majority party median than traditional conferees. The fourth and final hypothesis alters the conditions under which Hypothesis 1 holds (and thus affects Hypotheses 2 and 3 as well): The Speaker augments conference delegations for bills originating from moderate unrepresentative committees, but not those originating from extreme unrepresentative committees. Packing is only necessary if a committee chairman proposes a slate of conferees that would negotiate a conference report that is distant from M. Unfortunately these proposed slates are not public knowledge and we cannot test this empirically. However, we expect the proposed slates of extreme unrepresentative committees such as C 2 to be close to M, for two reasons. First, the conference report cannot pass a floor vote unless it is closer to F than the status quo. Thus, C 2 cannot propose an extreme slate without risking the death of its bill at the conference report approval stage. Second, although this theory makes no claims about the Speaker s strategic use of minority party conferee appointments, members of the minority party are placed on every conference committee. These members ideal

12 points are almost uniformly on the opposite side of F from M, and thus pull the conference median from C 2 toward the F-M interval. Section III We test these hypotheses with data from House conference delegations appointed during the 97 th -106 th Congresses. Summary statistics for the dataset appear in Table 1. Our test of Hypothesis 1 employs LOGIT estimation of the probability of the Speaker appointing outsider conferees with the conference delegation. The key independent variables are dummy variables indicating whether the majority delegation on the bill s committee of origin is unrepresentative of the majority party. We determine which standing committees are unrepresentative by comparing the median DW-NOMINATE scores of committees majority contingents to the median DW-NOMINATE score of the majority party as a whole using Wilcoxon difference-of-medians tests; results are presented in Figure Each cell represents an individual test between a standing committee and the majority party median. A blank cell indicates that we could not reject the hypothesis that the committee delegation is representative of the majority party at the 95% confidence level. Marked cells indicate that the committee is unrepresentative. A plus sign indicates that the committee is more conservative than the party median, and a minus sign indicates that the committee is more liberal than the party median; all indicated differences are significant at the.05 level. Which sign (plus or minus) denotes a moderate or extreme committee changes depending on whether the majority in the House is Democratic or Republican at the time. For the 16 Our results match almost exactly with a similar analysis performed in Cox and McCubbins (1993; pp ) for the Congresses where our analysis overlaps with theirs. The few differences can likely be attributed to our use of DW-NOMINATE as compared to their use of D-NOMINATE.

13 congresses with a Democratic majority (97 th -103 rd ), any committee which is more conservative than the party (+) is moderate, and any committee which is more liberal (-) is more extreme; for the congresses with a Republican majority (104 th -106 th ), the opposite holds. Cells indicating moderate standing committees are shaded in Figure 2 for easier identification; marked cells which are not shaded identify extreme standing committees. Any conference bill with at least one unrepresentative committee of origin is coded as being unrepresentative in the direction of that committee of origin, resulting in the dummy variables moderate-unrepresentative and extreme-unrepresentative. 17 The reference group includes conference bills whose originating committees are representative of the majority party. Hypotheses 1 and 4 predict that the coefficient on moderate-unrepresentative will be positive and significant, and that of extreme-unrepresentative will be insignificant. Other independent variables control for other aspects of bill passage which might influence whether the Speaker inserts outsider conferees. Committees of referral indicates the number of committees to which each bill is referred, and varies from one to ten. Party Vote is a dummy variable coded 1 if the final passage vote fell along party lines, and 0 otherwise. Divided Congress is a dummy variable coded 1 if the conference took place during a congress in which the chambers were controlled by different majority parties, and 0 otherwise. Democratic Congress is a dummy variable coded 1 if the conference took place during a congress in which the Democratic party held a majority in the House, and 0 otherwise. 17 For singly-referred bills, moderate-unrepresentative and extreme-unrepresentative are mutually exclusive categories. However, for multiply-referred bills, it is possible for a single bill to be coded as both extreme and moderate. Indeed, some bills are coded as both. For both regressions in Table 2 we included these bills in the analysis. As a robustness check, we ran the same regressions while excluding bills coded as both moderate and extreme. Results conformed even more strongly to hypotheses 1, 2, and 4.

14 Results of the LOGIT estimation are in Table 2. The results conform closely to hypotheses 1 and The probability of the Speaker placing an outsider conferee on a conference committee is significantly higher when the committee of origin is moderateunrepresentative than when the committee of origin is representative of the majority party, as predicted in our first hypothesis. On the other hand, there is no significant difference between extreme committees of origin and representative committees of origin, as predicted by hypothesis 4. We used CLARIFY (Tomz et al 2003) to estimate the probabilities of outsider conferees being appointed to a conference committee given different levels of multiple referral. Results are in Figure 3. For all levels of referral, the probability with which the Speaker places outsider conferees on bills coming out of moderate-unrepresentative committees is from 0.10 to 0.15 higher than other bills. We test Hypothesis 2 using OLS regression in which the dependent variable is the number of conferees appointed. The key independent variables are a series of five dummy variables indicating whether the conference delegation contains non-traditional conferees and what type of committee the bill came out of. Moderate committee with outsider conferees is coded 1 if the bill came out of a moderate-unrepresentative committee and includes outsider conferees, and 0 otherwise. Other variables are coded similarly, as their names imply. The reference group for these variables includes conference committees involving bills which come from representative committees and which have no outsider conferees. All other independent variables are the same as in the LOGIT model estimated in Table 1. Results are presented in Table In estimating both Model 1 and Model 2, we exclude the five cases in which the Speaker appointed untraditional conferees, but only inserted members of the minority party. These are HR 3471 (100 th ), HConRes 67 (104 th ), HR 3103 (104 th ), and S4 (104 th ), in which the Speaker inserted a single member of the minority party; and HR 483(104 th ), in which the Speaker inserted four members of the minority party.

15 Hypothesis 2 predicts that delegations augmented by the Speaker are larger than those which are not. The LOGIT estimations in Table 2 indicate that bills with moderate committees of origin get augmented. These combine to predict that, in table 3, the coefficient for moderate committee with outsider conferees be positive and significant, and be significantly larger than the coefficient for moderate committee with no outsider conferees. The results in Table 3 confirm this hypothesis. Conference delegations for bills which come out of a moderate-unrepresentative committees and have outsider conferees are almost 34 conferees larger, on average, than the reference group. Delegations for the bills which come out of moderate-unrepresentative committees and have no outsider conferees are not significantly different than the reference group. The difference between the two coefficients is significant at p< using an F test, indicating that conference delegations are larger when nontraditional conferees are involved. Hypothesis 4 predicts that the coefficients for both extreme committee variables should not be significant, and not significantly different from each other. Neither is significant, and the difference between them is not significant using an F test. One unpredicted result is that conference committees on bills which come from a representative committee and have outsider conferees are about 22 conferees larger than conference bills with no outsider conferees. So far, the evidence suggests that the Speaker is more likely to augment a conference delegation when a traditional delegation is likely to produce an unfavorable conference report i.e., when a committee of origin is moderate-unrepresentative. Further, these delegations are larger than other delegations, indicating that the Speaker is including outsider members in addition to, rather than instead of, traditional conferees. However, this is not enough to prove that packing is going on; one question still remains. Is the Speaker really using outsider conferees to pack conference delegations, or is he using them for another purpose? If the

16 Speaker inserts outsider members for the purposes of packing, then nontraditional conferees should look more like the majority party median than the traditional conferees. Packing occurs when the Speaker inserts members onto moderate-unrepresentative committees. Hypothesis 3 predicts that when bills coming out of moderate-unrepresentative committees have outsider conferees, outsider conferees should be consistently more extreme than traditional conferees. On the other hand, when the Speaker inserts traditional conferees onto other committees (i.e representative committees or extreme committees), he does so for some other purposes (we briefly explored these reasons at the end of Section 1). As a result, there should be no identifiable pattern of ideological differences between traditional and nontraditional conferees. Hypothesis 4 predicts that extreme unrepresentative committees should look like representative committees in this capacity. We test these predictions by comparing median DW-NOMINATE scores of traditional conferees to those of outsider conferees who are appointed for the same bill. We could not use the Wilcoxan difference-of-medians test to obtain significance levels because the number of people in each group is too small; we would be unable to reject the hypothesis of similarity in any but the largest conferences. Instead, we tallied the number of times outsider conferees are more extreme than traditional conferees in each type of committee, and present results in Table 4. Where there is a moderate committee of origin, outsider conferees are more extreme more than 75% of the time. The 10 cases in which traditional members are more extreme do not constitute compelling evidence against the packing theory. Even when a bill comes out of a moderate unrepresentative committee, the Speaker might appoint outsider conferees for reasons other than ideological packing. However, the high rate of extreme outside appointees stands in stark contrast to bills with other types of originating committees. When there is an extreme or

17 representative committee of origin, traditional and outsider conferees are each more extreme with approximately equal probability. In these cases, there is no identifiable pattern of ideological appointment. Section IV The empirical evidence suggests that, when it comes to conference committees, the Speaker does indeed exercise his Rule I-derived appointment authority to ensure that conference reports conform to majority party needs. Though the pattern of originating-committee member appointment is generally consistent with the conventional wisdom, that school of thought cannot explain the pattern by which Speakers supplement conference committees with delegates who are not members of the bill s originating committee. In this paper, we present a theory that predicts that pattern and empirical tests bear out those predictions. When so-called traditional appointment procedures are likely to produce a conference report which is unfavorable to the majority party, the Speaker alters the conference delegation. This occurs when a conference bill comes out of a standing committee which is more moderate than the majority party as a whole, since these bills are most likely to threaten the majority agenda. The alteration typically consists of appointing originating-committee members as well as outsider conferees; or as we have called it, packing the conference committee. Though the Speaker may not omit traditional conferees, he does make sure that the outsider conferees are more likely to produce a conference report which is more to the Speaker s liking than the traditional conferees. He does this by appointing outsider conferees who are more extreme and thereby more representative of the majority party than their traditional counterparts from the moderate committee.

18 Despite his clear authority to appoint conferees of his choosing, the Speaker violates tradition only when necessary and in the least intrusive way possible. Packing occurs only when the Speaker judges that appointing conferees according to the traditional manner is likely to result in a conference report that does not represent the best interests of the majority party. Some committees which are not representative of the majority party (those which are representative in the extreme direction) are spared having their committees packed. Additionally, this method of controlling conference committees is less extreme than another strategy available to him, that of simply replacing the traditional conferees with others who are more to his liking. This evidence suggests that, even though the Speaker is in control of conference delegations, he avoids exercising this control in a manner which encumbers members of his party. That the Speaker intrudes on traditional appointment procedures minimally (in terms of both how often it occurs and how severe the intrusion), indicates that the Speaker acts with multiple goals in mind when making conference appointments. On the one hand, he attempts to maximize the electoral value of conference appointments to those who receive them. On the other, he must judge what kind of conference report a particular delegation is likely to report back to the chamber, and be sure that his party s label is protected accordingly.

19 References Bach, Stanley Resolving Legislative Differences in Congress: Conference Committees and Amendments Between the Houses. Congressional Research Service. Report no Brown, William Holmes House Practice: A guide to the Rules, Precedents and Procedures of the House. Washington, D.C.: U.S. Government Printing Office. Clapp, Charles L The Congressman: His Work as he Sees It. Washington D.C. Cox, Gary W. and Mathew D. McCubbins Legislative Leviathan: Party Government in the House. Los Angeles: University of California Press Agenda Power in the U.S. House of Representatives, 1877 to 1986 In David Brady and Mathew D. McCubbins, eds, Party, Process, and Political Change in Congress: New Perspectives on the History of Congress. Stanford: Stanford University Press. Dion, Douglas and John D. Huber Sense and Sensibility: The Role of Rules. American Journal of Political Science 41: Downs, Anthony An Econcomic Theory of Democracy. New York: Harper Press Ferejohn, John A Pork Barrel Politics. Palo Alto: University of Stanford Press Krehbiel, Kieth Information and Legislative Organization Ann Arbor: The University of Michigan Press Pivotal Politics: A Theory of U.S. Lawmaking Chicago. University of Chicago Press Longley, Lawrence D. and Walter J. Oleszek Conference Committees in Congress. New Haven: Yale University Press Mayhew, David R Congress: The Electoral Connection. New Haven: Yale University Press Oleszek, Walter J Congressional Procedures and the Policy Process, Fourth Edition. Washington, D.C.: Congressional Quarterly Press Poole, Keith T. and Howard Rosenthal Congress: A Political-Econcomic History of Roll Call Voting. New York: Oxford University Press. Pressman, Jeffrey L House vs. Senate: Conflict in the Appropriations Process. New Haven: Yale University Press

20 Rohde, David A. and Kenneth A. Shepsle Democratic Committee Assignments in the House of Representatives: Strategic Aspects of a Social Choice Process. American Political Science Review. 67: Rogers, Lindsay Conference Committee Legslation. The North American Review. 215: Shepsle, Kenneth A. and Barry R. Weingast The Institutional Foundations of Committee Power. American Political Science Review. 81: Sinclair, Barbara Unorthodox Lawmaking: New Legislative Processes in the U.S. Congress. Washington D.C: Congressional Quarterly Press. Smith, Steven S An Essay On Sequence, Position, Goals, and Committee Power. Legislative Studies Quartely. 13: Smith, Steven S. and Christopher J. Deering Committees in Congress. Washington D.C: Congressional Quarterly Press. Tomz, Michael, Jason Whittenberg, and Gary King CLARIFY: Software for Interpreting and Presenting Statistical Results. Accessed via Vogler, David J The Third House: Conference Committees in the United States Congress. Evanston: Northwestern University Press.

21 Figure 1 The Effect of Speaker Discretion on Majority Party Conference Committee Medians F C 1 M C 2

22 Figure 2 Wilcoxon Difference-of-Medians Tests on DW-NOMINATE Scores of Majority Party Contingents, Floor vs. Standing Committee Contingents, 96 th -104 th Congresses Agriculture Appropriations - Armed Services Banking Budget Commerce Education & Labor Foreign Affairs Government Operations - n/a 20 n/a n/a House Administration Interior & Insular Affairs Judiciary Merchant Marine & Fisheries + n/a 22 n/a n/a Post Office & Civil Service Public Works & Transportation Rules Science Small Business Veterans Affairs Ways & Means Notes: + = Contingent median is significantly greater (more conservative) than the party median at the.05 level. - = Contingent median is significantly less (more liberal) than the party median at the.05 level. Shaded cells identify majority-party committee contingents which are significantly more moderate than the majority party floor median. 19 The change to Republican control in the 104 th also led to a change in the names and structures of several of the standing committees in the House. In the cases of Armed Service, House Administration, and Foreign Affairs, the column for the 104 th Congress reflects information from the roughly equivalent committees (i.e. National Security, House Oversight, and International Relations). 20 Data unavailable. 21 In the 103 rd Congress, Interior and Insular Affairs was replaced by Natural Resources. The Wilcoxon test for the 103 rd was actually done using the majority contingent from Natural Resources. 22 Abolished by the Republicans when they assumed the majority in the 104 th Congress.

23 Figure 3 23, 24 Probability of Insertion by Number of Committees of Referral probability of insertion no referral committee was moderate at least one referral committee was number of referral committees 23 We calculated estimations using Monte Carlo simulation via CLARIFY (Tomz et al 2003). Simulations were performed twice, and probability estimates derived from both sets of simulations. No estimate varied by more than.006 between the two sets of simulations. 24 Ninety-two percent of observations (624) fall within the range reflected in the figure. Twelve conference committee bills (1.8%) had five or more committees of referral; forty-three (6.3%) were not referred.

24 Table 1 Summary Statistics For Conferences in Dataset Congress Years Total conferences Conferences including nontraditional conferees Percent of conferences including nontraditional conferees % % % % % % % % % % total %

25 Table 2 The Effects of Moderate Committees of Origin on the Makeup of Conference Committees Dependent Variable = 1 if the Speaker appoints an outsider conferee, 0 otherwise Moderate committee.689** (.261) Extreme committee.013 (.346) Party vote.221 (.226) Number of committees of referral.341** (.102) Democratic Congress.776** (.271) Divided Congress -.850*** (.249) Constant -2.45*** (.288) N 674 log likelihood pseudo-r Notes: *p<.05; **p<.01; ***p<.001 Standard errors in parentheses

26 Table 3 The Effects of Moderate Committees of Origin on the Size of Conference Committees Dependent Variable = Number of people appointed to a conference delegation Moderate committee with 33.9*** outsider conferees (3.36) Moderate committee without.807 outsider conferees (2.36) Extreme committee with.143 Outsider conferees (4.52) Extreme Committee without outsider conferees (2.88) Representative Committee with 22.0*** outsider conferees (2.22) Party vote 7.53*** (1.61) Number of committees of referral 1.95* (.795) Democratic Congress 2.74 (1.84) Divided Congress (1.65) Constant 6.98*** (1.89) N 674 Adjusted R Notes: *p<.05; **p<.01; ***p<.001 Standard errors in parentheses

27 Table 4 Ideological Differences Between Traditional and Outsider Conferees Number of times non-traditional conferees are more extreme than traditional conferees Number of times traditional conferees are more extreme than non-traditional conferees Proportion of time nontraditional conferees are more extreme Moderate Committees % Extreme Committees % Representative Committees %