The Honorable Thomas M. Bakk
The Honorable David W. Hann
Re: Constitutional and Practical Issues with Senate File 154
Dear Majority Leader Bakk, Minority Leader Hann, and members of the Senate:
On behalf of the Center for Competitive Politics, I am writing you today to respectfully submit the following comments regarding the constitutional and practical impact of the provisions contained in Senate File 154, which proposes amendments to Minnesota’s campaign finance laws.
The Center for Competitive Politics is a nonpartisan, nonprofit 501(c)(3) organization that promotes and protects the First Amendment political rights of speech, assembly, and petition. It was founded in 2005 by Bradley A. Smith, a former Chairman of the Federal Election Commission. In addition to scholarly and educational work, the Center is actively involved in targeted litigation against unconstitutional laws at both the state and federal levels. For instance, we presently represent nonprofit, incorporated educational associations in challenges to state campaign finance laws in Colorado, Delaware, and Nevada. We are also involved in litigation against the state of California.
If Senate File 154 becomes law as written, there is a high likelihood that the law will be found unconstitutional if challenged in court. Any potential legal action will cost the state a great deal of money defending the case, and will distract the Attorney General’s office from meritorious legal work. Additionally, it is probable that the state will be forced by the courts to award legal fees to successful plaintiffs. Legal fee awards are often expensive, and can cost governments well over one hundred thousand dollars.
Various sections of this measure are unconstitutionally vague. Senate File 154 regulates an expansive amount of speech, including books and websites. The bill’s disclosure provisions go farther than any form of disclosure approved by the U.S. Supreme Court, and are drafted in such a way as to impose onerous burdens on small organizations.
I. The modified definition of “expressly advocating” is vague, and a similar definition adopted in Iowa was declared unconstitutional by the United States Court of Appeals for the Eighth Circuit, which also covers Minnesota.
The U.S. Supreme Court’s decision in the seminal campaign finance case of Buckley v. Valeo[1] famously limited the speech that could trigger political committee status under federal law to “express words of advocacy of election or defeat, such as ‘vote for,’ ‘elect,’ ‘support,’ ‘cast your ballot for,’ ‘Smith for Congress,’ ‘vote against,’ ‘defeat,’ [and] ‘reject.’”[2] These would become known as “magic words” of express advocacy. In the case of McConnell v. FEC,[3] which reviewed the constitutionality of the Bipartisan Campaign Reform Act of 2002 (BCRA), the Court recognized that speech not containing these magic words might still be the equivalent of express advocacy, but nevertheless reaffirmed the central rule that statutes regulating speech may not be vague.
Additionally, a case decided in the United States Court of Appeals for the Eighth Circuit, which covers Minnesota, reviewed an Iowa definition of express advocacy that was similar to the provision in Section 1, Subdivision 16a(2) of S.F. 154. The Court’s opinion noted that “[t]o avoid uncertainty, and therefore invalidation of a regulation of political speech, the Supreme Court in Buckley, established a bright-line test….There is no way for [the Iowa Right to Life Committee] to know ahead of time whether its speech does or does not meet the definition and therefore subjects them to government reporting and disclosure requirements. The possible intent and effect attributed to the speech creates uncertainty… [W]e find that the State’s definition of express advocacy creates uncertainty and potentially chills discussion of public issues… [and] we affirm the district court’s grant of preliminary injunctive relief.”[4]
The Iowa definition that was challenged read as follows:
b. When taken as a whole and with limited reference to external events such as the proximity to the election, could only be interpreted by a reasonable person as containing advocacy of the election or defeat of one or more clearly identified candidate(s) … because:
(1) The electoral portion of the communication is unmistakable, unambiguous, and suggestive of only one meaning; and
(2) Reasonable minds could not differ as to whether it encourages action to elect or defeat one or more clearly identified candidate(s) … or encourages some other kind of action.[5]
The proposed new additional definition of “expressly advocating” in S.F. 154 reads as follows:
(2) that a communication, when taken as a whole and with limited reference to external events, including the proximity to the election, is not susceptible to any other interpretation by a reasonable person other than that as advocating the election or defeat of one or more clearly identified candidates.
The First Amendment problem with this language is that it introduces “external events” and outside information into analyzing the language of the communication to determine whether it is express advocacy. Which external events? What is the significance of proximity to the election? How many days constitute “proximity to the election?”
As the Supreme Court said in Buckley, vague laws that regulate speech “put the speaker in these circumstances wholly at the mercy of the varied understanding of his hearers and consequently of whatever inference may be drawn as to his intent and meaning. Such a distinction offers no security for free discussion. In these conditions it blankets with uncertainty whatever may be said. It compels the speaker to hedge and trim.”[6]
While it is not clear how the Court would rule on the proposed language, certainly the proposed provision could be made clearer. For example, the following language would achieve the provision’s goals with a more precise definition of “expressly advocating”:
that a communication, in the context of only the communication itself, can have no other reasonable meaning than to urge the election or defeat of one or more candidates.
Current Minnesota law defines such activity with the same precision employed by the Supreme Court in Buckley: it requires actual “words or phrases of express advocacy.”[7] The amendment proposed in S.F. 154 would greatly expand the law’s definition. Thus, rather than clarifying the law, it would confuse speakers and spawn litigation about the proper scope of campaign finance regulation. Instead of ensuring that political speech is uniformly and constitutionally regulated, this addition will muddy the waters, replacing a crisp rule with a more amorphous standard. Moreover, by eliminating a bright-line test for regulated speech, it invites political gamesmanship and partisan polarization concerning messages that should be judged on their merits by voters, not by lawyers and public officials. This will inevitably require speakers to hire expert attorneys in this highly specialized area of law, or, for the smaller organizations that cannot afford such help, risk enforcement actions that could drive their voices from the public debate.
If the Legislature wishes to add to the definition of expressly advocating, it should tread carefully and write as precise a provision as possible.
II. The legislation’s “electioneering communication” definition is stunningly broad and highly susceptible to legal challenge.
The bill’s over-inclusive definitions cause S.F. 154 to sweep far more broadly than is constitutionally appropriate.
Section 4 defines an “electioneering communication” as a communication referring to a clearly identified candidate distributed within thirty days of a primary election or sixty days of a general election if the communication “can be received by more than 1,500 persons” in a state legislative (or other) district.
The legislation does not require that a communication include an appeal to vote before it can be subject to regulation and disclosure rules. Electioneering communication provisions, such as those at the federal level, have been upheld in some circumstances, but generally only as applies to broadcast communications, such as television and radio ads. The McConnell ruling also upheld such a provision only after an enormous volume of fact finding by Congress.
Additionally, S.F. 154’s extension of electioneering communication language to “printed material,” including newspapers, periodicals, books, and leaflets as well as phone calls and billboards, without any evidence that such communications pose a risk of corruption, would put Minnesota on dangerous constitutional ground.
For example, if a corporation funded a book (which is obviously “printed material”), which incidentally named a candidate for office, that corporation could find itself subject to Minnesota’s electioneering communications laws if the book were published or distributed within the 30 or 60-day windows. This could conceivably force major publishers such as Penguin, Simon & Schuster, or McGraw-Hill to file electioneering communications reports.
Although forcing book publishers to be wary of when they print and distribute books is in and of itself an obvious constitutional harm, Minnesota’s statute also applies to special elections – which are, by their nature, usually unplanned. Books that discuss the workings of Minnesota government, or even mention the wrong state senator, could be accidentally regulated under the law.
The bill also provides that communications published by “the news media” are not electioneering communications. However, even this safe-harbor is vaguely described: it does not define “the news media.” Some news outlets are now projects of § 501(c)(4) organizations – such as Think Progress, a widely read blog that is a project of the Center for American Progress Action Fund.[8] Would such publications be exempt if the same information is published in printed material?
In short, nearly any mention of a candidate in a covered communication may be regulated under the legislation. This stunningly broad definition would appear to cover nonpartisan voter guides and listings of votes cast by lawmakers if distributed to the public by covered organizations during periods that would be regulated by the bill. This broad coverage will silence many speakers during the electioneering communication period and deprive voters of vitally important nonpartisan information.
Also, unlike the federal electioneering communications provisions, there is no objective way for a speaker to determine whether a communication “can be received by more than 1,500 persons” in a district. The federal electioneering communications statute required the Federal Communications Commission to establish and maintain an online electioneering communications database,[9] so that speakers can know whether the communication can be heard by 50,000 or more persons in a state or district. This proposal provides no such directive to any state agency, nor is there any objective way to determine the reach of a billboard or sign, both of which are explicitly covered by the definition.
One of the reasons why the federal electioneering communications statute has survived constitutional review is because it narrowly regulates a specific type of ad: broadcast communications. The same analysis would plainly not extend to a statute that seeks to regulate books or (potentially) communications published by a news outlet affiliated with a § 501(c)(4) organization, released shortly before a state election.
III. The legislation’s overbroad disclosure requirements run contrary to longstanding Supreme Court First Amendment precedent and are highly vulnerable to a legal challenge.
Section 4, Subdivision 4(b)(6) triggers some form of disclosure from organizations that spend more than $5,000 out of their general treasuries to fund electioneering communications. However, the language of the provision is unclear as to what form of disclosure is demanded, and could be challenged as unconstitutionally vague.
As written, the law compels the disclosure of “the name, address, and amount attributable to each person that paid the association membership dues or fees, or made donations to the association that, in total, aggregate more than $1,000 of the money used by the association for electioneering communications.”[10] It is unclear if this statute compels the generalized disclosure of anyone who can be attributed to have donated more than $1,000, or simply requires the disclosure of those whose donation over $1,000 was used to directly finance a communication, or communications in general – that is, earmarked contributions. Such a vague provision would inevitably invite lengthy litigation.
However, changing the text of the statute to compel generalized donor disclosure would likely be unconstitutional. In Citizens United v. FEC,[11] the U.S. Supreme Court upheld an electioneering communications disclosure regime that only compelled the disclosure of earmarked contributions.[12] The Supreme Court has never upheld any other donor disclosure requirement for speech that is not express advocacy.
This “earmark-only” disclosure provision was specifically designed to prevent corporations from disclosing all of their funders as a condition of engaging in protected First Amendment political speech. Contrary to claims by those who advocate for greater regulation of political speech, generalized disclosure for electioneering communications has never been upheld by the United States Supreme Court. Indeed, the Supreme Court has never upheld generalized disclosure as it pertains to genuine, non-pejorative issue speech – which would be regulated by the proposed law.[13]
Conversely, adopting language only requiring the disclosure of those contributions specifically intended for electioneering communications would be constitutional, pursuant to a nearly forty-year-old unbroken chain of U.S. Supreme Court litigation.[14]
Such an earmarking provision would also cure another problem with the bill – the fact that it could compel the generalized disclosure of donors to Section 501(c)(3) organizations engaged in legitimate nonpartisan voter information activity. Section 501(c)(3) organizations are prohibited under federal tax laws from engaging in any electoral advocacy, and as such, the state has no interest in the donors to such groups. But such groups are permitted to educate the public through neutral, nonpartisan voter guides and similar materials.
This is a reason why many states have enacted electioneering communications statutes that exempt neutral communications, or prevent § 501(c)(3) organizations from being regulated under such statutes.[15] Presently, the Center for Competitive Politics represents a § 501(c)(3) organization in a suit against Delaware’s[16] electioneering communications regime – which compels generalized donor disclosure from organizations engaged in neutral, nonpartisan issue speech.
IV. The inclusion of cross-referenced websites in the determination of whether an electioneering communication is present has been flatly rejected by the Supreme Court in 2007, and will perversely result in less informative disclosure.
There are two significant problems with Section 4, Subdivision 1(b)’s language that: “[i]f an electioneering communication clearly directs recipients to another communication, including a Web site, on-demand or streaming video, or similar communications, the electioneering communication consists of both the original electioneering communication and the communication to which recipients are directed and the cost of both must be included when determining if disclosure is required under this section.”
The United States Supreme Court squarely and overtly rejected such cross-referencing in FEC v. Wisconsin Right to Life.[17] In that case, which invalidated certain federal regulations on issue speech, the Court considered a communication’s “specific and repeated cross-reference” to a website, but explicitly rejected consideration of such communication in determining the communication’s status for purposes of regulation. The Court found that “[a]ny express advocacy on the website, already one step removed from the text of the ads themselves, certainly does not render an interpretation of the ads as genuine issue ads unreasonable,”[18] reasoning that “newspaper ads and websites are not reasonable alternatives to broadcast speech in terms of impact and effectiveness.”[19]
Such a provision is not only foreclosed by Supreme Court jurisprudence, but it will also result in less, rather than more disclosure. Indeed, an organization’s website is often prominently featured on the face of an electioneering communication that organization sponsors. A curious voter may visit that website to learn more about the communication’s message, and may thereby judge the credibility and reliability of its message. This proposed legislation will discourage organizations from featuring such information on the face of communications and voters will be hampered from engaging in such independent investigations.
V. The electioneering communications exemption for board rules and advisory opinions implicitly acknowledges the considerable overbreadth posed by this proposed law.
Under S.F. 154 Section 4, Subdivision 1(c)(4), “[e]lectioneering communication does not include: any other communication specified in board rules or advisory opinions as being excluded from the definition of electioneering communications.” This blanket grant of authority at least implicitly recognizes that the new rules governing electioneering communications proposed in this legislation impose a not-insignificant burden on speakers; otherwise, allowing the rules to be relaxed without legislative action would be unnecessary. Indeed, such preemptive recognition of the law’s potential unintended consequences counsels in favor of a narrower law, rather than a provision for administrative authority to amend it. It is possible that the Minnesota Campaign Finance and Public Disclosure Board will feel that its hands are tied by the broad language in the bill, and not permit any additional substantive exemptions.
VI. The electioneering communications exemption for issue speech mentioning incumbents will inhibit protected issue speech about challengers.
Section 4, Subdivision 1(c)(5)(i)-(iii) also exempts from the definition of electioneering communications “a communication that: refers to a clearly identified candidate who is an incumbent member of the legislature or a constitutional officer; refers to a clearly identified issue that is before the legislature in the form of an introduced bill; and is made when the legislature is in session.” The Center commends the Legislature for this apparent attempt to exempt genuine issue speech from Minnesota’s electioneering communications regime. Indeed, this carve-out recognizes that legislation is often about issues, and speech about issues is undoubtedly protected by the First Amendment. Sitting legislators propose, discuss, and vote on legislation. Thus, debate about issues being considered by the Legislature necessitates the mention of those legislators.
This provision, however, is crafted in a way that will inhibit speech about challengers, while allowing the same speech about incumbents. This makes no sense. A group seeking to influence policy would want both the incumbent and the challenger to endorse its position on legislation, which may not come up for a vote until after a new Legislature or governor is sworn in after an election.
VII. The 24-hour reporting requirement for electioneering communications will severely burden less sophisticated speakers, and will increase the likelihood of inaccurate disclosure reports.
The statute appears to require less-formalized organizations, which engage in electioneering communications, to fill out reports within 24 hours of triggering the statute.[20] For less sophisticated speakers, who are not used to involving themselves in politics – and therefore have not registered a political fund with the Minnesota Campaign Finance and Public Disclosure Board, nor formed a PAC – aggregating records, prorating contributions, and filing for the first time will inevitably be a difficult endeavor. Lengthening the reporting time to involve less immediate disclosure will shield these less sophisticated actors from inadvertently filing incorrect reports – likely in an endeavor to comply with the burdensome deadline – or from incurring fines for late filing. Beyond these concerns, the reports are so burdensome that substantial First Amendment questions are raised by the provision.
VIII. The electioneering communications report provision compelling the disclosure of “any person exercising direction or control” over a disbursement is vague.
Section 4, Subdivision 4(b)(1)(ii) requires an electioneering communications report to contain the name of “any person exercising direction or control over the activities of the association with respect to the disbursement.” This provision is capable of multiple readings, and may be unconstitutionally vague. If a communication is a television advertisement, does this provision require reporting the name of the person directing the actors in the advertisement? If a communication is a newspaper ad, would it require the disclosure of the name of the person writing the copy for the ad? For a billboard advertisement, would it require the disclosure of the individual responsible for physically placing an advertisement directly on a rented billboard, their supervisor, or the president of the billboard rental company? These vagueness issues are likely to inhibit compliance with this provision.
IX. The prorating system is confusing, a paperwork nightmare, and would inevitably burden small organizations attempting to comply with the disclosure regime, exposing the state to a potentially successful legal challenge.
The proposed prorating system, whereby an organization that spends some of its general treasury funds on electioneering communications must “prorate the total disbursements made for electioneering communications during the calendar year over all general treasury money received during the calendar year,”[21] and then disclose the full name and street address of contributors whose prorated donation is over $1,000, is a paperwork nightmare. Organizations that seek to do more than one round of electioneering communications in a given calendar year will constantly be forced to recalculate figures and add, modify, or remove names and addresses from disclosure reports. The burdens on a small organization would be tremendous, and would likely require obtaining outside compliance assistance. Imposing such a burden on small organizations is likely unconstitutional. As the Court noted in Citizens United, “[t]he First Amendment does not permit laws that force speakers to retain a campaign attorney…before discussing the most salient issues of our day.”[22]
Further, it is unclear that this system would provide the state with much relevant information. Given that the system encompasses large amounts of neutral, nonpartisan speech, and could be read as a generalized donor disclosure regime, lots of “junk” will be disclosed. The name and address of a $100,000 donor to a § 501(c)(3) neutral, nonpartisan get-out-the-vote organization with a budget of $1 million would be publicly disclosed if that organization spends 2% of its funds on distributing sample ballots. This “junk disclosure” will crowd out actually relevant disclosure – the names and addresses of people giving large sums to PACs, parties, and political candidates.
Lastly, Section 4, Subdivision 4(d) appears to extend the period of donor disclosure, in certain circumstances, from the past calendar year to the past two years. At a certain temporal point, the government’s informational interest in disclosure must yield to donor privacy. This is especially true in this instance, as it is unclear under a system of generalized disclosure how a $1,250 donation to an organization in January 2015 – before this specific law was even considered by the Legislature – could be relevant to understanding a communication aired in October 2016.
X. Because of the aforementioned prorating mechanism, the poorly drafted disclosure requirements could lead to disclosure of the wrong donors.
The bill provides for disclosure of donations earmarked for electioneering communications, but only if the donation is deposited in “a segregated bank account” used to pay for the communications. Consider what would happen, for example, if an organization with a $200,000 budget received a $10,000 donation earmarked for an electioneering communication and deposited the funds in the operating account. If it spent only that $10,000 for an electioneering communication, that donor would not be disclosed. Instead, other donors who gave significantly more to the organization for general support would be disclosed. In this example, a donor who gave $15,000 to the organization because she supported its mission, and did not know the organization would make an electioneering communication, would be disclosed. As currently constructed, the $10,000 donor who actually gave the money for an electioneering communication would not be disclosed, but the $15,000 donor, who has no knowledge of the communication, and might even disagree with its message, would be disclosed to the public.
This would occur because of the prorating mechanism in the legislation, described above. In the case of the donor who earmarked the $10,000 contribution, it would consist of 5% of the organization’s budget, so 5% of his donation would be subjected to the over $1,000 pro-rated disclosure threshold. 5% of that donation is $500, so the donor’s identity would not be disclosed, as his donation is under the $1,000+ threshold. The $15,000 donation would represent 7.5% of the organization’s budget. 7.5% of $15,000 would be $1,125, or more than the over $1,000 disclosure threshold, triggering disclosure of the donor’s private information.
XI. In general, the disclosure provisions in S.F. 154 are likely to chill constitutionally protected political speech.
If passed, this bill would require nonprofit corporations wishing to protect the privacy of their contributors to choose between communicating with the public about candidates and issues in close proximity to an election and sacrificing their contributor’s privacy rights, or opting to stay silent. This has severe civil rights implications.[23] This is why even anonymous political activity has been protected in certain contexts.[24]
Those who will be hurt the most by the chilling effects of this provision will be those who advocate for unpopular causes. Indeed, a principal reason for the constitutional protection for freedom of speech is the Founders’ acute awareness that majorities seek to suppress those who speak against the crowd.
XII. The statement of attribution may account for a significant portion of a communication’s duration, which would stifle speech and trample upon First Amendment rights.
Section 4, Subdivision 7(a)(2) forces speakers to state, at the end of a broadcast communication: “The preceding communication was paid for by the [association name].” This provision may run afoul of the First Amendment and is absurd. If the organization’s ad disclaimer instead said the same thing, but with fewer words or syllables, such as “[Association name] paid for this ad,” it would be illegal if this bill became law. In the context of speech restrictions, the state’s “regulatory technique may extend only as far as the interest it serves… [and cannot] completely suppress information when narrower restrictions on expression would serve its interest as well.”[25]
In this case, the “paid for by” disclaimer could swallow a large chunk of a 15-second radio or television communication. But merely requiring that a similar message be present on screen for a certain period of time would serve the state’s interest in providing the disclaimer – but without forcing speakers to take as much as five or more seconds of a 15-second ad to comply with the bill’s speech mandate.
XIII. The immediacy of S.F. 154’s effective date will, by nature of the legislation, compel the disclosure of citizens’ private information, some of whom gave in 2014 and had no knowledge of this legislation or its implications.
The legislation is scheduled to go into effect on July 1, 2015, and will reach communications immediately on and following that date. However, as mentioned supra, this may inadvertently force the disclosure of the names and addresses of contributors who gave to organizations in 2014 – before the consideration of this bill. Those individuals contributed to organizations without any expectation of being placed at risk of public exposure. At the very least, the law should not apply to contributions made prior to the effective date of this bill.
* * *
Thank you for allowing me to submit comments on Senate File 154. I hope you find this information helpful. Should you have any further questions regarding these issues or any other campaign finance proposals, please do not hesitate to contact me at (703) 894-6835 or by e-mail at mnese@campaignfreedom.org.
Respectfully yours,
Matt Nese
Director of External Relations
Center for Competitive Politics
https://www.ifs.org/wp-content/uploads/2015/05/2015-02-17_Senate-Comments_MN_SF-154_EC-Disclosure.pdf
[1] 424 U.S. 1 (1976).
[2] Id. at 80 n. 108, incorporating by reference id. at 44 n. 52.
[3] 540 U.S. 93 (2003).
[4] Iowa Right to Life Committee, Inc. v. Williams, 187 F.3d 963, 969-970 (8th Cir. 1999).
[5] Iowa Right to Life, 187 F.3d at 969 (quoting Iowa Admin. Code r. 351-4.100(1)(b)).
[6] Buckley v. Valeo, 424 U.S. 1, 43 (1976) (quoting Thomas v. Collins, 323 U.S. 516, 535 (1945)).
[7] Minn. Stat. § 10A.01, Subd. 16a.
[8] “About Us,” Think Progress. Retrieved on February 17, 2015. Available at: http://thinkprogress.org/about; “About the Center for American Progress Action Fund,” Center for American Progress Action Fund. Retrieved on February 17, 2015. Available at: http://www.americanprogressaction.org/about/capaf-mission/ (2015).
[9] “The Electioneering Communications Database,” Federal Communications Commission. Retrieved on February 17, 2015. Available at: http://apps.fcc.gov/ecd/ (February 28, 2014).
[10] Sec. 4, Subd. 4(b)(6).
[11] 558 U.S. 310 (2010).
[12] 78 Fed. Reg. 72899, 72911 (Dec. 26, 2007).
[13] Citizens United, 558 U.S. at 368.
[14] Buckley v. Valeo, 424 U.S. 1, 80 (1976).
[15] See, e.g. Conn. Gen. Stat. § 9-601b(b)(13) (excluding “[a] lawful communication by any charitable organization which is a tax-exempt organization under Section 501(c)(3) of the Internal Revenue Code of 1986, or any subsequent corresponding internal revenue code of the United States”); 10 Ill. Comp. Stat. Ann. 5/9-1.14(b)(4) (excluding “[a] communication by an organization operating and remaining in good standing under Section 501(c)(3) of the Internal Revenue Code of 1986”); Iowa Code § 68A.401A (limiting reporting for communications merely mentioning a candidate to § 527 organizations).
[16] See Delaware Strong Families v. Biden, 13-01746 (D. Del. 2014).
[17] 551 U.S. 449 (2007).
[18] FEC v. Wis. Right to Life, Inc., 551 U.S. at 473.
[19] Id. at 477 n. 9 (citing McConnell v. FEC, 251 F. Supp. 2d, at 569-573, 646 (Kollar-Kotelly, J.)).
[20] Sec. 4, Subd. 4(a).
[21] Sec. 4, Subd. 4(c).
[22] Citizens United v. FEC, 558 U.S. 310, 324 (2010).
[23] See NAACP v. Alabama ex rel. Patterson, 357 U.S. 449 (1958); Bates v. City of Little Rock, 361 U.S. 516 (1960).
[24] See McIntyre v. Ohio Elections Commission, 514 U.S. 334, 337-338 (1995).
[25] Central Hudson Gas & Elec. Corp. v. Public Service Comm’n, 447 U.S. 557, 565 (1980).