Summary
discussing Entergy II , 810 F.3d at 292 (quoting 29 U.S.C. § 160(e) – (f) )
Summary of this case from STP Nuclear Operating Co. v. Nat'l Labor Relations Bd.Opinion
No. 18-60522
03-20-2020
DISH NETWORK CORPORATION, Petitioner/Cross-Respondent, v. NATIONAL LABOR RELATIONS BOARD, Respondent/Cross-Petitioner.
Eric Shumsky, Benjamin Aiken, Orrick, Randall Smith, Herrington & Sutcliffe, L.L.P., Washington, DC, for Petitioner Cross-Respondent. David S. Habenstreit, David Casserly, Kira Dellinger Vol, National Labor Relations Board, Appellate & Supreme Court Litigation Branch, Washington, DC, Timothy L. Watson, National Labor Relations Board, Region 16, Fort Worth, TX, for Respondent Cross-Petitioner. Matthew G. Holder, David Van Os & Associates, P.C., San Antonio, TX, for Intervenor.
Eric Shumsky, Benjamin Aiken, Orrick, Randall Smith, Herrington & Sutcliffe, L.L.P., Washington, DC, for Petitioner Cross-Respondent.
David S. Habenstreit, David Casserly, Kira Dellinger Vol, National Labor Relations Board, Appellate & Supreme Court Litigation Branch, Washington, DC, Timothy L. Watson, National Labor Relations Board, Region 16, Fort Worth, TX, for Respondent Cross-Petitioner.
Matthew G. Holder, David Van Os & Associates, P.C., San Antonio, TX, for Intervenor.
Before SOUTHWICK, WILLETT, and OLDHAM, Circuit Judges.
ANDREW S. OLDHAM, Circuit Judge:
DISH negotiated with a labor union for more than four years. Eventually, DISH got tired and determined the parties were at an impasse. The question presented is whether the National Labor Relations Board had substantial evidence to gainsay DISH's impasse determination and penalize the employer for refusing still more years of negotiation. It did not. I.
It's easy to see how the parties tired of this years-long negotiation. We do our best to summarize the events that turned a relatively simple dispute into a contentious multi-round National Labor Relations Board ("NLRB" or "Board") proceeding and cross-petitions for review in federal court.
A.
DISH Network Corporation ("DISH") provides its customers with satellite TV. Local branches serve as hubs for technicians who conduct installations and repairs. Before 2009, DISH generally paid its employees by the hour with minor incentives.
In 2009, DISH experimented with a new compensation scheme for employees at two of its Dallas-Fort Worth locations: Farmers Branch and North Richland Hills. Under the Quality Performance Compensation ("QPC") payment scheme, employees were paid a lower hourly wage with greater performance-based incentives. DISH's employees hated QPC. They hated it so much that they unionized to oppose it at both of the affected branches. The Communication Workers of America, AFL-CIO ("the Union"), became the designated collective bargaining representative of the affected employees.
Bargaining over an initial contract began in July of 2010. For a time, it appeared the parties might agree to eliminate QPC and replace it with a different incentive-based payment plan (known as "Pi"). According to one of the negotiators, as of March 22, 2013, "We were at Pi—they were at Pi."
Things went haywire in July 2013. That's when the Union changed its position on QPC. Instead of opposing the QPC compensation scheme—which was the entire basis for the Union's existence—the union now demanded to keep it. Why, you might wonder? In a word: Technology. As the district court explained in a related case:
QPC paid employees based on performance metrics that DISH was unable to adjust due to collective bargaining. As DISH improved its processes and technologies, DISH technicians were able to complete tasks more quickly and thus earn more under the QPC incentive program. As a result, the wages of unit technicians increased by about $17,000 between 2013 and 2015. In 2015, Union members were making, on average, $19,000 more than their non-unionized peers.
Kinard ex rel. NLRB v. DISH Network Co. , 228 F. Supp. 3d 771, 775 (N.D. Tex. 2017) (quotation omitted). As the district court noted, the gap between Union employees and others grew over time. Union technicians working under QPC made 14 percent more than their non-union peers in 2013, then 41 percent more in 2014, and 43 percent more in 2015. Eventually, the technicians who were making more under QPC were working less than their non-union peers. By 2015, the gap was about 200 hours on average.
The parties' first appearances in federal court concerned injunctive relief granted during the NLRB's adjudication of unfair labor charges against DISH. See Kinard ex rel. NRLB v. DISH Network Corp. , 890 F.3d 608, 610 (5th Cir. 2018).
In addition to creating these horizontal disparities, QPC also produced vertical anomalies. One high-earning technician with an interest in management acknowledged that moving up the company ladder would have decreased his pay. So, as one might imagine, the Union fought hard to keep QPC. DISH tried to get rid of it. After the Union's change of heart, the parties made progress on some other issues. But on QPC, they made little headway.
B.
By November 2014, the parties had conducted approximately 25 face-to-face bargaining sessions over the course of more than four years. And they were no closer to an agreement on QPC than when they started. So, at a meeting on November 18, 2014, DISH made a "final proposal" that included the elimination of QPC. The final proposal said, "[DISH] rejects continuation of QPC." Not surprisingly, that session also adjourned without an agreement. More bargaining sessions had been scheduled for early December, but the Union had to cancel those meetings. The Union asked to reschedule, and DISH's lead negotiator responded by saying that DISH had "provided a final offer for [the Union] to accept or reject." DISH further emphasized that, if the Union couldn't meet as scheduled or offer a counterproposal, DISH would "consider the bargaining to be at an impasse."
The Union responded with a counterproposal. Under the terms of the counterproposal, all technicians employed by DISH at the time would retain QPC for the length of the contract. So everyone who had QPC would keep QPC going forward. Any new hires, however, would be paid according to a different plan.
DISH rejected the Union's counterproposal and insisted that its final offer was indeed its "last, best[,] and final offer." After finding out that the Union's lead negotiator was unavailable for the rest of December, DISH's lead negotiator asked the Union to take DISH's final offer to its members for a vote. He stated, "Once we know whether DISH's final offer is accepted or rejected, we can discuss if further bargaining is warranted." The Union refused to take the final offer to its members and asked for another bargaining session. The next day, December 31, 2014, DISH's lead negotiator replied and explained that his partner would take over for the rest of the negotiations. He also said that DISH's new representative would "be getting back to [the Union] sometime after the new year."
C.
A year's silence ensued—there was no contact between the parties until after the next new year. In January 2016, DISH's new lead negotiator sent a letter to ask whether the Union would accept DISH's final offer. The letter also stated that since that final offer was indeed "[DISH's] final offer, it does not appear at this point that further bargaining would be productive."
In response, the Union again asked for yet another bargaining session. DISH's replacement negotiator said he understood the Union's response as a rejection of DISH's final offer. He also observed that the parties had been bargaining since 2011, and that in his view, DISH and the Union were "at a standstill." He asked the Union to explain whether it disagreed; otherwise, DISH would "implement its last, best[,] and final proposal." In response, the Union demanded another meeting. In early April 2016, DISH said that "further bargaining would be futile," and DISH would implement its final offer—and nix QPC—no later than April 23, 2016.
On April 23, 2016, DISH did exactly that. With QPC gone, unit technicians saw their wages drop significantly. Seventeen employees quit.
D.
The Union brought a complaint to the NLRB and alleged that DISH committed unfair labor practices. An administrative law judge ("ALJ") determined DISH unlawfully declared an impasse. In the ALJ's view, the Union's November 2014 counterproposal—that existing employees could keep QPC but new hires could not—constituted a "white flag." The ALJ asserted that DISH's technicians had a "very high [annual] attrition rate" that ranged from 116 percent to 13 percent. The ALJ repeatedly cited the 116 percent figure to support his premise that the relevant attrition rate was "very high." Given such high attrition rates, the ALJ surmised that under the Union's counterproposal, DISH "would have attained most of what it wanted on wages in the short term," and "eventually abolishing QPC would have become an easier selling point." That is, if a high percentage of employees left and were replaced by new employees who didn't receive QPC, then QPC would quickly become less expensive for DISH and less important to the Union. According to the ALJ, the Union's counterproposal to phase out QPC prevented impasse. And because there was no impasse, the ALJ said, DISH violated Section 8(a)(5) of the National Labor Relations Act ("the Act" or "NLRA") by eliminating QPC. The ALJ also found that DISH violated Section 8(a)(3) of the Act when it constructively discharged 17 employees. On his theory, DISH did so by presenting the employees with a Hobson's Choice of quitting or "continuing to work under greatly diminished conditions that flowed from the violation of [the employees'] rights."
The NLRB affirmed all of the ALJ's findings and conclusions. But the Board noted that the ALJ had not explicitly applied the proper test for impasse analysis. So the Board applied that test and reached the same conclusion—there was no impasse in negotiations. In its impasse analysis, the Board acknowledged that by December 2014, the parties had bargained hard for four years. Moreover, the Board recognized that QPC "remained the most important issue of disagreement." But according to the Board, "[e]ven if the parties may have been near a valid impasse then," the Union's white flag changed everything. As to constructive discharge, the Board's decision simply included a footnote adopting the ALJ's rationale and finding.
One Board member dissented. He would have found that an impasse existed at least by April 23, 2016, and that as a result of that impasse, DISH "lawfully implemented the terms of its final offer." Because that conduct was lawful, he would have found "no support" for constructive discharge under the theory proffered by the ALJ and accepted by the majority of the Board.
DISH petitioned for review of the Board's conclusions on unlawful implementation and constructive discharge. The Board cross-petitioned for enforcement of its order. II.
The NLRB may be the only agency that needs a court's imprimatur to render its orders enforceable. NLRB v. Thill, Inc. , 980 F.2d 1137, 1142 (7th Cir. 1992) ("Unlike the orders of other agencies, the Board's orders are not self-executing."); see also Mitchellace, Inc. v. NLRB , 90 F.3d 1150, 1159 (6th Cir. 1996) (similar); E.I. Du Pont De Nemours & Co. v. NLRB , 682 F.3d 65, 71 (D.C. Cir. 2012) (Randolph, J., concurring in part and concurring in the judgment) (same). Congress has long limited the Board's powers in this way, though the Board wasn't always alone in being so limited. As the Seventh Circuit explained, "the curious impotence of unenforced orders of the Board is the result of a decision by the Congress that enacted the Wagner Act to give the Board it was creating the same procedures as the Federal Trade Commission then had." NLRB v. P*I*E Nationwide, Inc. , 894 F.2d 887, 892 (7th Cir. 1990). And for "both agencies, the denial of teeth to the agency's orders was a swap for procedural informality." Ibid. The Administrative Procedure Act brought more formality to administrative proceedings; thereafter, Congress gave the FTC enforcement power. Ibid. But Congress never did the same for the NLRB. Ibid. Thus, what was true in 1938 remains true today: "The Board is given no power of enforcement. Compliance is not obligatory until the court, on petition of the Board or any party aggrieved, shall have entered a decree enforcing the order as made, or as modified by the court." In re NLRB , 304 U.S. 486, 495, 58 S.Ct. 1001, 82 L.Ed. 1482 (1938) ; see also 29 U.S.C. § 160(e) –(f). And now, as then, when a court "enforces" an order, "[t]he order issued by the court is an injunction, enforceable by contempt." P*I*E Nationwide, Inc. , 894 F.2d at 893.
This case turns on the Board's finding that DISH prematurely declared an impasse. We hold the Board lacked substantial evidence to so find.
Ultimately, "[w]hether an impasse exists depends on whether, in view of all the circumstances of the bargaining, further discussions would be futile." Gulf States Mfg. Inc. v. NLRB , 704 F.2d 1390, 1398 (5th Cir. 1983). That's a fact-specific inquiry. Carey Salt Co. v. NLRB , 736 F.3d 405, 411 (5th Cir. 2013).
We review the Board's findings of fact for substantial evidence. Id. at 410. Congress has defined substantial evidence in terms of "the record considered as a whole ." 29 U.S.C. § 160(e) – (f) (emphasis added). It follows, then, that "a flawed reading of the record" provides no substantial evidence for a finding. Carey Salt , 736 F.3d at 420. And because the evidence must be viewed in light of the whole record, "[t]he substantiality of evidence must take into account whatever in the record fairly detracts from its weight." Universal Camera Corp. v. NLRB , 340 U.S. 474, 488, 71 S.Ct. 456, 95 L.Ed. 456 (1951). Finally, the evidence "must be substantial , not speculative, nor derived from inferences upon inferences." Brown & Root, Inc. v. NLRB , 333 F.3d 628, 639 (5th Cir. 2003). The Board's no-impasse finding flunks these standards.
A.
First, it's a long- and well-settled proposition of substantial-evidence review that the NLRB cannot build its decision on a foundational error of fact. For example, in Carey Salt , we addressed an ALJ's misreading of the record and stated that the "ALJ's flawed factual findings regarding the company's ‘movement’ do not provide substantial evidence in support of the Board's no-impasse finding." 736 F.3d at 421. We went on to note that the Board's analysis, built on the ALJ's flawed factual foundation, "was erroneous, and, therefore, no substantial evidence therein can support the Board's no-impasse finding." Id. at 421 n.16.
So here. The Board's decision rested on its determination that the Union's November 2014 counterproposal was a "white flag" of surrender. But the "white flag" characterization in turn rested on an unsound factual foundation from the ALJ.
The ALJ found an attrition rate of 116 percent is "very high"—but that rate occurred at a nonunionized , non-QPC branch. If the ALJ had considered the relevant data, he would've found that unionized QPC technicians generally had lower attrition rates than their non-union, non-QPC peers. And the relevant data thus supported DISH: QPC workers had a marginal incentive not to leave the company because they could make more money while working less. The relevant data likewise undermined the "white flag" characterization of the counterproposal: The comparatively low attrition at unionized QPC branches meant that the Union wasn't giving up all that much by agreeing to phase out QPC for new employees. The Board then built its no-impasse decision on the ALJ's factual error. In the Board's view, "very high" attrition rates made the Union's concession of no QPC for new employees a "white flag" that warranted more negotiations:
Particularly significant to our determination that impasse had not been reached was the Union's substantial movement on the QPC issue and [DISH's] refusal to meet and confer even once over this meaningful counterproposal by the Union on the most important issue separating the parties.
(Emphasis added). That characterization of the "meaningful counterproposal" as "substantial movement" flows straight from the ALJ's misreading of the record. The ALJ erred, and the Board doubled down. Two wrongs can't make the Board right.
B.
Relatedly, "[t]he substantiality of evidence must take into account whatever in the record fairly detracts from its weight." Universal Camera , 340 U.S. at 488, 71 S.Ct. 456. To survive substantial evidence review, then, the Board has to consider "contradictory evidence or evidence from which conflicting inferences could be drawn." Id. at 487, 71 S.Ct. 456. Take Entergy Miss., Inc. v. NLRB , 810 F.3d 287 (5th Cir. 2015), for example. In that case, the NLRB concluded that certain employees did not exercise independent judgment. Id. at 296–97. Even if that conclusion could be supported by the portions of the record cited by the Board, we held the Board lacked substantial evidence merely because it failed to grapple with countervailing portions of the record. Id. at 297–98. After all, we must review the Board's decisions "on the record considered as a whole ." 29 U.S.C. § 160(e) – (f) (emphasis added).
Given that the Board's white-flag theory depended on unit technicians maintaining "a very high attrition rate," the Board had an independent obligation to consider evidence that those attrition rates were relatively low and falling over time. That's especially true because unionized QPC employees were making more money while working fewer hours than their non-union, non-QPC peers.
It's no answer to say that the ALJ at one point cited the correct attrition rates for the unit branches for 2014 and 2015 and also described those rates as "high." That's because the ALJ's white-flag theory depended on high attrition rates making it "probable that new hires receiving non-QPC rates would soon become the majority .... [and] eventually abolishing QPC would have become an easier selling point." But the actual attrition rates at the unit branches simply don't suggest the same imminent change or easy elimination of QPC.
DISH presented evidence to explain why those factors contributed to declining attrition rates at union branches. DISH's Regional Director testified that unit technicians had relatively lower attrition rates because "they were making really good money and [so] they were staying because of that." And employees knew QPC was a good deal. One unit technician reckoned an employee would thrive under Pi only if he could "sell[ ] ketchup popsicles to a lady with white gloves." Suffice it to say that QPC gave unit technicians plenty of reason to stick around, and the evidence shows they did so.
And DISH gave the Board plenty of notice that the ALJ used the wrong attrition data. DISH raised the problem in its briefing before the Board in bolded, underlined, and italicized font. But DISH's printed shout fell on deaf ears. The Board's decision simply didn't address the mistake the ALJ made or any of DISH's arguments on that point. That oversight compounds the original error. See Lord & Taylor v. NLRB , 703 F.2d 163, 169 (5th Cir. 1983).
C.
Even if we set aside the Board's reliance on the wrong attrition rates, its decision still would not rest on substantial evidence. That's because "[s]uspicion, conjecture, and theoretical speculation register no weight on the substantial evidence scale." NLRB v. Mini-Togs, Inc. , 980 F.2d 1027, 1032 (5th Cir. 1993). For example, in Brown & Root , an employer refused to hire 48 employees after a series of contentious meetings over their plans to unionize. In finding the employer violated the NLRA, the Board pointed to specific statements and specific meetings to show the employer acted with "anti-union animus." Brown & Root , 333 F.3d at 641. We nonetheless granted the employer's petition for review because the Board's string of inferences and speculation did not sum to substantial evidence. See ibid. & n.10.
This case is easier. The ALJ's speculation involved an inferential chain that began with a flawed premise. First, in light of DISH's "very high" attrition rates, the ALJ asserted that DISH would have saved a lot money by accepting the Union's counterproposal. Next, the ALJ posited that those same "very high" attrition rates "meant that in a short time, the majority of the [union workers] would have likely ... turned over and no longer earn[ed] QPC wages." And that, in turn, "would have likely set in motion the wholesale elimination of QPC in future bargaining for a successor contract." The ALJ's musings on potential cost savings and shifting intra-Union dynamics are ill-founded speculation about what could have happened. It warrants "no weight on the substantial evidence scale." Mini-Togs , 980 F.2d at 1032.
The same is true of the supposition that if DISH "had been willing to meet about this substantial giveback, the give and take of bargaining might have led everyone closer to an agreement." (Emphasis added). The bare possibility that something might have clicked during later negotiations does not offer any support for the Board's finding. Cf. TruServ Corp. v. NLRB , 254 F.3d 1105, 1116 (D.C. Cir. 2001) (rejecting Board's reliance on "its intuitive belief that, upon further bargaining, each side would have made additional concessions"). As Judge Sentelle once put it, " ‘You never know’ is no substitute for substantial evidence." Erie Brush & Mfg. Corp. v. NLRB , 700 F.3d 17, 23 (D.C. Cir. 2012).
III.
The NLRB offers several counterarguments. Some are properly before us. Others are not. We reject them all.
A.
We start with the Board's procedurally proper arguments. The Board contends that DISH acted in "bad faith" by refusing to negotiate after the Union budged on QPC. It's true that an employer cannot act in bad faith and then pretend the parties are at an impasse. See 29 U.S.C. § 158(d). We so held in Carey Salt , for example. There, we found that the employer's bad faith provided an independent basis for the Board's conclusion that no impasse had taken place. See Carey Salt , 736 F.3d at 413.
But to preclude impasse, bad faith must precede impasse. That's because once a good-faith impasse occurs, it suspends "the duty to bargain in good faith." Id. at 425 ; accord Serramonte Oldsmobile, Inc. v. NLRB , 86 F.3d 227, 232 (D.C. Cir. 1996). Here, there is no dispute that DISH bargained in good faith right up to the point that the Union rejected DISH's "final offer" and made a counterproposal. Of course, "merely labeling an offer as ‘final’ is not dispositive," TruServ Corp. , 254 F.3d at 1115, since negotiators sometimes engage in gamesmanship by deploying multiple "final offers." See Chi. Typographical Union No. 16 v. Chi. Sun-Times, Inc. , 935 F.2d 1501, 1508 (7th Cir. 1991). But nothing in the record before us suggests that DISH's final offer was anything other than DISH's "last, best, and final offer" made after years of "the kind of good-faith, hard bargaining that characterizes impasse." TruServ Corp. , 254 F.3d at 1116. And the rejection of this kind of final offer is a telltale sign of impasse. Id. at 1115–16 ; cf. Carey Salt , 736 F.3d at 416. All of the Board's allegations of bad faith come after DISH's final offer was rejected. So the question is whether DISH had some additional obligation to continue negotiating in good faith after that point. We have never interpreted the NLRA to impose such an obligation, and we refuse to do it today.
For much the same reason, we reject the Board's equivocal statement that "[t]o the extent that [DISH] also conditioned further bargaining on the Union's submitting [DISH's] offer to unit employees for a ratification vote ... [DISH] also evinced a lack of good faith." Again, such conduct must precede impasse to preclude impasse.
The NLRB next contends that the Union's willingness to move a little on QPC precludes impasse, since impasse requires both parties to be unwilling to compromise. See Huck Mfg. Co. v. NLRB , 693 F.2d 1176, 1186 (5th Cir. 1982) ("[F]or a deadlock to occur, neither party must be willing to compromise."). Yet not every concession precludes impasse. See Carey Salt , 736 F.3d at 422 (noting that it's not true that "any kind of extended concession, despite rejection and remote chances of fueling future talks, precludes impasse" (emphasis added)); Grinnell Fire Prot. Sys. Co. v. NLRB , 236 F.3d 187, 200 (4th Cir. 2000) ("The Board has repeatedly held that inconsequential modifications that fail to address the heart of the employer's demands cannot forestall impasse."); E. I. Du Pont & Co. , 268 NLRB 1075, 1076 (1984). The Board's contrary position would allow one party to prolong negotiations by responding to every "final offer" by conceding just a wee bit more. That would be particularly untenable in a case like this one because every prolongation would give the Union everything it wanted—namely, preservation of QPC—for the length of the prolongation.
Lastly, the Board highlighted the Union's repeated demands for another bargaining session as a reason for finding no impasse. But "a vague request by one party for additional meetings, if unaccompanied by an indication of the areas in which that party foresees future concessions, is equally insufficient to defeat an impasse where the other party has clearly announced that its position is final." TruServ Corp. , 254 F.3d at 1117. In this case, DISH had long been clear that its final offer was indeed final. The Union's repeated requests for additional meetings did not suggest any change in its position on QPC. Under the Act, parties are not required "to engage in fruitless marathon discussions." NLRB v. Am. Nat'l Ins. Co. , 343 U.S. 395, 404, 72 S.Ct. 824, 96 L.Ed. 1027 (1952). DISH was not obligated to do so here.
B.
The rest of the NLRB's arguments are barred by SEC v. Chenery Corp. , 332 U.S. 194, 67 S.Ct. 1760, 91 L.Ed. 1995 (1947). It's "a simple but fundamental rule ... that a reviewing court, in dealing with a determination or judgment which an administrative agency alone is authorized to make, must judge the propriety of such action solely by the grounds invoked by the agency." Id . at 196, 67 S.Ct. 1760. That means we look to what the agency said, not what it might have said. And it means we "may not accept appellate counsel's post hoc rationalizations for agency action; Chenery requires that an agency's discretionary order be upheld, if at all, on the same basis articulated in the order by the agency itself." Burlington Truck Lines, Inc. v. United States , 371 U.S. 156, 168–69, 83 S.Ct. 239, 9 L.Ed.2d 207 (1962).
We recognize that the NLRB, in particular, struggles with this rule. In this case, the NLRB's brief used figures conspicuously absent from the actual decision of the Board to support the Board's view of the Union's counterproposal. "At an attrition rate of 30.5% ... only 33.6% of unit employees would still be under QPC at the expiration of the 3-year contract." That's simply not what the Board or the ALJ said. So we can't affirm on those grounds. That's one Chenery problem.
According to one professor who spent four years working in the NLRB's Appellate Court Branch, the frequency of the practice manifested itself in shorthand shoptalk: The "Board's appellate attorneys typically refer to a rationale not explicitly contained in a Board decision as a ‘post hoc.’ " Jeffrey M. Hirsch, Defending the NLRB: Improving the Agency's Success in the Federal Courts of Appeals , 5 FIU L. Rev . 437, 448 n.45 (2010). Prof. Hirsch also suggested several ways for the Board to improve its decisions so it would stop "forcing the Board's appellate attorneys to create justifications, which—as courts are often quick to note—they are not supposed to do under Chenery ." Id. at 448.
Appellate counsel next defends the Board's no-impasse finding by pointing to DISH's contemporaneous communications with the Union, suggesting those show neither party thought impasse existed. But the Board's analysis of the parties' contemporaneous understanding focused entirely on the Union's perspective. It's too late in the game for the NLRB to add a gloss to that part of its decision.
Likewise, Chenery precludes us from affirming on the basis of two grounds suggested by the Union—the passage of time and presence of a new negotiator for DISH. We've said those factors can be relevant to impasse analysis. See Gulf States Mfg. , 704 F.2d at 1399 (noting "the mere passage of time may also be relevant" to a finding on impasse); Raven Servs. Corp. v. NLRB , 315 F.3d 499, 505 (5th Cir. 2002) (noting a change in key personnel may impact impasse analysis). Here, however, the passage of time weighed in favor of impasse. First, the Board noted that as of December 2014, "the parties had bargained in numerous sessions for more than 4 years over a first collective-bargaining agreement," and that suggested "the parties may have been near a valid impasse ...." And at no point did the Board's analysis suggest the presence of a new negotiator weighed against finding impasse. Maybe, if the Board's decision had relied on the arguments now marshalled in the briefs, "we would have a far different case to decide. But as it is, we cannot accept appellate counsel's post hoc rationalizations for agency action; for an agency's order must be upheld, if at all, on the same basis articulated in the order by the agency itself." Fed. Power Comm'n v. Texaco Inc. , 417 U.S. 380, 397, 94 S.Ct. 2315, 41 L.Ed.2d 141 (1974) (quotation omitted).
IV.
Having addressed the factual basis for the Board's decision, we dispose of its legal conclusions in short order. We begin with unlawful implementation and turn next to constructive discharge.
A.
The Board says DISH violated § 8(a)(5) of the Act by unlawfully implementing its final offer in the absence of a good-faith impasse. But as we said in Carey Salt , "when an employer and union bargain to impasse, the employer may unilaterally implement changes in contract terms, so long as the changes were previously offered during negotiations." Carey Salt , 736 F.3d at 411. The Board's finding that DISH acted unlawfully in implementing its final offer depended on the Board's finding that there was no impasse in negotiations. Both fall together.
Another provision of the Act, § 8(a)(1), prohibits employers from interfering with employees' exercise of their rights under the Act. 29 U.S.C. § 158(a)(1). Such interference constitutes an unfair labor practice. Ibid. As a result, "an employer who violates Section 8(a)(5) also commits a ‘derivative’ violation of Section 8(a)(1)." Sara Lee Bakery Grp., Inc. v. NLRB , 514 F.3d 422, 427 n.3 (5th Cir. 2008). That's why the Board found that DISH "violated Section[s] 8(a)(5) and (1) by unilaterally changing unit employees' terms and conditions of employment." (Emphasis added).
B.
Next, constructive discharge. On that issue, the Board purported to "adopt, for the reasons stated by the judge, his finding that [DISH] violated Sec. 8(a)(3) and (1) by constructively discharging 17 employees ...." But as the Board itself said, that conclusion relied on DISH's "unlawful unilateral reductions in [employees'] wages and health benefits." So the Board's constructive discharge finding rested on its unlawful implementation finding, which in turn rested on its no-impasse finding. Again, they all fall together.
The ALJ framed his constructive discharge finding solely in terms of § 8(a)(3). But the Supreme Court has explained that "[a]lthough §§ 8(a)(1) and (a)(3) are not coterminous, a violation of § 8(a)(3) constitutes a derivative violation of § 8(a)(1)." Metro. Edison Co. v. NLRB , 460 U.S. 693, 698 n.4, 103 S.Ct. 1467, 75 L.Ed.2d 387 (1983). Because we reject the constructive discharge finding in its entirety, this variation between the ALJ's findings and the Board's decision is immaterial.
Because the premise of the constructive discharge finding fails, we need not consider whether the ALJ's theory of a Hobson's Choice constructive discharge conflicts with our decision in Elec. Mach. Co. v. NLRB , 653 F.2d 958 (5th Cir. 1981).
* * *
DISH's petition for review is GRANTED, and the corresponding portions of the Board's cross-petition for enforcement are DENIED. Those portions of the Board's order that DISH did not challenge are ENFORCED. See Sara Lee Bakery , 514 F.3d at 429.