OPINION OF BREYER, J.
RANDALL V. SORRELL
548 U. S. ____ (2006)
SUPREME COURT OF THE UNITED STATES
NOS. 04-1528, 04-1530 AND 04-1697
NEIL RANDALL, et al., PETITIONERS
WILLIAM H. SORRELL et al.
VERMONT REPUBLICAN STATE COMMITTEE, et al., PETITIONERS
WILLIAM H. SORRELL et al.
WILLIAM H. SORRELL, et al., PETITIONERS
NEIL RANDALL et al.
on writs of certiorari to the united states court of appeals for the second circuit
[June 26, 2006]
Justice Breyer announced the judgment of the Court, and delivered an opinion in which The Chief Justice joins, and in which Justice Alito joins except as to Parts II–B–1 and II–B–2.
We here consider the constitutionality of a Vermont campaign finance statute that limits both (1) the amounts that candidates for state office may spend on their campaigns (expenditure limitations) and (2) the amounts that individuals, organizations, and political parties may contribute to those campaigns (contribution limitations). Vt. Stat. Ann., Tit. 17, §2801 et seq. (2002). We hold that both sets of limitations are inconsistent with the First Amendment. Well-established precedent makes clear that the expenditure limits violate the First Amendment. Buckley v. Valeo, 424 U. S. 1, 54–58 (1976) (per curiam). The contribution limits are unconstitutional because in their specific details (involving low maximum levels and other restrictions) they fail to satisfy the First Amendment’s requirement of careful tailoring. Id., at 25–30. That is to say, they impose burdens upon First Amendment interests that (when viewed in light of the statute’s legitimate objectives) are disproportionately severe.
Prior to 1997, Vermont’s campaign finance law imposed no limit upon the amount a candidate for state office could spend. It did, however, impose limits upon the amounts that individuals, corporations, and political committees could contribute to the campaign of such a candidate. Individuals and corporations could contribute no more than $1,000 to any candidate for state office. §2805(a) (1996). Political committees, excluding political parties, could contribute no more than $3,000. §2805(b). The statute imposed no limit on the amount that political parties could contribute to candidates.
In 1997, Vermont enacted a more stringent campaign finance law, Pub. Act No. 64, codified at Vt. Stat. Ann., Tit. 17, §2801 et seq. (2002) (hereinafter Act or Act 64), the statute at issue here. Act 64, which took effect immediately after the 1998 elections, imposes mandatory expenditure limits on the total amount a candidate for state office can spend during a “two-year general election cycle,” i.e., the primary plus the general election, in approximately the following amounts: governor, $300,000; lieutenant governor, $100,000; other statewide offices, $45,000; state senator, $4,000 (plus an additional $2,500 for each additional seat in the district); state representative (two-member district), $3,000; and state representative (single member district), $2,000. §2805a(a). These limits are adjusted for inflation in odd-numbered years based on the Consumer Price Index. §2805a(e). Incumbents seeking reelection to statewide office may spend no more than 85% of the above amounts, and incumbents seeking reelection to the State Senate or House may spend no more than 90% of the above amounts. §2805a(c). The Act defines “[e]xpenditure” broadly to mean the
“payment, disbursement, distribution, advance, deposit, loan or gift of money or anything of value, paid or promised to be paid, for the purpose of influencing an election, advocating a position on a public question, or supporting or opposing one or more candidates.” §2801(3).
With certain minor exceptions, expenditures over $50 made on a candidate’s behalf by others count against the candidate’s expenditure limit if those expenditures are “intentionally facilitated by, solicited by or approved by” the candidate’s campaign. §§2809(b), (c). These provisions apply so as to count against a campaign’s expenditure limit any spending by political parties or committees that is coordinated with the campaign and benefits the candidate. And any party expenditure that “primarily benefits six or fewer candidates who are associated with the political party” is “presumed” to be coordinated with the campaign and therefore to count against the campaign’s expenditure limit. §§2809(b), (d).
Act 64 also imposes strict contribution limits. The amount any single individual can contribute to the campaign of a candidate for state office during a “two-year general election cycle” is limited as follows: governor, lieutenant governor, and other statewide offices, $400; state senator, $300; and state representative, $200. §2805(a). Unlike its expenditure limits, Act 64’s contribution limits are not indexed for inflation.
A political committee is subject to these same limits. Ibid. So is a political party, ibid., defined broadly to include “any subsidiary, branch or local unit” of a party, as well as any “national or regional affiliates” of a party (taken separately or together). §2801(5). Thus, for example, the statute treats the local, state, and national affiliates of the Democratic Party as if they were a single entity and limits their total contribution to a single candidate’s campaign for governor (during the primary and the general election together) to $400.
The Act also imposes a limit of $2,000 upon the amount any individual can give to a political party during a 2-year general election cycle. §2805(a).
The Act defines “contribution” broadly in approximately the same way it defines “expenditure.” §2801(2). Any expenditure made on a candidate’s behalf counts as a contribution to the candidate if it is “intentionally facilitated by, solicited by or approved by” the candidate. §§2809(a), (c). And a party expenditure that “primarily benefits six or fewer candidates who are associated with the” party is “presumed” to count against the party’s contribution limits. §§2809(a), (d).
There are a few exceptions. A candidate’s own contributions to the campaign and those of the candidate’s family fall outside the contribution limits. §2805(f). Volunteer services do not count as contributions. §2801(2). Nor does the cost of a meet-the-candidate function, provided that the total cost for the function amounts to $100 or less. §2809(d).
In addition to these expenditure and contribution limits, the Act sets forth disclosure and reporting requirements and creates a voluntary public financing system for gubernatorial elections. §§2803, 2811, 2821–2823, 2831, 2832, 2851–2856. None of these is at issue here. The Act also limits the amount of contributions a candidate, political committee, or political party can receive from out-of-state sources. §2805(c). The lower courts held these out-of-state contribution limits unconstitutional, and the parties do not challenge that holding.
The petitioners are individuals who have run for state office in Vermont, citizens who vote in Vermont elections and contribute to Vermont campaigns, and political parties and committees that participate in Vermont politics. Soon after Act 64 became law, they brought this lawsuit in Federal District Court against the respondents, state officials charged with enforcement of the Act. Several other private groups and individual citizens intervened in the District Court proceedings in support of the Act and are joined here as respondents as well.
The District Court agreed with the petitioners that the Act’s expenditure limits violate the First Amendment. See Buckley, 424 U. S. 1. The court also held unconstitutional the Act’s limits on the contributions of political parties to candidates. At the same time, the court found the Act’s other contribution limits constitutional. Landell v. Sorrell, 118 F. Supp. 2d 470 (Vt. 2000).
Both sides appealed. A divided panel of the Court of Appeals for the Second Circuit held that all of the Act’s contribution limits are constitutional. It also held that the Act’s expenditure limits may be constitutional. Landell v. Sorrell, 382 F. 3d 91 (2004). It found those limits supported by two compelling interests, namely, an interest in preventing corruption or the appearance of corruption and an interest in limiting the amount of time state officials must spend raising campaign funds. The Circuit then remanded the case to the District Court with instructions to determine whether the Act’s expenditure limits were narrowly tailored to those interests.
The petitioners and respondents all sought certiorari. They asked us to consider the constitutionality of Act 64’s expenditure limits, its contribution limits, and a related definitional provision. We agreed to do so. 545 U. S. ___ (2005).
We turn first to the Act’s expenditure limits. Do those limits violate the First Amendment’s free speech guarantees?
In Buckley v. Valeo, supra, the Court considered the constitutionality of the Federal Election Campaign Act of 1971 (FECA), 86 Stat. 3, as amended, 2 U. S. C. §431 et seq., a statute that, much like the Act before us, imposed both expenditure and contribution limitations on campaigns for public office. The Court, while upholding FECA’s contribution limitations as constitutional, held that the statute’s expenditure limitations violated the First Amendment.
Buckley stated that both kinds of limitations “implicate fundamental First Amendment interests.” 424 U. S., at 23. It noted that the Government had sought to justify the statute’s infringement on those interests in terms of the need to prevent “corruption and the appearance of corruption.” Id., at 25; see also id., at 55. In the Court’s view, this rationale provided sufficient justification for the statute’s contribution limitations, but it did not provide sufficient justification for the expenditure limitations.
The Court explained that the basic reason for this difference between the two kinds of limitations is that expenditure limitations “impose significantly more severe restrictions on protected freedoms of political expression and association than” do contribution limitations. Id., at 23. Contribution limitations, though a “marginal restriction upon the contributor’s ability to engage in free communication,” nevertheless leave the contributor “fre[e] to discuss candidates and issues.” Id., at 20–21. Expenditure limitations, by contrast, impose “[a] restriction on the amount of money a person or group can spend on political communication during a campaign.” Id., at 19. They thereby necessarily “reduc[e] the quantity of expression by restricting the number of issues discussed, the depth of their exploration, and the size of the audience reached.” Ibid. Indeed, the freedom “to engage in unlimited political expression subject to a ceiling on expenditures is like being free to drive an automobile as far and as often as one desires on a single tank of gasoline.” Id., at 19, n. 18.
The Court concluded that “[n]o governmental interest that has been suggested is sufficient to justify the restriction on the quantity of political expression imposed by” the statute’s expenditure limitations. Id., at 55. It decided that the Government’s primary justification for expenditure limitations, preventing corruption and its appearance, was adequately addressed by the Act’s contribution limitations and disclosure requirements. Ibid. The Court also considered other governmental interests advanced in support of expenditure limitations. It rejected each. Id., at 56–57. Consequently, it held that the expenditure limitations were “constitutionally invalid.” Id., at 58.
Over the last 30 years, in considering the constitutionality of a host of different campaign finance statutes, this Court has repeatedly adhered to Buckley’s constraints, including those on expenditure limits. See McConnell v. Federal Election Comm’n, 540 U. S. 93, 134 (2003); Federal Election Comm’n v. Colorado Republican Federal Campaign Comm., 533 U. S. 431, 441 (2001) (Colorado II); Nixon v. Shrink Missouri Government PAC, 528 U. S. 377, 386 (2000) (Shrink); Colorado Republican Federal Campaign Comm. v. Federal Election Comm’n, 518 U. S. 604, 610 (1996) (Colorado I) (plurality opinion); Federal Election Comm’n v. Massachusetts Citizens for Life, Inc., 479 U. S. 238, 259–260 (1986); Federal Election Comm’n v. National Conservative Political Action Comm., 470 U. S. 480, 491 (1985) (NCPAC); California Medical Assn. v. Federal Election Comm’n, 453 U. S. 182, 194–195 (1981) (plurality opinion).
The respondents recognize that, in respect to expenditure limits, Buckley appears to be a controlling—and unfavorable—precedent. They seek to overcome that precedent in two ways. First, they ask us in effect to overrule Buckley. Post-Buckley experience, they believe, has shown that contribution limits (and disclosure requirements) alone cannot effectively deter corruption or its appearance; hence experience has undermined an assumption underlying that case. Indeed, the respondents have devoted several pages of their briefs to attacking Buckley’s holding on expenditure limits. See Brief for Respondent-Cross-Petitioner Vermont Public Interest Research Group et al. 36–39 (arguing that “sound reasons exist to revisit the applicable standard of review” for expenditure limits); Brief for Respondent-Cross-Petitioner William Sorrell et al. 28–31 (arguing that “the Court should revisit Buckley and consider alternative constitutional approaches to spending limits”).
Second, in the alternative, they ask us to limit the scope of Buckley significantly by distinguishing Buckley from the present case. They advance as a ground for distinction a justification for expenditure limitations that, they say, Buckley did not consider, namely that such limits help to protect candidates from spending too much time raising money rather than devoting that time to campaigning among ordinary voters. We find neither argument persuasive.
The Court has often recognized the “fundamental importance” of stare decisis, the basic legal principle that commands judicial respect for a court’s earlier decisions and the rules of law they embody. See Harris v. United States, 536 U. S. 545, 556–557 (2002) (plurality opinion) (citing numerous cases). The Court has pointed out that stare decisis “ ‘promotes the evenhanded, predictable, and consistent development of legal principles, fosters reliance on judicial decisions, and contributes to the actual and perceived integrity of the judicial process.’ ” United States v. International Business Machines Corp., 517 U. S. 843, 856 (1996) (quoting Payne v. Tennessee, 501 U. S. 808, 827 (1991)). Stare decisis thereby avoids the instability and unfairness that accompany disruption of settled legal expectations. For this reason, the rule of law demands that adhering to our prior case law be the norm. Departure from precedent is exceptional, and requires “special justification.” Arizona v. Rumsey, 467 U. S. 203, 212 (1984). This is especially true where, as here, the principle has become settled through iteration and reiteration over a long period of time.
We can find here no such special justification that would require us to overrule Buckley. Subsequent case law has not made Buckley a legal anomaly or otherwise undermined its basic legal principles. Cf. Dickerson v. United States, 530 U. S. 428, 443 (2000). We cannot find in the respondents’ claims any demonstration that circumstances have changed so radically as to undermine Buckley’s critical factual assumptions. The respondents have not shown, for example, any dramatic increase in corruption or its appearance in Vermont; nor have they shown that expenditure limits are the only way to attack that problem. Cf. McConnell v. FEC, 540 U. S. 93. At the same time, Buckley has promoted considerable reliance. Congress and state legislatures have used Buckley when drafting campaign finance laws. And, as we have said, this Court has followed Buckley, upholding and applying its reasoning in later cases. Overruling Buckley now would dramatically undermine this reliance on our settled precedent.
For all these reasons, we find this a case that fits the stare decisis norm. And we do not perceive the strong justification that would be necessary to warrant overruling so well established a precedent. We consequently decline the respondents’ invitation to reconsider Buckley.
The respondents also ask us to distinguish these cases from Buckley. But we can find no significant basis for that distinction. Act 64’s expenditure limits are not substantially different from those at issue in Buckley. In both instances the limits consist of a dollar cap imposed upon a candidate’s expenditures. Nor is Vermont’s primary justification for imposing its expenditure limits significantly different from Congress’ rationale for the Buckley limits: preventing corruption and its appearance.
The sole basis on which the respondents seek to distinguish Buckley concerns a further supporting justification. They argue that expenditure limits are necessary in order to reduce the amount of time candidates must spend raising money. Brief for Respondent/Cross-Petitioner Vermont Public Interest Research Group et al. 16–20; Brief for Respondent/Cross-Petitioner William H. Sorrell et al. 22–25. Increased campaign costs, together with the fear of a better-funded opponent, mean that, without expenditure limits, a candidate must spend too much time raising money instead of meeting the voters and engaging in public debate. Buckley, the respondents add, did not fully consider this justification. Had it done so, they say, the Court would have upheld, not struck down, FECA’s expenditure limits.
In our view, it is highly unlikely that fuller consideration of this time protection rationale would have changed Buckley’s result. The Buckley Court was aware of the connection between expenditure limits and a reduction in fundraising time. In a section of the opinion dealing with FECA’s public financing provisions, it wrote that Congress was trying to “free candidates from the rigors of fundraising.” 424 U. S., at 91; see also id., at 96 (“[L]imits on contributions necessarily increase the burden of fundraising,” and “public financing” was designed in part to relieve Presidential candidates “from the rigors of soliciting private contributions”); id., at 258–259 (White, J., concurring in part and dissenting in part) (same). The Court of Appeals’ opinion and the briefs filed in this Court pointed out that a natural consequence of higher campaign expenditures was that “candidates were compelled to allow to fund raising increasing and extreme amounts of money and energy.” Buckley v. Valeo, 519 F. 2d 821, 838 (CADC 1975); see also Brief for United States et al. as Amici Curiae in Buckley v. Valeo, O. T. 1975, Nos. 75–436 and 75–437, p. 36 (“Fund raising consumes candidate time that otherwise would be devoted to campaigning”). And, in any event, the connection between high campaign expenditures and increased fundraising demands seems perfectly obvious.
Under these circumstances, the respondents’ argument amounts to no more than an invitation so to limit Buckley’s holding as effectively to overrule it. For the reasons set forth above, we decline that invitation as well. And, given Buckley’s continued authority, we must conclude that Act 64’s expenditure limits violate the First Amendment.
We turn now to a more complex question, namely the constitutionality of Act 64’s contribution limits. The parties, while accepting Buckley’s approach, dispute whether, despite Buckley’s general approval of statutes that limit campaign contributions, Act 64’s contribution limits are so severe that in the circumstances its particular limits violate the First Amendment.
As with the Act’s expenditure limits, we begin with Buckley. In that case, the Court upheld the $1,000 contribution limit before it. Buckley recognized that contribution limits, like expenditure limits, “implicate fundamental First Amendment interests,” namely, the freedoms of “political expression” and “political association.” 424 U. S., at 15, 23. But, unlike expenditure limits (which “necessarily reduc[e] the quantity of expression by restricting the number of issues discussed, the depth of their exploration, and the size of the audience reached,” id., at 19), contribution limits “involv[e] little direct restraint on” the contributor’s speech, id., at 21. They do restrict “one aspect of the contributor’s freedom of political association,” namely, the contributor’s ability to support a favored candidate, but they nonetheless “permi[t] the symbolic expression of support evidenced by a contribution,” and they do “not in any way infringe the contributor’s freedom to discuss candidates and issues.” Id., at 21, 24.
Consequently, the Court wrote, contribution limitations are permissible as long as the Government demonstrates that the limits are “closely drawn” to match a “sufficiently important interest.” Id., at 25. It found that the interest advanced in the case, “prevent[ing] corruption” and its “appearance,” was “sufficiently important” to justify the statute’s contribution limits. Id., at 25–26.
The Court also found that the contribution limits before it were “closely drawn.” It recognized that, in determining whether a particular contribution limit was “closely drawn,” the amount, or level, of that limit could make a difference. Indeed, it wrote that “contribution restrictions could have a severe impact on political dialogue if the limitations prevented candidates and political committees from amassing the resources necessary for effective advocacy.” Id., at 21. But the Court added that such “distinctions in degree become significant only when they can be said to amount to differences in kind.” Id., at 30. Pointing out that it had “no scalpel to probe, whether, say, a $2,000 ceiling might not serve as well as $1,000,” ibid., the Court found “no indication” that the $1,000 contribution limitations imposed by the Act would have “any dramatic adverse effect on the funding of campaigns,” id., at 21. It therefore found the limitations constitutional.
Since Buckley, the Court has consistently upheld contribution limits in other statutes. Shrink, 528 U. S. 377 ($1075 limit on contributions to candidates for Missouri state auditor); California Medical Assn., 453 U. S. 182 ($5,000 limit on contributions to multicandidate political committees). The Court has recognized, however, that contribution limits might sometimes work more harm to protected First Amendment interests than their anticorruption objectives could justify. See Shrink, supra, at 395–397; Buckley, supra, at 21. And individual Members of the Court have expressed concern lest too low a limit magnify the “reputation-related or media-related advantages of incumbency and thereby insulat[e] legislators from effective electoral challenge.” Shrink, supra, at 403–404 (Breyer, J., joined by Ginsburg, J., concurring). In the cases before us, the petitioners challenge Act 64’s contribution limits on that basis.
Following Buckley, we must determine whether Act 64’s contribution limits prevent candidates from “amassing the resources necessary for effective [campaign] advocacy,” 424 U. S., at 21; whether they magnify the advantages of incumbency to the point where they put challengers to a significant disadvantage; in a word, whether they are too low and too strict to survive First Amendment scrutiny. In answering these questions, we recognize, as Buckley stated, that we have “no scalpel to probe” each possible contribution level. Id., at 30. We cannot determine with any degree of exactitude the precise restriction necessary to carry out the statute’s legitimate objectives. In practice, the legislature is better equipped to make such empirical judgments, as legislators have “particular expertise” in matters related to the costs and nature of running for office. McConnell, 540 U. S., at 137. Thus ordinarily we have deferred to the legislature’s determination of such matters.
Nonetheless, as Buckley acknowledged, we must recognize the existence of some lower bound. At some point the constitutional risks to the democratic electoral process become too great. After all, the interests underlying contribution limits, preventing corruption and the appearance of corruption, “directly implicate the integrity of our electoral process.” McConnell, supra, at 136 (internal quotation marks omitted). Yet that rationale does not simply mean “the lower the limit, the better.” That is because contribution limits that are too low can also harm the electoral process by preventing challengers from mounting effective campaigns against incumbent officeholders, thereby reducing democratic accountability. Were we to ignore that fact, a statute that seeks to regulate campaign contributions could itself prove an obstacle to the very electoral fairness it seeks to promote. Thus, we see no alternative to the exercise of independent judicial judgment as a statute reaches those outer limits. And, where there is strong indication in a particular case, i.e., danger signs, that such risks exist (both present in kind and likely serious in degree), courts, including appellate courts, must review the record independently and carefully with an eye toward assessing the statute’s “tailoring,” that is, toward assessing the proportionality of the restrictions. See Bose Corp. v. Consumers Union of United States, Inc., 466 U. S. 485, 499 (1984) (“[A]n appellate court has an obligation to ‘make an independent examination of the whole record’ in order to make sure that ‘the judgment does not constitute a forbidden intrusion on the field of free expression’ ” (quoting New York Times Co. v. Sullivan, 376 U. S. 254, 284–286 (1964))).
We find those danger signs present here. As compared with the contribution limits upheld by the Court in the past, and with those in force in other States, Act 64’s limits are sufficiently low as to generate suspicion that they are not closely drawn. The Act sets its limits per election cycle, which includes both a primary and a general election. Thus, in a gubernatorial race with both primary and final election contests, the Act’s contribution limit amounts to $200 per election per candidate (with significantly lower limits for contributions to candidates for State Senate and House of Representatives, see supra, at 3). These limits apply both to contributions from individuals and to contributions from political parties, whether made in cash or in expenditures coordinated (or presumed to be coordinated) with the candidate. See supra, at 3–4.
These limits are well below the limits this Court upheld in Buckley. Indeed, in terms of real dollars (i.e., adjusting for inflation), the Act’s $200 per election limit on individual contributions to a campaign for governor is slightly more than one-twentieth of the limit on contributions to campaigns for federal office before the Court in Buckley. Adjusted to reflect its value in 1976 (the year Buckley was decided), Vermont’s contribution limit on campaigns for statewide office (including governor) amounts to $113.91 per 2-year election cycle, or roughly $57 per election, as compared to the $1,000 per election limit on individual contributions at issue in Buckley. (The adjusted value of Act 64’s limit on contributions from political parties to candidates for statewide office, again $200 per candidate per election, is just over one one-hundredth of the comparable limit before the Court in Buckley, $5,000 per election.) Yet Vermont’s gubernatorial district—the entire State—is no smaller than the House districts to which Buckley’s limits applied. In 1976, the average congressional district contained a population of about 465,000. Dept. of Commerce, Bureau of Census, Statistical Abstract of the United States 459 (1976) (Statistical Abstract) (describing results of 1970 census). Indeed, Vermont’s population is 621,000—about one-third larger. Statistical Abstract 21 (2006) (describing Vermont’s population in 2004).
Moreover, considered as a whole, Vermont’s contribution limits are the lowest in the Nation. Act 64 limits contributions to candidates for statewide office (including governor) to $200 per candidate per election. We have found no State that imposes a lower per election limit. Indeed, we have found only seven States that impose limits on contributions to candidates for statewide office at or below $500 per election, more than twice Act 64’s limit. Cf. Ariz. Rev. Stat. Ann. §16–905 (West Cum. Supp. 2005) ($760 per election cycle, or $380 per election, adjusted for inflation); Colo. Const., Art. XXVIII, §3 ($500 per election, adjusted for inflation); Fla. Stat. §106.08(1)(a) (2003) ($500 per election); Me. Rev. Stat. Ann., Tit. 21A, §1015(1) (1993) ($500 for governor, $250 for other statewide office, per election); Mass. Gen. Laws, ch. 55, §7A (West Supp. 2006) ($500 per year, or $250 per election); Mont. Code Ann. §13–37–216(1)(a) (2005) ($500 for governor, $250 for other statewide office, per election); S. D. Codified Laws §12–25–1.1 (2004) ($1,000 per year, or $500 per election). We are aware of no State that imposes a limit on contributions from political parties to candidates for statewide office lower than Act 64’s $200 per candidate per election limit. Cf. Me. Rev. Stat. Ann., Tit. 21A, §1015(1) (1993) (next lowest: $500 for contribution from party to candidate for governor, $250 for contribution from party to candidate for other statewide office, both per election). Similarly, we have found only three States that have limits on contributions to candidates for state legislature below Act 64’s $150 and $100 per election limits. Ariz. Rev. Stat. Ann. §16–905 (West Cum. Supp. 2005) ($296 per election cycle, or $148 per election); Mont. Code Ann. §13–37–216(1)(a) (2005) ($130 per election); S. D. Codified Laws §12–25–1.1 (2004) ($250 per year, or $125 per election). And we are aware of no State that has a lower limit on contributions from political parties to state legislative candidates. Cf. Me. Rev. Stat. Ann., Tit. 21A, §1015(1) (1993) (next lowest: $250 per election).
Finally, Vermont’s limit is well below the lowest limit this Court has previously upheld, the limit of $1,075 per election (adjusted for inflation every two years, see Mo. Rev. Stat. §130.032.2 (1998 Cum. Supp.)) for candidates for Missouri state auditor. Shrink, 528 U. S.em>. 377. The comparable Vermont limit of roughly $200 per election, not adjusted for inflation, is less than one-sixth of Missouri’s current inflation-adjusted limit ($1,275).
We recognize that Vermont’s population is much smaller than Missouri’s. Indeed, Vermont is about one-ninth of the size of Missouri. Statistical Abstract 21 (2006). Thus, per citizen, Vermont’s limit is slightly more generous. As of 2006, the ratio of the contribution limit to the size of the constituency in Vermont is .00064, while Missouri’s ratio is .00044, 31% lower. Cf. App. 55 (doing same calculation in 2000).
But this does not necessarily mean that Vermont’s limits are less objectionable than the limit upheld in Shrink. A campaign for state auditor is likely to be less costly than a campaign for governor; campaign costs do not automatically increase or decrease in precise proportion to the size of an electoral district. See App. 66 (1998 winning candidate for Vermont state auditor spent about $60,000; winning candidate for governor spent about $340,000); Opensecrets.org, The Big Picture, 2004 Cycle: Hot Races, available at http://www.opensecrets.org/ bigpicture/hotraces.asp?cycle=2004 (as visited June 22, 2006, and available in Clerk of Court’s case file) (U. S. Senate campaigns identified as competitive spend less per voter than U. S. House campaigns identified as competitive). Moreover, Vermont’s limits, unlike Missouri’s limits, apply in the same amounts to contributions made by political parties. Mo. Rev. Stat. §130.032.4 (2000) (enacting limits on contributions from political parties to candidates 10 times higher than limits on contributions from individuals). And, as we have said, Missouri’s (current) $1,275 per election limit, unlike Vermont’s $200 per election limit, is indexed for inflation. See supra, at 17; see also Mo. Rev. Stat. §130.032.2 (2000).
The factors we have mentioned offset any neutralizing force of population differences. At the very least, they make it difficult to treat Shrink’s (then) $1,075 limit as providing affirmative support for the lawfulness of Vermont’s far lower levels. Cf. 528 U. S., at 404 (Breyer, J., concurring) (The Shrink “limit . . . is low enough to raise … a [significant constitutional] question”). And even were that not so, Vermont’s failure to index for inflation means that Vermont’s levels would soon be far lower than Missouri’s regardless of the method of comparison.
In sum, Act 64’s contribution limits are substantially lower than both the limits we have previously upheld and comparable limits in other States. These are danger signs that Act 64’s contribution limits may fall outside tolerable First Amendment limits. We consequently must examine the record independently and carefully to determine whether Act 64’s contribution limits are “closely drawn” to match the State’s interests.
Our examination of the record convinces us that, from a constitutional perspective, Act 64’s contribution limits are too restrictive. We reach this conclusion based not merely on the low dollar amounts of the limits themselves, but also on the statute’s effect on political parties and on volunteer activity in Vermont elections. Taken together, Act 64’s substantial restrictions on the ability of candidates to raise the funds necessary to run a competitive election, on the ability of political parties to help their candidates get elected, and on the ability of individual citizens to volunteer their time to campaigns show that the Act is not closely drawn to meet its objectives. In particular, five factors together lead us to this decision.
First, the record suggests, though it does not conclusively prove, that Act 64’s contribution limits will significantly restrict the amount of funding available for challengers to run competitive campaigns. For one thing, the petitioners’ expert, Clark Bensen, conducted a race-by-race analysis of the 1998 legislative elections (the last to take place before Act 64 took effect) and concluded that Act 64’s contribution limits would have reduced the funds available in 1998 to Republican challengers in competitive races in amounts ranging from 18% to 53% of their total campaign income. See 3 Tr. 52–57 (estimating loss of 47% of funds for candidate Tully, 50% for Harvey, 53% for Welch, 19% for Bahre, 29% for Delaney, 36% for LaRocque, 18% for Smith, and 31% for Brown).
For another thing, the petitioners’ expert witnesses produced evidence and analysis showing that Vermont political parties (particularly the Republican Party) “target” their contributions to candidates in competitive races, that those contributions represent a significant amount of total candidate funding in such races, and that the contribution limits will cut the parties’ contributions to competitive races dramatically. See 1 id., at 189–190; 3 id., at 50–51; 8 id., at 139; 10 id., at 150; see also, e.g., Gierzynski & Breaux, The Role of Parties in Legislative Campaign Financing, 15 Am. Rev. Politics 171 (1994); Thompson, Cassie, & Jewell, A Sacred Cow or Just a Lot of Bull? Party and PAC Money in State Legislative Elections, 47 Pol. Sci. Q. 223 (1994). Their statistics showed that the party contributions accounted for a significant percentage of the total campaign income in those races. And their studies showed that Act 64’s contribution limits would cut the party contributions by between 85% (for the legislature on average) and 99% (for governor).
More specifically, Bensen pointed out that in 1998, the Republican Party made contributions to 19 Senate campaigns in amounts that averaged $2,001, which on average represented 16% of the recipient campaign’s total income. 3 Tr. 84. Act 64 would reduce these contributions to $300 per campaign, an average reduction of about 85%. Ibid. The party contributed to 50 House campaigns in amounts averaging $787, which on average represented 28% of the recipient campaign’s total income. Id., at 85. Act 64 would reduce these contributions to $200 per campaign, an average reduction of 74.5%. Ibid. And the party contributed $40,600 to its gubernatorial candidate, an amount that accounted for about 16% of the candidate’s funding. Id., at 86. The Act would have reduced that contribution by 99%, to $400.
Bensen added that 57% of all 1998 Senate campaigns and 30% of all House campaigns exceeded Act 64’s expenditure limits, which were enacted along with the statute’s contribution limits. 7 Trial Exhs. in No. 00–9159(L) etc. (CA2), Exh. 8, p. 2351. Moreover, 27% of all Senate campaigns and 10% of all House campaigns spent more than double those limits. Ibid.
The respondents did not contest these figures. Rather, they presented evidence that focused, not upon strongly contested campaigns, but upon the funding amounts available for the average campaign. The respondents’ expert, Anthony Gierzynski, concluded, for example, that Act 64 would have a “minimal effect on … candidates’ ability to raise funds.” App. 46. But he rested this conclusion upon his finding that “only a small proportion of” all contributions to all campaigns for state office “made during the last three elections would have been affected by the new limits.” Id., at 47; see also id., at 51 (discussing “average amount of revenues lost to the limits” in legislative races (emphasis added)); id., at 52–53 (discussing total number of campaigns receiving contributions over Act 64’s limit). The lower courts similarly relied almost exclusively on averages in assessing Act 64’s effect. See 118 F. Supp. 2d, at 470 (“Approximately 88% to 96% of the campaign contributions to recent House races were under $200” (emphasis added)); id., at 478 (“Expert testimony revealed that over the last three election cycles the percentage of all candidates’ contributions received over the contribution limits was less than 10%” (emphasis added)).
The respondents’ evidence leaves the petitioners’ evidence unrebutted in certain key respects. That is because the critical question concerns not simply the average effect of contribution limits on fundraising but, more importantly, the ability of a candidate running against an incumbent officeholder to mount an effective challenge. And information about average races, rather than competitive races, is only distantly related to that question, because competitive races are likely to be far more expensive than the average race. See, e.g., N. Ornstein, T. Mann, & M. Malbin, Vital Statistics on Congress 2001–2002, pp. 89–98 (2002) (data showing that spending in competitive elections, i.e., where incumbent wins with less than 60% of vote or where incumbent loses, is far greater than in most elections, where incumbent wins with more than 60% of the vote). We concede that the record does contain some anecdotal evidence supporting the respondents’ position, namely, testimony about a post-Act-64 competitive mayoral campaign in Burlington, which suggests that a challenger can “amas[s] the resources necessary for effective advocacy,” Buckley, 424 U. S., at 21. But the facts of that particular election are not described in sufficient detail to offer a convincing refutation of the implication arising from the petitioners’ experts’ studies.
Rather, the petitioners’ studies, taken together with low average Vermont campaign expenditures and the typically higher costs that a challenger must bear to overcome the name-recognition advantage enjoyed by an incumbent, raise a reasonable inference that the contribution limits are so low that they may pose a significant obstacle to candidates in competitive elections. Cf. Ornstein, supra, at 87–96 (In 2000 U. S. House and Senate elections, successful challengers spent far more than the average candidate). Information about average races does not rebut that inference. Consequently, the inference amounts to one factor (among others) that here counts against the constitutional validity of the contribution limits.
Second, Act 64’s insistence that political parties abide by exactly the same low contribution limits that apply to other contributors threatens harm to a particularly important political right, the right to associate in a political party. See, e.g., California Democratic Party v. Jones, 530 U. S. 567, 574 (2000) (describing constitutional importance of associating in political parties to elect candidates); Timmons v. Twin Cities Area New Party, 520 U. S. 351, 357 (1997) (same); Colorado I, 518 U. S., at 616 (same); Norman v. Reed, 502 U. S. 279, 288 (1992) (same). Cf. Buckley, supra, at 20–22 (contribution limits constitute “only a marginal restriction” on First Amendment rights because contributor remains free to associate politically, e.g., in a political party, and “assist personally” in the party’s “efforts on behalf of candidates”).
The Act applies its $200 to $400 limits—precisely the same limits it applies to an individual—to virtually all affiliates of a political party taken together as if they were a single contributor. Vt. Stat. Ann., Tit. 17, §2805(a) (2002). That means, for example, that the Vermont Democratic Party, taken together with all its local affiliates, can make one contribution of at most $400 to the Democratic gubernatorial candidate, one contribution of at most $300 to a Democratic candidate for State Senate, and one contribution of at most $200 to a Democratic candidate for the State House of Representatives. The Act includes within these limits not only direct monetary contributions but also expenditures in kind: stamps, stationery, coffee, doughnuts, gasoline, campaign buttons, and so forth. See §2801(2). Indeed, it includes all party expenditures “intended to promote the election of a specific candidate or group of candidates” as long as the candidate’s campaign “facilitate[s],” “solicit[s],” or “approve[s]” them. §§2809(a), (c). And a party expenditure that “primarily benefits six or fewer candidates who are associated with the” party is “presumed” to count against the party’s contribution limits. §2809(d).
In addition to the negative effect on “amassing funds” that we have described, see supra, at 18–21, the Act would severely limit the ability of a party to assist its candidates’ campaigns by engaging in coordinated spending on advertising, candidate events, voter lists, mass mailings, even yard signs. And, to an unusual degree, it would discourage those who wish to contribute small amounts of money to a party, amounts that easily comply with individual contribution limits. Suppose that many individuals do not know Vermont legislative candidates personally, but wish to contribute, say, $20 or $40, to the State Republican Party, with the intent that the party use the money to help elect whichever candidates the party believes would best advance its ideals and interests—the basic object of a political party. Or, to take a more extreme example, imagine that 6,000 Vermont citizens each want to give $1 to the State Democratic Party because, though unfamiliar with the details of the individual races, they would like to make a small financial contribution to the goal of electing a Democratic state legislature. And further imagine that the party believes control of the legislature will depend on the outcome of three (and only three) House races. The Act forbids the party from giving $2,000 (of the $6,000) to each of its candidates in those pivotal races. Indeed, it permits the party to give no more than $200 to each candidate, thereby thwarting the aims of the 6,000 donors from making a meaningful contribution to state politics by giving a small amount of money to the party they support. Thus, the Act would severely inhibit collective political activity by preventing a political party from using contributions by small donors to provide meaningful assistance to any individual candidate. See supra, at 19.
We recognize that we have previously upheld limits on contributions from political parties to candidates, in particular the federal limits on coordinated party spending. Colorado II, 533 U. S. 431. And we also recognize that any such limit will negatively affect to some extent the fund-allocating party function just described. But the contribution limits at issue in Colorado II were far less problematic, for they were significantly higher than Act 64’s limits. See id., at 438–439, and n. 3, 442, n. 7 (at least $67,560 in coordinated spending and $5,000 in direct cash contributions for U. S. Senate candidates, at least $33,780 in coordinated spending and $5,000 in direct cash contributions for U. S. House candidates). And they were much higher than the federal limits on contributions from individuals to candidates, thereby reflecting an effort by Congress to balance (1) the need to allow individuals to participate in the political process by contributing to political parties that help elect candidates with (2) the need to prevent the use of political parties “to circumvent contribution limits that apply to individuals.” Id., at 453. Act 64, by placing identical limits upon contributions to candidates, whether made by an individual or by a political party, gives to the former consideration no weight at all.
We consequently agree with the District Court that the Act’s contribution limits “would reduce the voice of political parties” in Vermont to a “whisper.” 118 F. Supp. 2d, at 487. And we count the special party-related harms that Act 64 threatens as a further factor weighing against the constitutional validity of the contribution limits.
Third, the Act’s treatment of volunteer services aggravates the problem. Like its federal statutory counterpart, the Act excludes from its definition of “contribution” all “services provided without compensation by individuals volunteering their time on behalf of a candidate.” Vt. Stat. Ann., Tit. 17, §2801(2) (2002). Cf. 2 U. S. C. §431(8)(B)(i) (2000 ed. and Supp. III) (similar exemption in federal campaign finance statute). But the Act does not exclude the expenses those volunteers incur, such as travel expenses, in the course of campaign activities. The Act’s broad definitions would seem to count those expenses against the volunteer’s contribution limit, at least where the spending was facilitated or approved by campaign officials. Vt. Stat. Ann., Tit. 17, §2801(3) (2002) (“[E]xpenditure” includes “anything of value, paid … for the purpose of influencing an election”); §§2809(a), (c) (Any “expenditure … intentionally facilitated by, solicited by or approved by the candidate” counts as a “contribution”). And, unlike the Federal Government’s treatment of comparable requirements, the State has not (insofar as we are aware) created an exception excluding such expenses. Cf. 2 U. S. C. §§431(8)(B)(iv), (ix) (2000 ed. and Supp. III) (excluding from the definition of “contribution” volunteer travel expenses up to $1,000 and payment by political party for campaign materials used in connection with volunteer activities).
The absence of some such exception may matter in the present context, where contribution limits are very low. That combination, low limits and no exceptions, means that a gubernatorial campaign volunteer who makes four or five round trips driving across the State performing volunteer activities coordinated with the campaign can find that he or she is near, or has surpassed, the contribution limit. So too will a volunteer who offers a campaign the use of her house along with coffee and doughnuts for a few dozen neighbors to meet the candidate, say, two or three times during a campaign. Cf. Vt. Stat. Ann., Tit. 17, §2809(d) (2002) (excluding expenditures for such activities only up to $100). Such supporters will have to keep careful track of all miles driven, postage supplied (500 stamps equals $200), pencils and pads used, and so forth. And any carelessness in this respect can prove costly, perhaps generating a headline, “Campaign laws violated,” that works serious harm to the candidate.
These sorts of problems are unlikely to affect the constitutionality of a limit that is reasonably high. Cf. Buckley, 424 U. S., at 36–37 (Coordinated expenditure by a volunteer “provides material financial assistance to a candidate,” and therefore “may properly be viewed as a contribution”). But Act 64’s contribution limits are so low, and its definition of “contribution” so broad, that the Act may well impede a campaign’s ability effectively to use volunteers, thereby making it more difficult for individuals to associate in this way. Cf. id., at 22 (Federal contribution limits “leave the contributor free to become a member of any political association and to assist personally in the association’s efforts on behalf of candidates”). Again, the very low limits at issue help to transform differences in degree into difference in kind. And the likelihood of unjustified interference in the present context is sufficiently great that we must consider the lack of tailoring in the Act’s definition of “contribution” as an added factor counting against the constitutional validity of the contribution limits before us.
Fourth, unlike the contribution limits we upheld in Shrink, see supra, at 16, Act 64’s contribution limits are not adjusted for inflation. Its limits decline in real value each year. Indeed, in real dollars the Act’s limits have already declined by about 20% ($200 in 2006 dollars has a real value of $160.66 in 1997 dollars). A failure to index limits means that limits which are already suspiciously low, see supra, at 14–17, will almost inevitably become too low over time. It means that future legislation will be necessary to stop that almost inevitable decline, and it thereby imposes the burden of preventing the decline upon incumbent legislators who may not diligently police the need for changes in limit levels to assure the adequate financing of electoral challenges.
Fifth, we have found nowhere in the record any special justification that might warrant a contribution limit so low or so restrictive as to bring about the serious associational and expressive problems that we have described. Rather, the basic justifications the State has advanced in support of such limits are those present in Buckley. The record contains no indication that, for example, corruption (or its appearance) in Vermont is significantly more serious a matter than elsewhere. Indeed, other things being equal, one might reasonably believe that a contribution of say, $250 (or $450) to a candidate’s campaign was less likely to prove a corruptive force than the far larger contributions at issue in the other campaign finance cases we have considered. See supra, at 15–17.
These five sets of considerations, taken together, lead us to conclude that Act 64’s contribution limits are not narrowly tailored. Rather, the Act burdens First Amendment interests by threatening to inhibit effective advocacy by those who seek election, particularly challengers; its contribution limits mute the voice of political parties; they hamper participation in campaigns through volunteer activities; and they are not indexed for inflation. Vermont does not point to a legitimate statutory objective that might justify these special burdens. We understand that many, though not all, campaign finance regulations impose certain of these burdens to some degree. We also understand the legitimate need for constitutional leeway in respect to legislative line-drawing. But our discussion indicates why we conclude that Act 64 in this respect nonetheless goes too far. It disproportionately burdens numerous First Amendment interests, and consequently, in our view, violates the First Amendment.
We add that we do not believe it possible to sever some of the Act’s contribution limit provisions from others that might remain fully operative. See Champlin Refining Co. v. Corporation Comm’n of Okla., 286 U. S. 210, 234 (1932) (“invalid part may be dropped if what is left is fully operative as a law”); see also Minnesota v. Mille Lacs Band of Chippewa Indians, 526 U. S. 172, 191 (1999) (severability “essentially an inquiry into legislative intent”); Vt. Stat. Ann., Tit. 1, §215 (2003) (severability principles apply to Vermont statutes). To sever provisions to avoid constitutional objection here would require us to write words into the statute (inflation indexing), or to leave gaping loopholes (no limits on party contributions), or to foresee which of many different possible ways the legislature might respond to the constitutional objections we have found. Given these difficulties, we believe the Vermont Legislature would have intended us to set aside the statute’s contribution limits, leaving the legislature free to rewrite those provisions in light of the constitutional difficulties we have identified.
We conclude that Act 64’s expenditure limits violate the First Amendment as interpreted in Buckley v. Valeo. We also conclude that the specific details of Act 64’s contribution limits require us to hold that those limits violate the First Amendment, for they burden First Amendment interests in a manner that is disproportionate to the public purposes they were enacted to advance. Given our holding, we need not, and do not, examine the constitutionality of the statute’s presumption that certain party expenditures are coordinated with a candidate. Vt. Stat. Ann., Tit. 17, §2809(d) (2002). Accordingly, the judgment of the Court of Appeals is reversed, and the cases are remanded for further proceedings.
It is so ordered.