Abstract
We relate legislators’ financial assets to their roll call voting on, and cosponsorship of, legislation to permanently repeal or significantly reduce the Estate Tax in the 109th Congress. Even after accounting for legislators’ party affiliations, their global opinions about taxation, and their constituents’ opinions about the Estate Tax, together with other confounding factors, we find that wealthier legislators were more likely to vote for and cosponsor bills to reduce and repeal the Estate Tax.
On June 12, 2011, The U.S. Supreme Court unanimously overturned a Nevada Supreme Court decision by ruling that a legislator’s vote is not free speech protected by the First Amendment and therefore can be regulated by conflict of interest laws. The case involved a Sparks, Nevada city council member, Michael Carrigan, who did not recuse himself on a casino proposal on which his campaign manager was a consultant. The issue was broad enough in its implications for legislatures at all levels that 14 states joined Nevada in its appeal to the Supreme Court.
In writing the majority opinion for the Court, Justice Scalia observed that Congress and the states have for two centuries imposed conflict of interest restrictions on elected officials: “The House rule—to which no one is recorded as having objected, on constitutional or other grounds . . . was adopted within a week of that chamber’s first achieving a quorum. The rule read as follows: ‘No member shall vote on any question, in the event of which he is immediately and particularly interested.’”
In striking contrast to this provision, New York Times journalists Eric Lipton and Eric Lichtblau have recently chronicled a number of House Representatives whose official activities appear to approach the conflict of interest threshold:
Representative Darrell Issa (R-CA) opposed the Treasury Department’s forced sale of Merrill Lynch in 2008 when Merrill had managed hundreds of millions of dollars in his investments and extended loans to him (Lichtblau, 2011);
Representative Mike Thompson (D-CA) has been the wine industry’s foremost champion in Washington, helping it secure tax breaks and beat back restrictions on direct wine sales; he also grows 20 acres of sauvignon blanc grapes on his farm (Lipton, 2011, July);
Representative Shelley Berkley (D-NV) has co-sponsored five House bills that would expand federal reimbursements for kidney care; her husband Dr. Larry Lehrner is a nephrologist with business interests in DaVita, a dialysis provider (Lipton, 2011, Sept.).
Are these isolated cases, or is apparently self-interested legislative behavior more widespread? This study employs the 2001 battle over the repeal of the Estate Tax as a window on whether legislators are motivated by self-interest when they act in their professional capacities. We construe “self-interest” narrowly—as actions that will directly result in legislators’ financial gain (the ability to maximize the portion of their estate that is passed on to their heirs). Estate Tax policymaking provides a valuable opportunity to gauge if self-interest influences the legislative behavior of Members of Congress (MCs). We relate the assets of MCs as reported in Financial Disclosure Reports to their cosponsorship of and roll call voting decisions on three Estate Tax bills in the 109th Congress.
To preview our results, we find that personal wealth influences the likelihood a MC will vote in support of and cosponsor legislation calling for the permanent repeal or substantial reduction in the Estate Tax. This obtains regardless of whether we measure Member assets continuously or use an indicator for whether a Member reported at least US$1 million in assets, making them subject to the tax prior to the consideration of the bills. We also conduct placebo, threshold, and sensitivity tests to probe the robustness of these findings.
Background
Scholarly research has identified three chief influences on legislators’ roll call and cosponsorship behavior: constituents’ preferences, legislators’ party affiliations, and their own preferences (e.g., Kessler & Krehbiel, 1996; Wawro, 2000). Constituents’ policy opinions shape legislator decision making both because constituents tend to choose like-minded representatives at election time (Bullock & Brady, 1983; Miller & Stokes, 1963) and because once they are in office, representatives usually (and perhaps chiefly) care about re-election (Arnold, 1990; Fiorina, 1974; Mayhew, 1974). Additionally, some argue that party leaders exert pressure on representatives through a highly organized and disciplined structure that serves to further party goals and Members’ re-election desires by minimizing defections (Aldrich, 1995; Ansolabehere, Snyder, & Stewart, 2001; Cox & McCubbins, 1993; Rohde, 1991).
Somewhat less attention has been paid to the importance of factors beyond constituents and political parties, especially legislators’ own characteristics. To be sure, this can be a difficult topic to investigate because legislators are relatively inaccessible to researchers. Still, there are studies showing that legislators’ ideologies, core values, and experiences shape their decision making (Burden, 2007; Gelpi & Feaver, 2002; Kingdon, 1981; Levitt, 1996; Reeher, 1996). One such study found that legislators’ religious denominations appear to shape their behavior on bills related to religious freedom, human cloning, and charitable choice (Burden, 2007). That same study found that MCs who use tobacco are less supportive of tobacco regulation, but was silent about whether this is motivated by self-interest or principle (Burden, 2007). A similar investigation demonstrated that legislators with female children were more supportive of reproductive rights bills than were legislators with male children (Washington, 2008). Finally, an expansive literature investigates the role that legislator race, ethnicity, and sex plays on their in-office behavior (Bratton & Haynie, 1999; Canon, 1999; Griffin & Newman, 2008; Rocca & Sanchez, 2008; Swers, 2002; Tate, 2003; Thomas, 1991; Whitby, 1997).
Among the studies that focus on the extent to which personal factors affect legislator decision making, few speak directly to the role of wealth (Matthews, 1954). To be sure, one study examined whether legislators’ retirement decisions were affected by a 1992 change in campaign finance law that disallowed the practice of pocketing unused campaign funds (Groseclose & Krehbiel, 1994). Others have examined roll call voting on congressional pay raises (or their equivalent; Bianco, Spence, & Wilkerson, 1996; Hibbing, 1983). 1 What these studies have not done is to ask if legislators who will benefit the most financially from legislation are more likely to support it. This is a difficult question to answer using pay raise legislation because the benefit Members receive (salary) does not vary across districts.
One study conducted prior to the adoption of current financial disclosure rules examined the relationship between MC assets in the banking industry and a 1975 vote on mortgage disclosure requirements, MC assets in the power and lighting industries and a 1975 vote on air pollution controls, and MC assets in oil and gas companies and a 1977 vote on tax policy for oil companies (Chappell, 1981). This study found that personal financial interests had no significant impact on roll call voting behavior on these bills after accounting for MC party, ideology, and the economic and demographic characteristics of each Member’s district.
The availability of public financial disclosure requirements for elected officials afforded researchers a new opportunity to revisit the role of financial self-interest. Responding to the Watergate scandal with the Ethics in Government Act of 1978, Congress sought to restore public confidence and integrity in government and its officials. The Act established formal procedures and administrative offices, including the Office of Government Ethics, to oversee and enforce ethical behavior in Washington. Title I of the Act requires Members of Congress, the President and Vice President, Supreme Court Justices, and other high-ranking government officials to submit financial disclosure reports annually. In these public reports, public officials disclose assets, liabilities, and other financial information. Importantly, financial disclosure has been at the center of attempts to discourage legislative conflicts of interest. As the House Ethics Manual states
[P]ublic disclosure of assets, financial interests, and investments has been required as the preferred method of regulating possible conflicts of interest of Members of the House and certain congressional staff. Public disclosure is intended to provide the information necessary to allow Members‘ constituencies to judge their official conduct in light of possible financial conflicts with private holdings. Review of a Member‘s financial conduct occurs in the context of the political process. As stated by the House Commission on Administrative Review of the 95th Congress in recommending broader financial disclosure in lieu of other restrictions on investment income: “In the case of investment income, then, the Commission‘s belief is that potential conflicts of interest are best deterred through disclosure and the discipline of the electoral process” (emphasis added).
2
After the 1978 disclosure requirements were put in place, one study examined the effect of financial interest in agriculture on roll call votes during the 96th Congress (Welch & Peters, 1983). It found that agriculture-related assets have a “negligible” direct effect and an “extremely modest” (Welch & Peters, 1983, p. 391) indirect effect on roll call voting via the effect of agricultural holdings on party affiliation. 3 Another postreform study showed that representatives from working-class occupations exhibit more liberal economic preferences than their white-collar counterparts (Carnes, 2012).
We relate legislators’ assets to their legislative activity with respect to the Estate Tax in the 109th House. The Bush Administration’s 2001 tax cuts set in motion a gradual, but temporary, phase out of the Estate Tax. In 2002, the Estate Tax only applied to estates valued at over US$1 million, however the provisions of the Economic Growth and Tax Relief Reconciliation Act (EGTRRA) simultaneously established a yearly incremental increase in this threshold and decreased the tax rate for estates above the threshold. Specifically, in 2002, estates valued at over US$1 million were subject to a 50%, tax rate, however, by 2009, estates valued at over US$3.5 million were only subject to a rate of 45%, and estates of less than US$3.5 million were not subject to the tax at all. In 2010, EGTRRA eliminated the tax altogether; however, the Act expired on January 1, 2011, at which time the Estate Tax was to return to 2001 levels and rates without further congressional action. 4 In 2005 and 2006, congressional Republicans were eager to extend the repeal of the tax beyond 2010. They were emboldened by a strong majority of the public who supported a permanent repeal, despite the tax only affecting the top 1% to 2% of the population (Bartels, 2008). Although they were not successful in 2005-2006, a compromise was reached in late 2010 with the passage of the Tax Relief Unemployment Insurance Reauthorization and Job Creation Act of 2010. This included a 2-year extension of lower Estate Tax rates and a higher exclusion level; estates were exempt up to US$5 Million with a tax rate of 35% on portions above the exclusion. 5 For reasons we detail below, we focus on the 2005-2006 effort to repeal the tax.
The flurry of activity in the 109th Congress to permanently repeal the Estate Tax affords an opportunity to revisit the question of whether legislators are motivated by their financial circumstances when they vote on tax policy. As the above summary shows, previous studies have examined the roll call relevance of legislators’ financial investments in a specific policy area. However, in these studies the connection between legislators’ financial holdings and the policies they are voting on is often quite attenuated. In other words, the policy outcome would have a small and indirect effect on legislators’ incomes, and only for those MCs with financial interests in the policy area. If personal financial interest does affect MC behavior, it is obvious why MCs would prefer this link to be obscured. The effect is likely to be indirect and difficult to uncover in the large amount of complex and particularized legislation considered. In contrast to previous research, our study takes advantage of a legislative circumstance where there is a clear nexus between legislator interest and policy outcome for all legislators with significant financial holdings. That is, if the Estate Tax were eliminated, wealthier legislators would be able to convey more assets to their descendants. This is, admittedly, a likely place to find an effect of self-interest on roll call behavior; indeed if we do not observe an effect in this context perhaps scholars should halt their efforts to uncover such behavior.
Data
Three bills to permanently eliminate or significantly reduce the Estate Tax reached a roll call vote in the 109th House of Representatives (HR; 2005-2006)—HR 8, HR 5638, and HR 5970. All three bills passed in the House. Only HR 8 and HR 5970 were debated on the Senate floor, and neither reached a vote on the merits. 6 We examine House Members’ decisions to support the passage of and cosponsor these bills based on their personal assets.
Dependent Variables: Roll Call Voting on, and Cosponsorship of, Estate Tax Repeal
On all three bills, a “yea” vote was a vote in favor of permanently repealing or reducing the Estate Tax. To increase the variance in our measure of Estate Tax roll call opposition, we created an opposition index, equal to the total number of “yea” votes a Member cast, divided by the number of Estate Tax roll calls in which the Member participated. 7 An opposition index score of 1 indicates that the Representative consistently voted for the permanent repeal or reduction of the Estate Tax, 0 consistent support for the tax.
In an attempt to muster additional support for the legislation among House Democrats, Republicans included a provision to raise the minimum wage in HR 5970. The strategy worked; more Democrats supported HR 5970 than supported the other two bills (though more Republicans refused to support HR 5970 than the other two bills). This increases the variance on our dependent variable not explained by party affiliation, but raises a concern about the inference we wish to draw from our analyses below. If we find that wealthier Members were generally more supportive of Estate Tax repeal, and less wealthy members were more opposed, we want to draw the inference that Members’ wealth affected their roll call behavior. But what if Members’ decision making was chiefly determined by the minimum wage provision? In this case, we might misinterpret support for the bill among wealthier Members who wanted to back an increase in the minimum wage as evidence of self-interested activity. The available evidence suggests this is unlikely. The few Democrats that voted in favor of HR 5970 and neither of the other bills on the Estate Tax are only moderately wealthy compared to other Democrats. Because wealthier Members tend to represent wealthier districts, where constituents are less in need of a minimum wage increase, these votes from less wealthy Democrats were likely motivated by constituent interests. The Republicans that did not support HR 5970 but did cast a vote in support of one or both of the other Estate Tax bills had a greater range of wealth. The lesser party cohesion on HR 5970 compared to the other votes illustrates how the binary nature of roll call votes does not usually reveal the strength of MCs preferences. Only when a second issue dimension is introduced in a bill are MCs forced to confront the strength of their preferences. In this light, the inclusion of the minimum wage provision in HR 5970 increases the variance in our dependent variable and allows us to observe if Members only weakly supported curtailing the Estate Tax (chiefly Republicans), or (in the case of Democrats) whether their overwhelming support for the minimum wage overcame their support for the Estate Tax.
While roll call votes are an important area of legislative activity, this is not the only area where the effect of assets might be observed. Roll call votes require little effort on the part of the legislator; it is reactive behavior and MCs have the least amount of freedom (Burden, 2007). Cosponsoring bills or speaking on the floor, in contrast, requires more initiative—it is “proactive” behavior (Burden, 2007; see also Wawro, 2000). Members are free to devote their time and energy to countless policy efforts, however, there is a practical limit to how many issue areas in which a Member can be active. Proactive behavior gives insight into the preferences and priorities of the Member. One recent study has shown that cosponsorship is a reflection of MC ideology and consistent with ideology measures generated from roll call voting (Aleman, Calvo, Jones, & Kaplan, 2009).
Therefore, an additional contribution of this analysis is to ask if MC assets affect the likelihood that a MC will cosponsor bills opposing the Estate Tax. In doing so, we probe whether self-interest operates in less visible channels of the lawmaking process. As with roll call vote data, the cosponsorship data is aggregated into an index. However, because HR 5970 attracted only four cosponsors, this bill was excluded from the cosponsorship score to maximize its variance, resulting in a cosponsorship index ranging from 0 to 2. Despite being a Republican initiative, there was Democratic cross-over on cosponsorship of these bills. Just over half of all MCs (51%) did not sponsor either bill; nearly two in five (38%) cosponsored one of the two, and the remainder (11%) cosponsored both bills.
Explanatory Variable: MC Assets
In accordance with the Ethics in Government Act of 1978, MCs are required to file annual financial disclosure reports. The reports include information about a Representative’s assets, liabilities, unearned income, and travel, along with other miscellaneous financial information. MCs are not required to include the value of their primary residence, and as a result the asset data are generally lower than their true value. 8 Exact values for each asset are not reported; instead, the report offers preset value ranges. For each asset, the filer must select the range in which the exact value of the asset lies (i.e., US$1,001-US$15,000, US$15,001-US$50,000 and so on). Ideally, our asset measures would reflect Members’ assets and liabilities (i.e., net worth). However, the CRP data report Member liabilities much less consistently than assets, so we rely solely on the latter.
The Center for Responsive Politics (CRP) regularly obtains these reports from the Senate Office of Public Records and the Office of the Clerk of the House. The Center calculates a Member’s “asset range” by summing the minimums and maximums for each asset listed. 9 We measure each Member’s assets as the midpoint of the 2005 CRP asset range (following Welch & Peters, 1983). 10 Measuring assets using ranges will obviously introduce measurement error in our key explanatory variable. However, this measurement error is highly likely to be random, and will thus attenuate the estimated effect of this explanatory variable on MC behavior (King, Keohane, & Verba, 1994). 11 Stated another way, the true effect of assets on behavior is likely to be stronger than our results suggest.
Our expectation is that as assets increase, so, to a point, will the likelihood of Estate Tax opposition. However, MCs with extremely large asset holdings will likely behave the same as those with smaller but still very large holdings, so we log our asset measure and adopt a linear-in-the-terms estimation approach. 12
Because the proposed reforms to the Estate Tax introduced new exemption thresholds, we also employ a binary asset measure. Namely, we estimate alternative specifications using an indicator for whether a Member’s assets exceeded US$1 million in 2006. We adopted this threshold because HR 8 called for permanent repeal of the entire tax while HR 5638 ad HR 5970 aimed to increase the threshold value of estates that would be exempt from the tax; on all three bills representatives with assets over US$1 million would have benefited financially.
Control Variables
We account for a number of factors that are likely to be related to both MC assets and MC legislative behavior. First, wealthier MCs may be more likely to hold antitax world views and these preferences are likely to influence their behavior on all tax votes, not just those that would tax their estates (Burden, 2007; Levitt, 1996). To control for MCs’ global opposition to taxes, our models incorporate MCs’ National Taxpayers Union (NTU) scores for the 109th Congress. This score is a ratio of the weighted support for the NTU’s position on all taxing and spending legislation, among those roll calls in which the MC participated. Notably, the NTU scores we use are adjusted to exclude the three roll calls we examine and reflect MCs’ global antitax opinion.
Next, if wealthier MCs are more likely to affiliate with one of the parties, and party members take distinct positions on the Estate Tax bills, without accounting for Members’ party affiliations we would exaggerate the importance of Member wealth. Indeed, our data show that Republican Members are on average much wealthier than Democrats, and strong (but imperfect) party line voting is present on our three roll calls. So, we control for MC party using a binary variable (Republican 1, Democrat 0). 13 By doing so, we are able to observe the direct effect of wealth on Estate Tax voting, over and above its indirect effect via party affiliation.
It is also necessary to account for the possible influence of constituent opinion on MC behavior (Miller & Stokes, 1963), since MC support for repeal may simply reflect attentiveness to constituents who oppose the Estate Tax. The availability of specific district-level opinion on the Estate Tax in the 2004 National Annenberg Election Survey (NAES) allows for a specific measure of constituent opinion. 14 The NAES asked 11,562 Respondents “Completely eliminating the Estate Tax, that is, the tax on property left by people who die—do you favor or oppose this?” with an additional 3650 respondents asked this base question and then a follow-up item regarding the intensity of their support or opposition. We combine these two groups using the latter group’s base response into a sample of 15,212 Respondents. Of these individuals, approximately 71% favored or strongly favored a “complete elimination” of the Estate Tax, and 29% opposed or strongly opposed. 15 We then map Respondents into congressional districts and calculate the proportion of NAES Respondents in a congressional district with a stated opposition to the Estate Tax, 16 and include this measure in our models.
MCs who represent wealthier districts will be more likely to oppose the Estate Tax if they are attentive not only to constituents’ stated opposition to the tax but also to objective conditions in the district—to constituents needs in addition to their wants. In keeping with this, we control for the per capita household income of each congressional district using data from the U.S. Census. 17 Relatedly, we control for the percent of district residents who live in rural areas, due to the frequent claim that the Estate Tax is particularly burdensome on residents of agricultural zones.
Our last set of controls relate to MC characteristics beyond party affiliation. We control for MC membership on the Ways and Means committee (from which all three bills emerged) because there is some evidence that committee membership impacts the decision to cosponsor legislation originating in that committee (Krehbiel, 1995). We also include MC age because older MCs may be more likely to oppose the Estate Tax, given they have had more time to accumulate assets and may be closer to passing those assets on to their descendants than are younger Members.
Results
We first describe the association between Estate Tax repeal roll call votes and MC assets. Dividing MCs into three similarly sized asset strata (below US$250,000, US$250,000-US$1,750,000, and more than US$1,750,000) 18 we find that as wealth increases, so does MC propensity to vote for anti-Estate Tax legislation (Table 1). Those with assets less than US$250,000 had a mean opposition score of .52 where an opposition score of 1 indicates the Member voted in favor of all bills to reduce or repeal the Estate Tax for votes in which they participated, and a score of 0 indicates the Member opposed all bills. Those in the middle strata, with assets between US$250,000 and US$1,750,000, had a mean opposition score of .59, and those Members with the highest incomes, above US$1,750,000, had an even higher mean opposition score of .69. These means are generally statistically distinguishable from one another (see table note).
Opposition to Estate Tax on Roll Calls, by MC Asset Strata.
Note: Results of difference of means tests: Stratas one and two p = .11, stratas two and three p = .01, and stratas one and three p = .001. MC = Member of Congress.
It is easy to identify particular Members who do not fit this pattern—even very wealthy Members who opposed the repeal (e.g., Nancy Pelosi, Rosa DeLauro, and Lloyd Doggett) or reduction of the Estate Tax and Members of more modest means who supported the repeal (e.g., Curt Weldon and Mark Sauder), but the general pattern in the data shows a tendency in the other direction.
An examination of Member assets and cosponsorship behavior yields a similar pattern. As MC assets increase, so does the likelihood of cosponsoring Estate Tax repeal legislation (see Table 2). The lowest asset group had a mean cosponsorship score of .43 (i.e., about half a bill), the middle asset group, a score of .57 and the highest assets group had a mean score of .73. All of these means are statistically distinguishable at a .05 significance level.
Mean Bill Cosponsorship Score, by MC Asset Strata.
Note: Results of difference of means tests: Stratas one and two p = .05, stratas two and three p = .01, and stratas one and three p = .0003. MC = Member of Congress.
Next, we account for the factors described above—MC antitax ideology, party affiliation, age, and Ways and Means Committee membership; and district opposition to the Estate Tax, per capita income, and percent urban. Our cosponsorship models also control for the total number of bills the MC cosponsored in the 109th Congress, following Wawro (2000) and others. For both dependent variables, we employ an ordered probit estimator due to the categorical nature of the roll call and cosponsorship indices. 19
The fully specified roll call vote model indicates that even after accounting for potential confounds, wealthier MCs in the 109th House were more likely to support the anti-Estate Tax bills (Table 3, column 1), and we can have a good deal of confidence in the direction of this estimated effect. When we employ an indicator for whether a Member reported at least US$1 million in assets, we once again find that wealthier MCs more likely to vote in favor of the bills (column 2), and we can be even more confident in the direction and magnitude of this effect. 20 To provide a sense of the size of this effect, we simulated the predicted probability of a MC supporting 1, 2, or all 3 bills for two asset levels—those with assets less than US$1 million and those with assets over US$1 million. According to these predicted probabilities, while a MC with less than US$1 million in assets had a 20% probability of supporting none of the bills, a Member with more than US$1 million in assets had just a 12% probability of supporting none of the bills. In contrast, while Members with less than US$1 million in assets had a 45% probability of supporting all three bills, Members with assets greater than US$1 million had a 59% probability of supporting all three bills, all else equal. It is important to note that this jump in predicted probability spans the point of indifference between support and opposition (50%), making the substantive significance of the result even greater than it appears at face value.
Member Assets and Voting on Estate Tax Repeal, 109th House.
Note: Cell entries are ordered probit estimates of MC support for three anti-Estate Tax roll calls. Standard errors in brackets. MC = Member of Congress; NTU = National Taxpayers Union.
p <.01. *p <.05, one-tailed test.
We also conducted threshold analyses to examine whether Members whose wealth was just above or below the US$1 million mark were particularly sensitive to their wealth when voting on these bills. They were. In bivariate comparisons, we found that the effect of being a millionaire on Estate Tax voting was much more significant among the 54 legislators whose wealth ranged from US$750,000 to US$1,250,000 (estimated effect .660) than among the entire sample (estimated effect .417). Both estimates were significant at a .05 level. When we added all the controls appearing in Table 3, the estimated effect of wealth remained larger among those whose wealth was closer to the threshold (.475) than among the entire sample (see Table).
Other results in these models are noteworthy. As expected, Republicans were much stronger opponents of the Estate Tax than were Democrats. In both specifications, legislators representing less urban districts were more likely when voting to support Estate Tax reform, confirming our expectations that the residents of these districts are more concerned about the Estate Tax. Indeed, when we regressed constituent opposition to the Estate Tax on district percent urban and generated expected values, we found that in a district that is 80% urban, 70% of the residents are anticipated to favor eliminating the Estate Tax, while in a district that is 20% urban, 80% of the residents are anticipated to favor eliminating the tax.
Somewhat deceptively, the results also seem to suggest that legislators representing wealthier districts were somewhat less supportive of the anti-Estate Tax bills. However, it is important to think carefully about the chain of causality linking constituent characteristics to legislator wealth to roll call votes. The proper way to think about this is that legislator wealth mediates the relationship between constituent characteristics and roll calls: wealthy voters who oppose taxes elect wealthy legislators who oppose taxes. Therefore, legislator wealth must be considered a “posttreatment variable” when analyzing the total effect of constituent wealth on legislator votes (see Gelman & Hill, 2006, Chapter 9). As such, estimates on constituency characteristics and opinion in a specification including legislator wealth will not capture the total effect of these characteristics.
Finally, before we proceed to our cosponsorship analyses we report the outcome of a placebo test. That is, we examined whether legislator wealth does not affect roll call voting on closely related issues on which wealthy legislators do not have a financial interest. Our contention that self-interest was truly driving opposition to the Estate Tax would be even more persuasive if this were the case. However, if legislator wealth is a strong predictor of tax bills where the link to their own finances is much more tenuous, then we would be less inclined to accept the self-interest argument. Our placebo tests modeled legislators’ National Taxpayers Union scores in the 109th House, net of the Estate Tax votes, as a function of legislator wealth. As above, we estimated two models using two different measures of legislator wealth—logged assets and a binary indicator for whether the legislator possessed more than US$1 million in assets. In both models, the estimated effect of legislator wealth on the placebo was positive but nowhere near statistically significant. Legislator wealth affected opposition to the Estate Tax but not behavior on tax issues less closely tied to Members’ financial futures.
Next we turn to answering whether self-interest affects cosponsorship. The results for the cosponsorship models are similar in many respects (see Table 4). First, personal wealth is positively associated with cosponsorship of bills to reform or repeal the Estate Tax. When we measure assets continuously, we find that wealthier Members were more likely to sponsor bills to limit the reach of the Estate Tax. When we use a millionaire indicator, the effect, while still positive, is no longer statistically significant in either the full sample or the subsample of legislators with assets between US$1.25 million and US$750,000. However, upon further examination, we found that Members with assets of more than US$2 million were reliably more likely than those less than US$2 million in assets to cosponsor these bills (p = .03, results available on request).
Member Assets and Cosponsorship of Estate Tax Repeal, 109th House.
Note: Cell entries are ordered probit estimates of MC cosponsorship of anti-Estate Tax bills. Standard errors in brackets. NTU = National Taxpayers Union; MC = Member of Congress.
p <.01. *p <.05, one-tailed test.
To provide a sense of the size of this effect, we simulated the predicted probability of a Member cosponsoring 0, 1, or 2 bills for two asset levels—those with less than US$1 million in assets and those with more than US$1 million. According to these predicted probabilities, while a MC with less than US$1 million in assets had a 57% probability of cosponsoring none of the bills, a Member with more than US$1 million in assets had a 50% probability of cosponsoring none of the bills. In contrast, 41% of MCs with less than US$1 million in assets cosponsored one of the bills, compared to 47% of MCs with more than US$1 million in assets. Wealthier MCs were more likely than less wealthy MCs to cosponsor Estate Tax limits in the 109th House.
Other findings in these models are worth noting and at times differ from the patterns observed in the roll call models. First, while we observed above that Members who hold antitax world views (as reflected in their NTU score) were not more likely to vote for Estate Tax repeal, these Members were more likely to cosponsor this legislation. In addition, Members of the Ways and Means Committee were more likely to cosponsor these bills, as we expected. Older representatives were somewhat less likely to cosponsor anti-Estate Tax legislation, but here again we may need to interpret age as a pretreatment variable, because older members will generally possess more assets.
Finally, legislators representing districts where public opinion was more opposed to the Estate Tax also were more likely to cosponsor Estate Tax repeal legislation, independent of the relationship between district wealth and legislator wealth. Indeed, because our measure of public opinion is not very reliable the true effect of public opposition to the Estate Tax on Member cosponsorship behavior is likely quite a bit larger than we have estimated. Still, the effect is substantial. Approximately 2 in 3 Members representing districts where 50% of residents favor eliminating the Estate Tax are anticipated to cosponsor neither of the two bills, and one in three is expected to cosponsor one of the two bills. In contrast, only half of the Members representing districts where 80% of the residents favor eliminating the tax are anticipated to sponsor none of the bills, with the remaining half anticipated to cosponsor one of the two bills. In their decision to cosponsor, Members have more freedom to be attentive to the variation in district opinion that does exist, even when in virtually all districts a majority of residents favor eliminating the tax. Namely, Members who represent districts where there is less, though still majority, support for eliminating the tax are less likely to cosponsor eliminating it.
Notably, in late 2010 a lame-duck session of Congress during a sluggish economic recovery was faced with expiring Bush era tax cuts—including the Estate Tax. President Obama and Congressional Democrats were eager to see extensions of economic stimulus provisions, while Republicans were concerned about tax cuts set to expire from the Bush tax cuts. A compromise was reached with the Tax Relief Unemployment Insurance Reauthorization and Job Creation Act of 2010, which included a 2-year extension of several key provisions of the Bush Tax Cuts, including personal income, dividend, and capital gains rates, in return for economic stimulus spending, including an extension of unemployment benefits. For 2011 and 2012, estates valued at less than US$5 million are exempt from the Estate Tax, and those exceeding this value are subject to a 35% tax rate. 21 Although there was considerable variation in partisan voting, the omnibus nature of this act does not allow us to cleanly model it based on our theory. Separating out personal financial motivation from many other policy considerations present in the act would be impossible. Furthermore, unlike the legislation we modeled, the provision with regard to the Estate Tax in 2010 was only an extension of a lower tax rate for 2 years, rather than a permanent change.
Conclusion
We have shown that the asset holdings of MCs affected the likelihood that they supported a reduction or repeal of the Estate Tax in the 109th House. This finding holds regardless of whether we model Members’ roll call voting decisions or their decisions to cosponsor the tax repeal bills, and it obtains even after accounting for the party affiliations and antitax views of MCs and their constituents’ opinions, as well as other potential confounds.
Some will find our key results cause for normative concern. Motivated by the same conflict of interest concerns that led the early House to bar legislators from acting on matters affecting their private interests, some will be alarmed that legislators own financial circumstances impact tax policy. Still, it is difficult to imagine what institutional reforms might be pursued to minimize conflicts of interest. Indeed, according to the House Ethics Manual, such conflicts may be “inherent to a representative system of government.” 22 One possibility is to modify the Congressional oath, which does not currently include any language regarding conflicts of interest. 23 Research in behavioral economics has shown that oaths have been surprisingly effective at modifying behavior (Ariely, 2008). Another thought is to amend the House Ethics Manual to require Members to disclose to the Committee potential legislative conflicts of interest. This practice would allow the Committee to issue an advisory opinion regarding the Member’s subsequent involvement in the legislation. Finally, voters have a role to play in encouraging legislators to avoid conflicts of interest. A number of Members under conflict-of-interest scrutiny lost their 2012 re-election bids, with many of the races sufficiently close that the unwelcome attention may have been pivotal (Lipton, 2012). If lawmakers sense that voters’ have become more sensitive to perceived conflicts of interest, they may be more cautious about creating such perceptions.
Others may be more sanguine about our key findings. In this view, Congress must act upon a myriad of matters which affect the nation’s citizens. They are citizens, and so enjoy the same benefits and suffer the same burdens as all citizens subject to the laws. Aristotle, Locke, Madison, and Jefferson all argued that lawmakers should be subject to the laws they make. This is thought to constrain legislators, preventing them from imposing burdens on citizens that legislators do not have to suffer themselves. As Madison put it in Federalist 57, “the House of Representatives . . . can make no law which will not have its full operation on themselves and their friends, as well as on the great mass of the society. This has always been deemed one of the strongest bonds by which human policy can connect the rulers and the people together.” This may be, but our analysis of estate tax lawmaking in the 109th Congress serves as a caution that making legislators subject to the laws also permits them to pursue policies that may further their own interests.
Before we discuss the additional implications of our key finding, we add one caveat. Specifically, we emphasize that personal legislator wealth is one factor among several that influence behavior. What is more, the substantive significance of legislator wealth on decision making is dwarfed by the importance of other factors, principally their political party affiliations. This fact does not diminish the import of our conclusion that legislator wealth affects behavior even after accounting for a number of potential confounding factors, but instead places it in a proper context.
Our finding will be of interest to scholars in a number of areas. First, where others have shown that legislators’ race, ethnicity, and gender can have an independent effect on their roll call decision making and other representational behaviors (e.g., Canon, 1999; Griffin & Newman, 2008; Swers, 2002) we show that legislator wealth can exert influence as well (Carnes, 2012).
Future work might build on our core result by investigating additional issue areas beyond the Estate Tax. Are legislators’ affected by their wealth when they cast votes in other, related domains such as taxes on capital gains and the progressivity of the tax structure. Future research might also pursue whether proactive legislator behavior beyond cosponsorship is affected by their wealth—floor speeches are one promising avenue for exploration because the costs of speechmaking are more substantial than the cost of cosponsorship.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.