167

ELECTION LAW JOURNALVolume 8, Number 3, 2009© Mary Ann Liebert, Inc.DOI: 10.1089/elj.2008.0025

Original Article

The Devil We Know? Evaluating the Federal ElectionCommission as Enforcer

Michael M. Franz

The Federal Election Commission is an agency that was designed to fail, and at that task, atleast, it has succeeded brilliantly.

Editorial, The Washington Post, 7/11/2003

The Federal Election Commission (FEC) (for-mally created in 1974 when Congress amendedthe Federal Election Campaign Act) is chargedwith the burdensome role of administering andenforcing federal election laws. Comprised ofsix voting commissioners, the FEC by law canhave no more than three members of the samepolitical party and in practice is balanced be-tween three voting Democrats and three vot-ing Republicans. Its performance has beenmuch maligned, however, and is often targetedfor complete overhaul or total elimination. Be-yond the Washington Post editorial, considersome recent attacks on the FEC:

• “A close look . . . at the FEC’s 27-year his-tory reveals an agency that puts protect-ing the interests of the Democratic and Re-publican parties ahead of policing election

laws or guarding public confidence in theintegrity of campaigns and elections”(Columnist Jim Drinkard, 2002).

• “The FEC is an ineffectual agency, struc-tured by Congress to be slow and ineffec-tive, composed of commissioners whose appointments are tightly controlled bymembers of Congress and the political par-ties they regulate, and hobbled by a lack ofreal enforcement powers and a chronic lackof funds” (Walsh and Eilperin, 2002: citingFred Wertheimer of Democracy 21).

• “The Federal Election Commission hasdone exactly what Congress intended it todo—nothing much to curb widespreadcampaign finance abuses. In fact, the feck-less six-member panel created the ‘soft-money’ monster of unregulated campaigncontributions. This glaring loophole hasessentially negated the reform’s goal ofsanitizing a corrupt system whereby bigcontributors gain ready access to lawmak-ers” (Editorial, San-Diego Union Tribune,7/21/2003).

• “It is clear that the FEC is a failed agencywith overtly partisan commissioners whooppose both new and longstanding cam-paign finance statutes. The FEC hasproved its ineffectiveness and its willing-ness to run roughshod over the will of theCongress, the Supreme Court, the Ameri-can people and the Constitution” (SenatorJohn McCain, 2004).

Michael Franz is assistant professor of Government andLegal Studies at Bowdoin College. He is co-author of Cam-paign Advertising and American Democracy (2007, TempleUniversity Press) and author of Choices and Changes: In-terest Groups in the Electoral Process (2008, Temple Uni-versity Press), the latter being an investigation of interestgroup electioneering from the mid-1970s until 2004. Hehas published in American Journal of Political Science, TheJournal of Politics, Political Behavior, The Sociological Quar-terly, and Social Science Quarterly. He owes special thanksto Sam Chapple-Sokol, who diligently coded the firstwave of the data. David C.W. Parker read an early draftof the article and offered invaluable suggestions, as didanonymous reviewers and the journal’s editors. Any andall errors remain the author’s.

FRANZ168

The FEC has also inspired fascinating analo-gies. Todd Lochner (2003) writes: “the FederalElection Commission has been characterized asa toothless tiger institutionally incapable andwholly unwilling to enforce the Federal Elec-tion Campaign Act, but others claim it is tooth-less like an anaconda whose overzealous pros-ecution of innocent third parties strangles thelife from First Amendment freedoms.” ScottThomas and Jeffrey Bowman (2000) note alsothat the FEC has been called “a lapdog,” “a self-licking ice cream cone,” and “nuts.”1

Amanda La Forge (1996) additionally opines,“The structure of the FEC may make it the weak-est of all federal agencies; certainly it is the weak-est of the so called ‘independent agencies’. . . .An even number of Commissioners generatesvote-splitting and effectively incorporates grid-lock into the enforcement process. The evennumber of representatives from each politicalparty also may indicate that Congress expectedthe Commissioners to behave in a partisan,rather than an impartial manner” (p. 4).

From these assessments, then, one might begreatly worried about the functioning of theAmerican campaign finance system. Indeed, arecent controversy in 2007 and 2008 over theappointment of four new commissioners to theFEC (a controversy that left the Commission in-operable for over six months) raised seriousquestions about the importance of maintainingthe current FEC structure.2 In response to all ofthe above assessments, however, there is littleempirical work examining systematically theenforcement behavior of the Federal ElectionCommission. We know very little about thescope of penalties levied by the FEC, its abilityto overcome a 3–3 partisan Commissioner bal-ance, or its approach to regulating the samecampaigns and candidates that fund the Com-mission and determine its make-up.

Fortunately, however, the FEC has begunposting detailed information about its enforce-ment matters. On its web site, for example, theFEC now lists all completed Matters Under Re-view (MUR), which are enforcement cases in-vestigated by the Commission. The data con-tain all of the relevant documents and votes foreach case, going back over 10 years to about1996. This article exploits these data and eval-uates the enforcement actions of the Commis-

sion in these MUR. I code these complaints ona variety of dimensions, including the issue fo-cus, the characteristics of the complainant andrespondent, and the final penalty levied by theCommission. I use the data to test a series ofhypotheses about predicted FEC behavior.

All told, the evidence points to an FEC thatis more severe in recent years, but with somebias towards incumbents and national parties.The evidence is far from a damning indictmentof the FEC’s enforcement pattern, however. In-deed, I conclude that the Commission deservesincreased enforcement powers (in contrast tomany who call for its elimination or reconsti-tution), including the ability to conduct ran-dom audits of regulated political committees.

The analysis in this article is the most com-prehensive and up-to-date account of FEC en-forcement cases in the current literature. Lochnerand Cain (1999) rely on a sample of MUR be-tween 1991 and 1993, and more recently Lochneret al. (2008) examine FEC enforcement between1999 and 2004. The analysis offered here goes far-ther than either previous investigation. For ex-ample, I include a more detailed review of thoseinitiating and those implicated in enforcementcases, and I spend considerable time discussingCommission voting patterns.3

The focus on the last 10 years also allows usa number of important advantages. First, we canevaluate the FEC enforcement action during one

1 In his dissertation on the FEC, Julian Salisbury (2002)also lists the tendency of FEC observers to describe theCommission with amusing metaphors.2 For more information on the FEC appointment contro-versy in 2008, I recommend Bob Bauer’s blog on campaignfinance issues (�http://www.moresoftmoneyhardlaw.com�). Click on “Enforcement” for a list of his postingson the matter.3 To be more specific, the analysis offered by Lochner andcolleagues (2008) is invaluable, but there are importantdifferences between that article and my analysis. Theydrop certain MUR in 2000 and 2001, which I include. Onthe other hand, they consider alternative enforcementmechanisms at the Commission (beyond MUR), which Ido not. I also include measures for case subject in mod-eling the outcome of MUR, and I predict the impact of arange of variables on the size of the assessed fine, not sim-ply the presence of a fine (as they do). All told, Lochneret al. are concerned mostly with the impact of structuralchanges on FEC enforcement, whereas I am concernedalso with whether the FEC treats certain respondents withparticular deference.

THE FEC’S ROLE AS ENFORCER 169

of the most controversial campaign finance pe-riods in a generation (that is, during the use ofsoft money, issue advocacy, and 527s). Second,we are afforded a number of unique natural ex-perimental conditions—namely a period whenthe FEC is initiating structural reforms and twobrief periods when the FEC operated with a par-tisan imbalance (because there were, five com-missioners instead of six).

THE FEC AS ENFORCER

What pattern of enforcement should we ex-pect from the FEC? This question suggests an-other: what might Congress want in the FEC?From first principles, we might expect Con-gress to create an agency that monitors dili-gently for election law violations, wheneverand wherever they occur. On the other hand,members of Congress—as once and futurecandidates—might wish the agency to treadvery lightly, at least with respect to incumbentsand their major party committees. After all, anaggressive agency with a strong enforcementmandate might quickly become an election lawFrankenstein. In the design of the FEC, then,we witness a classic problem in all bureaucraticdelegation: how do principals (Congress) del-egate authority to an agent (the FEC) that mon-itors an important problem in democratic pol-itics but does little harm to the principals?

Congress has granted the Commission con-siderable discretion to define the landscape ofcampaign finance laws. For example, the FECcan alter and clarify the regulatory context intwo principal ways—regulation changes andadvisory opinions. Regulations are changedboth when Congress alters the law (and tasksthe Commission with clarifying and fleshing outthe new rules) and when the Commission inde-pendently perceives current regulations to beunclear or inadequate (for example, how exist-ing election law restrictions conform to newtechnologies, such as the Internet). The FEC mayissue advisory opinions when candidates, par-ties, or interest groups are uncertain as to thescope of the law and ask the FEC for formalguidance on a specific question. The issuedopinion establishes a precedent for similar ques-tions asked by different political actors. With

both advisory opinions and regulations, Con-gress has almost no say in the Commission’s in-terpretation of existing election law. If dissatis-fied, Congress can sue the Commission infederal court or pass new legislation.4

The FEC also must respond to complaintsfrom individuals, candidates, parties, interestgroups, and other governmental agencies aboutthe alleged abuse of federal campaign financelaws. Once a person asserts a potential vio-lation, the FEC opens an enforcement proceed-ing, called a “Matter Under Review” (MUR).The FEC’s Office of General Counsel (OGC)oversees the initial investigation and makes areport to the six commissioners about the via-bility of the case. If four commissioners vote tofind reason to believe that a violation has oc-curred, the MUR is pursued further. The Com-mission might investigate and find a violation,but choose not to pursue a penalty, but if com-missioners wish to fine those implicated, theycan do so only if respondents willingly enter aconciliation agreement with the FEC.5 Such en-forcement power represents a risk for politicalactors—an investigation is sure to cause eventhe longest serving incumbent some headache.

How might Congress control the outcome ofsuch FEC power? We know from the bureau-cracy delegation literature that Congress canimpose both ex ante and ex post oversight of abureaucratic agency (McCubbins 1999). Con-trol ex ante occurs in the agency structure,which is designed (or constrained) to performwith certain powers and in certain ways in linewith the wishes of Congress or the president.6By contrast, ex post oversight happens whenCongress monitors and evaluates agency per-formance.7 There is a trade-off, however, be-

4 Visit �http://www.fec.gov� and click on “Laws & Reg-ulations” for more information.5 If the FEC is unable to obtain a conciliation agreement,it still has the option of pursuing the case in federal civilcourt.6 See also McCubbins, Noll, and Weingast (1987, 1989).7 There are a number of ex post strategies. McCubbins andSchwartz (1984) distinguish, for example, between “firealarm” and “police patrol” oversight—the former occur-ring when Congress reacts to an overt crisis; the latterwhen Congress constantly monitors for wrong-doing inthe bureaucracy.

tween the forms of enforcement. A predomi-nantly ex ante approach creates the potential forregulators to “drift” from the wishes of Con-gress, but an ex post approach requires a sig-nificant investment of time and resources bythe principals.

With respect to the FEC, Congress employsa mix of both oversight strategies. Consider itsex ante oversight first. As Oldaker (1986) notes,during the initial phase of FEC design, Con-gress worked to maintain significant controlsover agency performance. He writes, “If thefoxes in Congress were giving up their self-su-pervisory role, they were not going to turn thekeys over to just anyone” (p. 134). Congress de-signed the agency to be bipartisan, for instance,and mandated that decision-making be basedon a majority of four Commissioners.8 The FECadditionally does not have a powerful chair-person, but rotates the position annuallyamong the six commissioners. Julian Salisbury(2002, p. 176) writes: “The Commission hasbeen poorly led. Yet this enduring characteris-tic was more the product of Congressional self-interest and less the product of individual fail-ure at the FEC” (p. 179). In 1979, Congressworked to further weaken the Commission’sinfluence by stripping its ability to conduct ran-dom audits of registered political committees.

In many ways, then, the extent of the FEC’slegislative and enforcement discretion is lim-ited by the ex ante constraints on its power.With no random audit authority, the Commis-sion is forced to rely on third party complaintsor inadvertent discovery by agency staff to ini-tiate an investigation. (Furthermore, concilia-tion agreements must be accepted by accusedrespondents; this precondition provides someopportunity to force the Commission’s handinto either dropping the investigation or push-ing the matter into federal court.) Its bipartisanstructure with rotating leadership hampers theCommission’s ability to form a clear and con-sistent perspective on enforcing campaign fi-nance laws. And with its modest budget, theextent to which the general counsel’s office canswiftly investigate multiple violations is weak-ened. On this latter point, Salisbury notes thatFEC staff increased from 223 to 302 between1978 and 1998, but the amount of spending incampaigns (and by proxy, the potential for in-creased campaign finance law violations) in-

creased by nearly 500 percent in the same time-frame (2002, p. 122). These figures notably ex-clude the explosion of issue advocacy and 527sin late 1990s, both being controversial forms ofelectioneering which put further pressure onFEC enforcement (Franz 2008).

As a consequence, the MUR process has beenthe focus of considerable criticism. For exam-ple, Ken Gross (1991, p. 283)—former associategeneral counsel at the FEC—notes: “Many par-ticipants in the political process believe that theelection laws need not be heeded. The chanceis slim that violations will be detected; party-line deadlocks reduce the chance that the Com-mission will investigate violations of the law,and any resulting penalty will probably comelong after the election.”9

In addition to these structural (ex ante) con-straints, though, Congress also relies on signif-icant ex post controls. These include, perhapsmost significantly, yearly budgeting authority.Congress also holds significant sway in com-missioner appointment powers. AlthoughCongress formally approves commissionersnominated by the president, an informal normallows Congress to effectively choose new com-missioners. In addition, Salisbury (2002, p. 114)convincingly recounts congressional pressureson the FEC to improve performance, usuallymeasured as case completions and conciliationagreements. He writes, “One might reasonablyassert that Congress is attempting to influencethe type of violations and cases the FEC ulti-mately enforces”—namely, smaller violations

FRANZ170

8 Originally, Congress established that two of the six com-missioners were to be appointed by the president and theother four by the House and Senate. Following Buckley v.Valeo, 424 U.S. 1 (1976), Congress responded by giving thepresident power to appoint all six commissioners with ad-vice and consent from the Senate. The FEC originally alsohad two non-voting ex officio members, the Clerk of theHouse and the Secretary of the Senate. In 1993, however,the D.C. Appeals Court in FEC v. NRA Political VictoryFund (6 F.3d 821) ruled that their participation on theCommission was unconstitutional.9 Gross also questions whether MUR enforcement violatesrespondents’ due process rights. Since 2003, in responseto such criticisms, commissioners have adopted new poli-cies relative to its enforcement process (such as the pub-lication of all MUR materials on the FEC web site). Visit�http://www.fec.gov/law/policy.shtml� for more in-formation and documentation concerning these changes.The FEC conducted a similar review of enforcement inearly 2009.

THE FEC’S ROLE AS ENFORCER 171

that are easier to investigate.10 The Democracy21 Task Force report on the FEC, released in2002, also tells of considerable (and aggressive)pressures from Congress to improve perfor-mance (pp. 71–84).

Under such conditions, then, we can makesome reasonable inferences about the Com-mission’s enforcement of election laws. Giventhe pre-existing ex ante limits on enforcementpowers, for example, we might have reason to expect its overall performance to be un-satisfactory to many. Nevertheless, Congressshould feel relatively unthreatened with itsgenerous grant of power to the FEC in therealm of regulations, advisory opinions, andenforcement—after all, the grant of powercomes with few opportunities for the Franken-stein outcome noted earlier.

Given the concomitant threat of ex postoversight, though, we might expect to findcertain patterns in Commission enforcementbehavior. With a watchful Congress oftenbent on improved enforcement numbers, withthe looming threat of budget cuts, and withcongressional appointment powers, it is nostretch to predict that the Commission willtreat its benefactors with some deference. AsWilliam Oldaker (1986, p. 136) notes: “theFEC will be forced to continue to walk atightrope—vigorously enforcing the lawspassed by Congress, while stopping short ofincurring congressional wrath to such an ex-tent that Congress decides to put the FEC outof business.” Lochner and Cain (1999) add:“The FEC . . . is somewhat unique in that itdirectly regulates the very people responsiblefor the agency’s budget” (p. 1899). (See also“Symposium on the Federal Election Com-mission,” 1994.)

I term my first set of expectations the “re-spondent bias hypothesis.” The hypothesisstates that FEC enforcement decisions will fa-vor incumbents and the major DC-based partycommittees, the latter because of their close tiesto incumbents.11 This outcome allows Congressto grant agency discretion in the enforcementof campaign finance laws, but remain insulatedfrom aggressive intrusion on its members’ ownbehavior. In other words, ex post controls (bud-getary or otherwise) are always available toCongress should the FEC drift away from thepolicy preference of Congress (or assert pow-

ers inconsistent with the ex ante limitationsplaced on it).

The hypothesis specifically expects the fol-lowing. First, incumbents and national partycommittees should be less likely than other re-spondents to enter conciliation agreements,and pay lower fines when they do. Second,votes on MUR involving incumbent and na-tional party committees should induce consen-sus among commissioners as they jointly deferto principals.

There is some existing evidence of this phe-nomenon. Franz (2008) investigated a relatedregulatory process at the FEC, evaluations ofadvisory opinions. When looking at all opin-ions issued between 1977 and 2003, Franzfound a significant bias in favor of candidates.More specifically, when controlling for a num-ber of factors predicting FEC approval of pro-posed campaign behavior, candidates (and in-cumbents particularly) were far more likely tohave their request approved.

There are at least two possible objections tothis hypothesis. First, some delegation schol-ars worry about congressional preferenceschanging over time, making it harder for bu-reaucrats to locate Congress’s current outcomegoals (Epstein and O’Halloran 1994, p. 713).Others note that the initial creation of anagency can be often fraught with conflictswithin Congress, weakening the match be-tween bureaucratic outcomes and congres-sional wishes (Wood and Bohte 2004). Theseshould be less of a concern for campaign fi-

10 On a larger scale, Congress often considers plans for acomplete FEC reorganization, overhaul, or elimination—on its own an ex post and overt threat to the FEC. For ex-ample, in both 2003 and 2006, Senators John McCain (R-AZ) and Russ Feingold (D-WI), and Representatives ChrisShays (R-CT) and Marty Meehan (D-MA), proposed theFederal Election Administration Act, which would re-place the FEC with a new Election Administration com-posed of an independent three-member commission andadministrative law judges.11 Commissioners often have close ties to incumbents andparty committees. For example, Michael Toner (whoserved from 2002–2007) was former chief counsel to theRepublican National Committee, and Thomas Josefiak(1985–1991) was former legal counsel to the National Republican Congressional Committee. David Mason(1998–2008) was a former staffer to Republicans JohnWarner and Trent Lott, and current Commissioner Cyn-thia Bauerly was a former staffer to Democrats CharlesSchumer and Amy Klobuchar.

nance law enforcement, as presumably all in-cumbents (whether Democrats or Republi-cans) would prefer an incumbent bias.

Second, some might object to an expectationof incumbent bias because parties often haveconflicting opinions on campaign finance lawenforcement. Republicans, for example, mightprefer less enforcement, and a weaker FEC,consistent with its goal of less federal power.In contrast, Democrats might prefer a strongerFEC because of its general faith in governmentpower. Stated simply, the FEC conceivablymight be unable to assert an incumbent orparty bias because ideological commissionersdiverge on these matters.

Beyond the hypothesis of “respondent bias,”we might also form expectations about the in-fluence of certain changes in agency design.Since 1996 there have been three key changesin the nature of the Commission. First, Congressenacted term limits for commissioners in theTreasury and General Government Appropria-tions Act of 1998. Those nominated beginningin 1998 are now eligible for only one term onthe Commission. This has produced consider-able turnover at the FEC. For example, threenew commissioners took office in July 1998, andfour took office in July 2008. All together, 13new commissioners have taken office since1998. Compare this to 15 different commission-ers serving between 1975 and July 1998.

Second, the FEC initiated a pilot program tohandle minor reporting violations more swiftlyand outside the MUR process. This is called theAdministrative Fines Program (AFP), and Con-gress authorized the change beginning on July14, 2000. Although the initial authorization ranout at the end of 2001, Congress has regularlyextended the program, most recently in De-cember 2008 (until the end of 2013)—an obvi-ous example of ex post control. Under the pro-gram, if a candidate, party, or PAC committeefails to file its reports on time, the Commissioncan automatically levy a penalty, although italso allows the respondent a chance to protestthe fine.12 By most impressionistic accounts,the structural change has been a success, mov-ing minor violations outside of the MUR pro-cess and allowing more time for more seriousand consequential enforcement cases (Smithand Hoersting, 2002).

Finally, the FEC established an additionalenforcement tool through the Alternative Dis-pute Resolution Office, which went into effect(initially as a pilot program) in November2000. This process (called ADR) allows for amediated approach to “minor and inadvertentviolations” (Smith and Hoersting, 2002, p. 4).13

According to the FEC’s 2005 annual report:“The program’s success in reaching its goal ofexpediting the resolution of enforcement mat-ters is evident in changes that have taken placeboth at the Commission and within the regu-lated community. For example, ADR has es-tablished a presence among the regulated com-munity, with members of the election barrequesting that matters be considered by theprogram [as opposed to the traditional MURprocess].”14

We might expect a number of different em-pirical patterns in the handling of MUR sincethese changes went into effect. I refer to thesepotential changes as the “structural change hy-pothesis.” First, because of both the AFP andADR programs, we should expect to see moreconciliation agreements and higher assessedpenalties in MUR since 2000. This is becausethe remaining enforcement cases are more con-sequential and substantive. Second, both pro-grams should also lead to a decrease in thenumber of dismissed cases and a reduction inthe total time spent on processed MUR. Bothtrends point to an MUR process that is moreefficient. Third, because of increased turnover

FRANZ172

12 The penalty is determined by a number of factors, in-cluding the number of previous late filings and the levelof activity on the report.13 According to Smith and Hoersting (2002): “The pro-gram exempts certain cases from consideration underADR, most notably matters also subject to criminal in-vestigations; allegations or prima facie evidence of know-ing and willful violations; violations of the PresidentialFund Acts; matters covered under a concurrent MURwithin the General Counsel’s Office; repeat offenders; re-spondents who fail to respond to Commission inquiries,and certain types of complex legal cases” (p. 4).14 It should be noted that the changes described abovecome after another major change prior to the time frameunder consideration here. In 1993 the FEC initiated an en-forcement prioritization program that allowed its Officeof General Counsel to preview all enforcement cases, giv-ing the Commission the ability to weed out more effec-tively complaints that appeared to have no merit (Mann2005, pp. 236–237).

THE FEC’S ROLE AS ENFORCER 173

in commissioners, and because the AFP andADR programs weed out minor enforcementmatters, we should see more contentious votesover time. (This is because the remaining, moresubstantial, cases might induce principled con-flict over whether a violation was actuallycommitted.)

Finally, with any agency-induced shift inprocedure comes the possibility of “bureau-cratic drift,” in which agency behavior divergesfrom the wishes of Congress (Epstein andO’Halloran 1994). With the discretion grantedto the FEC in implementing the ADR and AFP,it is conceivable that an incumbent or party ad-vantage has declined in recent years. Becauseof FEC term limits, however, any noted drift islikely to be minor; Congress would put mucheffort into appointing commissioners who fol-low Congressional wishes. This is in contrastto long-serving commissioners who mightmore gradually “drift” from their original man-date. As such, the final expectation of the struc-tural change hypothesis is that altered behav-ior has not adversely affected any respondentbias.15

DATA

The data for this article come from the FEC’sonline Enforcement Query System.16 The datawere coded between June and July of 2006, andthey include all Matters Under Review (MUR)listed on the web page at the time. This repre-sents 747 completed enforcement cases. Thevast majority of enforcement cases involved al-leged infractions in the elections of 1996–2004,but a few cases involved lingering issues fromolder elections.17

The data were coded on a number of impor-tant variables—including the names and identi-ties of the MUR complainants and respondents,the subject of the complaint, final penalties anddispositions on all respondents, and all of theCommission votes relative to the case. The dataallow for a number of different structures. Forexample, in one structure I disaggregated theMUR to the level of individual respondents.Some enforcement cases involve only one re-spondent, while a handful involves over 100. Themedian number of respondents per MUR is four.

Table 1 shows the distribution of the datadisaggregated to the level of the respondent;there are over 5,500 implicated in the 747 MUR.As we can see, over 55 percent of these re-spondents were non-candidate individuals(i.e., treasurers of campaigns and parties,CEOs, union leaders, and regular citizens). 18percent of the respondents were campaigns,while just under 6 percent were Democratic orRepublican Party committees (local, state, ornational). 9.5 percent were corporations.

The pattern demonstrates that the FEC is ag-gressive at holding responsible the individualwho committed the violation. For example, inMUR 5044, the FEC inadvertently discovered areporting violation by The Loose Group, a PACregistered with the Commission. The FEC pur-sued a conciliation agreement with the groupand its treasurer, Frank Stevenson, who wasconsidered responsible for physically makingthe report to the Commission. Both Stevensonand the group agreed to pay a $2,100 fine. InMUR 5647, the 2004 congressional campaigncommittee of Virginia Foxx (at the time a can-didate for an open seat in North Carolina, oneshe would win that year) agreed to pay a fineof $17,000 for reporting violations concerningmedia expenditures. Notably, her campaigntreasurer (Carolyn Aldridge) also was assessedthe fine.

The distributions in Table 1 are important.The FEC does not neglect individuals in en-forcement proceedings while pursuing organi-zations. This is at least modest evidence for theclaims of some that one weakness in the FECenforcement process is its “overdeterrence,” orits tendency to aggressively enforce the lawagainst “small fish” or for minor violations.

15 There is one alternative to this characterization of termlimit effects. In the pre-term limit phase, commissionersmay have attended to an incumbent and party bias toguarantee re-appointment. Once this incentive is nolonger present, term limited commissioners might con-ceivably be free to go after any and all election law vio-lators. If this alternative is true, we should expect someevidence of bureaucratic drift away from any respondentbias.16 The System can be accessed at �http://www.fec.gov/em/mur.shtml�.17 The FEC is in the process of posting information on allcompleted MUR, but the data here consist of all uploadedcases as of summer 2006.

Table 2 shows the distribution of com-plainants. I split these numbers in two ways—complainants as a percentage of the over 700MUR, and as a percentage of the over 5,500 re-spondents. Individuals who file a complaint ac-count for 32 percent of all MUR and 25 percentof all respondents.18 The discrepancy in thesenumbers demonstrates that “lone wolf” com-plainants tend to initiate enforcement mattersin cases involving a disproportionately lowernumber of respondents than other com-plainants.

Campaigns account for only 10 percent of allMUR (and 11.5 percent of all respondents)while Democratic or Republican committeesaccount for nearly 20 percent (and 28 percent

FRANZ174

TABLE 1. PROFILE OF RESPONDENTS

Respondent Frequency Percent

Individuals (non-candidates) 3,063 55.62

Campaigns or Candidates 1,005 18.24Incumbents 312 5.66Challengers 645 11.71

Democratic or Republican party committees 321 5.83National committees 115 2.10State committees 115 2.10Local committees 91 1.65

All other groups (i.e., advocacy groups, third parties) 594 10.7

Corporations 524 9.5

Total 5,507 100

TABLE 2. PROFILE OF COMPLAINANTS

With MUR as the With respondent as theunit of analysis unit of analysis

Complainant Frequency Percent Frequency Percent

Individuals (non-candidates)* 229 32.11 1,393 25.27

Campaigns or Candidates 70 9.82 635 11.52Incumbents 15 2.10 154 2.79Challengers 55 7.71 481 8.73

Democratic or Republican party committees 136 19.07 1,560 28.30National committees 61 8.56 880 15.97State committees 67 9.40 629 11.41Local committees 9 1.26 51 0.93

Corporations 10 1.40 178 3.23

Interest/advocacy groups 61 8.56 610 11.07

Third parties/other 22 3.01 97 1.76

FEC 171 23.98 958 17.38

Department of Justice 14 1.96 81 1.46

Total 713 5,512

*Refers to individuals acting without any other complainant.

18 I define individuals here as those without any overtconnection to an organization, campaign, or party. For ex-ample, in MUR 4895, George Anderson of Rome, Geor-gia filed a complaint against Congressman Bob Barr, alleging expenditure and reporting violations. Mr. Anderson had no evident affiliation with a candidate orcampaign.

THE FEC’S ROLE AS ENFORCER 175

of all respondents—making them dispropor-tionately more likely to initiate complaints incases involving larger numbers of respon-dents). The FEC initiated in nearly 24 percentof all MUR, but the Department of Justice ini-tiated only about 2 percent (most often refer-ring a potential civil violation discovered in acriminal investigation).

Table 3 shows the final judgment for almost5,300 respondents. Nearly 33 percent of all re-spondents had the complaint against them dis-missed, while 27 percent of the respondentcomplaints were resolved with “no reason tobelieve” the law was violated. 23 percent wereconsidered potential violations, but no furtheraction was taken, meaning no fine was levied.Altogether, just over 83 percent of the over5,000 respondents faced no financial penalty.This might strike many as too large a number,under the assumption that enforcement mat-ters should produce fines in a greater numberof cases. In contrast, 16 percent of respondentsvoluntarily entered a conciliation agreementwith the FEC and agreed to pay some fine. Avery small percentage of cases (less than 1 per-cent) were pursued further into federal court.

It should be noted that in cases involvingmany respondents, it is quite common for theFEC to find fault with some respondents andnot others. For example in MUR 5305, a caseinvolving 25 respondents, the FEC entered intoa conciliation agreement with six of them, levy-ing a combined fine of $159,000. Four other re-spondents were found not to have committedany violation, while fifteen had potentially vi-olated the law but no conciliation agreement

was reached. This diversity of enforcement res-olutions points in part to an FEC that astutelyinvestigates all dimensions of a complaint.More respondents mean more investigativework, and there appears to be little evidencethat the FEC is inclined to treat complaints withblanket judgments.

As for penalties, Table 4 shows the distribu-tion of final penalties levied against 740 re-spondents.19 28 percent pay less than $5,000,but nearly 30 percent pay over $50,000. Thesefindings stand in contrast to the MUR studiedby Lochner and Cain (1999). Their analysis of196 dispositions from a sample of MUR de-cided in the early 1990s showed that less than10 percent of respondents paid a fine of morethan $3,000. At the very least, these updateddata show that the FEC has been more willingto impose far higher fines in recent years.

In addition, the Enforcement Query Systemlists the case subject for each MUR. This subjectassignment is difficult to interpret, however, asthe FEC tends to use fairly broad categories toencompass a large diversity of potential viola-tions. For example, 55 percent of the 747 MUR

19 The distribution excludes one extremely high penalty—$3.8 million levied against Federal Home Loan MortgageCorporation in MUR 5390. The fine is so far in the tail ofthe distribution that I exclude it from the analysis that fol-lows. I also make no adjustments for shared penalties inthe table. In other words, if two respondents agree jointlyto pay $25,000, the penalty is counted twice. When I alternatively count shared penalties only once, the distri-butions are affected as follows: penalties of less than$30,000 account for 72 percent of the total (as opposed to63.5%).

TABLE 3. DISTRIBUTION OF DISPOSITIONS

Disposition Frequency Percent Cumulative

Dismissed 1,736 32.79 32.79

No reason to believe 1,411 26.65 59.44

Reason to believe/No further action 1,253 23.67 83.11

Conciliation 854 16.13 99.24

Court case authorized 35 0.66 99.90

Other 5 0.09 99.99

Total 5,294 100

*Total does not match Table 1 because some respondents have no listed disposition.

were coded as contribution violations, in someform or another. Without reading the specificsof each case, we cannot be sure if contributionrefers to an excessive in-kind gift, an actualcheck transfer, or an inappropriate coordina-tion. Therefore, the case subject classificationsare not nearly as fine as they could be. To com-pensate, I rely on three broad categories to clas-sify MUR—contributions, expenditures, andreporting violations. 55 percent involve contri-butions, 49 percent are about reporting vio-lations, and 17 percent concern expenditures(note that it is possible for a case to involve oneor all three of these, which is why the percent-ages do not add to 100).

The structure of the data also allows for dis-aggregating to the level of the Commissionvote. There were just under 1,500 separate FECvotes on all included MUR. The median num-ber of votes per MUR is one, but the average isjust above two. One important structuralchange in the FEC make-up included two pro-longed periods of time with only five commis-sioners (from October 1995 to August 1998 andAugust 2005 to January 2006). This was the con-sequence of a delay between a vacancy and Sen-ate approval of a replacement. Table 5 showsthe distribution of vote outcomes under condi-tions of five and six voting commissioners.

In both cases, a unanimous full member voteis the most common (6–0 votes occur 55 per-cent of the time with six commissioners, and5–0 votes happen 54 percent of the time withfive commissioners). Of course, with only fivevoting commissioners, there can be only onedissention, abstention, or absence to avoidstalemate, and this accounts for 41 percent ofthe five-member votes. With six members,however, a single vote deviation is less com-mon (24 percent), with just over 10 percent ofthe votes involving only four affirmative votes.

FRANZ176

TABLE 4. DISTRIBUTION OF PENALTIES

# of Final Penalty Dispositions Percent Cumulative

�$5,000 207 27.9 27.97

$5,000–$9,999 122 16.49 44.46

$10,000–$14,999 53 7.16 51.62

$15,000–$19,999 41 5.54 57.16

$20,000–$24,999 35 4.73 61.89

$25,000–$29,999 12 1.62 63.51

$30,000–$49,999 53 7.16 70.68

$50,000–$99,999 94 12.70 83.38

$100,000� 123 16.62 100

Total 740

*Total does not match the 854 conciliation agreements in Table 3 because some respon-dents have no listed penalty.

TABLE 5. DISTRIBUTION OF MUR VOTES

Five SixVotes Commissioners Commissioners

0-6 — 1 (�0%)1-4 0 (0%) 1 (�0%)1-5 0 (0%) 8 (0.5%)2-3 0 (0%) 6 (0.4%)2-4 — 14 (1%)3-1 5 (3.6%) 0 (0%)3-2 1 (0.7%) 6 (0.4%)3-3 — 32 (2.4%)4-0 32 (23%) 58 (4.3%)4-1 25 (18%) 36 (2.7%)4-2 — 44 (3.3%)5-0 75 (54%) 318 (23.6%)5-1 — 75 (5.6%)6-0 — 739 (55%)Total 138 1,342

THE FEC’S ROLE AS ENFORCER 177

Of course, the larger story of Table 5 is the in-frequency of 3–3 ties or any stalemate (3–2 or3–1 votes, for example).

Deadlocked votes are worth some brief dis-cussion, however. One might argue that dead-locked votes, while rare, involve highly con-troversial and political enforcement cases. Tosome extent, this is true. Of the 38 enforcementcases involving at least one failed vote, 10 in-volved a complaint initiated by Democratic orRepublican party committees, and 15 involvedrespondents who were party committees. 32 ofthe cases implicated a federal candidate in thealleged wrong-doing, and 14 of these werepresidential candidates. Respondents in dead-locked cases were also less likely to pay fines.16 percent of respondents in cases with nodeadlocked vote ended up entering a concilia-tion agreement; only 9 percent of respondentsin cases where at least one vote failed endedup paying a fine. Furthermore, of the 32 votesthat split 3–3, 19 involved Democratic com-missioners opposing Republican commission-ers, while 13 included at least one dissent fromboth parties. It is beyond the scope of this arti-cle to draw deep inferences into the causes andconsequences of these deadlocked votes, rareas they are, but they very likely signal to po-litical actors the campaign finance behaviorthat is most susceptible to regulatory and en-forcement delay or confusion.

LOCATING RESPONDENT BIAS

In this section I look for evidence of a re-spondent bias in MUR evaluations. To findsuch evidence, I ran two multivariate models.The first is a Heckman selection model that pre-dicts simultaneously both the probability thata respondent enters into a conciliation agree-ment and conditional on such an agreement,the final penalty levied by the FEC. The unit ofanalysis is the respondent (see Table 1). In theselection equation, then, the dependent vari-able is coded as 0 for all respondents where theFEC dismissed the charge, investigated butfound no reason to believe a violation tookplace, or found reason but chose not to pursuea penalty. The variable is coded 1 if the FECpursued a conciliation agreement and suc-

ceeded in convincing the respondent to pay afine. (Respondents whose cases were autho-rized for a civil suit, or who were handled insome other fashion—see Table 3—are not in-cluded in this analysis.)

The second model is a binomial regression inwhich the unit of analysis is each vote and thedependent variable is the number of yes votesout of six commissioners voting.20 With thismodel, the initial challenge is correctly classi-fying vote outcomes. As Table 5 shows, five affirmative votes might either imply one dis-sention, one abstention, or one absence. Simi-larly, four affirmative votes might imply twoof those possibilities or some combination ofthem. Should I treat a 4–2 vote differently froma 4–1 vote with one absence or a 4–1 vote withone abstention? In the end, because there is solittle variation within these categories I can ef-fectively treat the number of affirmative votesas the same (regardless of the reasons for two,three, four, or five yes votes). This also makessome sense theoretically. We do not know whya commissioner chooses to abstain from a par-ticular vote, or what the difference in motiva-tion is for a commissioner to dissent or simplynot show up. Furthermore, because the finalnumber of yes votes is what determines theoutcome (and commissioners know this in ad-vance), the more important question is whatdrives yes votes up or down.

In both models I control for a number of vari-ables that might predict values on the depen-dent variables. For example, I specify eachmodel with seven complainant binary vari-ables; specifically, I consider whether the com-plainant was an incumbent candidate, chal-lenging candidate, national or state partycommittee, the FEC, an advocacy group, or anindividual. This variable is measured at thelevel of the MUR, so they take on a value of 1if a respondent was implicated in an MUR ini-tiated by any of the above. Second, I includesix respondent binary variables; specifically, Iconsider whether the respondent was an in-cumbent or challenger, a national or state party

20 I drop all votes where there were only five sitting com-missioners (as shown in Table 5). I discuss the results ofa 5-commission vote model in the text below.

committee, a corporation, or an individual. Seeagain the distributions in Tables 1 and 2; by in-cluding these variables, the base category foreach is all other respondents and complainants.

If incumbent and party complainants are fa-vored, we should expect to see a stronger co-efficient in both stages of the Penalty model—as their complaints are taken more seriously. Ifincumbent and party respondents are favored,we should see significant negative coefficientsin both stages—as their cases are resolved moreleniently.

I also include three variables for the case sub-ject of each MUR. In particular, I specify themodels with three binary variables—was therespondent implicated in a case involving a po-tential contribution violation, a reporting prob-lem, or excessive or potentially illegal cam-paign expenditures? These variables are notmutually distinct, as some respondents are al-

leged to have violated two or all of these cate-gories.

Finally, I account for the total number of re-spondents implicated in each MUR (to accountfor complaints where a respondent is groupedwith others), the year of the initial vote on theMUR (or the year of the vote in the Votemodel), the total number of votes taken on theMUR (only in the Penalty model), whether thefinal penalty was shared across respondents(only the second stage of the Penalty model),and whether the vote outcome was the first orlast vote (only the Vote model).

The results of the multivariate models are re-ported in Tables 6 and 8. Consider first the re-sults in Table 6. Two results in particular pointto possible “overdeterrence” issues—an out-come not specifically predicted in my hy-potheses, but a common charge leveled at theCommission (Lochner 2003). Salisbury (2002),

FRANZ178

TABLE 6. PREDICTING CONCILIATION AGREEMENTS AND PENALTIES

Probability(Conciliation) Penalty

Year of initial vote 0.012 (0.024) 8,630.30 (2,755.94)**Penalty shared 92,341.98 (19,914.76)**Total respondents �0.012 (0.005)* 4,009.72 (772.21)**Number of votes 0.053 (0.009)** �3,809.68 (914.56)**

Complainant typeIncumbent 0.005 (0.332) �38,464.59 (33,052.17)Challenger �0.746 (0.265)** �35,119.31 (32,217.97)National Party �1.051 (0.272)** 19,057.82 (37,288.41)State Party �0.633 (0.216)** 1,351.22 (33,837.41)FEC/DOJ 0.657 (0.209)** 12,240.22 (28,557.15)Interest/advocacy group �0.514 (0.269)� �143,812.50 (49,430.55)**Individual �0.689 (0.237)** �10,026.73 (31,208.65)

Respondent typeIncumbent 0.035 (0.151) �21,970.85 (14,687.38)Challenger 0.439 (0.109)** �31,323.83 (12,825.90)*National Party �0.166 (0.237) 96,215.71 (33,225.58)**State Party 0.778 (0.142)** �13,149.55 (16,408.94)Corporation 0.227 (0.202) 56,646.83 (26,467.11)*Individual 0.136 (0.107) �14,853.59 (10,561.90)

Case SubjectContribution �0.224 (0.107)* 4,409.60 (12,906.50)Expenditure �0.511 (0.140)** �11,805.07 (17,136.59)Reporting 0.001 (0.110) �23,437.46 (14,172.59)+Constant �23.676 (48.146) �17,300,000 (5,513,210)

N 4958 718

� p�.10 *p�.05 **p�.01Entries are coefficients and (standard errors) from both stages of a Heckman selection model. Respondents are

only included in the right column estimation if they entered a conciliation agreement.

THE FEC’S ROLE AS ENFORCER 179

for example, notes that “the historical reputa-tion of the Commission [is] as a ‘nit picker’” (p.131). The results here show that as the agencyexpends more time on a case (measured byproxy as the number of votes on the MUR), re-spondents are more likely to enter a concilia-tion agreement, but they ultimately pay a lowerfine than in cases with fewer Commissionvotes. Secondly, MUR involving reporting re-quirements are more likely than cases involv-ing contributions or expenditures to result in aconciliation agreement, but such cases result inlower fines.

Both patterns suggest a skew of FEC re-sources; penalties are lower for the types ofcases (reporting requirements) most likely toreach conciliation, and for enforcement matterscommanding a good deal of Commission ef-fort. Clearly, many in the reform community(as well as inexperienced candidates and PACswho fear making minor mistakes) would wishfor an opposite pattern—one in which theCommission spends most of its time on casesthat result in higher fines and on pressing mat-ters that most threaten the integrity of the cam-paign finance system.

How do the results conform to the respon-

dent bias hypothesis? This is seen a bit moreclearly in Table 7, in which I show predictedeffects from the Penalty model under variouscomplainant-respondent match-ups. In gen-eral, the results point to possible respondentbias. First, in terms of entering conciliationagreements, both incumbents and nationalparty committees fare far better as respondentsthan challengers, state parties, corporations, orindividuals. Across any complainant type, in-cumbent and national party respondents havethe lowest probability of entering a conciliationagreement. For example, when advocacygroups file complaints against challengers andcorporations, they have a respective 0.26 and0.20 chance of securing a conciliation agree-ment. Those probabilities for complaintsagainst incumbents and national parties are0.15 and 0.11.

Still more, incumbents as complainants havea far higher probability of securing a concilia-tion agreement. Incumbents tend to go afterchallengers, corporations, and individuals (andalmost never accuse other incumbents or partycommittees), and they succeed anywhere from1 in 3 complaints to nearly half the time, de-pending on the respondent.

TABLE 7. PREDICTED PROBABILITY OF A CONCILIATION AGREEMENT AND PREDICTED PENALTY

Respondents

National Complainants Incumbents Challengers parties State parties Corporations Individuals

Incumbents 0.45 0.37 0.34$66,951 $154,922 $83,421

Challengers 0.10 0.19 0.07 0.14 0.12$79,649 $70,297 $197,836 $158,267 $86,767

National parties 0.06 0.12 0.04 0.20 0.08 0.07$133,827 $124,474 $252,013 $142,648 $212,444 $140,944

State parties 0.12 0.22 0.09 0.34 0.17 0.14$116,120 $106,767 $234,307 $124,941 $194,738 $123,237

FEC/DOJ 0.55 0.70 0.47 0.81 0.63 0.59$127,009 $117,656 $245,196 $135,830 $205,627 $134,126

Individuals 0.11 0.21 0.08 0.32 0.15 0.13$104,742 $95,389 $222,929 $113,563 $183,360 $111,859

Advocacy Groups 0.15 0.26 0.11 0.20 0.17�$29,043 �$38,396 $89,143 $49,574 �$21,925

*Blank entries are for instances where there are less than 10 cases of such pairings in the data**Penalties are estimated from results in Table 6 for a conciliation agreement where the final fine is shared among

respondents.

Compare also the difference between in-cumbents and challengers. Incumbents have a0.45 probability of forcing a conciliation agree-ment against challengers. But challengers haveonly a 0.10 probability of forcing a conciliationagreement against incumbent respondents.When incumbents accuse corporations of a vi-olation, they have a 0.37 chance of forcing apenalty, but challengers have only a 0.14chance in the same situation.21

Both sets of results do suggest a Commissioninclined to favor incumbents and national par-ties. This is potentially damaging evidence thatthe Commission is not achieving its mission ofnon-biased enforcement. Some other patterns,however, belie conclusive evidence of bias. For example, when national parties force aninvestigation they are significantly less likely to see the complaint reach the penalty phase.Across any respondent type, in fact, the pre-dicted probability of a conciliation agreementis lowest when national party committees arecomplainants. Additionally, national partycommittees pay some of the largest fines. (Cor-porations pay the next largest fines.) As for in-cumbents who enter conciliation agreements,they pay roughly the same fines as chal-lengers.22

All told, the evidence is only moderately sug-gestive of a respondent bias. In other words,the success of incumbents and national partiesat more strongly resisting conciliation agree-ments (and incumbents’ success at forcingpenalties when they initiate) could point to anFEC that favors them. It could also indicate,however, that the incumbents and national par-ties are simply more efficient campaigners andless prone to making glaring mistakes. Con-sider again the evidence from Tables 1 and 2.Together, incumbents and national party com-mittees represent only about 7.5 percent of allrespondents. Challengers and state/local par-ties account for about 14 percent of all respon-dents. Perhaps the more experienced cam-paigners have less to fear from the FEC notbecause of a regulatory bias, but because oftheir prior expertise as campaigners.23

On the other hand, however, the “overde-terrence” evidence offered above could pointto a less overt respondent bias. Because of theex post oversight of Congress, one that pushes

the FEC to achieve more conciliation agree-ments, it’s possible that the agency has com-petently responded by going after weakercrimes more likely committed by more inexpe-rienced candidates and political actors.

We can also assess potential respondent biasby examining how the FEC votes on certain en-forcement matters. The hypothesis of respon-dent bias in this realm expects MUR involvingincumbents and national parties to induce con-sensus, as Commissioners jointly work to helpthese beneficiaries avoid penalties or concilia-tion agreements. The results of the binomial re-gression model are displayed in Table 8. Theevidence is completely unsupportive of the ex-pectation. In fact, votes involving incumbentsand national party committees are more con-tentious than votes involving challengers, stateparties, corporations, or individuals.

Furthermore, we see some evidence of aCommission that operates with a good degreeof consensus on other matters. For example, if

FRANZ180

21 It should be noted that these incumbent/challenger dif-ferences contrast with findings presented by Lochner etal. (2008), who find no distinctions among candidates asto the probability of paying a fine. They have only 287candidates in their model, however, because they excludeMUR between July 2000 and December 2001. My modelincludes 931 candidates between 1996 and 2006.22 These results also help us sort out the ability of the FECto respond to what Ginsberg and Shefter (1990) call the“RIP” strategy. This is where partisans in contemporarypolitics tend to pursue a tactic of “revelation,” “investi-gation,” and “prosecution” to hurt political opponents.The enforcement procedure at the FEC is ideal for such atactic. Imagine a candidate launching a complaint againsthis opponent so as to secure publicity; the slow turn-around time at the Commission insulates the complainantfrom embarrassment should the matter be dismissed. Allelse equal, then, with press coverage as the goal, politi-cally-motivated MUR should have the least merit. If theFEC is more independent in its response (meaning it re-sists being a party to such tactics), these politically-moti-vated cases should be less likely to reach the conciliationphase. This is true for national parties especially. Fur-thermore, because national party committees pay suchlarge fines and FEC-initiated MUR are the most success-ful, we might interpret the evidence as suggestive of acompetent FEC, not necessarily one compliant with theRIP tactic.23 Some might argue that the bias is located earlier in theprocess, at the enforcement priority stage (when the OGCweeds out trivial cases). This would put the mechanismof bias, however, at the stage in which the Commissionis generally viewed as most aggressive and professional.

THE FEC’S ROLE AS ENFORCER 181

the vote is the final one on the MUR (that is,on the vote to close the case), the number of yesvotes goes up. And if the vote is on a case yield-ing a conciliation agreement (the base categoryin the “most severe action” subsection of inde-pendent variables), the number of yes votesgoes up. This indicates that the Commissioncloses cases and levies fines with little publicconflict. (It should be noted that because con-flict behind closed doors is unobserved, it iscertainly possible that these consensus votesare the consequence of considerable compro-mise and debate prior to the vote. Even if suchsituations are common, however, it is notablethat the Commission can resolve them withunanimity in so many instances.)

On the other hand, the FEC is most overtlycontentious in cases initiated by national par-ties, in cases involving incumbents and na-tional parties as respondents, and when the fi-nal action is to find reason to believe a violationoccurred but to take no action. Consider thepredicted yes votes in Figure 1. When the FECinitiates a MUR against a corporate respondentand secures a conciliation agreement, the aver-age number of yes votes on the case is over5.5—near consensus. When an incumbentpushes a case involving a challenger, however,and the case is ultimately dismissed, we can ex-pect about one dissention or abstention. This issimilarly true for a challenger complaint filedagainst an incumbent. In the final case, when anational party files a complaint against theother party, and the FEC chooses not to pursuea conciliation agreement but still recognizes apotential violation, the average vote is only4.4.24

We should derive two important inferencesfrom these results. First, even in the most po-litical of enforcement matters, the FEC is almostnever handcuffed by its bipartisan structure.More to the point, in almost no cases is the pre-dicted number of votes below 4 (the necessarymajority to take action in any enforcement mat-ter). Second, the evidence is a refutation of therespondent bias expectation. Cases involvingincumbents and national parties do not induceconsensus, but conflict.

We can assess this from a different perspec-tive by looking for a partisan relationshipamong commission “no” votes. In other words,in contentious cases involving candidates or na-tional parties, do Democratic and Republicancommissioners support respondents and com-plainants of their own party? Of the 1,342 votesin the data set involving all 6 commissioners(see again Table 5), there are 254 votes withsome level of Republican dissent but no Demo-cratic objections; 222 votes where a Democratic

TABLE 8. PREDICTING COMMISSION VOTES

Number of Yes Votes

Year of vote �0.056 (0.015)**Total respondents �0.002 (0.002)First vote on MUR 0.058 (0.146)Last vote 0.810 (0.109)**

Most severe actionDismissal �0.339 (0.156)*No reason to believe �0.316 (0.155)*Reason to believe/no action �0.660 (0.108)**

Complainant typeFEC/DOJ �0.287 (0.186)Incumbent �0.426 (0.264)Challenger �0.279 (0.221)National Party �0.386 (0.200)+State Party 0.068 (0.213)Interest/advocacy group �0.151 (0.205)Individual �0.204 (0.188)

Respondent typeIncumbent �0.437 (0.126)**Challenger �0.248 (0.105)*National Party �0.714 (0.103)**State Party �0.026 (0.117)Corporation 0.120 (0.095)Individual 0.101 (0.086)

Case subjectContribution �0.188 (0.091)*Expenditure 0.173 (0.110)Report 0.128 (0.085)Constant 113.91 (30.77)

N 1297

�p�.10 *p�.05 **p�.01Entries are coefficients and (standard errors) from a

generalized linear model with a binomial link function.Dependent variable is the number of commissioner “yes”votes.

24 I re-ran the Vote model in Table 7 for the two time pe-riods when the Commission had only five votes. Becausethis involves only about 140 votes, I do not report the re-sults in a table. All told, the results are only weakly sug-gestive. All of the effects from Table 8 drop out. It thusappears that Commission votes during an imbalance pe-riod are more idiosyncratic.

commissioner voted “no” but all GOP commis-sioners voted affirmatively; and 121 voteswhere commissioners from both sides regis-tered a dissent. (As with the previous analysis,any means of not voting “yes”—a dissention,recusal, or absence—is treated as a “no.”)

I ran two models predicting the presence ofa dissenting Republican or Democratic vote.The model specification is identical to the oneshown in Table 8, but I also added dummy vari-ables for Democratic and Republican com-plainants and respondents.25 Table 9 showspredicted probabilities in four instances, de-pending on the party of the complainant or re-

spondent and the presence or absence of a con-ciliation agreement (holding all other controlvariables constant).

The results demonstrate a clear partisan pat-tern. For example, regardless of the presence of

FRANZ182

TABLE 9. PREDICTED PARTISAN DISSENT AMONG COMMISSIONERS

Prob.(Democratic dissention)

No fine in FineMUR levied No fine Fine

Republican complaint/ 0.38 0.39 0.28 0.22Democratic respondent

Democratic complaint/ 0.22 0.18 0.15 0.20Republican respondent

Estimates are from two separate models predicting the probability that a Republican orDemocratic commissioner fails to vote “yes.” Full model results are not shown, but areavailable from the author on request. See text for more discussion of the model specifica-tion.

Prob.(Republican dissention)

FIG. 1. Predicted commissioner votes. RTB/NFA means the final action on the case was a “reason to believe” a violation occurred, but “no further action” was taken to pursue conciliation agreement. Estimates are from the resultsin Table 8.

25 I also ran the models as a binomial regression, as inTable 8; the substantive results are the same. For ease ofdiscussion, and because the vast majority of cases involveonly one Democratic or Republican dissention, I preferthe dichotomous dependent variable. The partisan inde-pendent variables are for respondents or complainantswho are either federal candidates or national party com-mittees.

THE FEC’S ROLE AS ENFORCER 183

a fine, Democratic commissioners are far morelikely to object in enforcement cases initiatedby a Republican (at about twice the rate ofwhen Democrats file the complaint). More still,the likelihood of a Democratic dissention isgreatest when a Republican complaint elicits afine against a Democratic respondent, and leastwhen a Democratic complaint secures a finefrom a Republican respondent. For Republicancommissioners, the partisan affiliation of therespondent and complainant matter little whena fine is involved (see the right column of re-sults), but the differences are stark when nofine is levied. On that score, Republican com-missioners dissent least often when Democra-tic complaints against Republicans result in nofine, but at twice the rate in the opposite case.

All told, the evidence in this section is mixedwhen it comes to the expectations of respon-dent bias. While incumbents and national par-ties see some advantage (over other respon-dents) in terms of their success as respondents,it does not come without some level of conflictin Commission votes. More still, that conflict isat least partly driven by the partisan disposi-tion of the commissioners. The vote patternsspecifically suggest that the FEC fails as anagency designed, as critics cynically suggest, todo incumbents and national parties the leastharm. I think it safe to conclude as a conse-quence that experienced political actors bene-fit in the MUR process not from an explicitCommissioner conspiracy but from the specificadvantage of knowing best the rules of thegame.

ASSESSING STRUCTURAL REFORM

Is there evidence of the structural reform hy-pothesis? Recall, this hypothesis expects con-ciliation agreements and realized penalties togo up in recent years and dismissals to godown. It also expects more contentious votes atthe FEC as the agency sees more turnover incommissioners and devotes more attention toconsequential enforcement actions.

The strong effect of the “Year” measure inthe second phase of the Penalty (Table 6) modelis initial persuasive evidence in favor of thestructural change hypothesis. The results indi-

cate that the FEC levies significantly higherpenalties in conciliation agreements brokeredmore recently. All else being equal, the pre-dicted fine for a conciliation agreement in 1996was $40,407 (for a fine shared by at least tworespondents); the predicted fine for an identi-cal agreement in 2006 was $118,080.26 There isalso evidence that the FEC in recent years hasspent more time on cases with at least somemerit. Table 10 shows that the FEC now dis-misses fewer cases. For example, for all MURinitiated in 1999 or before, the FEC dismissednearly 50 percent. This number steadily de-clined to only about 12 percent of all MUR ini-tiated in 2004, though it increased in the fol-lowing two years.

At the same time, however, the results in thefirst phase of the Penalty model do not showan increased tendency for MUR to result in con-ciliation. Table 10 shows the by-year break-down, which also highlights a jump in “no rea-son to believe” and “reason to believe”evaluations. Lochner et al. (2008) make a simi-lar claim about the steadiness of conciliationagreements (see their Table 5, p.227), and theycare very little about judgment distributionsamong cases with no fines.

How do we interpret this evidence? Indeed,we might establish increased efficiency onlywhen conciliation agreements are more fre-quent (indications of an FEC focused only oncases that merit a penalty) and when outrightdismissals are less common. To that effect, theFEC has only accomplished the latter. More op-timistically, of course, at least the Commissionhas winnowed down the number of purelyfrivolous cases in the MUR process. And per-haps more importantly, penalties are signifi-cantly higher—a pattern that alone might sig-nal to political actors a more forcefulCommission.

Shifting to the vote analysis, the significanceof “Year” in Table 8 indicates a more con-tentious Commission—another implication ofthe structural reform hypothesis. All else beingequal, the number of “yes” votes on a typical

26 These predicted dollar figures are in real dollars, buteven when adjusting for inflation, the statistical effect isstill present and strong.

MUR vote declined by just over 0.5 votes be-tween 1996 and 2006. This is unlikely to makestalemates more common, but consider this in-creased conflict in the context of the predictedvotes from Figure 1. On FEC enforcement mat-ters involving party complaints and party re-spondents (the most contentious cases), in-creased conflict in Commission voting couldmore often result in stalemate (or at least comeclose). (It should also be noted that there isanecdotal evidence that the FEC has becomemore partisan since the appointment of newcommissioners in 2008. A number of high pro-file Advisory Opinions and MUR stalemated inlate 2008 and early 2009. Rick Hasen haspointed this out on his blog, electionlaw-blog.org).

Finally, the structural change hypothesis ex-pects little evidence of “bureaucratic drift.”More specifically, as the FEC is given more timeand energy to investigate higher priority en-forcement matters, it is possible that the agencywould grow more aggressive in its enforce-ment (turning more watchful eyes at incum-bents and national parties). But as commis-sioners now are selected for only one term,increased turnover should allow Congress to

select commissioners unlikely to be overly ag-gressive. I looked for any temporal change inthe Penalty phase model by splitting the sam-ple into pre and post ADR/AFP time frames.In both the pre and post phases, the effects forincumbents and national parties are un-changed. Even with certain structural changesin the MUR process, then, both are still dis-proportionately more successful.

It should be noted that we can only make acautious inference about how structural re-forms relate to the total time spent on eachMUR. A scatter plot (not shown here) of the to-tal number of days between the first and lastvotes on an MUR on the date of the initial voteshows a relative decline in the total time de-voted.27 This might imply that the FEC’s struc-tural reforms have sped up the enforcementprocess. However, it is also possible that a sig-nificant number of enforcement matters are on-going, and it is highly likely that these ongo-

FRANZ184

TABLE 10. DISTRIBUTION OF FINAL DISPOSITION BY YEAR

Reason No reason to believe/

Year Dismissal to believe No fine Conciliation Total

pre-1999/1999 675 233 230 242 138048.91% 16.88% 16.67% 17.54%

2000 351 164 223 99 83741.94% 19.59% 26.64% 11.83%

2001 222 277 200 95 79427.96% 34.89% 25.19% 11.96%

2002 114 98 124 147 48323.60% 20.29% 25.67% 30.43%

2003 196 298 126 97 71727.34% 41.56% 17.57% 13.53%

2004 57 136 206 88 48711.70% 27.93% 42.30% 18.07%

2005/2006 107 202 47 47 40326.55% 50.12% 11.66% 11.66%

Total 1,722 1,408 1,156 815 5,101

Percents are row percentages for each year.

27 A lowess fit of the data shows that earlier cases (initi-ated before 2000) often took over 500 days on average forcompletion. In later years that number was essentially cutin half.

THE FEC’S ROLE AS ENFORCER 185

ing MUR are disproportionately those initiatedin the last few years. Excluding them (which isa necessary function of the MUR process—onlyconcluded cases are made public) limits whatwe can definitively know about whether thecurrent system is more time efficient.

Unfortunately, though, time efficiency is animportant indicator of an enforcement processthat works. Lochner (2003) writes, “Theoreti-cally, a large fine might provide a strong mea-sure of deterrence if it were levied shortly be-fore an election in a way that significantlydepleted a candidate’s war chest or preventeda PAC from making a timely contribution to afavored candidate. In reality, however, the timelag between detection of a possible infractionand the punishment imposed—to say nothingof the time lag between commission of an in-fraction and its detection—is quite large” (p. 5).Determining the extent to which structural re-forms can help with this problem is an impor-tant focus of future research.

DISCUSSION

The evidence in this article suggests an FECmore functional than many often claim. It istrue that incumbents and national parties oftenhave little to fear with the MUR process. Thisalone may be enough to justify continued op-position to the FEC. But I contend that the re-sults are by no means an unequivocal affirma-tion of large-scale bias. Instead, the agencyfollows through most consistently on its owndiscoveries of wrong-doing (see the predictedeffects in Table 7); it secures large fines fromcorporations and national parties, and fargreater fines in the wake of structural reforms;and it almost never deadlocks. It seems likelythat incumbents and national parties are sim-ply better and more efficient campaigners, lessprone to making mistakes—and that this is thesource of the empirical patterns noted in thisarticle.

A number of proposals have been consideredto reform or overhaul the FEC. These include:

• Reducing the FEC from six members toone powerful administrator

• Changing the current structure of a rotat-

ing chair to a more permanent, more pow-erful chair

• Using administrative law judges ratherthan partisan commissioners

All of these plans are premised on the need fordeep change in the way election laws are en-forced, and they should all be taken seriously.But with deep changes come also a requirementto handle sticky normative challenges. Astronger FEC, for example, runs the risk of pun-ishing innocent campaigns and candidates (arisk that goes up if the FEC is compelled to re-solve election-related matters before elections).A single administrator would also have signif-icant powers in and over American elections.A compelling argument can be made that anyregulatory control in the area of electionsshould be subject to serious restrictions ofpower—lest the regulatory agency overreachwith its enforcement. After all, elections are themechanism through which power is asserted indemocracy. If a non-elected bureaucrat wieldssignificant powers in this area (one meant toscare candidates into compliance), it might cre-ate more problems than it solves. Of course, aweaker FEC runs the risk of allowing too muchbehavior that might violate existing electionstatutes—and this must be guarded against as well.

Not to be forgotten in this debate is the roleof sincere political differences over the scope ofcampaign finance laws. In fact, a significantnumber of “infractions” are only such in theeyes of some. For example, it is not clear that527s should register with the Commission asPACs; or that “issue ads” are really election-re-lated. What is an issue ad, after all? Even theSupreme Court has recently recognized the am-biguity in this debate, along with the challengeof establishing regulations that respect FirstAmendment rights.28 There are real questionsalso about appropriate limitations on Internet

28 In FEC v. Wisconsin Right to Life 551 U.S. 449 (2007), theSupreme Court ruled that to protect First Amendmentrights of free speech and association, some ads aired closeto Election Day should not be classified as “electioneer-ing” if a reasonable person can interpret the message assomething other than an unambiguous appeal to vote foror against a candidate for office.

campaigning and blogging; and in many otherareas.

Some will propose significant reforms, pre-ferring to stem the exploitation of loopholes.That is a legitimate stand to take. Some will seemore limited need for such aggressiveness, pre-ferring less bureaucratic oversight in the realmof elections. This is also legitimate. But the factthat those at one end of the spectrum of a prin-cipled debate are dissatisfied is hardly indica-tive of a weak Commission.

The data here do not resolve these questions,but they highlight that some regulatorychanges can prove fruitful. In the last 20 years,the FEC has gotten better at levying higherfines and expending fewer resources on frivo-lous cases. Keep in mind this comment fromelection lawyer Bob Bauer:

The first question in judging the effective-ness of enforcement is what we wish tohave enforced. On the basics, the law hasbeen plentifully enforced. Contributionslimits are enforced; disclosure require-ments are enforced; most, for that matter,of the Code of Federal Regulations imple-menting the federal law is enforced. Notalways within the same election cycle asthe occurrence of the violation, and not al-ways with the degree of mercilessness inthe setting of fines that critics call for: but,still, enforced.29

The best possible plan to reform the FEC maynot be a complete re-working of the Commis-sion (or its elimination), but a minor re-toolingof it. On that score, the focus should be a con-tinued trend towards efficiency at the Com-mission. The simplest reform is one that hasbeen proposed for years: return to the FEC itspower to conduct random audits. With it, I pro-pose doubling the Commission’s budget to hiremore staff. This would also help the Commis-sion increase its turn-around time in process-ing enforcement cases. I agree with many thatquickening the enforcement process should bea priority.

Such “tweaking” is of a piece with others’recommendations. For example, Todd Lochner(2003) proposes a “scarlet letter” approach to

FEC infractions, which would allow the FEC touse local media to advertise election law viola-tors. He recommends little else by way of dras-tic restructuring. In fact, in his discussion of“underdeterrence” at the FEC—this being theinability to stem or punish serious election lawviolations—Lochner argues that any shift to-ward more aggressive enforcement is likely tocompel political actors to “strategically shift”to campaign practices outside the scope of elec-tion laws, further weakening the very founda-tion of regulated elections.30

The results here are clear. Incumbents are in-frequently accused of wrongdoing (see againthe distribution of respondents in Table 1), andunlikely to face conciliation agreements (Table7). But as noted, this likely has more to do withthe professionalism of incumbents than it doesthe bias of commissioners. This should also ex-plain the lack of “bureaucratic drift” in recentyears toward more penalties inflicted on in-cumbents and national party committees.

In short, the random audit power is unlikelyto put incumbents and national parties at seri-ous risk, but it should give all political com-mittees some additional pause in seeking toskirt the law. Lochner et al. (2008, p. 231) won-der, though, if random audit power wouldhave any effect on the frequency of outside-ini-tiated complaints, which they see as a signifi-cant drag on efficient FEC enforcement. This isa realistic concern. So long as the FEC is forcedto expend enforcement resources on frivolousor politically-motivated complaints, efficiencywill be reduced. Nonetheless, granting the au-thority to randomly audit regulated commit-tees seems far more feasible than many largescale plans that would likely be stalled in Con-gress. Keep in mind: a larger budget would alsohelp the Commission prioritize and process en-

FRANZ186

29 Accessed at �http://www.moresoftmoneyhardlaw.com/�. August 21, 2008.30 There are other small reforms that might bear fruit aswell. For example, Congress might mandate that anyagency commissioner have a background in enforcementor administrative oversight (see Democracy 21’s report onthe FEC, p. 61). I am more persuaded by such small re-forms than I am by grand projects that seem likely to cre-ate a whole host of new problems once implemented.

THE FEC’S ROLE AS ENFORCER 187

forcement cases more quickly—and weed outpolitically motivated ones.

In the end, campaign finance reformers havetwo tasks when seeking to reformulate regula-tory politics. One must balance the political fea-sibility of any solution with a reform that cre-ates as few new problems as possible. The FECmay indeed be the devil we know, but as itseeks to improve (and shows realized signs ofdoing so) it may still end up being better thanany devil we don’t.

REFERENCES

Democracy 21 Task Force. 2002. “No Bark, No Bite, NoPoint.” Accessed at �http://www.democracy21.org/�.August 23, 2008.

Drinkard, Jim. “Agency that Referees Elections ProtectsParties First,” USA Today, 11/12/2002, p. A1.

“Editorial,” San-Diego Union Tribune, 7/21/2003, p. B6.“Editorial,” The Washington Post, 7/11/2003, p. A20.Epstein, David and Sharyn O’Halloran. 1994. “Adminis-

trative Procedures, Information, and Agency Discre-tion,” American Journal of Political Science. 38(3): 697–722.

Franz, Michael. 2008. Choices and Changes: Interest Groupsin the Electoral Process. Philadelphia: Temple UniversityPress.

Ginsberg, Benjamin and Martin Shefter. 1990. Politics byOther Means: The Declining Importance of Elections inAmerica. New York: Basic Books.

Gross, Kenneth. 1991. “The Enforcement of Campaign Fi-nance Rules: A System in Search of Reform,” Yale Lawand Policy Review. 9(2): 279–300.

La Forge, Amanda. 1996. “The Toothless Tiger—Struc-tural, Political and Legal Barriers To Effective FEC En-forcement: An Overview and Recommendations,” Ad-ministrative Law Journal. 10: 351–383.

Lochner, Todd. 2003. “Overdeterrence, Underdeterrence,and a (Half-Hearted) Call for a Scarlet Letter Approachto Deterring Campaign Finance Violations,” ElectionLaw Journal. 2(1): 23–41.

Lochner, Todd, Dorie Apollonio, and Rhett Tatum. 2008.“Wheat from Chaff: Third-Party Monitoring and FECEnforcement Actions,” Regulation and Governance. 2(2):216–233.

Lochner, Todd and Bruce Cain. 1999. “Equity and Effi-cacy in the Enforcement of Campaign Finance Laws,”Texas Law Review. 77 (7): 1890–1942.

Mann, Thomas. 2005. “The FEC: Administering and En-forcing Federal Campaign Finance Law,” in CampaignFinance Reform: A Sourcebook (Anthony Corrado,

Thomas E. Mann, Daniel R. Ortiz, Trevor Potter, andFrank J. Sorauf, eds.), Washington, D.C: The BrookingsInstitution.

McCain, John. “Paying for Campaigns: McCain Eyes NextTarget,” USA Today, 11/4/2004, p. 27A.

McCubbins, Matthew. 1999. “Abdication or Delegation?Congress, the Bureaucracy, and the DelegationDilemma,” Regulation. 22(2): 30–37.

McCubbins, Matthew, Roger G. Noll, and Barry R. Wein-gast. 1987. “Administrative Procedures as Instrumentsof Political Control,” Journal of Law, Economics and Or-ganization. 3(2): 243–277.

McCubbins, Matthew, Roger G. Noll, and Barry R. Wein-gast. 1989. “Structure and Process, Politics and Policy:Administrative Arrangements and the Political Controlof Agencies,” Virginia Law Review. 75(2): 431–482.

McCubbins, Matthew and Thomas Schwartz. 1984. “Con-gressional Oversight Overlooked: Police Patrols vs. FireAlarms,” American Journal of Political Science. 28(1):165–179.

Oldaker, William. 1986. “Of Philosophers, Foxes, and Fi-nances: Can the Federal Election Commission Ever Doan Adequate Job?” Annals of the American Academy ofPolitical and Social Science. 486: 132–145.

Salisbury, Julian. 2002. “U.S. Campaign Finance Regula-tion: An Examination of the Federal Election Commis-sion Enforcement Process,” Unpublished Dissertation.Accessed at �http://www.fec.gov/agenda/agendas2003/notice2003-09/comments.shtml�. August 1, 2008.

Smith, Bradley and Stephen Hoersting. 2002. “A Tooth-less Anaconda: Innovation, Impotence And Overen-forcement At The Federal Election Commission,” Elec-tion Law Journal. 1(2): 145–171.

“Symposium on the Federal Election Commission.” 1994.Journal of Law and Politics. 10: 369–412.

Thomas, Scott and Jeffrey Bowman. 2000. “Obstacles toEffective Enforcement of the Federal Election Cam-paign Act,” Administrative Law Review. 52(2): 575–608.

Walsh Edward and Juliet Eilperin, “FEC’s Structure is ’In-effectual,’ Critics Say”, The Washington Post, 5/13/2002,p. A2.

Wood, B. Dan and John Bohte. 2004. “Political Transac-tion Costs and the Politics of Administrative Design,”The Journal of Politics. 66(1): 176–202.

Address reprint requests to:Dr. Michael M. Franz

Bowdoin College203 Hubbard Hall

Brunswick, Maine 04011

E-mail: mfranz@bowdoin.edu