HALL, Senior Circuit Judge.

The City of San Diego bans contributions exceeding $250 to any committee supporting or opposing a candidate for City Council office. San Diego Municipal Code (SDMC) § 27.2935 (2005). This limit applies to recall efforts because the term "candidate" includes a "City officeholder who becomes the subject of a recall election." Id. § 27.2903.

In this case, Citizens for Clean Government ("Citizens") argues that the contribution limit is unconstitutional as applied to the signature-gathering phase of a recall election. San Diego asserts its ordinance is valid under Buckley v. Valeo, 424 U.S. 1, 96 S.Ct. 612, 46 L.Ed.2d 659 (1976), and the district court agreed. It denied Citizens' request for preliminary injunctive relief, and this court affirmed. The parties then stipulated to a final judgment against Citizens. We now vacate and remand.

I.

The recall election of a City Councilman in San Diego ("the City") proceeds in two phases. In the first phase, recall proponents must gather the signatures of at least 15 percent of the voters in the council district. SDMC § 27.2703 (2000). They may begin collecting signatures 21 days after they publish a notice of intent to circulate a recall petition, id. § 27.2708, and must collect the required number of signatures within 60 days of the notice. Id. § 27.2715. In total, recall proponents have 39 days to collect the signatures required to file the recall petition. In the second phase of a recall, which occurs only after the requisite number of signatures has been collected and each signature verified, the City calls a special election in which voters choose whether to recall their councilman and select a replacement from a slate of successors. Id. §§ 27.2722, 27.2725.

In 2003, Citizens was formed as an unincorporated association to recall Scott Peters, the City Councilman for District 1. To succeed at obtaining a special election, Citizens had to collect over 12,000 signatures on its recall petition, and it planned to do so by employing paid petition circulators. Citizens ultimately failed to collect the required number of signatures in time. The following year, Scott Peters was re-elected in the 2004 city general election.

Citizens filed this lawsuit for declaratory and injunctive relief on June 20, 2003, the day after it published its notice of intent to circulate a recall petition. It alleged that the City's campaign contribution limit, then codified at SDMC § 27.2941, violated its right to freedom of speech and freedom of association under the First Amendment of the United States Constitution.

On July 3, 2003, the district court denied Citizens' request for preliminary injunctive relief. Citizens, it held, had not shown a likelihood of success on the merits of its claim that the City's ordinance was unconstitutional under Buckley, the Supreme Court's seminal campaign finance decision. On November 14, 2003, this court affirmed. The parties stipulated to a final judgment against Citizens on October 6, 2004, and Citizens timely filed this appeal on November 5, 2004.

Because the recall campaign against Scott Peters did not obtain enough signatures to reach the ballot, and Councilman Peters was ultimately re-elected in 2004, we must address the question of whether this case is moot. The Supreme Court has held that the following two factors may preclude a finding of mootness: "(1) the challenged action was in its duration too short to be fully litigated prior to its cessation or expiration, and (2) there [is] a reasonable expectation that the same complaining party [will] be subjected to the same action again." First Nat'l Bank of Boston v. Bellotti, 435 U.S. 765, 774, 98 S.Ct. 1407, 55 L.Ed.2d 707 (1978) (quoting Weinstein v. Bradford, 423 U.S. 147, 149, 96 S.Ct. 347, 46 L.Ed.2d 350 (1975) (alteration in original)). In short, the question is whether the alleged injury is "capable of repetition, yet evading review." S. Pac. Terminal Co. v. ICC, 219 U.S. 498, 515, 31 S.Ct. 279, 55 L.Ed. 310 (1911). We have previously noted that election cases tend to fall within this exception. See Alaska Right to Life Comm. v. Miles, 441 F.3d 773, 779 (9th Cir.2006); Cal. Pro-Life Council, Inc. v. Getman, 328 F.3d 1088, 1095 n. 4 (9th Cir.2003).

Indeed, both factors are present here. First, unless a recall campaign succeeds in reaching the ballot, the contribution limitation is relevant for only those 60 days between the publication of a notice of intent and the deadline for collecting the requisite number of signatures. This amount of time is too short to litigate a challenge to the ordinance to finality in court. Second, if Citizens attempts a recall again, it will be subject to the same contribution limits. See Moore v. Ogilvie, 394 U.S. 814, 816, 89 S.Ct. 1493, 23 L.Ed.2d 1 (1969) (rejecting mootness argument in voting rights case despite occurrence of election). Citizens' claim, accordingly, is not moot.

Because this appeal relates to a permanent injunction, we are not constrained by the more limited standard of review that applied at the preliminary injunction phase of this litigation. See Walczak v. EPL Prolong, 198 F.3d 725, 730 (9th Cir.1999). While we review the district court's decision to deny permanent injunctive relief for abuse of discretion, we review de novo any legal conclusions underlying that decision. See Jones v. City of L.A., 444 F.3d 1118, 1125-26 (9th Cir.2006). Though the district court correctly chose to apply Buckley's reduced scrutiny, we hold that it erred by deciding, apparently as a matter of law, that the City had a sufficient interest justifying the application of its contribution limits to the signature-gathering phase of a recall election. We hold that the City must provide evidence demonstrating a sufficiently important government interest, such as the risk of corruption, in this context.

II.

Limits on contributions to political campaigns are permissible under the First Amendment "as long as the Government demonstrates that the limits are `closely drawn' to match a `sufficiently important government interest.'" Randall v. Sorrell, ___ U.S. ___, ____, 126 S.Ct. 2479, 2491, 165 L.Ed.2d 482 (2006) (plurality opinion) (quoting Buckley, 424 U.S. at 25, 96 S.Ct. 612). This standard is sometimes referred to as "less rigorous" scrutiny. McConnell v. FEC, 540 U.S. 93, 136, 124 S.Ct. 619, 157 L.Ed.2d 491 (2003).

Under this test, the Supreme Court and this circuit have tended to uphold limits on contributions to candidate campaigns. See, e.g., id.; Nixon v. Shrink Mo. Gov't PAC, 528 U.S. 377, 120 S.Ct. 897, 145 L.Ed.2d 886 (2000); Mont. Right to Life Ass'n v. Eddleman, 343 F.3d 1085 (9th Cir.2003). But see Randall, 126 S.Ct. at 2494-95 (striking down state contribution caps as too stringent). By contrast, neither the Supreme Court nor this court has found an interest sufficiently important to justify limits on contributions to ballot measure campaigns. See, e.g., Citizens Against Rent Control/Coalition for Fair Housing v. City of Berkeley, 454 U.S. 290, 299, 102 S.Ct. 434, 70 L.Ed.2d 492 (1981); Bellotti 435 U.S. at 765, 98 S.Ct. 1407.

Citizens, which likens recall petitions to ballot measure petitions, argues that the Berkeley case implicitly establishes a strict scrutiny standard for contribution limits in ballot measure campaigns. The cases provide no conclusive support for this reading, and we decline to adopt that interpretation here. First, in Berkeley, the Court avoided any direct statement regarding the standard of review. And, to the extent the standard can be inferred, the Court seemed to apply the level of scrutiny described in Buckley. See Berkeley, 454 U.S. at 299, 102 S.Ct. 434 ("It is clear, therefore, that [the ordinance] does not advance a legitimate governmental interest significant enough to justify its infringement of First Amendment rights.") (emphasis added). Two concurrences in Berkeley also suggest that Buckley's less rigorous scrutiny applies. Id. at 301, 102 S.Ct. 434 ("I must assume that the Court is following our consistent position that this type of governmental action is subjected to less rigorous scrutiny than a direct restriction on expenditures.") (Marshall, J., concurring); id. at 302, 102 S.Ct. 434 ("Berkeley must demonstrate that its ordinance advances a sufficiently important governmental interest and employs means closely drawn to avoid unnecessary abridgment of First Amendment freedoms.") (Blackmun, J., and O' Connor, J., concurring) (quoting Buckley, 424 U.S. at 25, 96 S.Ct. 612) (internal punctuation omitted).

As both the Supreme Court and this court have recently suggested, the act of contribution, rather than the context in which contribution occurs, determines the standard of review. In a case addressing limits on corporate campaign contributions, the Court stated that

the level of scrutiny is based on the importance of the "political activity at issue" to effective speech or political association. . . . [R]estrictions on political contributions have been treated as merely "marginal" speech restrictions subject to relatively complaisant review under the First Amendment, because contributions lie closer to the edges than to the core of political expression.

FEC v. Beaumont, 539 U.S. 146, 161, 123 S.Ct. 2200, 156 L.Ed.2d 179 (2003) (citations omitted). The Court reiterated that lesser scrutiny is appropriate because "the transformation of contributions into political debate involves speech by someone other than the contributor." Id. at 161-62, 123 S.Ct. 2200 (quoting Buckley, 424 U.S. at 21, 96 S.Ct. 612 (punctuation omitted)).

Similarly, this court has read Buckley to stand for the principle that "[contribution limits do not significantly burden speech because the communicative content of the act of contributing is largely symbolic, and therefore is not diminished by limits on the amount of the contribution." Jacobus v. Alaska, 338 F.3d 1095, 1108 (9th Cir.2003). Because it restricts only this indirect speech, a contribution limit "does not in any way infringe the contributor's freedom to discuss candidates and issues." Id. (quoting Buckley, 424 U.S. at 21, 96 S.Ct. 612). Accordingly, in light of Beaumont and its predecessors, we have applied Buckley's less rigorous scrutiny to contribution caps. See Mont. Right to Life, 343 F.3d at 1092 (limits on contributions made by political action committees to candidates); Jacobus, 338 F.3d at 1109 (limits on "soft money" contributions to political parties).

We find no precedent holding that contributions to ballot measure campaigns convey a different type or degree of speech from contributions to candidates or parties. We therefore hold that the district court's preliminary decision to apply Buckley's less rigorous scrutiny was correct.

III.

We now turn to the district court's holding that the City has a sufficiently important state interest to justify imposing contribution limits during the signature-gathering phase of a recall effort. The paradigmatic sufficient state interest under Buckley is the prevention of corruption, or the appearance of corruption, in the political process. Randall, 126 S.Ct. at 2488-90; McConnell, 540 U.S. at 143, 124 S.Ct. 619. Limits on political contributions serve the government's interest in preventing corruption because they reduce the risk of quid pro quo arrangements and mitigate "the appearance of corruption spawned by the real or imagined coercive influence of large financial contributions on candidates' positions and on their actions if elected to office." Buckley, 424 U.S. at 25, 96 S.Ct. 612.

Corruption, as the Court has defined it more recently, can encompass more than straightforward quid pro quo transactions:

Just as troubling to a functioning democracy as classic quid pro quo corruption is the danger that office-holders will decide issues not on the merits or the desires of their constituencies, but according to the wishes of those who have made large financial contributions valued by the officeholder. Even if it occurs only occasionally, the potential for such undue influence is manifest. And unlike straight cash-for-votes transactions, such corruption is neither easily detected nor practical to criminalize.

McConnell, 540 U.S. at 153, 124 S.Ct. 619. Notably, the Court rejected any narrower conception of corruption as "crabbed" and "ignoring] precedent, common sense, and the realities of political fundraising." Id. at 152, 124 S.Ct. 619.

Though it has defined corruption flexibly, the Court has also rejected the argument that a state may limit contributions simply because they may sway the outcome of an election. See Bellotti, 435 U.S. at 790, 98 S.Ct. 1407 ("[T]he fact that advocacy may persuade the electorate is hardly a reason to suppress it. . . ."). Rather, contribution limits must target some "greater or more imminent danger to the public interest." Id. at 792, 98 S.Ct. 1407; see also Mont. Chamber of Commerce v. Argenbright, 226 F.3d 1049, 1057-58 (9th Cir.2000) (affirming district court's finding that corporate contributions to ballot measure campaigns posed no "imminent threat to the democratic process").

Between these two poles of clearly valid and clearly invalid anti-corruption interests, legislators are free to craft new arguments about corruption provided they acknowledge that "[t]he quantum of empirical evidence needed to satisfy heightened judicial scrutiny . . . will vary up or down with the novelty and plausibility of the justification raised." Shrink, 528 U.S. at 391, 120 S.Ct. 897. Because the regulations at issue in Shrink were similar to those in Buckley, the state's asserted interest was neither novel nor implausible. Id. at 393, 120 S.Ct. 897. Therefore, the Court declined to impose, let alone articulate, a stringent evidentiary burden. Id. at 393-95, 120 S.Ct. 897.

Despite the flexibility implied by its sliding scale approach, the Court has "never accepted mere conjecture as adequate to carry a First Amendment burden." id. at 392, 120 S.Ct. 897. In finding a sufficiently important state interest in Shrink, the Court relied on evidence provided by the state of Missouri and in the record below, including news accounts and affidavits from the legislative history. Id. at 393-94, 120 S.Ct. 897. By introducing similar evidence in a case in this circuit, the state of Montana successfully justified its laws limiting candidate contributions by individuals and political action committees. See Mont. Right to Life, 343 F.3d at 1093.

While the laws at issue in Montana Right to Life and Shrink were more easily analogized to those in Buckley because they limited contributions to candidates, the Supreme Court also emphasized the evidentiary burden when it evaluated "coordinated expenditures" made by political parties by arrangements with specific candidates. See FEC v. Colo. Republican Fed. Campaign Comm., 533 U.S. 431, 121 S.Ct. 2351, 150 L.Ed.2d 461 (2001). These expenditures are treated as contributions, and, therefore, limits on coordinated expenditures are entitled to Buckley's less rigorous scrutiny. See id at 456, 121 S.Ct. 2351. In Colorado Republican, the Court found that "substantial evidence," including declarations from party operatives and legislators, demonstrated how candidates, donors and parties "test the limits" of the existing law, and why a less deferential standard of review would further undermine Congress's anti-corruption efforts. Id. at 457, 121 S.Ct. 2351. Importantly, after the standard of review was settled, the only "bone of contention" in Colorado Republican was whether the government had marshaled adequate evidence of its interest. See id. at 456, 121 S.Ct. 2351.

Because the government has the burden of demonstrating its state interest, Shrink, 528 U.S. at 387-88, 120 S.Ct. 897, any empirical evidence it offers must overcome any evidence to the contrary presented by the plaintiff. See id. at 394, 120 S.Ct. 897 ("There might, of course, be need for a more extensive evidentiary documentation if [the plaintiffs] had made any showing of their own. . . .") . In this circuit, we have upheld the district court's decision to overturn limits on corporate spending because the government failed to counter the plaintiff's evidence that the state's electoral process was functioning healthily. See Argenbright, 226 F.3d at 1055, 1057; cf. id. at 1058-59 (McKeown, J., concurring) (citing Shrink's discussion of evidence). We specifically remarked upon the excellent briefing and record prepared by both sides in Argenbright, id. at 1055, and, following both Supreme Court and Ninth Circuit precedent, we again emphasize the importance of factual development.

The district court appeared to determine as a matter of law that the City had a sufficiently important interest in limiting contributions to recall petition campaigns. The district court made one reference to Municipal Code section 27.2901, which states that one purpose of the City's campaign finance laws is "to avoid the corruption or the appearance of corruption brought about when candidates for elective City office accept large campaign contributions." The district court's findings that the City's ordinance furthered a sufficient government interest, however, rested on hypothetical situations not derived from any record evidence or governmental findings. At oral argument on appeal, the City also conceded that there is no evidence in the record of any corruption, or the potential for corruption in the recall context. It instead pointed out a recent pattern of corrupt conduct in local politics. Its brief described several hypothetical types of corruption that could arise from allowing unlimited recall contributions.

We cannot hold that hypothetical, accompanied by vague allusions to practical experience, demonstrate a sufficiently important state interest. And, while relevant to the analysis, the City's statement of purpose only vaguely articulates a risk of corruption. In Jacobus v. Alaska, where we upheld a state's soft money contribution limits, we relied on legislative findings made on the basis of a state-commissioned report, 338 F.3d at 1099, as well as reasoning from the Supreme Court specifically addressing the possibility of corruption related to party spending in federal elections. Id. at 1113 (citing Colo. Republican, 533 U.S. at 461, 121 S.Ct. 2351). Here, by contrast, the City offers no evidence of deliberation on the issue of campaign finance in recall elections, and it has no recourse to legal authority addressing these exact issues because none exists.

We make no statement about whether the City, on remand, will be able to develop and introduce evidence establishing a sufficiently important interest in limiting contributions to recall campaigns. We hold only that the district court erred by failing to require evidence clarifying the analogy between the state interest in Buckley and the one asserted here. We therefore VACATE the denial of a permanent injunction and declaratory relief and REMAND for further evidentiary development in accordance with this opinion.

VACATED AND REMANDED.