CRT Antitrust Case Overview

The CRT case involves one of the most significant price-fixing conspiracies ever prosecuted in the United States. A global cartel of electronics manufacturers — including major Japanese, Korean, and Chinese companies — coordinated to fix prices for cathode ray tubes used in televisions and computer monitors from March 1995 through November 2007. After years of litigation and more than $547 million in settlements with other defendants, the remaining defendants — Irico Group Corporation and Irico Display Devices Co., Ltd., both Chinese state-owned enterprises — faced a default judgment after being sanctioned for extensive discovery misconduct that the Special Master described as reflecting a “malign strategy” of evidence destruction and delay.

With liability established, the court held a three-day evidentiary hearing in May 2025 to determine damages. Three experts testified: Econ One’s Dr. Phillip Johnson on behalf of direct purchaser plaintiffs, Dr. Janet Netz on behalf of indirect purchaser plaintiffs (IPPs), and Margaret Guerin-Calvert on behalf of Irico.

Static Model or Dynamic?

Dr. Johnson and the IPP expert each built regression models to estimate the “but-for” prices that would have prevailed absent the conspiracy, with damages calculated as the difference between actual and but-for prices applied to the volume of commerce.

One methodological divide was between static and dynamic specifications. The IPP expert’s static model treated each period as independent. Dr. Johnson’s dynamic models incorporated a lagged price variable, reflecting the economic reality that cartel members set prices not for a single month but for a quarter or multiple quarters at a time, and that price targets were often anchored to prior actual or target prices.

Dr. Johnson examined and analyzed voluminous meeting notes and communications showing that cartel members set forward-looking price targets that carried over across periods. His statistical tests confirmed that the influence of target prices on actual prices extended significantly beyond the immediate quarter.

Judge Tigar found Dr. Johnson’s reasoning and analysis compelling and adopted his model stating: “Having reviewed the evidence, including the experts’ testimony, the Court concludes, as Johnson did, ‘that the effect of target prices went beyond the immediate quarter.’ No testimony was presented regarding why a lagged price variable should not be included, or that a static model should be adopted over a dynamic one. The Court therefore finds it appropriate to adopt a dynamic model to estimate the overage effect in this case.”

Single Regression Model or Separate?

The second major methodological question was whether to estimate overcharges using a single regression or separate regressions for the two different product types: color picture tubes (CPTs) and color display tubes (CDTs).

Dr. Johnson’s base model used a single regression, modeling some factors’ effects as common to both CDTs and CPTs and others having distinct effects. This reflected evidence of common economic factors affecting the two product types — both were CRT products, sold by the same cartel members, and subject to many of the same, but not all economic forces. Acknowledging that the choice not to fully disaggregate was a close judgment call in this instance, Dr. Johnson also presented separate regressions as a robustness check, which yielded higher estimates of overcharges. He characterized his single-regression specification as the more conservative approach and stood behind it as a reasonable and well-grounded estimate.

The court concluded that the evidence of differentiation between CPTs and CDTs was sufficient to warrant the separate regressions.

The court therefore adopted Dr. Johnson’s dynamic model, modified to run separate regressions for CPTs and CDTs, as generating the most reasonable estimate of overcharges for both the direct and indirect purchaser classes.  Notably, Dr. Johnson’s own work had already put those modified overcharge estimates on the table.

Rejecting Annual Conduct Variables

The defense expert had proposed replacing Dr. Johnson’s two-period conspiracy specification with annual dummy variables — one for each calendar year. The court rejected this modification on multiple grounds.

First, the court recognized, there was no economic justification for the annual segmentation. Where there was an observed change in cartel conduct that would be expected to change its effectiveness, Dr. Johnson divided the conspiracy period accordingly. Disaggregating by calendar year, by contrast, was, as Dr. Johnson put it, “an arbitrary choice” with “nothing special about it from the conspiracy perspective.”

Second — and more critically — the defense expert conceded that she could not disentangle whether any portion of her annual coefficients reflected conspiracy conduct versus market factors unrelated to the cartel impact. Such a model component confounds overcharges with unrelated market noise and undermines the reliability of the estimates the expert’s model is asked to produce. The court found this to be a decisive flaw. Judge Tigar found that,

Third, that lack of reliability was demonstrated by Guerin-Calvert’s own results. Her model modifications generated eight negative annual overcharge estimates — implying, implausibly, that the cartel suppressed rather than elevated prices in certain years. She zeroed these out to be “conservative,” but the court found that this adjustment itself revealed her skepticism about the reliability of her own model.

Takeaways for Practitioners

Several lessons emerge from Judge Tigar’s careful analysis. A model’s design choices should be grounded in the facts of the specific case to isolate the conduct effect. The dynamic specification survived scrutiny because it was anchored in the actual mechanics of how this cartel operated — not because dynamic models are inherently superior. Similarly, the separate-regression approach prevailed, not as a matter of general preference, but because the evidence showed meaningful economic differences between CPTs and CDTs had a substantial effect on the overcharge estimates. Lastly, the annual conduct variables were rejected because they confounded the conduct effect.

The best model isolates the effects of the alleged conduct from other factors rather than conflating them. The court was rightly not persuaded by a specification that could not be interpreted as an estimate of the overcharge even by the expert who proposed it.

And when a court must choose a single model to apply across plaintiff classes — as Judge Tigar concluded was necessary here — the model that can be coherently defended on economic and factual grounds, and that remains robust to sensitivity testing, will carry the day.

Econ One’s expert economists bring precisely this kind of rigorous, case-specific approach to damages estimation in antitrust matters. Learn more about our antitrust services at https://econone.com/services/antitrust.