Revolving Door Lobbyists and the Affordable Care Act Richard L. Hall and Geoffrey M. Lorenz University of Michigan Abstract Students of democratic institutions have long criticized the “revolving door” that moves government officials into lucrative positions as lobbyists, presumably giving the interest groups that hire them an advantage in the competition for access. The migration is especially common among congressional staff. Critics allege that revolvers gain access by exploiting personal relationships with former employers and co-workers. We argue that their advantage is more general than that. Positioned at the intersection of two policy-relevant networks, revolvers enjoy a comparative advantage in generating political information, but the value of that information to the member, we hypothesize, will be conditional on partisan alignment between member and group. Employer-based relationships and policy knowledge vary too little to explain much of the revolver’s advantage. Analyzing data on access from interviews with lobbyists on the Affordable Care Act, we provide the first direct evidence of a revolver-access relationship. Draft: April 20, 2015 (Very Rough Draft – Please do not circulate outside of the Workshop.) 1 Students of democratic institutions consider the “revolving door” one of the most disturbing patterns in contemporary American politics, and this pattern is nowhere more evident than the United States Congress. Former members and their staff, upon leaving Capitol Hill, accept lucrative salaries to go back and lobby on it, presumably giving the interest groups that hire them an advantage in the competition for legislative influence. There is growing evidence that revolving door lobbyists improve the fortunes of their employers, but the means by which they do so remain opaque. Most assume that the key variable is access and that revolving door lobbyists are better at gaining it. Access enables interest groups to make their case for or against a bill or make requests for specific provisions, increasing the probability that their preferences will be reflected in policy. To date, there have been no direct tests of the revolver-access mechanism, however. Nor do we fully understand the reasons that revolvers would enjoy a comparative advantage in this regard, though the critics allege that revolving door lobbyists exploit their personal relationships with former employers and co-workers to open doors on Capitol Hill. We hypothesize that the revolver’s advantage is more general than that. Legislators grant access to acquire information, and revolvers are better able to acquire one type of it. Positioned at the intersection of two policy-relevant networks, revolvers can more easily generate political intelligence about the preferences and strategies of key actors. The value of that information to the member, we hypothesize, is conditional on the partisan alignment of the member and group. We also argue that personal ties to former employers and co-workers will account for little of the revolving door advantage in access. Others have shown that those relationships sometimes have a high payoff, but over ninety percent of all revolvers are former staffers, and rarely will a staffer have employer-based relationships with more than a very small fraction of the current members of Congress. Neither is it common for revolving door lobbyists to have greater policy expertise than their non-revolving peers. Only former committee counsel might add value in policy expertise, and that advantage will be restricted to issues arising from their former committee’s jurisdiction. When it is present, however, the policy expertise of the lobbyist should be valuable to the member. We investigate these hypotheses using highly granular data, collected from face-to-face surveys of interest group representatives who lobbied House committees on the Patient Protection and Affordable Care Act (ACA). Our results provide the first direct evidence that revolving door lobbyists enjoy an advantage in gaining access to members of Congress, and we find that the advantage is conditional in the ways that we hypothesize. We also find that campaign contributions are related to access, but inferences about the causal nature of that relationship remain elusive. Congressional Staff and the Revolving Door The “revolving door” is a metaphor for a specific labor market pattern. Individuals who work in a policymaking capacity in government leave for positions as lobbyists outside it, working on some of the same issues and with some of the same people that they did as public servants. Institutional critics have long alleged that this pattern is commonplace, but until the 2 study by Heinz et al., The Hollow Core (1995), the only data on the practice were unsystematically generated or otherwise anecdotal and thin (1995, 106). Heinz et al. reconstructed the career paths of 1700 Washington lobbyists, finding that forty-five percent had federal government experience. Ethics reforms adopted since the time of that study have not slowed the migration. LaPira and Thomas (2014) find that the percentage of lobbyists with government experience is higher now than it was at the time of the Heinz et al. survey. Career moves from Capitol Hill positions to advocacy jobs are especially common, but by far most of those moves involve congressional staff In 2011, former congressional staff out-numbered former members of Congress as registered lobbyists by more than 10 to 1. If the migration to K-Street occurs frequently, only recently have we begun to understand what difference it makes. Blanes i Vidal et al. (2012) demonstrate that revolvers suffer an abrupt drop in revenue when their former bosses leave Capitol Hill, with those losses greater when the departing member held major committee assignments and enjoyed high seniority. LaPira and Thomas (2015) find that interest groups assign revolvers to more important legislation. We thus can conclude that interest groups believe that revolvers do better by their clients. Luechinger and Moser (2012) find that investors share this belief. Richter et al. (2009) show that these beliefs are well-founded; organizations employing revolving door lobbyists on tax policy generate higher returns on investment (Richter et al. 2009). Lazarus and McKay (2012) find that universities employing revolving door lobbyists receive more government funding than those that do not. In the most comprehensive study of lobbying to date, Baumgartner et al. (2009) find that interest group coalitions that employ revolvers exhibit a higher probability of policy success (2009, 208). The micro-level mechanisms by which revolvers do better for their clients are not well understood, but the presumption is that access is critical. Access is a precondition for influence, and revolvers are thought better able to gain it. The most common assumption is that they exploit their personal ties to former employers and office co-workers to gain access that other lobbyists cannot. This claim remains untested, however. Lobbying Disclosure Act (LDA) reports tell us who is lobbying and on which issues. They also provide data on lobbying expenditures and revenue. But they provide no data on the access to individual offices for which the interest group money is spent. Of the studies that measure lobbyist-legislator interactions directly (Wright 1990; Austen-Smith and Wright 1992, 1994; Caldeira and Wright 1998; Hojnacki and Kimball 1999, 2001; Kalla and Broockman forthcoming; Langbein 1986), none examine the relationship between access and past government employment. If revolvers are better able to give their clients entry into the process, scholars have been unable to test hypotheses about whether, how much, and under what conditions they do so. The Comparative Advantage of Revolving Door Lobbyists Access to an office is a product of two decisions, not one: the lobbyist’s decision to seek access and the legislator’s decision to grant it. The former, the decision of whom to lobby, has received greater emphasis in the literature (e.g.. Austen-Smith and Wright 1994; Hojnacki and Kimball 1998; 1999), but to understand revolving door access, greater emphasis belongs on the 3 latter. We assume, over-simply, that lobbyists invariably want access, but that legislators selectively grant it.1 The reason for the legislator's selectivity is that her supply of access is scarce. The typical legislator spends remarkably little time on legislative affairs (e.g., Weissert 1989), her issue-attention is limited (e.g., Sulkin 2005), her staff is spread thin (e.g., Zeller 2015), and she often has long lists of supplicants, so it seems unlikely that she would grant access uneconomically. The problem for the legislator is allocating how much access to whom. What is it about revolvers that would lead a legislator to favor them as a class of lobbyists in the competition for her time and attention? In theorizing about these decisions, we begin from the premise that lobbying is an exercise in information transmission. Legislators look to lobbyists for information, both for determining their preferences over alternative policies (e.g., Ainsworth; Austen-Smith 1993; Wright 1996) and supporting work on their legislative priorities (e.g., Baumgartner and Leech; Baugartner et al. 2009; Hall and Deardorff 2006). The question thus becomes: What information can revolvers better collect and convey that makes them more valuable to legislators than lobbyists without government experience? The literature identifies at least three types of information that lobbyists provide (see, e.g., Wright 1996). The theoretically most prominent of these stems from their interest in reelection. Lobbyists regularly collect district-specific, issue-relevant information that they might use to influence legislators.2 As Hansen puts it, lobbyists are effective insofar as they “determine the kinds of information about constituents that are available and the kinds of information that are not” (1991, 3; see also, e.g., Kingdon 1981; Hojnacki and Kimball 1998). The question here is whether revolvers as a subclass of lobbyists will be better informed in this respect. As former insiders, they will have learned something about other members’ constituencies, but non-revolving lobbyists will surely do the same, often with support staff who conduct district-level research and maintain files on individual legislators. The most plausible exception is the revolver with respect to the one or few offices where he formerly worked. Such ties have received the most attention and harshest condemnation from congressional observers, and there is strong evidence that they have important consequences for lobbyists, affecting the issues on which they work (Bertrand, Bombardini, and Trebbi 2011) and the revenue they generate (Blanes I Vidal, Draca, and Fons-Rosen 2012). But if these ties do matter, it is not clear why they matter. Indeed, the personal nature of the relationship may have little to do with it. Their ability to provide information that is at once constituency-relevant and policy-specific provides one reason they might have a comparative advantage, with respect to other lobbyists generally and other revolvers specifically The theoretical problem with this reasoning is that little of that information will be the private information of the lobbyist with respect to the office, that is, better or different than that of the member and her current staff. Members, after all, constantly monitor the district in myriad ways, including regular trips home, 1 Unless otherwise noted, we use the term legislator to refer to what Shepsle and Salisbury call the “legislator’s enterprise (fill in). 2 Organizations use outside as well as inside lobbying to transmit information about constituents’ preferences (e.g., Kollman 1998). We assume that the use of outside lobbying is orthogonal to the use of revolving door lobbyists, recognizing that the realism of that assumption needs to be investigated. 4 meetings with constituency groups, allocating staff to district offices, conducting polls, and tracking issue-specific constituency mail. Thus, on somewhat different grounds than the critics, we hypothesize the relationship that many have criticized: H.3. Revolvers will have greater access to the office(s) where they formerly worked. A second type of information in which lobbyists presumably traffic is policy expertise. Legislators are uncertain about the connections between policies and outcomes. They deal with a wide range of issues almost daily and must rely on a limited staff, very few of whom are themselves policy specialists. In contrast, lobbyists tend to focus on one policy at a time, learning the policy-relevant arguments and evidence as thoroughly as possible. They may not be experts themselves, but as Heclo (1978) observes, they “are experts at using experts.” The relevant point here is that there is no general reason to expect that revolvers will have more policy expertise than non-revolving lobbyists. If anything, the generalist role most of them played as staffers would suggest otherwise. There is a narrow exception to this. Congressional committee staff often go deep into the policy evidence and analysis related to the issues they cover, so they should be better able to convey specialized policy knowledge. The exception is narrow because their comparative advantage will be conditional on the match between their former committee’s jurisdiction and the issue at hand. With respect to the dozens of issues that may be relevant to their organization’s mission, as individual lobbyists they will have no special advantage. H.2. Former committee staff will have greater access on issues arising out of their former committee’s jurisdiction. This advantage need not be institution-specific, however. Expertise on Medicare policy, for instance, might be acquired as a committee staffer for House Ways and Means or Senate Finance. To this point, then, it is not clear that revolving door lobbyists should get better access except under fairly narrow conditions. If they have a more general comparative advantage it should derive from a different type of information that revolvers as a class are better able to collect and convey, what we will refer to as “legislative intelligence,” meant in two senses. Having played staff roles themselves, revolving door lobbyists will better understand the internal workings of the legislative process, knowledge that the non-revolving lobbyist only acquires at greater cost and through less direct means (see esp. Wright 1996, 93). More importantly, revolvers should be better able to gather contemporaneous intelligence. In making informed choices, legislators need to know not only what policies are preferable but which ones are viable, which depends on what the key actors are “thinking or planning” (Wright 1996, 92, 93). The key actors are mainly situated in two different networks. Like other lobbyists, the revolver is connected to interest group networks that generate information relevant to policy viability (Ainsworth 1997; Carpenter et. al. 1998; Heaney 2014; Heaney and Lorenz 2013; DeGregorio 1997). But unlike other lobbyists, the revolver begins with a set of well-developed connections to a congressional network and emeritus status within it. He thus has sources of information about both interest groups’ and legislators’ preferences and objectives, a combination that no other legislative informant enjoys. In Blanes I Vidal, Draca, and Fons-Rosen’s terminology (2012), he draws on a form of “relational capital” more 5 extensive and less personal than his connections to former employers. H.3. Former staffers will have greater access to the offices of current members. If H.3 tends to be true, however, it is not likely that for the individual lobbyist it will forever be true. His distinctive connectedness to internal political networks should decay over time. Former staff members may have better lines of communication to current staff members, but staff turnover in Congress is high. After a decade or so, the revolver will know almost no one he worked with at the staff level. Even if there is a self-sustaining element to the connections that a lobbyist begins with, over time the revolver and non-revolver should become difficult to distinguish. as a lobbyist should improve one’s access on Capitol Hill, H.4: The advantage of congressional background will be decreasing over time. The Conditional Value of Information Theoretical work on lobbying as information transmission takes it as axiomatic that the value of information to the legislator will depend on her perception of its credibility (see e.g., Austen-Smith 1995; Wright 1998). Lobbyists can reveal their information selectively, perhaps deceptively. Legislators know this and thus look to sources they believe they can trust. Those beliefs can depend on a number of factors, including lobbying costs and interest group competition (Grossman and Helpman 1994, 2001), but a recurring theme in the literature is that legislators look first to those with similar policy beliefs. In his study of legislators’ voting decisions, for instance, Kingdon finds this pattern in both member-to-member (1981, 72-82) and interest group lobbying: “[Congressmen] know which organizations are generally in agreement with them and which ones are generally opposed and they use this bit of past history to help determine which side to be on in the current instance” (1981, 187). Similarly, Hall and Deardorff (2006) argue that legislators’ priors about interest group agreement will affect their priority-setting decisions. Austen-Smith (1995) states the general point: “[T]he amount of information an informed agent (lobbyist) can credibly transmit to an uninformed decision maker (legislator) is increasing in the extent to which these individuals' preferences over final consequences are coincident. That is, the more like the legislator the lobbyist is, the more valuable that lobbyist will be to the legislator on informational grounds” (566). In the contemporary Congress, similarity and difference in preferences are divided sharply by party. Interest groups often try to broker partisan conflicts (Heaney 2006), but Koger, Masket, and Noll show that interest groups separate into two distinct party networks, concluding that “these actors are best understood as networks of co-operating allies” (2009, 634). Partisan alignment of groups and members should thus appear in the patterns of interaction among them, the latter looking to the former for information they can trust. Such patterns should hold for all lobbyists, but it should be particularly true for legislative revolvers. As former legislative staffers, their intelligence-gathering capacity will come from their better connections to partisan networks, whose participants they interacted most while on Capitol Hill and on whom they can still depend for information. H.4. The revolver’s advantage in gaining access will be greater when there is partisan alignment between member and group. 6 Finally, we expand our focus from the individual lobbyist attributes to the staff capacity of the interest group. Organizations vary widely in the number of lobbyists they employ and hence their ability to strategically assign lobbyists to individual offices. They vary more specifically in the extent to which they populate their staff with revolving door lobbyists. However, the advantage of large staffs will depend on the number of issues across which their lobbyists are spread. We thus expect that the access of an organization will be increasing in the number of lobbyists per issue, but hypothesize more specifically that: H.5. Access will be increasing in the proportion of those lobbyists that have congressional backgrounds. Lobbying House Committees on the Affordable Care Act The principal data for our exploration come from face-to-face interviews with individuals who lobbied in the House of Representatives on health care reform in the 111th Congress, one of the most important domestic initiatives in a generation. That the political environment of the Affordable Care Act was densely populated by interest groups thus surprised no one. A century of health policy history foreshadowed it. Social scientists seeking to explain the repeated failures of major reform proposals often placed responsibility at the feet of private sector interests. By the time of the 2008 elections, the stakes were especially high. The legislation under consideration would regulate a sector of the economy that accounted for one sixth of GDP, about $8,000 per person, and employed one in every ten Americans in the private sector. The Medicare provisions alone would immediately affect the program’s 46 million beneficiaries and thousands of health services providers. The number of low-income individuals eligible for Medicaid would expand dramatically. The insurance regulations, the revenue provisions, and the individual and business mandates would affect almost everyone else directly or indirectly. Every group in America, it seemed, had something at stake in this debate. Over a thousand organizations filed Lobbying Disclosure Reports in 2009 that specifically mentioned the Affordable Care Act, and many other advocates – especially foundations and other 501(c)(3) organizations – seldom file disclosure reports. 7 Table 1. Lobbying on the Affordable Care Act: Sample of Organizations Type of Organization Snowball (%) Sample (%) Cases (%) Health professionals 23 15% 9 18% 5 17% Pharmaceuticals/health products 21 14% 9 18% 4 14% Hospitals/nursing homes 10 7% 4 8% 4 14% Health insurance 18 12% 6 12% 4 14% Public interest groups/think tanks - health 36 24% 8 16% 6 21% Public interest groups/think tanks - general 24 16% 9 18% 4 14% General business and labor organizations 21 14% 5 10% 2 7% Total 153 50 29 For substantive and practical reasons, we focused the data collection on the House and, within the House, on the three committees of jurisdiction. In the Senate, the main legislative negotiations occurred not in committee but in the ad hoc “Gang of Six,” an informal working group of three Democrats and three Republicans organized by Finance Committee Chair Max Baucus (D-MT).3 That so much of the legislative attention was focused on six senators made the process unusual and suggested that the variance on lobbying would be highly truncated. The legislative process in the House was relatively conventional in that the deliberations were committee-centered and, because all bills in committee are marked up under the equivalent of an open rule, the participation of individual members was procedurally inclusive. The same House bill was marked up in the three committees that shared jurisdiction, 145 members in all,4 with scores of amendments offered by members of both parties. The bill considered on the House floor, in turn, was largely the product of the committee process, for like most major bills in the House, it came to the floor under a highly restrictive rule.5 The sample of organizations is not random with respect to the population of groups that filed LDA reports on this issue. That list is not only unwieldy; most groups who listed the ACA in their issue reports were scarcely visible on Capitol Hill, their lobbying activity across members on this bill difficult to measure.6 More importantly, the LDA does not require think tanks and other 501(c)(3) organizations to file reports, even though advocates from these groups 3 The Senate bill was eventually adopted by the House, though a number of House-initiated changes were adopted through a second, budget reconciliation bill negotiated prior to enactment. See Jacobs and Skocpol 2012. 4 The three committees were Ways and Means, Energy and Commerce, and Education and Labor. 5 Of course, the final bill, even after the reconciliation amendments, was much more the product of the Senate than the House, but that fact was not known or anticipated until the very end of the process. 6 We recognize that groups select into that category. Why some groups and not others, on some issues and not others, engage in significant lobbying activity is an important matter that political scientists have not yet been able to model. Our analysis exhibits the same limitation. 8 play a significant role in most major policy issues, including this one. We thus generated the set of groups using a snowball process, beginning with a list of organizations culled from newspaper, magazine, and online reports. We then conducted interviews with House committee staff from both parties during which we showed them the working list of organizations and asked them to suggest others that lobbied on healthcare reform. We did the same as we interviewed lobbyists. This cumulative process generated a list of 153 organizations, which, as Table 2 summarizes, included insurance companies, pharmaceuticals, hospital associations, healthcare professionals, general labor and business organizations, and think tanks and other nonprofits. From this list we sampled 50 and completed interviews with 29. As the table indicates, the final dataset includes a mix of groups, roughly representative of the ones that actively lobbied House members regarding the bill. The central purpose of the interviews was to gather data on access at the level of the lobbyist-legislator dyad. Early in each meeting we administered a survey instrument designed to capture the number of “phone or face-to-face conversations” between the organization’s lead lobbyist(s) on this issue and each legislative office. The instrument closely follows the one developed by Wright (1986) and used in most individual-level studies of lobbying to date (e.g., Austen-Smith and Wright 1994; Caldiera and Wright 1998; Hojnacki and Kimball 1998; Wright 1990), with the exception that we employed categories that required the respondent to estimate the actual number of contacts with each legislator or their staff.7 The specific ranges were: “none (0),” “once or twice (1-2),” “a few (3-5),” “several (5-10),” “many (10-20),”and “repeated contacts (>20).” When respondents checked off “repeated contacts,” we asked them to estimate the actual number of contacts with that office on that issue. We thus have a measure of the issue- specific conversations between lobbyist and legislator, which can be analyzed either as an ordinal or a count variable. As Figure 1 shows, the values are skewed toward zero, a pattern that is consistent with our assumption that legislators grant access selectively. 7 The question wording was: “Please estimate the number of phone or face-to-face conversations you had with the following members or their staff regarding the healthcare reform bill in 2009 through its adoption in 2010.” 9 Figure 1. Lobbyist Contacts with Individual Members 50 45 40 A s n 35 o it a 30 v r e s 25 b o f 20 o % 15 10 5 0 none 1 to 2 3 to 5 6 to 10 11 to 20 > 20 To test the first three hypotheses, we require accurate data on the attributes of lobbyists, which we gathered from career information reported on two online databases, the Center for Responsive Politics at www.opensecrets.org and CQ’s proprietary database, First Street. We then checked that information against the lobbyists’ online biographies. The variables include: (1) whether the lobbyist had previous congressional experience and (2) whether he had House experience specifically;8 (3) whether he had staffed a health policy committee; (4) whether he had worked for a specific member in the sample; and (5) the number of years he had worked as a registered lobbyist. We also gathered data on the number of lobbyists employed by each group, to capture the lobbying capacity of the organization, and the proportion of lobbyists that were revolvers. To test hypotheses H.4 and H.5, we collected data on the contributions of the organization’s employees to the respective parties’ campaign committees and candidates, as reported to Federal Election Commission. Our measure of the group-member partisan alignment is simply the percentage of employee contributions going to the same party as the member.9 8 None of the revolving door lobbyists in our sample were former members of Congress. 9 Bonica (2010) has used employee contribution data to estimate the ideal points of similar interest groups on the same scale as legislators, but doing that at the group level was not possible for all organizations in our sample. 10