CUDAHY, Circuit Judge.

This case presents an interesting question concerning the balance between an employer’s right of expression and its employees’ right of association. Approxi *1310 mately two months before a vote on unionization, the employer, Wiljef Transportation, Inc. (Wiljef), read to its employees a corporate by-law which states:

Section 2 — Corporate Dissolution. Wil-jef Transportation, Inc. hereby expresses as a matter of corporate policy that operations will cease and the corporation will be dissolved in the event of unionization of its employees. As hereby authorized by the Board of Directors, this by-law may be announced to the employees of Wiljef Transportation, Inc. at any time deemed appropriate by the Board.

The by-law was adopted in 1979, and the announcement occurred in 1988. In the ensuing union representation election, the employees rejected unionization. The issue in this case is whether the announcement of the by-law constituted a “permitted prediction” of plant closure or a “proscribed threat.” The NLRB held that the announcement was a threat in violation of section 8(a)(1) of the National Labor Relations Act (NLRA), and Wiljef appealed to this court. See 29 U.S.C. § 158(a)(1).

In NLRB v. Gissel Packing Co., 395 U.S. 575, 89 S.Ct. 1918, 23 L.Ed.2d 547 (1969), the Supreme Court addressed the tension between an employer’s right to announce the probable consequences of unionization and the employees’ right to organize without threat of retaliation. On the one hand, the employer’s right to communicate its views to its employees is firmly established in the First Amendment and is recognized in section 8(c) of the NLRA, which provides that “the expressing of any views, argument, or opinion ... shall not constitute or be evidence of an unfair labor practice ... if such expression contains no threat of reprisal or force or promise of benefit.” 29 U.S.C. § 158(c); see also Gissel, 395 U.S. at 617, 89 S.Ct. at 1942. On the other hand, the exceptions to the freedom of expression recognized in section 8(c) reflect the right of employees to associate free of coercion by the employer. Section 8(a)(1) of the NLRA codifies that right by declaring that it is an unfair labor practice to interfere with, restrain or coerce employees exercising their right to organize in unions. The difficulty in cases attempting to relate these two rights is in determining when speech becomes essentially coercive rather than factually informative or predictive so as to fall outside the protection of the First Amendment and violate the NLRA.

The Supreme Court in Gissel provided a framework in which to analyze this relationship between arguably conflicting rights. The employer in Gissel stated to the employees that the company’s financial situation was precarious, that a strike could result in the closing of the plant and that the employees would have difficulty finding re-employment because of their age and specialized skills. The Court held that the statements constituted prohibited coercive expression. In reaching that conclusion, the Court noted that

an employer may ... make a prediction as to the precise effects he believes unionization will have on his company. In such a case, however, the prediction must be carefully phrased on the basis of objective fact to convey an employer’s belief as to demonstrably probable consequences beyond his control or to convey a management decision already arrived at to close the plant in case of unionization. See Textile Workers v. Darlington Mfg. Co., 380 U.S. 263, 274 n. 20, 85 S.Ct. 994, 1001 n. 20, 13 L.Ed.2d 827 (1965). If there is any implication that an employer may or may not take action solely on his own initiative for reasons unrelated to economic necessities and known only to him, the statement is no longer a reasonable prediction based on available facts but a threat of retaliation based on misrepresentation and coercion, and as such without the protection of the First Amendment.

395 U.S. at 618-19, 89 S.Ct. at 1942. The parties in the case before us have offered two contradictory interpretations of the Gissel analysis. Wiljef argues that the Supreme Court established two categories of protected speech. The first category is speech involving a prediction by the employer of the future consequences of unionization. In order to fall within the protection of the statute and the Constitution, the prediction must have some objective basis *1311 in economic reality. But, if the prediction is based upon factors that are within the employer’s absolute control, as, for example, a prediction of a non-economic nature, it becomes a threat of reprisal rather than an honest forecast.

The second category in Wiljef’s paradigm includes the announcement of a management decision already made to close the plant in the event of unionization. According to Wiljef, expression of this type needs no independent objective basis and requires no economic justification. In effect, the existence of a decision already made by management provides the credibility that the objective economic factors would normally provide. Under Wiljef’s interpretation, then, an employer’s announcement of management decisions will never constitute an unfair labor practice and is always accorded First Amendment protection.

In contrast, the NLRB argues that even expression of a management decision constitutes coercion unless objective factors indicate that the management decision is likely to be implemented. Under this interpretation, expression would not be insulated from sanction merely because it took the form of a by-law. Other objective evidence would have to be introduced indicating that the management decision was real and not a sham designed to manipulate the election process. The issue in this case, then, is whether announcement of a by-law purporting to reflect a management decision is per se protected under the First Amendment, or instead is protected only if objective factors demonstrate that the by-law represents a bona fide management decision. This is a novel question in that the case law generally involves predictions by employers of the economic consequences of unionization, which must have an objective basis in order to be protected.

The real issue, however, remains credibility and bona fides. A by-law purporting to be a management decision to close a business in the event of unionization is not protected expression unless objective factors demonstrate that it is really controlling on the question of closure. This holding is consistent with the Supreme Court’s statements in Gissel and Darling-ton, and preserves the balance between free expression and the right to organize.

First, the Gissel Court’s reasoning requires that both predictions of economic consequences and management decisions have objective indicators of credibility. As noted, the Court held that “the prediction must be carefully phrased on the basis of objective fact to convey an employer’s belief as to demonstrably probable consequences beyond his control or to convey a management decision already arrived at to close the plant in case of unionization.” 395 U.S. at 618, 89 S.Ct. at 1942 (emphasis added). As the emphasized portion of this holding should make clear, the requirement that the prediction have an objective basis applies to expressions of both economic consequences and management decisions. The language of the Court provides no support for limiting the objective basis requirement to economic predictions. Moreover, it is clear from the Court’s analysis that the crucial element disqualifying such an employer announcement from protection is found where “an employer may or may not take action solely on his own initiative.” Id. A by-law stating corporate policy — which might be rescinded, amended or ignored — is not by itself enough to immunize an employer from scrutiny for coercive actions. And, in any event, the by-law at issue here, in the absence of any objective indicia of credibility, is not the sort of “management decision already arrived at” contemplated by the Court in Gissel.

The Gissel Court cited as support for its holding footnote 20 of its earlier decision in Textile Workers Union v. Darlington Mfg. Co., 380 U.S. 263, 274 n. 20, 85 S.Ct. 994, 1001 n. 20, 13 L.Ed.2d 827 (1965). In Dar-lington, the Court held that an employer may actually close its business for any reason, including its anti-union animus. 380 U.S. at 272, 85 S.Ct. at 1000. In footnote 20, the Court declared that “[njothing we have said in this opinion would justify an employer’s interfering with employee organizational activities by threatening to close his plant, as distinguished from an *1312 nouncing a decision to close already reached by the board of directors or other management authority empowered to make such a decision.” 380 U.S. at 274 n. 20, 85 S.Ct. at 1001 n. 20. The Court then noted that an employer might be encouraged to make a “definitive decision to close” on the theory that its announcement before the election might prevent unionization and the decision to close would never have to be implemented. Id. The Court concluded, however, that such a possibility is only likely to occur in marginal businesses because more profitable enterprises would not take the risk that the employees would call the employer’s bluff. The Court saw no way of eliminating the possibility of closure short of ordering the employer to remain in business if the employees voted to unionize, a step which it rejected. Ultimately, the Court concluded that the problem was not of sufficient significance to justify a change in the basic analysis. Id.

Although it is dicta, the language in footnote 20 gives some indication of the Court’s position on the issue before us. The citation to it in Gissel adds to its importance. At first glance, the language of footnote 20 might appear to support Wiljef’s position in this case. In the first sentence of the footnote, the Court implies that a decision by the board of directors to close in the event of unionization would be protected speech. Similarly, the Court in Gissel recognized that an employer may announce a management decision to close which has already been made. In principle, at least, there could be a standing, firm decision to close in the event of unionization. The context of the Court’s statements, however, indicates that the announcement in the case before us is not protected under the First Amendment.

In Darlington, the Court envisioned a management decision that would be “definitive” and that would put the employer at risk to a sufficient extent. Essentially, the Court feared that employers with marginal businesses would run the risk of being forced to close if the employees unionized, in the hope that the specter of losing their jobs would convince the employees not to unionize. The Court’s focus on marginal businesses indicates that it believed the decision to close must have some binding effect. Otherwise there would be no reason even for prosperous businesses to refrain from “bluffing.” The Court’s subsequent statement that the only available remedy would be to order the business to remain open even if the employees voted to unionize also indicates the Court’s belief that the “threat” to close was real and would be carried out. On the other hand, the Court’s language does not mandate protection for announcements of management decisions to close that are easily revocable or that are unlikely to be implemented. Management “decisions” of this kind involve virtually no risk and could be used to manipulate the situation even by prosperous businesses. However, announcement of a “definitive” management decision to close would be protected under the First Amendment.

The question then becomes whether a by-law is sufficiently definitive and binding to make its announcement a prediction grounded in reality rather than merely a threat. Wiljef has provided no objective evidence that this management “decision” would be carried out if the employees ever voted to unionize. The by-law at issue in this case was adopted years before it was invoked and it speaks only in terms of “corporate policy.” Although some mandatory language is used (“operations will cease and the corporation will be dissolved”), no plans have been drawn up for the dissolution and by-laws are easy to ignore or rescind. In fact, Wiljef has not provided any evidence other than the relatively old by-law that management had ever seriously considered the possibility of dissolution. Faced with a vote on unionization and the allegedly “binding” by-law, one might expect management to be considering its alternatives and the practical demands of going out of business.

In its brief to this court, Wiljef appears to rely almost exclusively on its argument that Gissel and Darlington protect announcement of what purport formally to be management decisions even if there is little *1313 evidence that they are seriously intended. Wiljef does not really argue in the alternative that, if objective indicia of reliability are required, it can meet' that standard. As noted, Wiljef s interpretation of Gissel and Darlington is clearly not mandated by the language of those cases. In fact, as a policy matter, such an interpretation would make it “open season” for employer threats of plant closure having little basis in reality. If merely enacting a by-law were enough for a continuing announcement of closure, many corporations would undoubtedly enact such by-laws. Little risk would be involved since the by-law would encourage employees not to unionize. But, if a union won an election, the company could simply ignore the by-law and remain open.

A more exacting scrutiny of what purport to be “management decisions” is necessary to determine whether an employer's statement is an accurate assessment of reality or an attempt to coerce. If expression by an employer constitutes an accurate and factual assessment of the likely consequences of unionization, the employer must be allowed to provide this information to employees, and the employees have a strong interest in acquiring this information so that their decision for or against a union will be informed. On the other hand, statements designed merely to coerce do not provide reliable information and, in fact, are deceptive. Wiljef has provided no evidence indicating that the by-law should be taken seriously here. It certainly can be used as an easily manipulable tool to thwart unionization.

Absent some persuasive evidence of other measures indicating that Wiljef intends to implement the corporate policy described in the by-law, the announcement of the by-law to the employees is coercive and in violation of the NLRA. Objective evidence to lend credibility to the by-law need not be based on economics and need not necessarily indicate circumstances beyond the employer’s control. If an employer has made a definitive decision to close its plant, and the definitive nature of the “decision” can be demonstrated through objective evidence, it does not matter that the decision is economically unsound or represents a free choice made by the employer. The information then is highly relevant to employees determining whether to unionize. Here no such evidence has been presented.

This circuit in NLRB v. Village IX, Inc., 723 F.2d 1360 (7th Cir.1983), appeared to apply an even stricter test than we are suggesting here. In that case, we declared:

Analytically ... the line is clear. To predict a consequence that will occur no matter how well disposed the company is toward unions is not to threaten retaliation; to predict a consequence that will occur because the company wants to punish workers for voting for the union — a consequence desired and freely chosen by a company rather than compelled by economic forces over which it has no control — is.

723 F.2d at 1367 (citation omitted). That approach would appear to preclude protection for management decisions that were based on anti-union animus, even if those management decisions were “definitive” and supported by objective evidence. Such a rigorous restriction on employer speech may not be in the interest of either the employer or the employee. That result, however, is not compelled by Village IX because that case dealt with employer predictions of the economic consequences of unionization, as do almost all cases in this area. Village IX provides the proper analysis for determining whether employer predictions are protected speech. Because the court in that case was not addressing the announcement of management decisions, it is not controlling here.

Finally, Wiljef challenges the relief awarded by the NLRB. The Supreme Court has held that relief ordered by the NLRB “will not be disturbed ‘unless it can be shown that the order is a patent attempt to achieve ends other than those which can fairly be said to effectuate the policies of the Act.’ ” Fibreboard Paper Prods. Corp. v. NLRB, 379 U.S. 203, 216, 85 S.Ct. 398, 405, 13 L.Ed.2d 233 (1964) (quoting *1314 Virginia Elec. & Power Co. v. NLRB, 319 U.S. 533, 540, 63 S.Ct. 1214, 1218, 87 L.Ed. 1568 (1943)). In the present case, the NLRB concluded that the by-law contained an impermissible threat, and thus ordered Wiljef to expunge the by-law, cease further coercive activity and post a notice to employees indicating that it had violated the law and would cease such violations. In light of our conclusion that Wiljef used the by-law in an attempt to coerce its employees and that no objective evidence indicated an intent to implement the by-law, the relief granted by the NLRB is proper.

The petition for review is denied and the order is enforced.