Cortina v. North American Title Company
In a 19-year-old misclassification class action, the Fifth District reverses a $43 million judgment for roughly 400 escrow workers on two independent grounds: the trial court’s appointment of a private referee to try the damages phase without the parties’ consent was unauthorized, and the liability-phase trial plan violated Duran’s limits on proving classwide liability by statistical sampling.
Cortina v. North American Title Company (2026) ___ Cal.App.5th ___ [No. F085389]
Partial publication. Part III of the Discussion is not certified for publication and is not citable; this analysis addresses the published portions (the nonconsensual reference and the Duran trial-plan holdings).
Wage & Hour
Class Actions
Misclassification
Reversed
In brief. The wage-and-hour claims of roughly 700 title-company employees in “Exempt” and “Nonexempt” classes were tried in a bifurcated bench trial that, after 19 years of litigation, produced a $43 million judgment for the ~400-member Exempt class. The Fifth District reversed. First, the trial court appointed a referee to conduct the entire second phase of trial “without the parties’ consent and over defendant’s strenuous objections” — a “nonconsensual reference of [a] scope and magnitude” that “appears unprecedented,” was “entirely unauthorized,” and “alone compels reversal.” Second, the liability-phase trial plan “contravene[d] the holdings of Duran” by using representative sampling to establish classwide liability and rejecting the employer’s exemption defenses wholesale based on misinterpretations of law. (Cortina v. North American Title Company (2026) No. F085389 (slip opn. at pp. 1–4).)
By Jonathan J. Delshad
Draft for review · source-verification pending
Facts and procedural posture
Filed 19 years ago, the case has generated “numerous interlocutory appeals and writ proceedings,” the most recent reaching the Supreme Court. (slip opn. at pp. 1–2, citing North American Title Co. v. Superior Court (2024) 17 Cal.5th 155.) The plaintiffs — roughly 700 employees in two classes (“Exempt” and “Nonexempt”) — alleged the employer misclassified them as exempt from overtime, meal-period, and rest-period requirements. (slip opn. at pp. 2–3.) The claims were tried in a bifurcated bench trial. (Id. at p. 2.) After phase one, the trial court decertified the Nonexempt class but ruled against the employer on certain defenses as to the ~400-member Exempt class; it then appointed a referee to conduct phase two, “without the parties’ consent and over defendant’s strenuous objections.” (Id. at p. 2.) The reference “lasted years” and “included live testimony from over 230 class members,” culminating in a $43 million judgment. (Ibid.)
The exemption defenses turned on how employees spent their time: under California’s “uniquely quantitative approach,” an employee is exempt only if more than half their time is spent on qualifying exempt work. (slip opn. at pp. 2–3, quoting Duran v. U.S. Bank National Assn. (2014) 59 Cal.4th 1, 27.) Plaintiffs alleged the entire Exempt class — despite varied job titles and offices over a 10-year period — “all performed the same allegedly nonexempt job the same way, all the time”; yet some cohorts, such as “Branch Managers,” “indisputably did perform exempt work on a regular basis.” (slip opn. at p. 3.) To bridge the gap, plaintiffs elicited testimony from about 15 percent of the Branch Manager cohort (roughly 24 of 156) and argued it was “representative of the entire group.” (Ibid.)
Issue
May a trial court delegate the entire damages phase of a class action to a private referee without the parties’ consent? And does a liability-phase trial plan that establishes classwide misclassification liability through representative sampling — while precluding the employer from contesting individual class members’ exempt status — comply with Duran v. U.S. Bank? (slip opn. at pp. 1–3.)
Holding
Reversed. (1) “A trial court’s authority to delegate matters to a referee without the parties’ consent is strictly circumscribed by the California Constitution and Code of Civil Procedure,” and the sweeping, nonconsensual reference here — years of proceedings and 230-plus witnesses culminating in a $43 million judgment — was “entirely unauthorized” and “appears unprecedented in our state jurisprudence.” “This error alone compels reversal.” (slip opn. at p. 2.) (2) The liability-phase trial plan independently required reversal: it “contravene[d] the holdings of Duran” by using statistical/representative sampling to “establish universal liability, regardless of the margin of error or variation among witness testimony,” while “prohibit[ing]” the employer “from further contesting the exemption status of individual class members,” and it rejected the employer’s exemption defenses “out of hand, on a classwide basis, due to misinterpretations of the applicable law.” (slip opn. at pp. 2–3 & slip opn. discussion, citing Duran, supra, 59 Cal.4th at p. 49.) On remand, the cause is returned for retrial of the named plaintiffs’ individual claims, with the trial court free to entertain a new certification motion. (slip opn. at p. 4.)
Reasoning
1. A nonconsensual general reference is constitutionally constrained. The court treated the reference as the more fundamental error. A trial court may appoint a referee to try issues only within strict limits; a reference of the entire damages phase — without consent and over objection — exceeded that authority. (slip opn. at p. 2.) The magnitude of what occurred (a multi-year reference, 230-plus live witnesses, a $43 million judgment) underscored the point: such a “nonconsensual reference of [this] scope and magnitude … appears unprecedented,” and because the proceedings “were entirely unauthorized,” the judgment built on them could not stand. (Ibid.)
2. Duran limits proof of classwide liability by sampling. Turning to the liability phase, the court invoked Duran v. U.S. Bank, which famously called a wage-and-hour class action tried to verdict an “exceedingly rare beast,” and which holds that “[i]f sampling is used to estimate the extent of a party’s liability, care must be taken to ensure that the methodology produces reliable results.” (slip opn. at pp. 2–3, quoting Duran, supra, 59 Cal.4th at pp. 12, 27.) Eliciting testimony from 15 percent of a cohort and treating it as representative of the whole — to prove that no class member crossed the 51-percent exempt-work threshold — was the kind of scientifically unreliable extrapolation Duran rejects. (slip opn. at p. 3.)
3. The defendant must be allowed to contest the model. Central to the error was the preclusion of individualized defenses. Under Duran, “[i]f the trial proceeds with a statistical model of proof, a defendant accused of misclassification must be given a chance to impeach that model or otherwise show that its liability is reduced because some plaintiffs were properly classified as exempt.” (Duran, supra, 59 Cal.4th at p. 49, italics added; slip opn., Discussion.) The trial here did the opposite — using sampling “to establish universal liability, regardless of the margin of error or variation among witness testimony,” while barring the employer from contesting individual exempt status. (slip opn., Discussion, fn. 9.)
4. Margin of error and due process. The court drew on Duran’s and Bell v. Farmers Ins. Exchange’s treatment of statistical error: a 32-percent margin of error “was so large that the resulting damages award violated due process,” and even the 13-to-14-percent margin in Duran drew sharp scrutiny. (slip opn., Discussion, quoting Duran, supra, 59 Cal.4th at pp. 40, 46, and citing Bell v. Farmers Ins. Exchange (2004) 115 Cal.App.4th 715, 756–757.) Because “greater certainty is required for liability findings” than for damages, a statistical model purporting to establish classwide liability “would have to reliably show, at a minimum, that a significant majority of the class members were misclassified.” (slip opn., Discussion.) The employer’s affirmative defenses, the court held, were also wrongly rejected on a classwide basis through misinterpretations of the exemption law. (slip opn. at p. 3.)
Significance
Cortina is a cautionary capstone to one of California’s longest-running wage-and-hour class actions and a strong restatement of two structural limits on how such cases may be tried. First, on the rarely litigated question of judicial references, it holds that a trial court cannot offload the damages phase of a complex class action onto a private referee without the parties’ consent — a power “strictly circumscribed” by the Constitution and Code of Civil Procedure — and that doing so voids the resulting judgment. Second, it reaffirms Duran’s core teaching: classwide misclassification liability cannot be established by representative sampling that ignores variation and denies the employer the chance to contest individual class members’ exempt status, and statistical models carry due-process limits tied to their margin of error. For plaintiffs, the decision is a warning that an ambitious, sampling-driven trial plan can unravel an enormous verdict; for both sides, it underscores that misclassification classes remain difficult to try to verdict and frequently founder on manageability and proof problems.
Key quotes
“A nonconsensual reference of the scope and magnitude ordered in this case is not merely a rare occurrence. It appears unprecedented in our state jurisprudence. … The reference proceedings below were entirely unauthorized. This error alone compels reversal of the judgment.” (Cortina, supra, No. F085389 (slip opn. at p. 2).)
“If the trial proceeds with a statistical model of proof, a defendant accused of misclassification must be given a chance to impeach that model or otherwise show that its liability is reduced because some plaintiffs were properly classified as exempt.” (Duran v. U.S. Bank National Assn. (2014) 59 Cal.4th 1, 49, quoted in Cortina, supra, No. F085389.)
Practice pointer
Two lessons for trying — or defending — a misclassification class. On case management, do not consent to a general reference of the merits or damages phase unless you mean it: under Cortina, a nonconsensual reference of that scope is unauthorized and will void the judgment, so object on the record and preserve the constitutional and statutory limits. On proof, plaintiffs must build a trial plan that Duran can sustain — a reliable statistical methodology with a defensible margin of error, and a structure that lets the defendant impeach the model and contest individual exempt status; treating a 15-percent sample as conclusively “representative” of a varied class invites reversal. Defense counsel should press the variation among class members and cohorts (e.g., Branch Managers who plainly performed exempt work), demand the right to individualized defenses, and frame the margin-of-error and due-process limits from Duran and Bell. Both sides should recognize that, as Duran and now Cortina show, these classes are exceptionally hard to take to verdict. (slip opn. at pp. 1–4.)
Open questions
The court reversed and remanded for retrial of the named plaintiffs’ individual claims, expressly leaving the trial court “free to entertain a new certification motion on remand” as to the Exempt class — so whether a properly structured class can be certified and tried remains open. (slip opn. at p. 4.) The opinion also declines to resolve whether an “all-or-nothing” classwide liability determination is ever proper in a misclassification case, and (echoing Duran) what margin of error a liability model may tolerate, noting only that liability demands greater certainty than damages and that a “significant majority” showing would be the minimum. (slip opn., Discussion.) Finally, because part III of the Discussion is unpublished, the court’s analysis of the additional issues addressed there is not citable. (slip opn. at p. 1, fn. *.)