What Fifty Years of Regulation Did to the Money

Campaign finance law invites a simple question with a difficult answer: whether five decades of federal regulation reduced the money in federal elections, redirected the money into different hands and different forms, or left the scale of spending largely untouched while changing the paperwork attached to each dollar. A statute can hold on the precise transaction a statute prohibits and fail on the larger ambition that motivated the prohibition. Measurement therefore has to proceed promise by promise and channel by channel, because an aggregate total can conceal every distinction that decides whether regulation worked.

The architecture against which the record gets measured began with the Federal Election Campaign Act of 1971, the statute that supplied the foundation on which later amendments built the modern system. That system combined contribution limits, compulsory disclosure, public financing for presidential elections, and enforcement through a dedicated commission, and each element carried a distinct promise. Limitation promised that the size of a gift to a candidate or a committee would stop at a boundary set by law. Disclosure promised that the sources behind election spending would enter a public record open to voters, journalists, rival campaigns, and enforcement officials. Public financing promised an alternative supply of money for presidential candidates, offered in exchange for limits accepted voluntarily. Enforcement promised that the other promises would rest on investigation and sanction rather than on voluntary compliance alone.

The limitation promise met a constitutional boundary early, and the boundary shaped everything that followed. Contribution limits survived judicial review and remained a durable feature of the system across the whole period. Expenditure limits fared differently, and money spent independently of a candidate occupied a far less restricted position from the nineteen seventies onward. That asymmetry explains much of the later history without recourse to evasion or bad faith. A rise in independent spending therefore tested the design of the boundary as much as the force of any single prohibition, and the sections that follow treat movement across that boundary as evidence about design rather than as evidence of lawlessness.

The disclosure promise produced the evidentiary base for the entire inquiry, and the disclosure record also exposed the limits of disclosure as a form of control. Reports filed under the system made federal election money visible on a scale no earlier period could match, and every total and comparison in the sections that follow rests in large part on filings that disclosure compelled. Visibility and control, however, proved to be different achievements. A filing can name a committee and an expenditure with precision and name a donor whose own sources remain undisclosed, so that the trail ends at a lawful stopping point. The record through the 2020 cycle held both developments at once: an immense public archive of reported activity, and a widening distance between the spender named in a report and the source that supplied the funds behind the spender.

The public financing promise depended on a bargain that no candidate had a legal duty to accept, and that dependence gave the program a fragility the other elements lacked. A public financing offer binds nobody, because a candidate who calculates that private fundraising will exceed the public payment can decline the payment and the attached limit together, at no legal cost. The program therefore survived only for as long as the payment remained competitive against the private alternative, and the test of the program is a test of participation rather than of compliance, measured in acceptances and refusals by major party nominees and in taxpayer participation in the checkoff that stocked the fund.

The enforcement promise rested on a commission designed to require agreement across a partisan divide before consequential action could proceed, and that design contained both a safeguard and a vulnerability. The vulnerability appeared whenever agreement across the divide became scarce, because a body that cannot assemble the votes to act cannot investigate, cannot issue guidance, and cannot impose a penalty. Deadlock and institutional failure are not synonyms, and some divided votes involved contested questions of law on which reasonable commissioners could differ. The measurement question is empirical: how often division ended matters without action, how the pattern changed across the period, and whether the disclosure machinery continued to function during the years when contested enforcement faltered.

A final distinction governs the reading of every section that follows, the distinction between reduction and redirection. Reduction means that money left federal elections, or that a category of spending contracted and stayed contracted after a rule closed a channel. Redirection means that money changed route, changing the identity of the spender, the legal form of the organization, the degree of disclosure, or the relationship between the spender and any candidate, while the broader scale of spending held or grew. One channel in the record closed and stayed closed. Another channel recovered only a fraction of the money that a ban had displaced. A third channel expanded under court decisions that affirmed disclosure while removing limits, and the thinning of disclosure at the level of original source proceeded inside a reporting system that continued to function. The sections that follow take the promises in turn, totals first and channels after, and the verdict at the end weighs the whole record rather than any single episode.

The Baseline: What the System Promised and What the Totals Show

Any measurement of the impact of campaign finance law requires a standard fixed before the numbers are examined, because the numbers alone cannot say whether a legal system succeeded. A total can rise while a particular prohibition holds, and a total can fall while money moves into a less visible channel. A channel can close exactly as Congress directed, and the funds that once passed through that channel can reappear in another form, reported under another name or not reported at all. For that reason, the record assembled in the sections that follow is measured first against the purposes that the modern framework announced for itself, in the architecture of the statutes, in the institutions the statutes created, and in the obligations imposed on candidates, parties, committees, and other participants in federal elections. The standard is therefore an internal one. The question is whether the system built in the nineteen seventies performed the work that the system assigned to itself, and whether later revisions preserved that performance when the courts, the parties, and politically active organizations adapted to the rules.

The formal identities of that framework can be stated in one pass. The Federal Election Campaign Act, Public Law 92-225, signed on February 7, 1972, supplied the statutory foundation. The amendments of 1974, Public Law 93-443, reshaped that foundation into the modern system of contribution limits, disclosure, public financing for presidential elections, and enforcement through the Federal Election Commission. The Bipartisan Campaign Reform Act of 2002, Public Law 107-155, signed on March 27, 2002, and effective on November 6, 2002, then revised the framework again, most prominently by prohibiting national party committees from raising or spending soft money. Taken together, the enactments define the legal world within which the totals in this section were produced. Campaign finance law in the United States after 1974 was never a single rule against political spending as such. Campaign finance law was a set of channel specific decisions about who could give, who could receive, how much could pass between them, what had to be reported, what public money would be offered in return for limits, and which institution would administer the resulting obligations.

The first aim was limitation. The 1974 amendments placed limits on contributions, and that choice expressed a judgment that the size of a transfer from a donor to a candidate or committee was itself a proper object of regulation. A contribution limit does not prohibit political participation. A contribution limit draws a boundary around one form of participation, the direct financial support of a candidate or committee, and treats amounts beyond the boundary as presenting a risk that amounts within the boundary do not present in the same degree. When later spending rose outside the contribution system, the rise did not by itself prove that contribution limits had failed at the task assigned to them. Contribution limits governed contributions. Independent spending and spending by organizations operating outside the candidate structure raised separate questions, some of which the courts placed beyond the reach of limitation. Buckley v. Valeo, 424 U.S. 1 (1976), drew the most consequential of those lines when contribution limits and disclosure were sustained and expenditure limits were struck. From that point forward, the American system combined strict regulation of money given to candidates with far weaker authority over money spent independently in support of or in opposition to candidates. Any honest baseline has to preserve that asymmetry, because much of the later history consists of money finding the less restricted side of a line that the Supreme Court itself had drawn.

The second aim was disclosure. The 1974 architecture compelled participants in federal elections to report financial activity through a public system, and disclosure performed several functions at once. Disclosure allowed voters to see the sources of support behind a candidate, allowed journalists and opposing campaigns to examine those sources during an election, and allowed enforcement officials to test whether contributions and expenditures complied with the substantive rules. Disclosure also created the evidentiary base on which later research depended. Every total discussed in this section rests on reports, estimates, and classifications derived in large part from the disclosure system, and that dependence is itself part of the baseline. A legal regime can fail to reduce spending and succeed in making spending visible. A legal regime can also preserve formal reporting duties while an increasing share of politically consequential money moves through organizations that do not disclose their original donors. The baseline therefore separates visibility from volume, two aims that the later record pulled apart in ways that a single spending total cannot capture.

The third aim was public financing of presidential elections. The 1974 system offered public money to presidential candidates, through primary matching funds and general election grants, as an alternative source of support that carried conditions with the money. The logic of the offer was substitution. A candidate who accepted public funds accepted limits attached to those funds, and the public subsidy reduced the need to assemble an entire presidential campaign budget from private contributions. The mechanism depended on voluntary participation, and voluntary participation depended in turn on a candidate calculation that the grant and the certainty attached to the grant were worth more than the private fundraising that the accompanying limit would foreclose. If private fundraising capacity grew faster than the value of the public offer, candidates could decline the bargain without violating any rule. Declining public financing was not evasion, and treating refusal as noncompliance would misstate the law. The proper measure is whether the program continued to attract the candidates whose participation gave the program meaning, and later sections examine the cycle by cycle record of acceptance and refusal on those terms.

The fourth aim was enforcement through a commission. The Federal Election Commission was created as the institution that would administer disclosure, receive reports, issue guidance, and pursue violations of the contribution and spending rules. Enforcement gave the other three aims a practical foundation. A contribution limit without investigation and sanction is a statement of preference. A disclosure duty without review and penalty is a filing exercise. A public financing program without certification and audit is a transfer of money without the conditions that justified the transfer. The commission design also contained a structural tension that later became central to the record. The Commission had six commissioners, with no more than three from one party, and four affirmative votes were required for enforcement actions, rulemakings, and advisory opinions, so a split of three to three stopped action. Defenders of that design have described the even structure as a deliberate protection against partisan enforcement, and that counter reading belongs in the baseline rather than as an afterthought. A deadlock can reflect institutional failure, and a deadlock can also reflect a genuinely contested question of law on which a bipartisan majority could not be assembled. The baseline therefore asks whether the Commission acted on the matters placed before it, and whether the disclosure system continued to operate even when contested enforcement questions divided the commissioners.

Has total spending fallen since the modern rules began?

Judged in nominal dollars, the answer is no. Center for Responsive Politics estimates placed total federal election spending at about 5.3 billion dollars in the 2008 cycle and about 14.4 billion dollars in the 2020 cycle, a rise well over a doubling after inflation is taken into account.

The longer arc gives that answer its proper scale. In the 1999 to 2000 cycle, according to Ansolabehere, de Figueiredo and Snyder, candidates, parties and organizations raised nearly 3 billion dollars. That figure described amounts raised in a presidential cycle before the Bipartisan Campaign Reform Act of 2002 had taken effect and before the later outside spending channels had reached their 2020 scale. The Center for Responsive Politics estimated total federal election spending, in nominal dollars, at about 5.3 billion dollars in the 2008 cycle. The Center for Responsive Politics estimated the corresponding total, in nominal dollars, at about 6.5 billion dollars in the 2016 cycle. The Center for Responsive Politics estimated total federal election spending, in nominal dollars, at about 14.4 billion dollars in the 2020 cycle. Each of those three figures is a Center estimate of the federal election total for candidates, parties and outside groups, stated in nominal dollars rather than in dollars adjusted for inflation. The sequence has to be read with that label attached. Nominal totals from different cycles are not automatically comparable as measures of real resources, because a dollar in the 2008 cycle purchased more than a dollar in the 2020 cycle. No available adjustment reverses that direction.

The inflation qualification strengthens rather than weakens the conclusion. Consumer prices rose by roughly a fifth between 2008 and 2020, so the Center estimate of about 14.4 billion dollars in nominal dollars for the 2020 cycle represented well over a doubling in real terms against the Center estimate of about 5.3 billion dollars in nominal dollars for the 2008 cycle. Price change explained only a fraction of the increase. An increase of roughly a fifth in the general price level cannot convert a 2008 total of about 5.3 billion dollars into a 2020 total of about 14.4 billion dollars. After allowance for that price increase, the resources devoted to federal elections in the 2020 cycle remained well over twice the resources devoted in the 2008 cycle. The rise from the Center estimate of about 6.5 billion dollars in nominal dollars in the 2016 cycle to the Center estimate of about 14.4 billion dollars in nominal dollars in the 2020 cycle was also too large to be described as a routine adjustment to higher prices. Between those two presidential cycles, the estimated total more than doubled in nominal terms.

Comparison with the earlier baseline requires more care, because the nearly 3 billion dollars that Ansolabehere, de Figueiredo and Snyder reported as raised by candidates, parties and organizations in the 1999 to 2000 cycle and the Center estimates for later cycles do not rest on identical language. One figure described money raised, and the later Center figures estimated money spent across federal elections. Fundraising and spending are closely related in a completed cycle, but the categories are not interchangeable in precise accounting. The comparison is nevertheless informative at the level of magnitude. The federal system entered the twenty first century with a presidential cycle measured at nearly 3 billion dollars raised, and two decades later the Center for Responsive Politics estimated a presidential cycle at about 14.4 billion dollars in nominal dollars spent. Even allowing for differences in definition and for price change across two decades, the later figure does not support a claim that the modern rules produced a sustained reduction in the financial scale of federal elections. Campaign finance law after 1974 did not freeze the cost of competition at the level that prevailed when the Bipartisan Campaign Reform Act of 2002 became law. The totals kept rising through successive legal regimes and through cycles governed by different mixes of party money, candidate money, and outside money.

That finding settles less than first appears. Total spending is an aggregate, and an aggregate conceals the composition that determines how campaign finance law operated. A billion dollars raised by candidates under contribution limits and reported by named donors differs, as a matter of legal design, from a billion dollars spent by an independent organization financed by a small number of large contributors. A billion dollars spent by a party committee under federal rules differs from a billion dollars spent by a nonprofit organization that does not disclose its donors. The aggregate treats those dollars as equivalent units of election spending, and for measuring the overall financial scale of an election, equivalence is appropriate. For measuring legal impact, equivalence hides the central issue. The statutes regulated some channels more tightly than others, prohibited some transfers outright, required disclosure from some participants and not from others, and offered public money under conditions that candidates could refuse. A rising total is therefore compatible with several different accounts of legal effect. Regulation might have had little effect on behavior. Regulation might have redirected money from regulated channels into less regulated channels, changing the identity of the spender and the degree of disclosure without reducing the amount. Regulation might also have suppressed some spending that would otherwise have occurred while other forces pushed spending upward by a larger amount. The totals alone cannot choose among those accounts. The totals establish the baseline fact of growth, and the channels establish the mechanism.

The shape of the outside spending series gives the first strong indication of mechanism. OpenSecrets reported total outside spending by cycle, excluding party committees and covering independent expenditures, electioneering communications, and communication costs, at 51.6 million dollars in the 2000 cycle, at 27.7 million dollars in the 2002 cycle, and at 193.1 million dollars in the 2004 cycle. OpenSecrets reported the same category, excluding party committees, at 338.4 million dollars in the 2008 cycle, at 309.8 million dollars in the 2010 cycle, at 1,038.1 million dollars in the 2012 cycle, and at 1,411.3 million dollars in the 2016 cycle. OpenSecrets reported outside spending, excluding party committees, at 2,914 million dollars in the 2020 cycle. Each figure in that series is an OpenSecrets total in nominal dollars for the defined outside category. The endpoints capture the transformation. A category measured at 51.6 million dollars in the 2000 cycle reached 2,914 million dollars in the 2020 cycle. The increase moved outside spending from a small fraction of federal election activity into a central component of presidential cycle finance. In the 2020 cycle, the OpenSecrets outside total of about 2.9 billion dollars represented about one dollar in five of the Center estimate of about 14.4 billion dollars in total federal election spending. One dollar in five did not mean that candidate and party fundraising had ceased to matter. One dollar in five did mean that an analysis confined to candidates and parties would miss a large portion of the money that the legal system was supposed to organize.

The timing within the outside series also disciplines interpretation. The reported total fell from 51.6 million dollars in the 2000 cycle to 27.7 million dollars in the 2002 cycle, and any account that portrayed outside spending as rising mechanically with each successive cycle would fail against those two figures. The category then rose sharply by the 2004 cycle, eased between the 2008 cycle and the 2010 cycle in the OpenSecrets figures, and crossed one billion dollars in the 2012 cycle. That uneven path matters because campaign finance law changed at particular moments, court decisions altered the permissible forms of organization and spending at particular moments, and a midterm cycle and a presidential cycle present different incentives and different offices at stake. By the 2020 cycle, the cumulative change had become difficult to reconcile with a picture of a system centered exclusively on candidates and their authorized committees. The regulated center remained large, and the regulated center no longer exhausted the field.

Composition adds a further caution drawn from OpenSecrets analyses of the sources behind outside spending. Direct corporate treasury money formed a modest share of super PAC receipts, and the largest growth in outside funding came from wealthy individuals giving to super PACs, together with nonprofits that did not disclose donors. That composition matters because popular accounts of the post 2010 system often assign the entire transformation to corporate treasuries. The OpenSecrets composition evidence points toward a more specific mechanism. Legal permission for independent spending interacted with concentrated individual wealth and with nonprofit forms that preserved donor anonymity, and the resulting organizations could raise and spend amounts that the contribution limits applicable to candidates did not permit any single donor to give directly to a campaign. SpeechNow.org v. FEC supplied the legal basis for committees making only independent expenditures to raise funds without the contribution limits that would otherwise have applied, and delivered analysis of the period treats that decision, rather than Citizens United v. FEC standing alone, as the decision that made super PACs possible. Precision about that legal origin matters to the baseline, because the channel grew from a particular rule about contributions to independent expenditure committees, not from a general abolition of campaign finance law.

The baseline can therefore be stated as a set of linked findings. The 1974 architecture set out to limit contributions, compel disclosure, finance presidential elections with public money, and enforce the resulting rules through a commission. Total federal election spending did not fall across the cycles for which the Center for Responsive Politics supplied the estimates used in this section, and the Center estimate for the 2020 cycle, about 14.4 billion dollars in nominal dollars, stood well over double the Center estimate for the 2008 cycle, about 5.3 billion dollars in nominal dollars, after allowance for a price rise of roughly a fifth. Outside spending, as OpenSecrets defined that category while excluding party committees, grew from 51.6 million dollars in the 2000 cycle to 2,914 million dollars in the 2020 cycle, and that growth concentrated a much larger share of election money outside the candidate and party system that the statutes regulated most tightly. Those findings frame the impact question in the only form that the evidence can answer channel by channel. Did campaign finance law reduce the money in federal elections, did campaign finance law redirect the money into different organizations and different forms of disclosure, or did campaign finance law accomplish neither reduction nor meaningful redirection?

Soft Money: The Channel That Grew and the Ban That Held

The phrase soft money described money raised and spent outside the contribution limits and other federal rules that governed the rest of the federal election system. Hard money, by contrast, meant funds raised under the Federal Election Campaign Act limits, from permitted sources and in limited amounts, and reported under the federal disclosure system. The distinction mattered because campaign finance law rested on a boundary: money connected to federal elections belonged inside the federal rules, and money outside that boundary escaped the limits even when the spending shaped federal races. Soft money lived on the far side of that boundary, and the national political parties learned to raise soft money in quantities that rivaled the regulated system operating beside the exempt channel.

The opening came from Congress itself. The 1979 amendments to the Federal Election Campaign Act, signed on January 8, 1980, moved specified state and local party activity outside the federal definitions of contribution and expenditure. The activities at issue sounded modest and local: grassroots work, volunteer driven organizing, and party building by state and local party committees, the routine labor of maintaining a party between elections. Federal law had treated party spending through a single lens, and the amendments separated a class of party activity from that lens on the theory that state and local party building differed from federal campaigning. The separation created a second ledger. Money raised for the exempt activities did not count as a federal contribution, and spending on those activities did not count as a federal expenditure, so the federal limits that defined the regulated system did not reach the second ledger at all.

National party committees recognized the scale of the opening faster than the modest framing of the amendments suggested. A national party could raise money for party building without the amount limits that governed the federal accounts, direct resources toward state and local party activity, and support the organizational work that surrounded federal campaigns without charging the cost against the limited ledger. Fundraising followed the opportunity. Both the Democratic and Republican national party organizations built the channel into a central revenue source, soliciting large contributions that federal law forbade in a candidate account or in a federal party account. The result was a parallel finance system attached to the same parties: one ledger limited and capped under federal law, and a second ledger bounded by little beyond the willingness of donors to write large checks and the capacity of the parties to spend the proceeds on activities classified as party building.

The growth curve showed how completely the parties absorbed the channel. The Brennan Center figure put national party soft money at about 45 million dollars in 1988, a sum large enough to matter in a presidential cycle and small enough to read as a supplement to the regulated system. The Congressional Research Service, drawing on Federal Election Commission data, put the national party committee total at 262 million dollars in 1996 and at about 495 million dollars in the 2000 cycle. Read in sequence, the figures describe a channel that multiplied roughly elevenfold between 1988 and the 2000 cycle while the formal limits on the regulated ledger changed far more slowly. By the 2000 cycle, soft money no longer functioned as an auxiliary account at the edge of party finance. The national committees raised and spent sums through the exempt channel that stood in comparison with the committees’ regulated fundraising, and any account of how the 1996 and 2000 elections were financed that ignored the second ledger missed a central part of the money.

The proceeds bought the infrastructure of campaigns rather than the advertisements of the candidates in the first instance. Party building covered the unglamorous inputs on which federal campaigns depended: staff, offices, lists, registration drives, turnout operations, and generic advertising that promoted a party label without naming a federal contest as the subject. Voter drives registered and mobilized the electorate from which federal candidates drew their margins. None of that spending carried the name of a candidate as a necessary feature, and all of that spending strengthened the organization that nominated, supported, and campaigned beside federal candidates. The classification followed the activity rather than the effect. A registration drive in a competitive state counted as party building even when the value of the drive to a presidential or Senate campaign was obvious to every participant, because the federal definitions, after the 1979 amendments, asked how an activity was classified and how the money was allocated rather than whose election the activity helped to decide.

The most consequential purchase was broadcast time. Soft money paid for issue advertising that named federal candidates, described their records, praised or attacked their positions, and reached voters in the weeks before an election, while avoiding the express advocacy words that would have classified an advertisement as a federal expenditure. An advertisement could present the face, name, and record of a candidate and urge viewers to call the candidate and deliver an opinion, and the absence of a direct instruction to vote for or against the candidate kept the spending on the party building side of the line. The distinction between discussing an issue and advocating an election outcome collapsed in practice during the 1996 and 2000 cycles, when the exempt channel financed broadcast campaigns that voters experienced as election advertising in every respect except the legal label. Candidates benefited from advertising that the candidates had not financed from limited funds, parties gained a broadcast voice that the regulated ledger could not have financed at the same scale, and donors gained a route to election influence without the amount limits that governed direct giving.

The design question ran deeper than the totals of any single cycle. Campaign finance law after the 1974 amendments rested on the premise that limiting contributions to candidates and parties would limit the influence that large donations could purchase, and the soft money channel tested that premise at the foundation. A donor barred from giving more than a fixed sum to the federal account of a national party could give an unlimited sum to the nonfederal account of the same party, deal with the same party officials, and fund advertising that shaped the same elections. The formal limits survived intact on paper while a parallel route carried the largest checks around the limits. Congress faced a choice between treating the channel as a safety valve that kept large donations attached to accountable parties and treating the channel as a repeal of the contribution limits in practice. The settlement embodied in the 1979 amendments had reflected the first judgment. The legislation of 2002 embodied the second judgment, and the distance between the two judgments measured how far the channel had traveled from the modest activity that the 1979 amendments had set out to protect.

Congress closed the channel in the Bipartisan Campaign Reform Act of 2002, Public Law 107-155, signed on March 27, 2002. The statute prohibited national party committees from raising or spending funds outside the federal limits, which ended the legal basis of the second ledger rather than narrowing the permitted uses of the second ledger. A national committee could no longer accept a contribution that federal law excluded from the federal accounts, could no longer maintain a nonfederal account for the exempt activities, and could no longer spend money that sat outside the federal definitions. The prohibition took effect on November 6, 2002, the day after the 2002 midterm elections, so the 2002 cycle closed under the older rules and the 2004 cycle opened under the ban. The prohibition addressed the national committees directly, and the national committees were the center of the channel: the Congressional Research Service figures drawn from Federal Election Commission data measured national party committees, and the ban removed exactly the institutions that had raised the 262 million dollars recorded for 1996 and the about 495 million dollars recorded for the 2000 cycle.

The prohibition faced an immediate constitutional test, and the Supreme Court sustained the prohibition in McConnell v. FEC in 2003. The ruling arrived against the background of Buckley v. Valeo in 1976, the decision that had sustained contribution limits and disclosure while striking expenditure limits, and the McConnell outcome placed the soft money ban on the contribution side of that divide, the side of the doctrine where Congress had prevailed in 1976. The decision mattered beyond the survival of a single provision. A ban that the Court had struck down in 2003 would have restored the second ledger for the 2004 cycle and returned the parties to the finance structure of the 2000 cycle. A ban sustained in 2003 instead governed the 2004 cycle and every later cycle in the record through the 2020 cycle, and the parties planned fundraising, compliance, and spending on the assumption that the exempt channel had ended for good.

The record after 2003 separated the soft money ban from the pattern that governed much of campaign finance law during the same period. Provisions fell, channels opened, and money found routes around restrictions in area after area of the system, a pattern that the balance of the evidence in the wider record traces through outside spending and independent groups. The national party prohibition produced no comparable sequence. No later Supreme Court decision reversed the ban or reopened the national party channel. No later Congress restored the nonfederal accounts. National party soft money, measured at about 495 million dollars in the 2000 cycle in the Congressional Research Service figures drawn from Federal Election Commission data, did not reappear in the 2004 cycle or in any later cycle through 2020. The second ledger remained closed through the 2020 cycle, and the absence was total rather than partial: the committees that had built the channel reported no soft money receipts of the kind that the 1996 and 2000 figures had recorded, because the statute left no lawful form in which such receipts could exist.

Durability of that kind had several sources. The prohibition regulated the fundraising of the parties rather than the speech of outside speakers, and the Supreme Court sustained the ban on that footing in McConnell v. FEC in 2003. The parties also discovered, during the 2004 cycle, that the regulated ledger could carry more weight than the debate before the ban had assumed, and a prohibition that regulated institutions can operate under generates less pressure for reversal than a prohibition that regulated institutions cannot survive. The text of the statute contributed as well, because a prohibition covering the raising and the spending of funds outside the federal limits left no residual category through which a national committee could rebuild the channel under a different name. Whatever weight each explanation carries, the outcome distinguished the provision from the redirection pattern around the provision: a fundraising channel that had grown from the opening of 1980 to the totals of the 2000 cycle ended on a date fixed by statute, November 6, 2002, and the ending held in every cycle that followed through 2020.

The durability also had a measurement advantage that few provisions in campaign finance law enjoyed. Because the Congressional Research Service figures rested on Federal Election Commission data, the channel had a documented size before the ban and a documented absence after the effective date, and the comparison between the 2000 cycle and the cycles from 2004 through 2020 did not depend on inference about hidden flows. Money that moved outside the party committees after 2002 raised harder questions of measurement, and the wider record reflected those questions. Money inside the national committees raised no such question: the committees filed reports, the reports showed hard money totals, and the Campaign Finance Institute could compare the 2004 totals against every earlier cycle on a consistent basis. The ban therefore produced a rare outcome in the regulation of election money, a result that auditors could verify from the public record rather than a result that rested on estimates of activity beyond the reporting system. Verification strengthened the settlement of 2002 and 2003, because each cycle from 2004 through 2020 added a further reported confirmation that the closed channel had not reopened inside the institutions the statute governed.

Did the soft money ban work?

Yes, in the specific sense the statute defined. The ban closed the national party soft money channel, that channel did not return in any cycle from 2004 through 2020, and the national parties replaced the revenue with hard money raised under the federal limits in the 2004 cycle and afterward.

Adaptation, rather than disappearance, described the response of the parties. The national committees entered the 2004 cycle without the revenue source that had supplied the second ledger in the 2000 cycle, and the committees rebuilt fundraising around the federal account: solicitation within the amount limits, broader donor recruitment, and accounting that kept every dollar inside the federal definitions. The organizational functions that soft money had financed did not end with the channel that had financed the functions. Registration work, turnout operations, staff, and generic party support continued under the ban, funded from a different ledger and under different rules, and the committees that had depended most heavily on the exempt channel in the 2000 cycle contested the 2004 elections as fully financed organizations operating entirely within the federal limits.

The Campaign Finance Institute assessed the adjustment after the 2004 cycle and found that the national party committees had raised substantially more hard money in 2004 than in any earlier cycle. The composition of the increase carried as much significance as the size of the increase. The Institute found growth from both large donors and small donors, which meant that the committees had not simply replaced a small number of very large soft money checks with a small number of maximum hard money checks drawn from the same narrow circle. Fundraising inside the limits required volume: more solicitations, more contributors, and more donors giving amounts far below the sums that the exempt channel had accepted from a single source. The finding measured the institutional effect of the ban directly. A prohibition that had ended the largest revenue channel of the parties coexisted with the largest hard money totals that the committees had recorded to that point, and the breadth of the donor growth showed that the increase rested on an expanded fundraising operation rather than on a narrow substitution at the top of the donor scale.

The role of the parties changed form in concrete ways. Fundraising success depended on the size of the donor file rather than on access to a small number of donors able to give without limit. Compliance moved to the center of party finance, because every contribution to a national committee had to satisfy the federal limits and source rules that the second ledger had bypassed. The money that the parties raised bought the same categories of organizational work, registration, turnout, staff, and generic support, but the route from donor to expenditure ran through a regulated account structure from the first dollar to the last. Donors lost the option of giving a national party an unlimited sum, and the influence that accompanied very large party contributions lost the legal channel that the national committees had provided before November 6, 2002. Influence did not end, and money did not leave federal elections; both statements belong to the wider record through the 2020 cycle. The narrower statement belongs to the soft money record alone: the specific transaction at the center of the system before 2002, an unlimited contribution to a national party committee, ceased to exist as a lawful transaction after November 6, 2002, and no cycle through 2020 restored that transaction.

The soft money episode therefore supplied the durable exception inside a larger story of redirection. Campaign finance law repeatedly watched money move when a channel closed, and the movement after 2002 ran in large part through organizations outside the parties. The party channel itself behaved differently. Congress identified a defined institutional practice, prohibited the practice in terms that covered the raising and the spending of funds alike, defended the prohibition successfully in McConnell v. FEC in 2003, and watched the practice end. The figures mark the scale of the practice that ended: a channel measured by the Brennan Center at about 45 million dollars in 1988 and by the Congressional Research Service, from Federal Election Commission data, at about 495 million dollars in the 2000 cycle, reduced to absence in every cycle from 2004 through 2020. A ban can fail through displacement, and the wider record through the 2020 cycle showed displacement in abundance. The national party soft money ban showed the other possibility present in the record: a defined channel, closed on a fixed date, that remained closed through the final cycle in the record, the 2020 cycle.

The Tax-Code Detour: Redirection at Half Strength

The Bipartisan Campaign Reform Act of 2002, signed on March 27, 2002 and effective on November 6, 2002, prohibited national party committees from raising or spending soft money. The prohibition closed a channel that had carried about 495 million dollars through national party committees in the 2000 cycle, on Congressional Research Service figures drawing on Federal Election Commission data, and it raised an immediate question about substitution. Campaign finance law can forbid a named recipient from accepting a defined category of funds, and the 2002 statute did exactly that with unusual clarity. The donors who had supplied the soft money retained their wealth, their political preferences, and in many cases their relationships with the parties and candidates the money had supported. The answer was neither complete substitution nor complete disappearance. Money moved, but it moved at half strength, through a different donor population, into organizations with a different legal position and a different relationship to candidates and parties than the channel the statute had closed.

A 527 organization, in the sense that matters for the period after 2002, was a political organization formed under section 527 of the Internal Revenue Code that had not registered as a federal political committee with the Federal Election Commission. The defining feature for campaign finance purposes was that negative one. Because such an organization stood outside the registration regime that governed federal political committees, it stood outside the contribution limits that bound parties and candidates. It could accept contributions of unlimited size and direct its spending toward voter mobilization and issue advertising that influenced the environment of a federal election without itself becoming a party committee or a candidate committee. That position made the form attractive to donors whose ambitions had exceeded the hard money limits even before 2002. A donor could write a single large check, avoid the fragmentation that giving through regulated committees required, and support advertising or turnout work aligned with a party or a candidate without making a contribution to that party or candidate in the legal sense defined by the Federal Election Campaign Act.

The distinction between alignment and formal direction did much of the work in the 527 model. An organization that presented its purpose as voter education, registration, turnout, or discussion of public issues could operate in the space adjacent to a campaign while maintaining formal separation from the campaign itself. Voters encountering its advertising experienced no such separation, because an advertisement praising or attacking a candidate on a contested issue, aired in the weeks before an election in the states where that candidate competed, functioned as campaign communication regardless of the legal category in a disclosure filing. What the form changed was accountability and control. A party committee answered to a leadership with a continuing interest in a national brand, a bench of candidates, and elections beyond a single cycle. A freestanding 527 organization answered principally to its own founders and funders, could be created for a single cycle, and could dissolve after the votes were counted.

The migration became visible in the aggregate figures. On Center for Public Integrity and Center for Responsive Politics data, as reported at the time, spending by 527 organizations rose from about 171 million dollars in 2000 to 316 million dollars in 2002, and then to 653 million dollars in 2004, the first presidential cycle conducted entirely under the soft money prohibition. The sequence matters because the 2002 figure already stood well above the 2000 figure, in a cycle when the prohibition took effect only on November 6, after the election that closed the cycle. Donors and organizers were building the substitute channel before the old channel had fully closed, and the scale of the 2004 figure showed that the construction had succeeded on its own terms. Any account that treated the ban as having simply removed the money behind party soft money would fail against those totals. The statute changed the law governing national party committees, and political money responded by building a large structure just outside the committees the law had reached.

The same sequence, read carefully, already contains the qualification the closer analysis confirmed. The 653 million dollar figure for 2004 counted spending by 527 organizations as a broad category, on the Center for Public Integrity and Center for Responsive Politics data as reported at the time, while a narrower measurement of federally active 527 committees produced a smaller total for the same cycle. Treating every dollar spent by every organization in the tax category as a dollar of substituted party soft money would overstate the substitution before the comparison with the eliminated soft money even began. The headline total established that the channel grew rapidly. It did not establish that the growth replaced what had been lost, donor by donor or dollar by dollar.

Political posture made the channel legible to donors and voters alike. On the Democratic side, America Coming Together and the Media Fund stood among the prominent pro-Democratic organizations of the 2004 cycle. On the Republican side, Swift Boat Veterans for Truth and Progress for America stood among the prominent pro-Republican organizations. The symmetry of the examples carries an analytical point that partisan memory often obscures. The 527 channel was not the invention of one party evading a law the other party obeyed, and its benefits did not flow in a single direction. Operatives aligned with both parties examined the post 2002 rules, reached the same structural conclusion, and built organizations to exploit the same opening, and donors aligned with both parties supplied the funds.

Distance from any candidate also imposed a cost in coherence, and the cost helps explain the limits of the channel from its first presidential cycle onward. A party committee allocating soft money before 2002 decided among Senate races, House races, state parties, and presidential priorities from within a single institution that bore the consequences of the whole electoral map. A landscape of separate 527 organizations allocated money according to the preferences of separate founders and funders, with no institution charged with balancing the map as a whole, avoiding duplication between groups working the same states, or sustaining unglamorous operations in places where no major donor held a personal interest. Redirection through 527 organizations therefore changed not only how much money moved but what the money bought, favoring vivid advertising and clearly defined donor priorities over the routine party building work that the soft money regime had financed as a matter of institutional habit, and that difference in purpose followed the money into every later comparison between the old channel and its substitute.

Did tax-code organizations replace party soft money?

No. Federally active 527 committees raised and spent about 424 million dollars in the 2004 cycle, according to Stephen Weissman and Ruth Hassan for the Campaign Finance Institute, an increase of about 273 million dollars over 2002 that equaled less than half of the eliminated party soft money, and most former soft money donors never gave at all.

The Campaign Finance Institute analysis by Stephen Weissman and Ruth Hassan supplies the measured verdict because it asked the substitution question in its precise form. Rather than comparing a broad tax category total against a memory of party fundraising, Weissman and Hassan examined federally active 527 committees, the organizations whose activity actually belonged in a comparison with federal party soft money, and measured both the scale of their 2004 activity and its relationship to the donor base the ban had displaced. Their finding placed the federally active committees at about 424 million dollars raised and spent in the 2004 cycle, a substantial sum by any standard and a clear demonstration that the channel mattered. A channel that grows by 273 million dollars in a single cycle has absorbed a real redirection of political money, and no honest reading of the analysis can describe the 527 response as trivial, or as a failure of donors to find the new vehicle.

The force of the analysis lies in the comparison that followed. The increase of about 273 million dollars in federally active 527 activity between 2002 and 2004 equaled less than half of the party soft money that the Bipartisan Campaign Reform Act had eliminated. The arithmetic reframes the episode. Measured against its own past, the 527 channel had grown explosively, and measured against the regulated fundraising of parties and candidates it held a large position. Measured against the specific pool of money that reform had removed from national party committees, the substitute recovered a minority share. Weissman and Hassan gave the second observation its evidentiary foundation, and their finding is the load bearing fact in any assessment of the detour: redirection was real, substantial, and incomplete.

The donor evidence deepens the finding beyond aggregate arithmetic. Most former party soft money donors, on the Weissman and Hassan analysis, did not give to 527 committees at all. That fact dismantles the model in which a fixed cast of donors, committed to spending a fixed sum on federal politics, simply searches for the next available receptacle whenever Congress closes one. The 527 committees drew on a substantially different population of people and institutions, and many of the donors who had written the largest soft money checks to the parties declined the substitute. The structure of the choice helps explain the refusal, even where the record assigns no motive to any individual donor. A contribution to a national party committee purchased a relationship with a permanent institution, standing with its leadership, and a place in a continuing political enterprise whose memory extended beyond one election. A contribution to a new 527 organization purchased influence over a temporary vehicle with no candidates of its own, no certain existence in the following cycle, and a legal position that the Federal Election Commission might later reject. For a corporation, a trade association, or a wealthy individual weighing reputational exposure against political effect, the two expenditures were not equivalents at different addresses. They were different goods, and a large share of the old buyers declined the new one.

Donor discontinuity also helps explain why the redirection did not continue growing toward parity. Spending by 527 organizations, on Center for Public Integrity and Center for Responsive Politics data as reported at the time, fell from 653 million dollars in 2004 to 443 million dollars in the 2006 midterm cycle. Midterm cycles spend less than presidential cycles across nearly every category of campaign finance, so part of the decline reflects the calendar rather than the channel. Yet the 2006 figure did not suggest a substitute vehicle steadily consolidating the position party committees had vacated. It suggested a channel whose peak had arrived early, assembled rapidly for a presidential contest, and whose claim on donor loyalty remained contingent.

Enforcement then raised the cost of the experiment. In 2006 and 2007 the Federal Election Commission fined several of the largest 2004 groups, including Swift Boat Veterans for Truth, for operating as unregistered political committees. The significance of those settlements lies in the legal proposition the Commission asserted and the organizations accepted by settling, rather than in any penalty amount. A group that raised and spent money for the purpose of influencing federal elections, whatever its tax code designation, could be treated as a political committee subject to registration and to the contribution limits and source rules that registration carried. That proposition attacked the premise on which the 527 strategy rested, namely that tax status could determine campaign finance status. Once the Commission had acted against prominent organizations from the largest cohort the channel ever produced, a donor or organizer contemplating a comparable effort in a later cycle faced a different calculation. The vehicle that had looked in 2004 like a safe harbor outside the contribution limits looked by 2007 like a harbor whose entrance the regulator had charted and announced an intention to police, and legal uncertainty of that kind gives sophisticated donors a reason to direct money elsewhere when the election calendar offers alternatives.

The presidential cycle of 2008 supplied one such alternative in the form of candidate fundraising itself. Barack Obama declined the general election public funding grant for 2008, the first major party nominee to forgo general election public funding since the system began, on Congressional Research Service description, and the campaign built a fundraising operation that absorbed donor energy and donor money at a scale that competed directly with outside organizations for the attention of Democratic contributors. Across both parties, the cycle demonstrated the capacity of presidential campaigns to convert enthusiasm directly into contributions without an intermediary organization of uncertain legality.

A contribution to a candidate, within regulated limits, also carried a clarity that a contribution to a 527 organization could no longer match after the enforcement settlements of 2006 and 2007, and party channels offered large donors further regulated routes for supporting the same nominees. The pull of candidate fundraising in 2008 did not eliminate outside spending, and the cycle showed no general departure of money from federal politics. Its effect was narrower and more durable: donor preference shifted toward vehicles whose legal position was settled, whose recipient was unmistakable, and whose gratitude belonged to an identifiable campaign, at the expense of a vehicle whose position had been challenged by the regulator and whose benefits arrived indirectly. That shift in preference, visible in 2008, prepared the donor class for a further shift when a vehicle appeared that combined the scale of the 527 form with the legal clarity the 527 form lacked.

The arrival of the super PAC after 2010 completed the displacement by offering the donor class that 527 organizations had cultivated a legally cleaner instrument for the same purpose. The decision in SpeechNow.org v. FEC, 599 F.3d 686, decided by the United States Court of Appeals for the District of Columbia Circuit on March 26, 2010, held that contribution limits cannot be applied to a committee making only independent expenditures, and that holding, rather than Citizens United v. FEC alone, made super PACs possible. A super PAC could accept unlimited contributions, as a 527 organization had, and spend without limit on independent advertising, as donors had wanted in 2004, while standing inside the federal registration and disclosure system rather than outside it. The legal risk that had hung over the 527 model, the risk of being declared an unregistered political committee after the money had been spent, dissolved for an organization that registered from the start and confined itself to independent expenditures.

The 527 detour therefore holds a precise place in the history of campaign finance law, narrower than the language of loophole and more consequential than the language of failure. The episode demonstrated that a determined donor population and a skilled community of organizers could construct a substitute channel within a single cycle, raise hundreds of millions of dollars through it, and influence a presidential election by means of it. The episode equally demonstrated that adaptation has a measurable ceiling set by law, by institutional loyalty, and by the preferences of donors who distinguish among vehicles that a simple evasion model treats as identical. The substitute channel recovered less than half of the eliminated soft money on the increase that Weissman and Hassan measured, drew on a substantially different donor population, attracted federal enforcement within two years of its peak, lost donor attention to candidate fundraising in the following presidential cycle, and surrendered its donor class after 2010 to a registered vehicle whose legality rested on a court decision rather than on a tax code inference. Regulation moved the money, and the 527 years add the necessary measurement to that claim. The money moved at roughly half strength, through different hands, into organizations that lasted only until a cleaner instrument appeared. A reform that produces that outcome has neither abolished large scale political spending nor left the system it governed unchanged, and the distance between those two descriptions is the ground on which the post 2002 history of American campaign finance has been contested.

After the Twin Decisions: Super PACs, Nonprofits, and the Disclosure Gap

The change in outside spending after 2010 rested on two separate decisions, and campaign finance law cannot be described accurately when the two are collapsed into one. Citizens United v. FEC, decided in 2010, held that corporations and unions could not be barred from spending treasury funds on independent expenditures. The holding addressed the source of the money behind independent spending and removed a prohibition that had kept corporate and union treasury funds out of that channel. The same decision upheld disclosure and disclaimer requirements by a vote of 8 to 1, a second holding that mattered as much to the system that followed as the holding on expenditures. SpeechNow.org v. FEC, decided by the United States Court of Appeals for the District of Columbia Circuit in 2010, answered a different question. That decision held that contribution limits could not be applied to a committee that made only independent expenditures. The distinction is the foundation of the entire period, and the discussion in the campaign finance Supreme Court cases belongs beside a parallel account of the circuit decision, because neither decision alone produced the super PAC.

A super PAC existed because both questions received the answers given in 2010. Citizens United v. FEC permitted corporations and unions to spend treasury money independently, but the decision did not by itself free a committee organized solely for independent expenditures from contribution limits. SpeechNow.org v. FEC supplied that second step. After the circuit decision, a committee that confined activity to independent expenditures could accept contributions without the limits that applied to committees contributing directly to candidates. Attribution matters for more than doctrinal neatness. A narrative that credits Citizens United v. FEC alone misplaces the legal work and obscures the sequence through which the new committee form became available. The two decisions operated on different legal restrictions, were issued by different courts, and removed different obstacles. Together, the decisions changed both who could supply money for independent activity and how a committee devoted to that activity could raise money.

The disclosure holding in Citizens United v. FEC also complicates any simple account of deregulation. By a vote of 8 to 1, the Supreme Court sustained requirements designed to let voters know who paid for election related communications and to attach disclaimers to those communications. The result was a system in which independent spending expanded under a disclosure framework that the Court had affirmed rather than dismantled. Money moved through committees required to report donors and expenditures to the Federal Election Commission, and the reports created a public record of outside spending on a scale without precedent in earlier cycles covered by the record. At the same time, the reports reached only as far as the legal identity of the donor named in a filing. When that named donor was an organization with no obligation to disclose an original source, the chain ended at the filing line, even though the filing itself complied with the applicable rule. The gap therefore arose inside a functioning disclosure system, not outside one.

The scale of the new system appeared in the first cycle governed by both decisions. OpenSecrets reported that super PACs spent 62.6 million dollars in the 2010 cycle, a midterm cycle that offered only part of a year for organizations to form, raise funds, and spend under the new rules. The figure carried weight less as an endpoint than as a baseline. A committee form created by litigation in 2010 had, within the same cycle, become a channel for tens of millions of dollars in independent spending. The midterm setting also matters for interpretation, because presidential cycles draw larger totals across every category of federal election spending. Even with that caution attached, the 2010 total established that the SpeechNow.org committee form had moved from a legal possibility to an operating part of campaign finance law in a single cycle.

Two years later, the presidential cycle showed the form at a different scale. OpenSecrets reported that super PACs and hybrid PACs together spent 622.7 million dollars in the 2012 cycle. The category in that figure combined super PACs with hybrid PACs, and the combined scope has to travel with the number, because the 622.7 million dollars did not describe super PACs alone. Read beside the 2010 figure, with the different committee scope and the different cycle type both stated, the reported total showed a channel that had grown from an initial midterm presence into a central vehicle for outside spending in a presidential election. The growth described the size of the committee channel itself, measured through reported spending by the committees that the post 2010 rules made possible.

Super PACs formed part of a larger outside spending total that OpenSecrets measured across independent expenditures, electioneering communications, and communication costs, with party committees excluded from the total. Under that scope, OpenSecrets placed total outside spending at about 1.04 billion dollars in the 2012 cycle, about 1.41 billion dollars in the 2016 cycle, and about 2.9 billion dollars in the 2020 cycle. The series gives the section a disciplined way to describe expansion without attributing the entire outside total to any single committee type. Outside spending included super PACs and hybrid PACs, and it also included spending by other organizations operating outside candidate campaigns and outside party committees. Each figure in the series rested on the same stated scope, total outside spending excluding party committees, and each figure described a presidential cycle, so the comparison across 2012, 2016, and 2020 holds cycle type and scope constant in a way that a comparison against the 2010 midterm cannot. Within those limits, the movement from about 1.04 billion dollars to about 2.9 billion dollars across three presidential cycles showed outside spending nearly tripling in nominal dollars while the underlying reporting categories remained the categories through which the public record was assembled.

The 2020 figure also supplies a proportion that conveys scale more clearly than a raw total alone. OpenSecrets measured total outside spending excluding party committees at about 2.9 billion dollars in the 2020 cycle, and the Center for Responsive Politics estimated total federal election spending by candidates, parties, and outside groups at about 14.4 billion dollars in the same cycle. Placed together, those two figures with their different scopes make the outside total about one dollar in five of all federal election spending in 2020. The proportion has to be read with both scopes attached, because the numerator excluded party committees as spenders of outside money while the denominator included candidates, parties, and outside groups. With that reading in place, the proportion shows that outside spending in 2020 was not a marginal supplement at the edge of the federal system. A category accounting for roughly a fifth of estimated total spending occupied a structural position in the election, and the disclosure rules governing that category therefore governed the visibility of a substantial share of the money spent to influence the result.

Growth in the total did not settle the question of composition, and OpenSecrets analyses supplied a caution that runs against a common shorthand account of the 2010 decisions. Those analyses found that direct corporate treasury money formed a modest share of super PAC receipts. The largest growth in outside funding came from wealthy individuals giving to super PACs, together with nonprofits that did not disclose donors. The distinction between permission and use explains the finding. Citizens United v. FEC removed a ban on corporate and union treasury spending for independent expenditures, and a removal of a ban establishes what the law permits. A record of receipts establishes what contributors in fact supplied, in what proportions, through which committees, during a defined cycle. The permission changed in 2010 for corporations and unions, and the money that expanded most rapidly in the reported record came principally from wealthy individuals. Any account that treats the post 2010 surge as primarily a surge of direct corporate treasury contributions therefore moves beyond the OpenSecrets composition analyses and contradicts the modest share those analyses assigned to direct corporate treasury money in super PAC receipts.

The composition caution also changes how the role of nonprofit money should be understood. Nonprofits that did not disclose donors formed part of the growth identified by OpenSecrets analyses, alongside wealthy individuals, and nonprofit contributions carried a disclosure consequence that individual contributions to a super PAC did not carry in the same way. A wealthy individual who gave to a super PAC appeared in the super PAC report as the named donor, and a reader of the report could identify that individual as the source at the level the report recorded. A nonprofit that gave to a super PAC also appeared as the named donor, but the report identified the organization rather than the persons or organizations that had supplied funds to the nonprofit. The legal form was disclosed in both situations. The economic origin was disclosed at different depths. That difference in depth, repeated across many contributions and many committees, produced the disclosure gap named in the heading.

Can voters still trace where outside money comes from?

Voters could trace outside spending to a super PAC and to the donor named in its reports during every cycle from 2010 to 2020. Tracing stopped at that named donor when the donor was a non-disclosing nonprofit or a shell company, because the original source behind that donor faced no disclosure requirement.

Tracing worked in layers, and each layer had a different degree of reliability in the cycles from 2010 to 2020. The first layer identified the spender. A super PAC that paid for an independent expenditure filed reports with the Federal Election Commission, and those reports placed the committee name and the expenditure inside the public record. The second layer identified the contributor to the committee, because a super PAC disclosed donors and a report therefore connected a named contributor to the committee that later spent money independently of any candidate. The third layer asked who stood behind the named contributor. For an individual contributor, the second and third layers largely converged, because the named contributor was the source recorded by the system. For a non-disclosing nonprofit or a shell company, the third layer could disappear entirely. The report satisfied the second layer and gave no answer at the third, and a reader following the money reached a lawful stopping point rather than a missing filing.

The category at the center of that stopping point had a specific legal definition in the record. Politically active nonprofit organizations, principally organizations under section 501(c)(4) of the tax code and some trade associations under section 501(c)(6), could spend on elections without disclosing donors. Money spent through that route carried the label dark money. The definition matters because the label described a disclosure status, not a spending purpose or a partisan direction. A section 501(c)(4) organization and a section 501(c)(6) trade association occupied different places in the tax code and could support different political alignments, and the shared feature for campaign finance law was the absence of a donor disclosure obligation attached to election spending. Dark money therefore named money whose spender could be known in a report while the donors behind the spender remained outside the report. The term also covered money routed through such an organization into a super PAC, where the super PAC report disclosed the nonprofit as a donor and disclosed nothing about the nonprofit donors in turn.

The reported series shows how the scale of that channel changed across the period from 2006 to 2016. OpenSecrets reported that non-disclosing nonprofits reported less than 5.2 million dollars to the Federal Election Commission in the 2006 cycle. OpenSecrets reported that the corresponding figure rose to more than 300 million dollars in the 2012 cycle. OpenSecrets reported that the figure stood at about 181 million dollars in the 2016 cycle. The scope of the series has to be stated each time the figures appear, because the series covered amounts reported to the Federal Election Commission by non-disclosing nonprofits. The series did not purport to capture every dollar of election related spending by such organizations, and the distinction between a reported amount and a complete amount becomes central in the broader 2020 estimate. Within the reported scope, the movement from less than 5.2 million dollars in 2006 to more than 300 million dollars in 2012 marked a transformation in magnitude between a midterm cycle before the 2010 decisions and the first presidential cycle conducted entirely after those decisions. The decline to about 181 million dollars in 2016, measured on the same reported scope, showed that the reported channel did not move in a single direction across every cycle, even as the amounts remained far above the 2006 baseline.

Transfers between affiliated nonprofits compounded the opacity that a single nondisclosing contribution created. A contribution from one nonprofit to an affiliated nonprofit, followed by spending or a further contribution by the recipient organization, placed another organizational layer between an original donor and an eventual expenditure. Each transfer could be lawful, each organization could file the reports required for the activity that organization conducted, and the chain as a whole could remain untraceable to an original source because no link in the chain carried an obligation to disclose the donors behind a nonprofit. The shell company route produced a parallel result through a different organizational form. When a super PAC disclosed a shell company as a donor, the report named a legal entity whose filing gave a reader no path to the person or organization that had funded the entity for the contribution. In both routes, the deficiency lay in the depth of disclosure rather than in the existence of a report. A researcher, a journalist, or a voter reading a super PAC filing encountered a donor name that satisfied the form of disclosure while withholding the information that disclosure was designed to supply, namely the identity of the original source whose money entered federal election spending.

The 2020 cycle exposed the distance between reported dark money and estimated dark money more fully than any earlier cycle in the record. OpenSecrets estimated that dark money groups accounted for more than 1 billion dollars in the 2020 cycle when direct spending and contributions to super PACs were combined, and OpenSecrets placed about 660 million dollars of that total as money given to super PACs. The combined scope is essential to the figure. The estimate of more than 1 billion dollars did not describe direct outside spending alone, and the about 660 million dollars given to super PACs formed a component inside the broader total rather than an addition to that total. OpenSecrets also reported that the amount visible as direct outside spending in Federal Election Commission reports was far smaller than the broader estimate, because much nonprofit election spending was never reported to the Federal Election Commission at all. That explanation separates two different measures that a careless account can merge. A Commission report total captures spending that a filer reported under the applicable reporting rules. A broader estimate attempts to account for direct spending and for contributions routed into super PACs, including spending that never entered a Commission report. The distance between the broader estimate and the far smaller reported amount measured the disclosure gap in 2020 more directly than any single reported series could measure the gap.

The party direction of dark money changed across the same period, and a complete account has to report use by groups aligned with both parties in every cycle where the record supplies a split. OpenSecrets reported that dark money leaned heavily conservative through the 2016 cycle. The organizations that benefited from nondisclosing nonprofit spending during the early post 2010 cycles were, in the aggregate measured by OpenSecrets, principally on the conservative side, and the reported totals for 2012 and 2016 arose within that distribution. A description that ended in 2016 would therefore leave a reader with a picture of a channel used unevenly and associated mainly with conservative groups. That picture ceased to describe the aggregate record in the cycles that followed. OpenSecrets reported that 2018 produced the first midterm result in which dark money benefited Democrats more, and the presidential record then changed direction as well.

In the 2020 cycle, OpenSecrets tallied reported donations and spending and found that liberal groups directed more than 514 million dollars of dark money and conservative groups directed about 200 million dollars. The scope of that split was the dark money OpenSecrets could tally in reported donations and spending, and the two figures describe the tallied record rather than a complete census of every nondisclosed dollar in the cycle. Within that scope, 2020 was the first presidential cycle in which dark money benefited Democrats more than Republicans. The sequence matters for framing. The same legal channel that leaned heavily conservative through 2016 produced, first in the 2018 midterm and then in the 2020 presidential cycle, an aggregate advantage for Democratic aligned groups in the tallied record. Both parties aligned groups used the channel across the period, and the party that drew more from the channel varied by cycle. That variation cautions against treating dark money as the property of one party and against treating a single cycle split as a permanent feature of campaign finance law.

The gap coexisted with a disclosure record that captured a great deal. Candidate campaigns filed reports that identified contributors under the contribution and disclosure structure built by federal law. Political committees filed reports that placed receipts and expenditures in the public record. Super PACs disclosed donors at the level of the contributor named in the filing, and that disclosure allowed a reader to follow enormous sums from named individuals and named organizations into committees and then into independent expenditures reported to the Federal Election Commission. The OpenSecrets totals for outside spending, for super PAC spending, and for reported nonprofit spending all rested on that public record. Without filings by committees and spenders, no series could have shown total outside spending excluding party committees reaching about 1.04 billion dollars in 2012, about 1.41 billion dollars in 2016, and about 2.9 billion dollars in 2020. Disclosure in those cycles produced the very evidence used to identify the gap inside disclosure.

Bundled contributions supplied another area in which disclosure reached activity that a focus on dark money alone can obscure. Under the lobbying law enacted in 2007, bundled contributions by lobbyists were disclosed, so a contribution assembled by a lobbyist from other contributors entered the public record under a rule aimed at the intermediary role rather than only at the individual checks that composed the bundle. That disclosure channel connects campaign finance law to the Lobbying Disclosure Act of 1995, the statute through which lobbying activity and lobbyist identity entered a separate public reporting system. A lobbyist bundle and a nonprofit contribution to a super PAC presented opposite disclosure problems during the cycles covered by the record. The bundle involved money from identifiable contributors gathered by an identifiable intermediary, and the 2007 rule required disclosure of that bundling activity. The nonprofit contribution involved a disclosed organizational donor whose own donors remained undisclosed, and no equivalent look through rule supplied the original names. Disclosure therefore operated with real force in one channel and stopped at an organizational boundary in another, within the same federal system and during the same cycles.

A balanced assessment of the period from 2010 to 2020 therefore holds two findings together without allowing either finding to erase the other. The first finding concerns scale and composition. OpenSecrets measured total outside spending excluding party committees at about 2.9 billion dollars in 2020, after about 1.41 billion dollars in 2016 and about 1.04 billion dollars in 2012, and OpenSecrets analyses assigned the largest growth in outside funding to wealthy individuals rather than to direct corporate treasury money, which formed a modest share of super PAC receipts. The second finding concerns depth. A disclosure system affirmed in Citizens United v. FEC by a vote of 8 to 1 and applied through committee reports captured spenders and named donors across those cycles, and the same system permitted a named donor in the form of a section 501(c)(4) organization, a section 501(c)(6) trade association, or a shell company to terminate the inquiry. Money after 2010 became more visible in aggregate and less traceable at the level of original source, and both halves of that sentence describe the system that the 2010 decisions and the disclosure rules together produced.

The Channel Table: Closures and What Grew After

The sequence of closures and openings across the period fits into a single view, in which each regulatory event closed or opened a defined channel, another channel grew in response or in parallel, and the disclosure status of the channel that grew determined how much of the movement a voter or a researcher could trace in the public record.

Regulatory event	Year	Channel closed or opened	Channel that grew	Disclosure status of the channel that grew
Federal Election Campaign Act amendments	1974	Closed undisclosed and unlimited giving to federal candidates	Limited and disclosed contributions	Donors disclosed
Party building amendments signed in 1980 and named for 1979	1980	Opened party building spending outside the federal definitions	National party soft money	Sources reported to the Federal Election Commission and amounts unlimited
Bipartisan Campaign Reform Act	2002	Closed national party soft money	527 organizations and then politically active nonprofits	Donors to 527 organizations disclosed and donors to nonprofits undisclosed
Federal Election Commission enforcement against 527 committees	2006 to 2007	Closed unregistered 527 election spending at the scale reached in 2004	Election spending by organizations under section 501(c)(4)	Donors undisclosed
Citizens United and SpeechNow decisions	2010	Closed bans on corporate and union independent expenditures and contribution limits for committees making only independent expenditures	Super PACs	Donors disclosed except where the donor is a nonprofit that does not disclose donors

What the Money Buys: Votes, Access, and the Consumption Finding

Debate over campaign finance law often begins with a total, as if the size of the total settled what the money did. Concern grew with the totals for federal elections across the cycles covered in this series, on an assumption that spending translated directly into electoral victories, legislative votes, and policy results. The outcomes literature assembled for this article tested that assumption piece by piece, and the record that emerged was narrower and more conditional than the assumption allowed. Money correlated with winning, and money accompanied access. Yet studies that tried to isolate a causal effect on votes produced either a small estimate, an estimate that vanished under a stronger design, or a finding that pointed in a different direction from the study beside it.

Does the candidate who spends more win?

The candidate who spends more wins most House races, according to the record summarized on the fact sheet for this series. That pattern alone does not prove that spending caused the victories, because donations also follow expected winners and threatened incumbents raise the most money.

The correlation itself was never in dispute. In House races, the higher spending candidate won most contests, and the margin in spending between two candidates often tracked the margin in votes closely enough to look like a price list. A candidate who raised very little tended to lose, sometimes by a wide distance. Taken at face value, a table of spending and outcomes from any single cycle appeared to show money deciding elections in a routine fashion. The outcomes literature began, in large part, as an objection to the ease of that inference, because a table that recorded who spent more and who won could not by itself say which fact produced the other.

The objection rested on the direction of the flow. Donors did not distribute contributions at random, and contributors who wanted a relationship with a future officeholder, or who simply enjoyed backing a winner, gave disproportionately to candidates already expected to win. Party committees, political action committees, and individual contributors all had reasons to read polls, district partisanship, and incumbent status before writing a check, and the resulting contributions recorded those expectations as much as contributions created them. A strong candidate attracted money because victory looked likely, and then won for reasons that included candidate quality, district composition, and prior reputation. In that sequence, spending and winning shared a common cause in expected strength, and the raw correlation credited spending with work that expectation had already performed before the first advertisement aired.

Incumbency sharpened the problem from the other direction. Incumbents outraised challengers for structural reasons that had little to do with the persuasive power of any particular expenditure. An incumbent held the office, cast votes that organized interests monitored, appeared in local coverage without buying time, and offered contributors a continuing relationship rather than a speculative one. Challengers often began unknown, without a donor list, without a record to defend or promote, and without assurance that a contribution bought anything beyond support for a probable defeat. The funding gap therefore measured the advantages of office itself, folded into a dollar figure. When an incumbent outspent a challenger and won, the result could reflect the office, the district, challenger weakness, the spending, or any combination of the four, and a simple comparison of totals could not separate the strands.

The simultaneity problem named the deeper difficulty, and the name described a circle rather than a single bias. Spending influenced votes, if spending influenced votes at all, at the same time that expected votes influenced spending. An incumbent who looked safe raised modestly and coasted, recording low spending and a large victory. An incumbent who faced a credible threat raised urgently, spent heavily on defense, and either survived narrowly or lost despite the effort, recording high spending beside a poor result. A naive reading of those two cases concluded that incumbent spending backfired or accomplished nothing, because the largest incumbent expenditures clustered in the closest and most dangerous races. The spending did not create the danger. The danger created the spending, and the spending may also have limited the damage, a possibility that the raw figures could neither confirm nor exclude.

Challenger spending sat at the opposite end of the same circle. A challenger who raised nothing remained obscure and lost heavily, while a challenger who raised enough to advertise, travel, and hire staff became visible and competitive. The correlation for challengers therefore looked strong partly because money marked the threshold between a nominal candidacy and a real one. Disentangling the effect of the expenditure from the effect of the seriousness that produced the expenditure defeated the simplest statistical approaches, and that defeat set the agenda for decades of methodological argument among named researchers who read the same elections and reached incompatible conclusions.

Gary C. Jacobson of the University of California, San Diego, whose congressional elections research dated from 1978 onward, produced the estimates that became the standard starting point for the argument. In Jacobson research, challenger spending was strongly associated with vote share, while incumbent spending showed little or no association. The asymmetry had an intuitive account. Incumbents entered a campaign already known to constituents after years of casework, newsletters, local appearances, and prior elections, so additional incumbent advertising added little information and moved few voters. Challengers entered largely unknown, so challenger spending purchased recognition itself, and recognition was a precondition for any voter who might consider a change. Under that account, a dollar spent by a challenger and a dollar spent by an incumbent were different goods, and unequal effects followed naturally.

The same Jacobson estimates carried a warning inside the result. Incumbents raised and spent the most when threatened, so heavy incumbent spending served as a signal of vulnerability rather than strength, and the threatened incumbent dragged the estimated effect of incumbent spending downward, since high expenditure arrived packaged with a strong opponent or another liability that prompted the fundraising. The standard estimates therefore mixed the votes that incumbent spending might have gained with the votes that the incumbent was already in the process of losing when the spending decision was made.

Steven D. Levitt pursued the separation through repetition. In a 1994 study, Levitt compared House candidates who faced the same opponent in successive elections, a repeat candidate design that held constant many features that ordinarily confounded spending estimates. When the same two candidates met again, candidate quality, district character, and personal appeal changed far less than in a comparison between unrelated races, so changes in relative spending offered a cleaner test. Levitt reported that the estimated effect of spending fell by an order of magnitude under that design. In Levitt’s own summary of his estimates, an additional 100,000 dollars might be expected to add about a third of a percentage point to a candidate’s vote total. An effect of that size could matter in an exceptionally close race. The estimates did, however, shrink the effect far below the level implied by the raw correlation, and the shrinkage supported the view that most of the apparent power of money in ordinary comparisons belonged to the candidates and districts that attracted the money.

Alan S. Gerber answered with a different separation strategy and reached a different result. In a 1998 study published in the American Journal of Political Science, Gerber examined Senate elections and used candidate wealth as an instrument for spending, an approach designed to isolate variation in spending that came from a candidate’s personal resources rather than from expected closeness or opponent strength. When the endogeneity of spending was addressed through that instrument, Gerber found that the marginal effects of incumbent and challenger spending looked roughly equal, contrary to the asymmetry in the standard estimates associated with Jacobson. Gerber’s estimates implied that the average incumbent spending advantage translated to about six percentage points of vote share, a substantial effect by any measure used in congressional elections research. The Gerber result applied to the Senate context that the study examined, and within that context the finding restored incumbent spending to real electoral significance.

The three positions formed a genuine unresolved dispute, and the cause of the persistence was methodological rather than rhetorical. Jacobson’s standard estimates described associations across many races but remained exposed to simultaneity, because expected closeness drove spending decisions. Levitt’s repeat candidate design controlled for durable candidate and district traits by construction, but repeat matchups formed a selected and unusual subset of elections, and estimates from rematches might not generalize to first meetings between candidates. Gerber’s instrumental variables approach depended on candidate wealth affecting votes only through spending, an assumption that other researchers could contest, since wealth might also signal candidate background or correlate with personal networks that mattered independently. Each design repaired a flaw in the design before it and introduced a vulnerability of its own. Election data offered no experiment in which spending was assigned at random to comparable candidates, so the dispute survived because the evidence available to every participant left a defensible line of attack against every estimate on the table.

A second literature approached the money question from the contributor side and asked why the sums were so small relative to the stakes. Stephen Ansolabehere, John M. de Figueiredo, and James M. Snyder Jr. posed that question in an article titled Why Is There so Little Money in U.S. Politics, published in the Journal of Economic Perspectives, volume 17, number 1, in 2003. The article built on a puzzle associated with Gordon Tullock, posed about thirty years earlier. Tullock’s puzzle, as Ansolabehere, de Figueiredo, and Snyder presented the argument, held that the value of government policy to affected industries and groups dwarfed the amounts those interests spent on contributions, so an investment account of giving predicted far more money than American politics contained. Candidates, parties, and organizations raised nearly 3 billion dollars in the 1999 to 2000 cycle, a large absolute figure that the authors treated as puzzlingly small against the scale of federal taxation, spending, and regulation decided in Washington.

The answer that Ansolabehere, de Figueiredo, and Snyder developed recast the contribution itself. Contributions, the authors found, behaved mainly as political participation and consumption rather than as investment. A contributor gave much as a person bought a ticket, joined an organization, or supported a cause whose success provided satisfaction in itself, and the return arrived in the act of participation, in an expression of identity or conviction, rather than in a later policy payment that repaid the gift with interest. The consumption account explained features of giving that the investment account handled poorly, including the willingness of contributors to back losing candidates and to give amounts too small to purchase influence over any plausible calculation. Under the consumption account, the modest scale of contributions stopped being an anomaly. People spent on politics as people spent on other expressive goods, in amounts limited by the satisfaction of participation rather than by the enormous value of the policies at stake.

The authors also reassigned the identity of the typical contributor. Individuals, not special interests, were the main source of contributions in the account assembled by Ansolabehere, de Figueiredo, and Snyder. That finding cut against a familiar picture in which organized interests supplied the bulk of campaign money in expectation of a return, because a giving population dominated by individuals fit expression and participation more readily than it fit a coordinated purchase of outcomes. Individuals gave for reasons that included partisanship, ideology, candidate loyalty, and social connection, and those motives required no quid pro quo to sustain them. The composition of the money therefore mattered as much as the amount, and the authors’ evidence placed American campaign finance closer to a market for participation than to a market for policy for the period examined.

Measured against the economy, contributions also lacked the growth trajectory that an escalating auction for policy would have suggested. Ansolabehere, de Figueiredo, and Snyder found that contributions as a share of gross domestic product had not risen appreciably in over a century and had probably fallen. Campaigns cost more in nominal dollars across the twentieth century, and federal elections in the 1999 to 2000 cycle raised nearly 3 billion dollars, but the economy grew as well, and the political share of national output held roughly flat across a span that included the rise of mass media and the professional campaign. Nominal growth was real. Growth relative to the society paying the bills was, on the authors’ measurement, absent across more than a hundred years of elections.

The same article reviewed the prior literature on the most direct version of the purchase claim, the claim that contributions bought legislators’ votes. Ansolabehere, de Figueiredo, and Snyder found that only one in four studies in the prior literature supported the claim that contributions buy legislators’ votes. The authors further reported that controlling for unobserved constituent and legislator effects left little relationship between money and votes. Constituency opinion pushed a legislator toward particular positions, a legislator’s settled views pushed in the same direction, and contributors tended to give to legislators who already agreed with them. Once those aligned forces were held constant, the independent association between a contribution and a vote largely dissipated. Agreement attracted money, and money then received credit for agreement that preceded it.

Does a contribution buy a meeting with a member of Congress?

In the Kalla and Broockman experiment, identification as campaign donors made senior policymakers available between three and four times more often for a requested meeting. The experiment measured access offered by 191 congressional offices, and a meeting offered is different from a vote changed or a policy outcome delivered.

Joshua L. Kalla and David E. Broockman supplied the strongest experimental evidence in the literature with a study titled Campaign Contributions Facilitate Access to Congressional Officials: A Randomized Field Experiment, published in the American Journal of Political Science, volume 60, number 3, in 2016, on pages 545 to 558. The study was the first randomized field experiment on the question. Random assignment mattered because every observational study of access faced the same objection that applied to spending and votes: donors differed from nondonors in agreement, engagement, wealth, organization, and prior connection, so any greater access donors enjoyed might have reflected those differences rather than the contribution itself. By randomizing the information that an office received, Kalla and Broockman broke that circle for one specific question, whether identifying prospective meeting attendees as campaign donors changed the access that a congressional office offered.

The design was built with a political organization, CREDO Action, and operated through real requests. Working with CREDO Action, the researchers sought meetings between the organization members and 191 congressional offices. The offices were randomly assigned to be told, or not told, that the prospective attendees were campaign donors. Every other feature of the request proceeded through the ordinary process by which a constituent group asked to discuss an issue with an office, so the randomized disclosure was the difference between the two groups of offices by construction. The outcome was concrete and observable. An office either made a senior policymaker available for the meeting or routed the request to a less senior figure or declined the senior meeting, and the researchers could compare the rate at which each kind of response occurred across the two randomized groups without relying on recollection or self reporting by donors.

The result was large by the standards of the outcomes literature. When offices were told that the attendees were donors, senior policymakers, defined in the study as members of Congress and chiefs of staff, made themselves available between three and four times more often. The contrast with the spending literature was instructive. Decades of observational work on campaign expenditures had produced estimates that shrank, reversed, or diverged depending on the design, while a single randomized design on access produced an effect that was large, in one direction, and attributable to the donor identification itself because randomization had balanced the other characteristics of the requesting groups. The offices responded to a label attached to the request. The label carried the information that the people asking for time had given money in politics, and that information alone, randomly assigned, changed who in the office made time available.

Precision about the finding matters as much as the size of the finding. The experiment measured access offered, not votes changed. A meeting with a member of Congress or a chief of staff placed a group in the room where priorities were heard and follow up became possible, and access of that kind held obvious value for any organization trying to be heard. Access was nonetheless an input into a process rather than the output of the process. A senior policymaker could listen and vote against the visitors’ position, or listen and then follow constituency opinion, party leadership, or personal conviction to the opposite result. The design measured the door opening, not what happened to legislation after the door opened.

The access finding and the consumption finding fit together more comfortably than either fit with a simple purchase model. If contributions functioned mainly as participation, as Ansolabehere, de Figueiredo, and Snyder argued in 2003, then contributors were engaged participants whom offices had ordinary political reasons to meet, and donor status served as a visible credential of that engagement. Kalla and Broockman showed in 2016 that the credential worked even when it was the only thing randomized. Neither finding required a legislator to sell a vote. Both described a system in which giving expressed commitment, commitment earned a hearing, and a hearing stopped well short of a guaranteed result.

The limits of money at the electoral end of the system received a vivid demonstration in the 2020 presidential primaries. Tom Steyer and Michael Bloomberg together spent about 1.4 billion dollars in the 2020 primaries, according to the Center for Responsive Politics, and neither approached the Democratic nomination. Bloomberg’s only victory was the American Samoa caucuses. The combined spending bought advertising saturation, large staffs, and months of visibility that ordinary candidates could not match, and the spending failed to produce a delegate position anywhere near the nomination. The episode did not prove that money lacked influence in primaries generally. The episode did establish a ceiling in at least one conspicuous case, in which additional spending confronted voters who had formed judgments on other grounds.

A result from the preceding presidential cycle made the same point at the general election level. The better funded presidential campaign lost in 2016. That single outcome proved little on its own, since one election turned on the Electoral College, state level margins, and candidate characteristics that no spending figure captured. Placed beside the Steyer and Bloomberg experience in 2020, the 2016 result served as a reminder that the relationship between a funding advantage and victory was probabilistic and contested, never mechanical. Money bought the means of competition, including staff, travel, and communication with voters. Competitiveness purchased in that sense differed from victory purchased outright.

The synthesis that the evidence supported by the close of the 2020 cycle was therefore specific rather than general in either direction. On votes in Congress, Ansolabehere, de Figueiredo, and Snyder found in 2003 that only a minority of prior studies supported a purchase claim and that little relationship between money and votes survived controls for constituent and legislator effects. On votes in elections, Jacobson, Levitt, and Gerber left a dispute in which the effect of spending ranged from roughly a third of a percentage point per additional 100,000 dollars in Levitt’s summary, to about six percentage points for an average incumbent spending advantage in Gerber’s Senate estimates, with the methodological cause of the disagreement identified and unresolved. On access, Kalla and Broockman produced in 2016 the one experimental result in the set, and the effect was between three and four times more frequent availability of senior policymakers for attendees identified as donors. Money’s best documented purchase was a hearing, not an outcome, and the distinction between a hearing and an outcome is the distinction that the entire outcomes literature, read at the strength of its own designs, sustained.

The Quiet Death of Presidential Public Financing

Presidential public financing died without a repeal vote and without a court order striking the program from the statute books. The system built in the 1970s remained written into campaign finance law long after the candidates it was built for had stopped using it, and its decline therefore reads less like an abolition than like an abandonment. A fund continued to receive designations from taxpayers, and rules continued to govern payments that fewer and fewer serious candidates requested. The architecture outlasted the behavior that had given the architecture a purpose, and understanding that distance requires attention to the original bargain, to the decades when the bargain held, and to the sequence of refusals that broke it.

The architecture began with money set aside before the modern payment system existed. The tax checkoff and the Presidential Election Campaign Fund were created by the Revenue Act of 1971, Public Law 92-178, and the checkoff first appeared on tax returns for 1972. The design was deliberately modest at the level of the individual filer and deliberately large at the level of aggregation. A filer who marked the checkoff designated three dollars of existing tax liability to the fund. The designation added nothing to the tax owed and subtracted nothing from a refund, a feature that separated the act from a donation in any ordinary sense. The filer gave up no income, wrote no separate check, and joined no organization. The costlessness was the point. Congress had created a mechanism through which millions of filers, each directing a sum too small to alter a household budget, could together accumulate a public treasury for presidential campaigns, and the legitimacy of that treasury rested on breadth rather than on size. A campaign financed from the fund could claim support drawn from the taxpaying public as a whole, expressed through a routine administrative act repeated across filing seasons, rather than from a narrower circle of contributors writing checks of their own.

The payment system arrived with the 1974 amendments to the Federal Election Campaign Act, Public Law 93-443, which built the modern structure of contribution limits, disclosure, the Federal Election Commission, and presidential public financing in a single settlement. On the foundation of the fund, the amendments constructed two distinct forms of assistance aimed at two distinct stages of a presidential campaign. In the primaries, a candidate could receive matching funds, public payments that supplemented private fundraising for a candidate who accepted spending limits as a condition of participation. The matching design preserved a role for private effort. A candidate in the primaries demonstrated support by raising money, and the public payment followed that demonstration rather than replacing it, while the attached limits disciplined the scale of the contest the payments helped to finance. The arrangement rewarded breadth of support and imposed restraint at the same time, and acceptance was voluntary, a choice each candidate weighed against the freedom to spend without the program ceiling.

The general election grant operated on a different and more demanding logic. A major-party nominee who accepted the grant received full public funding for the general election campaign and, in exchange, forswore private fundraising for that campaign. Where the primary system matched private money, the general election system replaced it. The nominee campaigned on the grant alone, under the spending limit the grant carried, and the private contest for general election dollars ended for that candidate at the moment of acceptance. The severity of the bargain was inseparable from its appeal. A nominee freed from fundraising could devote the compressed weeks of a general election campaign to persuasion rather than to solicitation, and the public could watch a contest in which neither major candidate depended on private contributors for the resources that carried a message to the electorate. The grant also equalized the starting position of the two major-party nominees, because each received public funding on the same terms rather than entering the autumn with treasuries shaped by unequal success among donors. Public money first reached a presidential campaign in 1976, and the system that delivered it embodied a confident premise: that a fixed public payment, honestly sized at the moment of design, could sustain a competitive national campaign.

The era of participation lasted from 1976 through 2004 and gave the system the appearance of permanence. Every major-party nominee accepted the general election grant from 1976 through 2004, a span that covered contests of very different character, conducted by nominees of both parties, under conditions of incumbency and of open competition. Acceptance became a routine feature of the general election, so expected that the grant functioned as part of the background machinery of a presidential campaign rather than as a contested policy choice within it. Nominees planned around the payment, campaigned within its limit, and treated the forswearing of private general election fundraising as a normal discipline rather than as a sacrifice. The premise of the design held across that span because the relationship between the grant and the price of a national campaign held. A nominee could run a competitive campaign on the grant alone, reaching the electorate through the advertising, travel, organization, and communication that the payment could buy, without falling visibly behind an opponent bound by the same terms. Mutual acceptance reinforced itself. When both nominees took the grant, neither gained an advantage from refusal, and the competitive risk of participation stayed low, while the reputational risk of refusing public money for private fundraising remained untested because no major-party nominee had assumed it.

The first crack appeared inside that era rather than after it, and it appeared in the primaries. In 2000 George W. Bush declined primary matching funds while accepting the general election grant, splitting the program into parts and treating each part as a separate calculation. The Congressional Research Service describes him as the first person elected President since 1976 without participating in all elements of the program. The significance of the 2000 decision lay in its partial character. Bush preserved the general election bargain, with its full grant and its complete replacement of private fundraising, while discarding the primary bargain, where a candidate with formidable private fundraising capacity found the matching payment less valuable than freedom from the primary spending limits attached to it. The episode demonstrated that participation was not a single commitment to a philosophy of public financing but a series of stage-by-stage judgments about money and restraint, each reversible on its own terms. A candidate strong enough as a fundraiser could refuse assistance at the stage where private strength was greatest and accept assistance at the stage where the public payment remained competitive, and the system had no mechanism to answer a refusal of that kind except to record it.

The decisive break came in 2008, when Barack Obama declined the general election grant itself, the first major-party nominee to forgo general election public funding since the system began. John McCain accepted the grant of about 84.1 million dollars and its spending limit, and the 2008 general election therefore placed the two halves of the old bargain in direct view within a single contest. One nominee campaigned under the ceiling that acceptance imposed. The other campaigned free of that ceiling, relying on private fundraising for the general election that the program had been designed to displace. The Congressional Research Service describes Obama as the first person elected President since 1976 without accepting any public funds, a description that marks the distance traveled from 2000. Bush had declined one element and accepted another. Obama declined the central payment of the system, won the election without it, and established by demonstration that a nominee could refuse the grant, raise private money for a general election campaign, and prevail. Precedent in a voluntary system matters more than statutory text, because each refusal lowers the expected cost of the next refusal. After 2008, declining public money was no longer an untested gamble. It was a strategy with a winning example attached, and the reputational protection that universal participation had once given the program, the sense that acceptance was simply what nominees did, dissolved with that example.

The cycles that followed converted an exception into a pattern. In 2012 no major candidate accepted public funds, as the Congressional Research Service records, and the absence extended across the field rather than attaching to a single campaign. In 2016 and 2020 neither major-party nominee accepted the general election grant or primary matching funds, and public money in those cycles went to minor candidates only. The reversal was complete in functional terms. A program designed to finance the central contest of American politics, the general election campaign between the major-party nominees, had become a resource for candidates at the margins of that contest, while the nominees with a realistic path to the presidency financed campaigns entirely outside it. Minor candidate participation preserved the program as an operating system, with applications processed and payments made under the established rules, and that preservation can mislead an observer who counts activity rather than consequence. The measure of the system was never whether any candidate anywhere accepted public money. The measure was whether the candidates who might govern the country conducted campaigns on public funds under public limits, and by that measure the elections of 2016 and 2020 took place after the death of the arrangement, whatever administrative life the program retained.

The citizen side of the system told the same story on a longer and quieter timeline. Checkoff participation, as the Congressional Research Service reports from Treasury Department Financial Management Service figures, peaked in 1980, when 28.7 percent of filers designated funds to the Presidential Election Campaign Fund. That peak arrived early in the life of the system, and the line from that point descended across the following decades. Participation fell below 15 percent every year since 1993, according to the same Congressional Research Service reporting from Treasury Department figures, and the decline continued into the cycles when candidate participation collapsed. The figure reached 6.0 percent in 2012 and 5.4 percent in 2015, levels that transformed the checkoff from a mass act of designation into the habit of a small minority of filers. Fewer than six filers in a hundred directed three dollars of existing tax liability to the fund in 2015, and each such designation cost the filer nothing, which makes the decline difficult to explain as a response to price. The checkoff asked for no money, no membership, and no further effort beyond a mark on a return. Refusal required nothing either, and indifference, repeated across filing season after filing season, accomplished what organized opposition never had to attempt.

The fund itself accumulated a balance even as participation fell, because for years designations exceeded payouts to a shrinking pool of participating candidates, a paradox that captures the whole trajectory in a single accounting fact. Money entered the Presidential Election Campaign Fund faster than candidates drew money out of it. In a healthy version of the system, a falling participation rate among filers might have signaled a supply problem, a treasury draining toward emptiness as public demand consumed it. The opposite condition developed. Supply, diminished as it was, outran demand, because the demand side depended on candidates willing to accept spending limits, and that willingness vanished among major candidates after 2008. The accumulating balance was therefore not a sign of strength awaiting use. It was the residue of a bargain one side had stopped striking, funds gathered for campaigns that their nominees had chosen to finance elsewhere. A balance can grow while a program dies, provided the program collects automatically and pays out only by consent, and the Presidential Election Campaign Fund met both conditions across the years when its purpose drained away.

Why did candidates stop taking public money?

Candidates stopped taking public money because the grant and its attached spending limit, fixed by formula, fell far below what a major-party nominee could raise privately, so acceptance meant campaigning under a ceiling that refusal avoided at no legal cost beyond forgoing the public payment itself.

The mechanism deserves precise statement, because the death of public financing is often described as a change in political culture when the record supports a colder explanation grounded in arithmetic. The grant was fixed by formula. Its size followed a statutory calculation rather than a judgment, renewed each cycle, about what a competitive general election campaign required. The spending limit attached to the grant inherited the same fixity, since a candidate who accepted the payment accepted the ceiling as a single package. Private fundraising capacity followed no equivalent formula. A nominee with a national donor network, a compelling candidacy, and months of organized solicitation could raise sums governed by the scale of support and the intensity of the contest rather than by a calculation settled in statute. As long as the formula amount and the privately raisable amount stayed within reach of each other, the package remained attractive, because the grant arrived without the labor and obligations of raising it. When the two amounts separated, the package inverted. Acceptance no longer purchased freedom from fundraising at a fair price. Acceptance purchased a lower ceiling than refusal allowed, and the payment became compensation for a competitive handicap that grew with every cycle in which campaign costs advanced beyond the formula.

The totals recorded for federal elections give the separation a scale, as estimates by the Center for Responsive Politics placed total spending in federal elections at about 5.3 billion dollars in the 2008 cycle, about 6.5 billion dollars in the 2016 cycle, and about 14.4 billion dollars in the 2020 cycle, all in nominal dollars across candidates, parties, and outside groups. Those totals describe the whole federal system rather than a single presidential campaign, and their relevance here is directional rather than exact. The environment in which a nominee decided whether to accept a grant of about 84.1 million dollars, the amount John McCain accepted in 2008, was an environment of expanding campaign costs at every level, in which the price of advertising, organization, data, travel, and communication pressed upward while the public payment followed its formula. A nominee weighing the grant weighed a known and bounded sum against an open field of private possibility, in a contest where the opponent might refuse the same offer and spend beyond it. Under those conditions, refusal was not an ideological statement about campaign finance law. Refusal was the rational response of a campaign that expected to raise more than the grant provided and declined to surrender that expectation for a payment smaller than the advantage surrendered.

The contraction reached the edges of the program as well as its center. 2014 legislation (Gabriella Miller Kids First Research Act, Public Law 113-94) ended public funding of the party conventions. The conventions had occupied a supporting place in the architecture, public money assisting the gatherings at which nominees were formally presented, and the removal of that element by statute stands as the exception that clarifies the pattern. Where Congress acted, it acted on a peripheral component. The central component, the general election grant to major-party nominees, was never repealed, never struck down, and never formally closed. It ended by refusal rather than by statute, emptied by the accumulated decisions of candidates who found its terms uncompetitive and by the accumulated indifference of filers who stopped designating the three dollars that stocked its fund. An abandonment required nothing from anyone, and nothing is what the institution supplied.

What remains instructive in the episode is the dependence of a voluntary regulatory bargain on the continuing competitiveness of its terms. Campaign finance law in the public financing chapters offered money in exchange for restraint, and the exchange functioned across the era of participation because the money was worth taking and the restraint was worth accepting at the price paid. No scandal destroyed the program, no judgment invalidated it, and no Congress repealed it. The formula stayed fixed while the market for campaigns moved, the grant became a ceiling rather than a foundation, and the nominees walked away in sequence, first in part, then in full, then as a matter of course across 2012, 2016, and 2020. The checkoff box stayed printed on the return while the share of filers marking it fell below 15 percent every year since 1993 and reached 5.4 percent in 2015. The fund stayed open while its balance grew on designations that outpaced payouts to a shrinking pool. Every element of the system survived except the willingness that had made the elements into a system, and willingness, once lost to a demonstrated alternative, proved to be the one component that neither a formula nor a fund could supply.

Enforcement by Deadlock: The Commission That Stopped Deciding

The Federal Election Commission was built with a brake inside the engine. Under the structure described in the fact record for campaign finance law, the Commission has six commissioners, and no more than three may come from one party. Four affirmative votes are required to open an investigation, to find a violation, to issue an advisory opinion, or to make a rule. A vote of three to three therefore does not produce a weaker form of action, a delayed form of action, or a provisional form of action. A vote of three to three produces no action at all, and the matter stops where the tie leaves it.

That consequence was not an oversight tucked into an otherwise majoritarian design. A reader of the structure has to confront the plain arithmetic. Congress created an even numbered body, divided it evenly between the parties at its maximum partisan concentration, and then set the threshold for consequential decisions above the number that either party bloc could supply on its own. Every enforcement decision of consequence required at least one vote from across the partisan division. Every such decision could also be blocked by withholding that vote. The statute permitted deadlock as a regular state of the institution, and for long stretches the institution occupied that state.

Has the Federal Election Commission stopped enforcing the law?

No, not entirely, but enforcement narrowed sharply on contested matters between 2006 and 2016. The Commission continued receiving and publishing disclosure, yet deadlocked votes closed cases without action, penalties fell, advisory guidance stalled, and quorum losses during the 2020 cycle left hundreds of enforcement cases waiting unresolved for months.

The distinction inside that answer matters, because the Commission performs more than one function and the functions did not fail together. Disclosure is largely administrative. Reports arrive, staff process them, and the public record grows whether commissioners agree with one another or not. Enforcement is discretionary at its decisive points. Someone must vote to investigate, someone must vote to find reason to believe a violation occurred, and someone must vote to pursue a penalty or a settlement. Guidance sits between those poles, because an advisory opinion requires the same four votes even though no respondent stands accused of anything. Deadlock therefore struck selectively, and it struck hardest exactly where judgment, rather than intake, decided the outcome.

The bipartisan threshold had a defensible purpose, and that purpose deserves statement before the record of its effects. An agency that polices the fundraising and spending of candidates and parties holds power over the electoral process itself. A bare majority drawn entirely from one party could, in the worst case, direct investigations toward the opposing party and away from its own. Requiring four votes meant requiring some measure of cross party agreement before the machinery moved against a respondent. The design traded speed and decisiveness for a guard against partisan enforcement, and the trade was visible on the face of the statute from the beginning.

The cost of that trade depends on how often the guard becomes the outcome. If cross party agreement is ordinarily available, the threshold costs little and buys legitimacy for the decisions that emerge. If cross party agreement fails on a large share of contested matters, the threshold stops functioning as a check on enforcement and starts functioning as a substitute for enforcement, deciding cases by preventing decisions. The question that the deadlock record answers is an empirical one. Between 2006 and 2016, according to the most detailed contemporary accounting, the frequency of that failure rose by an order of magnitude.

That accounting came from Commissioner Ann Ravel, whose report Dysfunction and Deadlock appeared in February 2017. Ravel analyzed Commission votes in enforcement matters closed between 2006 and 2016, drawing on the Commission’s own data. Her central finding concerned substantive enforcement votes, the votes that determine whether a matter advances or ends. In 2006, deadlocks accounted for 2.9 percent of those votes. In 2016, deadlocks accounted for 30 percent. A mechanism that had been exceptional in the earlier year had become routine in the later one.

Ravel’s second measure looked at cases rather than individual votes, and it told the same story from a different angle. The share of enforcement cases in which commissioners deadlocked on at least one alleged violation rose from 4.2 percent in 2006 to 37.5 percent in 2016. A case can contain several allegations, and a deadlock on one of them can leave the rest to proceed, so this measure captures exposure to deadlock rather than total defeat by deadlock. Even read with that qualification, the movement is stark.

Her third measure captured the matters in which deadlock was fully dispositive. About 12.5 percent of matters in 2016 were closed without action because of deadlock, against none in 2006. Closed without action is the phrase that carries the weight in that finding. No investigation followed, no violation was found, and no penalty was imposed, but neither did the Commission exonerate the respondent on the merits after a completed inquiry. The file closed because the votes to proceed could not be assembled. Whatever the underlying conduct had been, the legal system charged with evaluating it produced silence as its final output.

The financial measure completed the picture. Civil penalties collected fell from 5.6 million dollars in 2006 to 595,000 dollars in 2016, figures Ravel drew from the Commission’s data for those years. Penalties are an imperfect measure of enforcement vigor, since a single large settlement can move an annual total and since some violations properly end in conciliation rather than fines. A decline of that scale, arriving alongside rising deadlock rates on two separate measures and a new category of cases closed without action, fits a pattern rather than standing as an isolated fluctuation. Less was being decided, fewer matters were proceeding, and the monetary consequences imposed on violators shrank accordingly.

Attribution is essential here, because Ravel did not write as a neutral auditor detached from the institution she described. She wrote as a departing Democratic commissioner, reporting on a Commission whose Republican members she had repeatedly failed to persuade, and her Republican colleagues disputed her characterization of the record. That posture does not invalidate figures drawn from the Commission’s own data, but it frames them. Ravel selected the measures, chose the comparison years, and supplied the words dysfunction and deadlock as the interpretive label. A reader who accepts her numbers can reasonably reserve judgment on her label, and her colleagues did exactly that.

Their counter reading begins with the design itself. On that reading, the even split was Congress’s deliberate guard against partisan enforcement, not a defect that later generations discovered. A tied vote means that half the Commission, drawn from the party other than the one that appointed the complaining side’s likely allies in a given matter, remained unpersuaded that the law had been broken or that action was warranted. Treating every tie as dysfunction assumes that enforcement action is the correct default and inaction the error. Defenders of the deadlocked votes rejected that assumption. From their position, a tie records a genuine failure of proof or of legal consensus, and the respondent benefits from that failure in the same way that a defendant benefits when a prosecution cannot meet its burden.

The counter reading gains further strength from the character of the questions involved. Several deadlocked matters, on the account offered by defenders of those votes, involved genuinely contested questions of law, where the statute and the precedents under campaign finance law did not point clearly in one direction. Campaign finance sits at the boundary between regulation of money and regulation of political speech, and that boundary had shifted repeatedly under constitutional challenge across the decades covered by this article. Commissioners confronting an unsettled question could disagree in good faith about what the law required. A tie on such a question reflects the state of the law as much as the state of the Commission, and resolving it by a bare majority of one party’s appointees would have settled the respondent’s liability without settling the underlying legal doubt.

The strongest form of the counter reading therefore runs as follows. A tied vote leaves the respondent unpenalized rather than convicted by a bare majority of the opposing party’s appointees. Between those two outcomes, defenders of the structure regard the first as the lesser evil and the one Congress chose when it wrote the four vote requirement. Enforcement that proceeds only when at least one commissioner from the other party agrees carries a legitimacy that party line enforcement could never claim. On this account, the figures Ravel assembled show the guard operating more often because the questions reaching the Commission grew harder and the partisan distance between the blocs grew wider, not because the guard itself had broken.

Enforcement was not the only function that deadlock reached. Advisory opinions occupy a quieter corner of the Commission’s work, and their deterioration reveals a different kind of harm. A person or organization contemplating political activity may ask the Commission in advance whether a proposed course of conduct complies with the law. The answer binds the Commission as to that requester and guides everyone else watching. When the Commission answers, regulated parties can act with confidence. When the Commission cannot answer, regulated parties must choose between abandoning lawful activity and proceeding under legal uncertainty.

The Brennan Center for Justice measured how often that failure occurred, and its figures supply independent corroboration for the guidance side of the story. Across the Commission’s first 32 years, from 1975 to 2007, the Commission deadlocked on at least one question in 4.9 percent of advisory opinion requests per year. From 2008 onward, the Brennan Center found, the Commission deadlocked on at least one question in 24.1 percent of advisory opinion requests per year. Guidance, which requires no accusation against anyone and punishes no one, proved nearly as vulnerable to the tie as enforcement itself.

The consequences of failed guidance fall on the compliant as much as on the accused. A respondent in an enforcement matter at least knows what conduct is at issue, because the conduct already happened. A requester of an advisory opinion is trying to learn the rules before acting, often before spending money or speaking at all. A deadlocked request leaves that person to guess at the boundary between lawful and unlawful conduct, or to hire counsel to guess on their behalf, or to stay silent. Multiplied across a quarter of all requests per year from 2008 onward, on the Brennan Center’s figures, the failure to answer became a standing feature of the regulatory environment rather than an occasional disappointment.

Deadlock within a functioning quorum is one condition. Loss of the quorum itself is another, and the 2020 cycle supplied it twice. According to the Congressional Research Service, the Commission lost its four member policymaking quorum for about nine months between August 2019 and mid 2020, and again for about six months between July and December 2020. Without four members, the Commission could not act as a policymaking body at all. The four vote threshold for enforcement, advisory opinions, and rules could not even be tested, because the body lacked the membership required to take those decisions.

The timing of the second loss compounded the institutional damage. The period between July and December 2020 covered most of the active heart of the 2020 cycle, the months in which campaigns raised and spent at their peak, complaints about that activity were filed, and questions about the rules governing it were most urgent. An enforcement agency absent during the contest it exists to police cannot recover the lost ground afterward by deciding the same cases later. Elections conclude on a fixed date. Voters cast ballots without whatever information a timely investigation or a timely advisory answer might have supplied, and no subsequent ruling restores that information to them at the moment it mattered.

When the quorum returned in December 2020, the accumulated cost stood in measurable form. The backlog reached 388 enforcement cases, a figure supplied by the Commission’s Office of General Counsel as of September 30, 2020. Each of those cases represented a complaint or a referral that had entered the system and waited without resolution. Some waited because commissioners had deadlocked. Others waited because no quorum existed to decide them in any direction. The distinction between those causes mattered inside the institution. To the people who had filed the matters and the people named in them, the experience was identical, a file that did not move.

Delay inside this system carries a hard legal edge, because enforcement matters are governed by a five year statute of limitations. The Congressional Research Service account records that some older matters had passed that five year limit and died unactioned during the quorum losses. Died unactioned is the most complete failure available to an enforcement system. A deadlocked case at least received votes. A case closed on the merits at least received a judgment. A case that expired under the statute of limitations received neither, and the conduct it alleged became legally unreachable regardless of what an inquiry might have found. Time, rather than any commissioner’s judgment, made the final decision.

Against that entire record stands a counterweight that any honest account must carry at full weight. Throughout the deadlock years, and throughout the quorum losses, the disclosure machinery never stopped. Reports were filed, processed, and published. Candidates, parties, and committees continued to report what they raised and what they spent, and the Commission continued to place that information before the public. The public record on which researchers and journalists rely continued to accumulate, cycle after cycle, even as contested enforcement faltered and advisory guidance stalled.

That continuity is not a minor administrative footnote. Disclosure was half of the design built by the 1974 amendments, paired with enforcement as the two instruments by which the law would discipline money in politics. One instrument informs voters and exposes conduct to scrutiny by opponents, journalists, and researchers. The other punishes violations after the fact. Between 2006 and 2020, on the evidence assembled in this section, those two instruments diverged sharply in performance. The half that informs voters kept operating. The half that punishes violations faltered, first by tie and then by absence.

The divergence also shapes what reform arguments the record can and cannot support. A claim that the Commission ceased to function between 2006 and 2020 overstates the evidence, because a central function demonstrably continued and produced the very data on which the claim relies. A claim that nothing serious changed understates the evidence by an equal margin, because penalties fell from 5.6 million dollars to 595,000 dollars across Ravel’s comparison, cases closed without action went from none to about 12.5 percent of matters, advisory deadlock rose to 24.1 percent of requests per year on the Brennan Center’s measure, and 388 enforcement cases waited in backlog as the quorum returned in December 2020. The record shows a partial failure with a precise shape, not a total collapse and not a passing quarrel among commissioners.

That shape is the section’s conclusion about impact. Campaign finance law between 2006 and 2020 was enforced through a structure that required cross party agreement for every consequential act, in a period when cross party agreement grew scarce. The result was neither the partisan enforcement the structure’s defenders feared nor the impartial enforcement the structure’s designers hoped to guarantee. The result was a growing volume of matters in which the law produced no answer at all, delivered late, or expired before an answer could come, while the disclosure record that might someday support a different judgment accumulated in full public view.

The Verdict the Record Supports

Five decades of federal regulation produced a record that resists both celebration and dismissal, and the resistance itself is a finding. Totals rose across every legal regime in the period. Disclosure generated an archive of reported activity without parallel in earlier American politics, and the archive recorded a growing share of money whose original source no filing named. A verdict that flattened that record into a slogan about money winning or reform failing would discard the distinctions that make the record instructive, and the claim that the evidence supports is narrower, more precise, and more demanding than either slogan.

Regulation moves money; it does not reduce it.

The totals establish the first part of the verdict before any channel enters the argument. The Center for Responsive Politics estimated total federal election spending, in nominal dollars across candidates, parties, and outside groups, at about 5.3 billion dollars in the 2008 cycle, at about 6.5 billion dollars in the 2016 cycle, and at about 14.4 billion dollars in the 2020 cycle. Consumer prices rose by roughly a fifth between 2008 and 2020, on the arithmetic carried in the fact record for the article, so the movement from the 2008 estimate to the 2020 estimate represented well over a doubling in real terms.

National party soft money reached about 495 million dollars in the 2000 cycle, on Congressional Research Service figures drawing on Federal Election Commission data for national party committees, and the Bipartisan Campaign Reform Act ended that channel in 2002. Federally active 527 committees then raised and spent about 424 million dollars in the 2004 cycle, on the Campaign Finance Institute analysis by Stephen Weissman and Ruth Hassan, and the increase of about 273 million dollars over 2002 equaled less than half of the soft money the ban had eliminated. Enforcement against unregistered 527 committees followed in 2006 and 2007, and election spending migrated toward nonprofit organizations that disclosed no donors. The decisions of 2010 then supplied a registered vehicle for unlimited independent spending: OpenSecrets reported super PAC spending of 62.6 million dollars in the 2010 cycle and combined spending by super PACs and hybrid PACs of 622.7 million dollars in the 2012 cycle, within a total outside spending figure, excluding party committees, that reached about 2.9 billion dollars in the 2020 cycle.

The disclosure trend ran alongside the channel sequence and gave redirection a second dimension, because money changed visibility as money changed route. OpenSecrets reported that nonprofits that disclosed no donors reported less than 5.2 million dollars to the Federal Election Commission in the 2006 cycle, more than 300 million dollars in the 2012 cycle, and about 181 million dollars in the 2016 cycle. For the 2020 cycle, OpenSecrets estimated that dark money groups accounted for more than 1 billion dollars when direct spending and contributions to super PACs were combined, with about 660 million dollars of that total given to super PACs. The system therefore grew more transparent in aggregate and less traceable at the level of origin at the same time.

The research on outcomes disciplines the claim further, because redirection mattered most at the point of access rather than at the point of electoral or legislative decision. The contribution literature, in the review by Stephen Ansolabehere, John de Figueiredo, and James Snyder in 2003, found that only one in four prior studies supported the claim that contributions bought legislators votes. Against that unsettled background, the randomized field experiment by Joshua Kalla and David Broockman in 2016 produced the firmest result in the record: identification of meeting attendees as campaign donors made senior policymakers available between three and four times more often across requests to 191 congressional offices. Money moved through new channels purchased a hearing with reliability, and a hearing was the surest purchase the system offered.

The counter reading deserves statement at full strength, because the record supplies genuine material for a defense of regulation. The soft money prohibition held. The channel that carried about 495 million dollars through national party committees in the 2000 cycle never returned in any cycle from 2004 through 2020, and the parties replaced the revenue with limited, disclosed, hard money fundraising that the Campaign Finance Institute found had reached record levels by 2004, with growth from large donors and small donors alike. A ban aimed at a defined institutional practice, written to cover raising and spending together, and sustained by the Supreme Court ended the practice outright.

Disclosure supplies the second element of the counter reading. Disclosure survived every constitutional challenge in the period, including the decision that freed corporate and union independent expenditures, in which the Supreme Court upheld disclosure and disclaimer requirements by a vote of 8 to 1. Every figure in the article, from the Center for Responsive Politics totals to the OpenSecrets outside spending series to the Commission data behind the deadlock analyses, rests on reports that disclosure compelled and on an archive that disclosure maintained. Redirection, properly understood, therefore demonstrates that design determines where money goes rather than that money sits beyond regulation altogether. The channels that escaped control escaped through specific legal decisions about independent spending and specific statutory silences about nonprofit donors, and different decisions or different statutory language would have produced a different map of channels, as the structural comparison of the Federal Election Campaign Act and the Bipartisan Campaign Reform Act sets out in detail.

Two elements of the record fit neither the redirection account nor the counter reading comfortably, and the verdict names the misfit directly. Presidential public financing and commission enforcement, two features of the design built in the nineteen seventies, ceased operating in practice without any repeal. Candidates declined the grant because a fixed formula payment and the attached spending limit fell below the amount a major party nominee could raise privately, a calculation visible in 2008, when John McCain accepted a grant of about 84.1 million dollars and Barack Obama declined public funding altogether, on Congressional Research Service accounts of that cycle. Enforcement followed a parallel path into disuse on contested matters: Commissioner Ann Ravel reported deadlocks in 30 percent of substantive enforcement votes in 2016 against 2.9 percent in 2006, and civil penalties fell from 5.6 million dollars in 2006 to 595,000 dollars in 2016, on Commission data analyzed in the Ravel report of February 2017. Abandonment and deadlock are a different failure mode from redirection. Participation ended, decisions stopped, and the statutory text remained in place as a description of a system that elections no longer used.

The record through the 2020 cycle therefore supports a verdict with three parts held together. Regulation moved money between channels, changed the identity and visibility of spenders, and left the aggregate scale of federal election spending larger at the end of the period than at the start of every regime within the period. Regulation also proved capable of ending a defined channel completely when a prohibition governed an institutional practice, covered raising and spending together, and survived constitutional review, and disclosure proved capable of surviving every challenge and of generating the public record on which all measurement depends. Regulation failed, where failure occurred, by abandonment and by deadlock in the two elements that depended on voluntary participation and on cross party agreement. A student of the period who keeps the three parts together holds a more accurate map than a reader who keeps any one part alone, and a careful record of totals, channels, and cases in sequence can keep your statute notes, citations, and case chronologies together free on VaultBook.

Frequently Asked Questions

Q: Did campaign finance law reduce total spending?

No. The Center for Responsive Politics estimated total federal election spending, in nominal dollars across candidates, parties, and outside groups, at about 5.3 billion dollars in the 2008 cycle, at about 6.5 billion dollars in the 2016 cycle, and at about 14.4 billion dollars in the 2020 cycle. The 2020 estimate stood at roughly double the 2016 estimate and at more than the two previous presidential cycles combined, on the Center estimates. Consumer prices rose by roughly a fifth between 2008 and 2020, so the increase represented well over a doubling in real terms against 2008. Totals rose under each legal regime in the period, through the soft money era, after the ban of 2002, and after the court decisions of 2010 reorganized outside spending.

Q: Where did money go after campaign finance law closed a channel?

Money moved to the nearest lawful channel that accepted large contributions. After the Bipartisan Campaign Reform Act closed national party soft money in 2002, federally active 527 committees raised and spent about 424 million dollars in the 2004 cycle, on the Campaign Finance Institute analysis by Stephen Weissman and Ruth Hassan. Federal Election Commission enforcement against unregistered 527 committees in 2006 and 2007 then raised the legal risk of that form, and election spending shifted toward nonprofit organizations that disclosed no donors. After the decisions of 2010, super PACs offered a registered vehicle for unlimited independent spending, and OpenSecrets reported total outside spending, excluding party committees, at about 2.9 billion dollars in the 2020 cycle. Each closure changed the route, the spender, and the degree of disclosure.

Q: Did the soft money ban work under campaign finance law?

Yes, within the scope the statute defined. The Bipartisan Campaign Reform Act prohibited national party committees from raising or spending soft money, effective November 6, 2002, and the Supreme Court sustained the prohibition in McConnell v. FEC in 2003. National party soft money, measured at about 495 million dollars in the 2000 cycle on Congressional Research Service figures drawing on Federal Election Commission data, did not return in any cycle from 2004 through 2020. The parties adapted inside the rules instead. The Campaign Finance Institute found that national party committees raised substantially more hard money in 2004 than in any earlier cycle, with growth from large donors and small donors alike. The ban ended a defined transaction, an unlimited contribution to a national party committee, and the ending held across every later cycle in the record.

Q: Does campaign finance law disclosure still cover big donors?

Disclosure covered big donors unevenly. Candidates, parties, and political committees filed reports that named contributors under the federal system, and super PACs disclosed the donors named in the reports that super PACs filed with the Federal Election Commission. Tracing stopped at a named donor when that donor was a nonprofit organization with no duty to disclose original sources, or a shell company whose filing named a legal entity and no person behind the entity. OpenSecrets estimated that dark money groups accounted for more than 1 billion dollars in the 2020 cycle when direct spending and contributions to super PACs were combined, with about 660 million dollars given to super PACs. Disclosure therefore captured the spender and the immediate contributor across the period, and captured the original source only where the contributor was a person or a disclosing organization.

Q: Does spending decide elections under campaign finance law?

The evidence left the question disputed. The candidate who spent more won most House races, and donors also gave to candidates already expected to win, so correlation alone could not separate cause from expectation. Gary Jacobson found challenger spending strongly associated with vote share and incumbent spending weakly associated, a pattern shaped by incumbents raising the most money when threatened. Steven Levitt, comparing repeat candidates in 1994, summarized the estimated effect as about a third of a percentage point for an additional 100,000 dollars. Alan Gerber, studying Senate elections in 1998 with candidate wealth as an instrument, estimated that the average incumbent spending advantage translated to about six percentage points of vote share. The better funded presidential campaign lost in 2016, and heavy spending in the 2020 primaries produced no nomination.

Q: What does research say money buys under campaign finance law?

Research documented access more firmly than outcomes. Joshua Kalla and David Broockman, in a randomized field experiment published in the American Journal of Political Science in 2016, worked with CREDO Action to seek meetings with 191 congressional offices, randomly assigned to hear or not hear that the prospective attendees were campaign donors. Offices told that attendees were donors made senior policymakers, members of Congress and chiefs of staff, available between three and four times more often. The experiment measured access offered, not votes changed. On votes, Stephen Ansolabehere, John de Figueiredo, and James Snyder reported in 2003 that only one in four prior studies supported the claim that contributions bought legislators votes, and that controls for constituent and legislator effects left little relationship between money and votes.

Q: Why did presidential public financing collapse under campaign finance law?

Candidates declined a bargain whose value had fallen behind private fundraising. Every major party nominee accepted the general election grant from 1976 through 2004, and George W. Bush declined primary matching funds in 2000 while accepting the general election grant. In 2008 Barack Obama declined the general election grant, the first major party nominee to forgo general election public funding since the system began, while John McCain accepted a grant of about 84.1 million dollars and the attached spending limit, on Congressional Research Service accounts. In 2012 no major candidate accepted public funds, and in 2016 and 2020 neither major party nominee accepted the general election grant or primary matching funds. Acceptance imposed a ceiling fixed by formula, and refusal carried no legal cost beyond loss of the payment.

Q: Does the election commission deadlock under campaign finance law?

Deadlock rose sharply on contested matters between 2006 and 2016. Commissioner Ann Ravel, in a February 2017 report analyzing Federal Election Commission data, found deadlocks in 2.9 percent of substantive enforcement votes in 2006 and in 30 percent in 2016, with the share of cases deadlocked on at least one alleged violation rising from 4.2 percent to 37.5 percent across the same years. Civil penalties collected fell from 5.6 million dollars in 2006 to 595,000 dollars in 2016, on the figures in the Ravel analysis. The Brennan Center for Justice found deadlock on at least one question in 4.9 percent of advisory opinion requests per year from 1975 to 2007 and in 24.1 percent per year from 2008 onward. Disclosure filings continued throughout the period.

Q: What share of federal election spending came from outside groups?

About one dollar in five in the 2020 cycle came from outside groups. OpenSecrets measured total outside spending, excluding party committees and covering independent expenditures, electioneering communications, and communication costs, at 2,914 million dollars in the 2020 cycle. The Center for Responsive Politics estimated total federal election spending by candidates, parties, and outside groups at about 14.4 billion dollars in nominal dollars in the same cycle. Placed together with the two scopes stated, the outside total of about 2.9 billion dollars represented roughly a fifth of all federal election spending. The share had grown across the period, from an outside total of 51.6 million dollars in the 2000 cycle on the OpenSecrets series, through about 1.04 billion dollars in 2012 and about 1.41 billion dollars in 2016.

Q: How much soft money did the national parties raise before the ban?

National party soft money grew from a supplement into a central revenue source before the ban. The Brennan Center figure placed national party soft money at about 45 million dollars in 1988. The Congressional Research Service, drawing on Federal Election Commission data for national party committees, placed the total at 262 million dollars in 1996 and at about 495 million dollars in the 2000 cycle. The sequence described a channel that multiplied roughly elevenfold between 1988 and the 2000 cycle while the limits on regulated party accounts changed far more slowly. Both the Democratic and Republican national party organizations built the channel into fundraising, soliciting large contributions that federal law barred from a federal party account, and spending the proceeds on party building activity classified outside the federal definitions.

Q: What happened to the national parties after the soft money ban?

The national parties rebuilt fundraising inside the federal limits and raised more regulated money than the parties had raised before the ban. The Campaign Finance Institute assessed the adjustment after the 2004 cycle and found that national party committees had raised substantially more hard money in 2004 than in any earlier cycle. Growth came from large donors and from small donors, so the increase rested on an expanded fundraising operation rather than on a narrow substitution of maximum hard money checks for unlimited soft money checks. Registration work, turnout operations, staff, and generic party support continued, funded from the federal account under federal limits and source rules. Donors lost the option of an unlimited contribution to a national committee, and compliance moved to the center of party finance.

Q: How much did super PACs spend in the first two cycles of the new form?

OpenSecrets reported super PAC spending of 62.6 million dollars in the 2010 cycle, a midterm cycle in which organizations had only part of a year to form and spend under the new rules. OpenSecrets reported combined spending by super PACs and hybrid PACs of 622.7 million dollars in the 2012 cycle, the first presidential cycle under the new form, with the combined committee scope attached to the figure. The legal basis rested on two decisions rather than one: Citizens United v. FEC barred bans on corporate and union independent expenditures in 2010, and SpeechNow.org v. FEC held that contribution limits could not apply to a committee making only independent expenditures. The second decision supplied the step that made super PACs possible as unlimited contribution committees.

Q: How can a super PAC donor stay hidden despite disclosure rules?

A disclosed donor can conceal an original source when the donor named in a super PAC report is an organization with no duty to name the persons behind the organization. A super PAC disclosed donors to the Federal Election Commission, and a report named a nonprofit organization or a shell company as the contributor when money arrived through either form. Politically active nonprofits, principally organizations under section 501(c)(4) of the tax code and some trade associations under section 501(c)(6), could spend on elections without disclosing donors. Transfers between affiliated nonprofits added further layers between an original donor and an eventual expenditure. The report satisfied the disclosure rule at the level of the immediate contributor and gave a reader no path to the original source.

Q: How much dark money was spent in the final presidential cycle in the record?

OpenSecrets estimated that dark money groups accounted for more than 1 billion dollars in the 2020 cycle when direct spending and contributions to super PACs were combined, with about 660 million dollars of that total given to super PACs. The amount visible as direct outside spending in Federal Election Commission reports was far smaller, because much nonprofit election spending never entered a Commission report at all. In the portion of dark money that OpenSecrets could tally in reported donations and spending for 2020, liberal groups directed more than 514 million dollars and conservative groups directed about 200 million dollars. OpenSecrets reported that 2020 was the first presidential cycle in which dark money benefited Democrats more than Republicans, after a period in which dark money leaned heavily conservative.

Q: Did corporations become the main funders of super PACs after the court decisions?

No. OpenSecrets analyses found that direct corporate treasury money formed a modest share of super PAC receipts across the period after 2010. The largest growth in outside funding came from wealthy individuals giving to super PACs, together with nonprofits that did not disclose donors. The distinction between permission and use explains the result. Citizens United v. FEC removed a ban on corporate and union treasury spending for independent expenditures, and removal of a ban established what the law permitted. Receipts recorded what contributors in fact supplied, and the reported record showed concentrated individual giving expanding faster than direct corporate treasury contributions. An account that treated the growth of outside spending after 2010 as primarily a surge of corporate treasury money moved beyond the composition evidence that OpenSecrets assembled.

Q: Do contributors give to buy policy or to take part in politics?

The strongest account in the literature treated most giving as participation. Stephen Ansolabehere, John de Figueiredo, and James Snyder, in the Journal of Economic Perspectives in 2003, found that contributions behaved mainly as political participation and consumption rather than as investment. Individuals, not special interests, formed the main source of contributions in the account the authors assembled. Contributors gave from partisanship, ideology, candidate loyalty, and social connection, and gave amounts and patterns, including support for losing candidates, that an investment account handled poorly. The authors also found that contributions as a share of gross domestic product had not risen appreciably in over a century and had probably fallen. Giving bought the satisfaction and standing of participation, on that account, rather than a policy return calculated against the value of legislation at stake.

Q: What is the Tullock puzzle about money in politics?

The Tullock puzzle asked why political money was so small rather than why political money was so large. Gordon Tullock posed the puzzle about thirty years before Stephen Ansolabehere, John de Figueiredo, and James Snyder took up the question in the Journal of Economic Perspectives in 2003. Government decisions on taxation, spending, and regulation carried enormous value for affected industries and groups, so an investment account of giving predicted contributions far larger than American politics contained. Candidates, parties, and organizations raised nearly 3 billion dollars in the 1999 to 2000 cycle, a large absolute sum that the authors treated as puzzlingly small against the scale of federal policy. The consumption account answered the puzzle by treating contributions as expressive participation, limited by the satisfaction of taking part rather than by the value of the policies under decision.

Q: How large was the enforcement backlog when the Commission regained a quorum?

The backlog stood at 388 enforcement cases when the quorum returned in December 2020, on the figure supplied by the Office of General Counsel of the Federal Election Commission as of September 30, 2020. The Congressional Research Service reported that the Commission lost the four member policymaking quorum for about nine months between August 2019 and mid 2020 and again for about six months between July and December 2020, periods in which the Commission could not open investigations, issue advisory opinions, or make rules. Delay carried a legal edge, because some older matters passed the five year statute of limitations during the quorum losses and died unactioned, beyond the reach of any later decision. Each waiting case represented a complaint or referral that had entered the system without resolution.

Q: What happened to the tax checkoff that stocked the public fund?

Participation in the tax checkoff declined across four decades while the fund balance accumulated. The Congressional Research Service, reporting Treasury Department Financial Management Service figures, placed the peak in 1980, when 28.7 percent of filers designated funds to the Presidential Election Campaign Fund. Participation fell below 15 percent every year since 1993, reached 6.0 percent in 2012, and reached 5.4 percent in 2015. A designation directed three dollars of existing tax liability to the fund and added nothing to tax owed and subtracted nothing from a refund, so decline reflected indifference rather than price. The balance grew even as participation fell, because designations exceeded payouts to a shrinking pool of participating candidates across the years after major party nominees stopped accepting public money.

Q: What should a student remember about the effects of campaign finance law?

A student should remember three findings held together. Regulation moved money between channels without reducing the aggregate scale of federal election spending across the period through the 2020 cycle, and each closure changed the spender, the legal form, and the degree of disclosure attached to the funds. Regulation also ended one defined channel completely, because the soft money prohibition governed an institutional practice, covered raising and spending together, survived constitutional review, and held in every cycle after the effective date. Disclosure survived every constitutional challenge and produced the public record on which all measurement in the field rests. Public financing and commission enforcement faded by abandonment and deadlock rather than by repeal, a separate failure mode. Money purchased access with documented reliability and purchased votes or outcomes with no comparable proof.