pricing-data-analysis-memo.docx

1039-00001 · Torchlight Mining Corp · documents

pricing-data-analysis-memo.docx DOCUMENT Litigation

65 / 88 ← → ✕

PRIVILEGED & CONFIDENTIAL ATTORNEY-CLIENT PRIVILEGED / WORK PRODUCT DOCTRINE NOT FOR DISTRIBUTION WITHOUT PRIOR WRITTEN AUTHORIZATION OF CALDERWOOD & HARKNESS LLP

PRICING DATA ANALYSIS MEMORANDUM

PREPARED AT THE DIRECTION OF COUNSEL — CALDERWOOD & HARKNESS LLP

November 1, 2022

TO: Elaine Forsythe, Partner, Calderwood & Harkness LLP

FROM: Dr. Vanessa Tremblay, Engagement Partner, Hargrove Forensic Advisory LLC Aaron Liu, Lead Associate, Calderwood & Harkness LLP Monica Reyes, Working Associate, Calderwood & Harkness LLP

RE: Pricing Data Analysis — Torchlight Mining Corp. U.S. Specialty Minerals Division Price Movements vs. Competitor Price Movements (2015–2022)

MATTER: DOJ Antitrust Division Grand Jury Investigation — Grand Jury No. GJ-22-0194 (N.D. Cal.); Calderwood & Harkness Matter No. 1039-00001

PRIVILEGE DESIGNATION: This memorandum and its attachments constitute attorney work product and attorney-client privileged communication prepared in anticipation of litigation and at the direction of counsel. This memorandum was prepared under Kovel protections pursuant to the engagement of Hargrove Forensic Advisory LLC by Calderwood & Harkness LLP on July 1, 2022. Disclosure of this memorandum or any portion thereof to any person or entity other than the client, Torchlight Mining Corp., or its authorized representatives, without the prior written consent of Calderwood & Harkness LLP, may constitute a waiver of the attorney-client privilege and/or work product protection. All recipients are instructed to treat this memorandum as strictly confidential and to limit further distribution to those individuals with a need to know in connection with the above-referenced matter.

Right-click to update Table of Contents

I. Executive Summary

This memorandum presents the findings of a joint analysis conducted by Calderwood & Harkness LLP and Hargrove Forensic Advisory LLC examining the pricing behavior of Torchlight Mining Corp. ("TMC") in the U.S. specialty-minerals market relative to the pricing behavior of its two principal competitors, Ridgecrest Minerals Inc. ("Ridgecrest") and Pinnacle Earth Resources Ltd. ("Pinnacle"), over the period January 2015 through September 2022. The analysis was undertaken at the direction of counsel as part of the internal investigation into the conduct forming the basis of the DOJ Antitrust Division's ongoing grand jury investigation, Grand Jury No. GJ-22-0194 (N.D. Cal.).

The three product categories at the center of the DOJ investigation — high-purity feldspar, industrial talc, and rare-earth concentrates — were analyzed across the full time period. These three product categories represent the principal product lines of TMC's U.S. Specialty Minerals Division and, according to the DOJ's theory of the case, the subjects of alleged price-fixing activity among TMC, Ridgecrest, and Pinnacle.

To facilitate meaningful comparison, the analytical period was divided into three discrete windows:

(a) Pre-Conspiracy Baseline (January 2015 – December 2016): The period preceding the alleged commencement of coordinated pricing behavior, used as a benchmark for normal pricing correlations attributable to common market factors.

(b) Alleged Conspiracy Period (January 2017 – August 2021): The period during which, according to the DOJ's investigation and the internal investigation findings, coordinated pricing activity occurred among the three companies, principally through VaultChat encrypted communications between Gavin Holt (TMC Vice President, Pricing Strategy) and Carl Edgerton (Ridgecrest) and Kevin Tsang (Pinnacle), as well as through side meetings at NASMC quarterly conferences.

(c) Post-Conspiracy Period (September 2021 – September 2022): The period following the departure of Gavin Holt from TMC (effective September 30, 2021) and the approximate cessation of the alleged conspiracy, used to evaluate whether pricing dynamics reverted to pre-conspiracy patterns.

Key Finding. The analysis identifies statistically significant pricing correlations between TMC and its competitors during the alleged conspiracy period (January 2017 – August 2021) that are materially elevated compared to both the pre-conspiracy baseline and the post-conspiracy period. Across all three product categories, Pearson correlation coefficients between TMC and competitor price series increased substantially during the conspiracy period — in some instances more than doubling — and declined to levels approximating the pre-conspiracy baseline following the end of the alleged conspiracy. These patterns are consistent across multiple analytical approaches, including time-series correlation analysis, structural break testing, regression analysis controlling for common cost drivers, and an event study linking pricing changes to documented VaultChat communications.

It is important to note, however, that the analysis identifies pricing patterns consistent with coordinated behavior during the conspiracy period. Price correlation alone is not dispositive of conspiracy; parallel pricing can also result from legitimate market dynamics, particularly in concentrated industries with relatively homogeneous products and common input cost structures. To address this, the analysis employed regression models controlling for common cost drivers — including energy costs, transportation indices, raw material prices, and downstream demand indicators — and found that the elevated correlations during the conspiracy period persist even after accounting for these market-level explanations. Nevertheless, the statistical evidence should be considered in conjunction with the documentary evidence (VaultChat communications), witness testimony (Sandra Chu, Thomas Eriksen), and other investigative findings before drawing conclusions regarding the existence or scope of the alleged conspiracy.

This memorandum supports the internal investigation findings to be presented to the TMC Board of Directors on January 20, 2023, and will inform the cooperation proffer and penalty negotiation strategy. The analysis will be refined in a formal Hargrove expert report addressing affected-commerce calculations and penalty analysis, which is scheduled for delivery in connection with the anticipated penalty negotiations with the DOJ Antitrust Division.

II. Scope and Methodology

A. Engagement Background

Hargrove Forensic Advisory LLC ("Hargrove") was retained by Calderwood & Harkness LLP on July 1, 2022, under Kovel protections, to provide forensic economic advisory services in connection with the DOJ Antitrust Division's grand jury investigation of pricing practices in the U.S. specialty-minerals market. The engagement was authorized by TMC's General Counsel, Rebecca Aldridge, and the Audit and Compliance Committee of the TMC Board of Directors, pursuant to the Board's resolution of June 15, 2022, authorizing the internal investigation.

The engagement partner at Hargrove is Dr. Vanessa Tremblay, who maintains her principal office at 233 South Wacker Drive, Suite 4800, Chicago, IL 60606. Dr. Tremblay holds a Ph.D. in Economics from the University of Chicago and has over twenty years of experience in antitrust economics, including extensive experience with price-fixing investigations and the economic analysis of coordinated conduct in concentrated industries. The Hargrove team also includes three senior economists and two data analysts, all of whom have been identified on the privilege log maintained by Calderwood & Harkness.

Hargrove was tasked with three principal assignments in connection with this engagement:

(1) Economic analysis of pricing patterns in the specialty-minerals market, including time-series analysis, correlation analysis, structural break analysis, regression analysis, and event studies linking pricing events to documentary evidence of coordination;

(2) Affected-commerce volume assessment, including the calculation of TMC's affected commerce during the alleged conspiracy period for purposes of penalty determination under the United States Sentencing Guidelines and the DOJ Antitrust Division's penalty guidelines; and

(3) Support for NPA penalty negotiations, including the preparation of economic analyses and presentations for use in negotiations with the DOJ regarding the appropriate monetary penalty, taking into account cooperation credits, remediation efforts, and mitigating factors.

This memorandum addresses task (1). Tasks (2) and (3) will be addressed in subsequent work product, as described in Section IX below.

B. Data Sources

The analysis draws upon five principal categories of data, each of which is described below. The data collection, processing, and quality-assurance procedures are described in greater detail in Appendix 6.

TMC Internal Data. The primary source of TMC pricing data consists of published price lists, transaction-level sales data, customer contract pricing (including both spot and long-term contract prices), and internal pricing memoranda from TMC's U.S. Specialty Minerals Division covering the period January 2015 through September 2022. These data were sourced from two principal repositories: (a) the approximately 487,000 documents produced to date by TMC in response to Grand Jury Subpoena No. GJ-22-0194-SD-003, which requested all documents relating to pricing decisions, competitor contacts, and trade association activities for the U.S. Specialty Minerals Division; and (b) TMC's internal enterprise resource planning (ERP) system (Harborline Software) and customer relationship management (CRM) system (Pinnacle Health IT), from which transactional data were extracted and processed. The collection and forensic processing of TMC's electronic data was performed by Redstone Digital Forensics Inc. (Neil Ashworth, Project Director), which was separately retained by Calderwood & Harkness under Kovel protections for digital forensics support.

Competitor Pricing Data. Because Hargrove does not have direct access to the internal pricing records of Ridgecrest or Pinnacle, competitor pricing data were assembled from multiple sources. The primary source of competitor benchmark pricing is the quarterly pricing surveys published by the North American Specialty Minerals Council (NASMC), headquartered in Arlington, VA. NASMC publishes average transaction prices for high-purity feldspar, industrial talc, and rare-earth concentrates on a quarterly basis, disaggregated by grade and region. These benchmarks provide a reliable, if somewhat coarse, measure of market-level pricing trends. Supplemental competitor pricing data were obtained from industry trade publications, including Mineral Markets Weekly, Industrial Minerals Review, and Specialty Minerals Quarterly. In addition, TMC's own records contained competitor price sheets, bid tabulation documents, and competitor pricing intelligence collected by TMC's sales personnel in the ordinary course of business, which provided more granular, transaction-level insight into Ridgecrest and Pinnacle pricing for specific products and time periods.

Market Data. To control for common cost drivers and market-level factors that could independently produce pricing correlation, Hargrove assembled a comprehensive set of market and cost data, including: (a) energy cost indices (U.S. Energy Information Administration natural gas and electricity price indices); (b) transportation cost indices (Bureau of Labor Statistics Producer Price Index for rail and truck transportation of minerals); (c) mining consumables and raw material input cost indices (PPI for industrial chemicals, explosives, and mining equipment); and (d) downstream demand indicators, including the Federal Reserve's industrial production index for chemical manufacturing (NAICS 325) and for glass and glass products (NAICS 3272), which serve as proxies for demand in the end-use markets for the three product categories. These data were sourced from publicly available federal databases and supplemented by industry reports from NASMC and the U.S. Geological Survey.

VaultChat Communications. A critical corroborative data source consists of the pricing-related communications exchanged between Gavin Holt (TMC) and Carl Edgerton (Ridgecrest) and Kevin Tsang (Pinnacle) via the VaultChat encrypted messaging application. These communications were identified, recovered (to the extent technologically possible), and catalogued in the VaultChat Message Analysis Memorandum dated September 15, 2022, prepared by Calderwood & Harkness as internal work product. That memorandum identified at least 37 documented exchanges between March 14, 2017, and July 9, 2021, in which pricing information — including specific prices, planned price-change timing, and customer allocation — was shared between Holt and one or both competitor counterparts. The event study component of the present analysis cross-references the dates and content of these VaultChat exchanges with TMC and competitor pricing events.

Witness Interview Data. The analysis also draws upon information obtained through Calderwood & Harkness's internal investigation witness interviews, conducted under Upjohn warnings. Two witnesses provided information particularly relevant to the pricing analysis:

• Sandra Chu (Director, Pricing Analytics, TMC U.S. Specialty Minerals Division): Chu stated during her interview on August 22, 2022, that she received pricing-adjustment instructions from Gavin Holt on at least 12 occasions between approximately mid-2017 and mid-2021. Chu described these instructions as unusual because they directed specific price levels or price-change timing without reference to the normal cost-based pricing models employed by the Pricing Analytics group. The dates of these instructions, as identified from Chu's calendar entries and email records, were provided to Hargrove for use in the event study analysis.

• Thomas Eriksen (Regional Sales Manager, Western Region, TMC U.S. Specialty Minerals Division): Eriksen stated during his interview on September 2, 2022, that he attended four NASMC quarterly conferences during the conspiracy period and observed Gavin Holt meeting privately with individuals he identified as representatives of Ridgecrest and Pinnacle at conference social events. Eriksen could not provide details of what was discussed in these meetings but noted that pricing guidance from Holt often changed in the days following NASMC conferences.

C. Analytical Methodology

Hargrove employed five principal analytical approaches, each designed to examine a different dimension of the pricing relationship between TMC and its competitors. The methodologies are summarized below; additional technical detail is provided in Appendix 6.

Time-Series Analysis. Monthly average prices for each product category (high-purity feldspar, industrial talc, and rare-earth concentrates) were computed for TMC and each identified competitor (Ridgecrest and Pinnacle) over the full January 2015 through September 2022 period. Monthly averages were computed from transaction-level data for TMC and from the best available data for competitors, as described in Section II.B above. The resulting time series were examined visually and statistically for trends, seasonality, and co-movement patterns across the three analytical windows.

Correlation Analysis. Pearson product-moment correlation coefficients and Spearman rank correlation coefficients were calculated for each TMC-competitor price series pair within each of the three analytical windows (pre-conspiracy, conspiracy period, and post-conspiracy). Pearson correlations measure the linear association between two price series, while Spearman correlations measure the monotonic association without assuming linearity. Both measures were employed to ensure robustness. The statistical significance of each correlation coefficient was assessed using standard t-tests, and Fisher's z-transformation was used to test whether the correlation coefficients differed significantly across time periods.

Structural Break Analysis. Chow tests were applied to the TMC-competitor price series to determine whether statistically significant shifts in the pricing relationship occurred at the hypothesized break dates of January 2017 (alleged conspiracy start) and September 2021 (alleged conspiracy end). In addition, Bai-Perron multiple structural break tests were applied without imposing a priori break dates to determine whether the data independently identify structural breaks consistent with the conspiracy timeline. The Bai-Perron methodology allows for the endogenous determination of break dates and the number of breaks, providing an important check against the risk of confirmation bias in the choice of break dates.

Regression Analysis. Ordinary least squares (OLS) regression models were specified for each product category to control for common cost drivers and market-level factors that could independently explain pricing correlations. The dependent variable in each model is the monthly average TMC price for the relevant product category. Independent variables include the competitor's monthly average price, energy cost indices, transportation cost indices, raw material and mining input cost indices, and downstream demand indicators. Conspiracy-period dummy variables and interaction terms were included to test whether the degree of price co-movement changed during the alleged conspiracy period after controlling for common market factors. Heteroskedasticity-robust standard errors (White standard errors) were employed, and models were tested for serial correlation using the Durbin-Watson statistic.

Event Study. An event study analysis was conducted to examine the temporal relationship between specific pricing events — dates on which TMC published price changes — and three categories of potentially relevant events: (a) documented VaultChat communications between Holt and competitor counterparts, (b) NASMC quarterly conference dates, and (c) competitor price-change announcements. The event window was defined as ±15 calendar days from the TMC price-change announcement date. The analysis examined the frequency with which VaultChat communications, NASMC conferences, and competitor price changes fell within the event window and compared this frequency to what would be expected under a null hypothesis of random timing.

Significance Levels. All statistical tests were conducted at the 95% confidence level (α = 0.05), with robustness checks at the 99% confidence level (α = 0.01). Results are reported with exact p-values where available to permit assessment of the strength of the evidence at multiple significance thresholds.

D. Limitations and Caveats

The analysis is subject to the following limitations, which should be considered in evaluating the findings and their implications.

First, the analysis is based on data available as of November 1, 2022. At the time of this analysis, the DOJ had not yet issued the anticipated second subpoena (GJ-22-0194-SD-007), which is expected to request additional categories of documents, including communications between TMC and Ridgecrest/Pinnacle outside the VaultChat platform. Additional data produced in response to that subpoena or obtained through other investigative channels may affect the findings presented herein.

Second, competitor pricing data is incomplete for certain periods and products. Most significantly, Pinnacle's rare-earth concentrate pricing data prior to mid-2016 is unavailable from the sources described in Section II.B, which limits the effectiveness of the pre-conspiracy baseline analysis for the TMC-Pinnacle rare-earth pair. In addition, the competitor pricing data assembled from public and semi-public sources is generally less granular than the TMC transaction-level data, which may introduce measurement error that could attenuate estimated correlations.

Third, the VaultChat communications were only partially recovered due to the application's self-destructing message features and the deletion of message histories from certain devices. The 37 documented exchanges identified in the VaultChat Message Analysis Memorandum may understate the total number of pricing-related communications between Holt and his competitor counterparts. Consequently, the event study results may undercount the instances of temporal alignment between communications and pricing events.

Fourth, and most fundamentally, correlation does not establish causation. Parallel pricing can result from legitimate market dynamics, particularly in concentrated industries with relatively homogeneous products and common cost structures. While the regression analysis controls for observable common cost drivers, it is possible that unobserved factors — such as common strategic responses to demand shocks, simultaneous receipt of market intelligence, or oligopolistic interdependence — could contribute to elevated correlations without any agreement or coordination. The statistical evidence presented in this memorandum should therefore be evaluated in conjunction with the documentary and testimonial evidence of the alleged conspiracy, not in isolation.

III. Market Overview

A. Industry Structure

The North American specialty-minerals market for high-purity feldspar, industrial talc, and rare-earth concentrates is a highly concentrated market in which a small number of producers account for a substantial share of total U.S. supply. This concentrated structure is relevant to the pricing analysis because, as noted in Section II.D, some degree of pricing correlation is expected in concentrated markets even in the absence of coordination, due to oligopolistic interdependence, common cost structures, and the transparency of pricing in markets with few participants.

Torchlight Mining Corp. ("TMC"). TMC is a publicly traded Canadian mining and minerals company with its corporate headquarters in Toronto, Ontario, and its U.S. operational headquarters in Phoenix, Arizona. TMC's U.S. Specialty Minerals Division operates mines and six downstream chemical-processing facilities across the United States, with significant operations in Nevada, Utah, Georgia, and North Carolina. For FY 2021 (TMC's fiscal year ending December 31, 2021), the U.S. Specialty Minerals Division reported revenue of approximately USD 687 million, representing approximately 20.3% of TMC's consolidated revenue of approximately USD 3.38 billion (CAD 4.27 billion at the average 2021 exchange rate). TMC is the largest North American producer of high-purity feldspar and among the top three North American producers of industrial talc and rare-earth concentrates.

Ridgecrest Minerals Inc. ("Ridgecrest"). Ridgecrest is a Delaware corporation headquartered in Denver, Colorado. Ridgecrest is the second-largest North American producer of high-purity feldspar and industrial talc, with mining operations primarily in Colorado, Wyoming, and Montana, and processing facilities in Colorado and Texas. Ridgecrest reported FY 2021 revenue of approximately USD 1.92 billion across all business lines, though its specialty-minerals operations account for a smaller share of total revenue than TMC's corresponding division. Ridgecrest is a portfolio company of Marlowe Asset Management Peak Capital Partners LP, a Denver-based private equity firm that acquired a controlling interest in Ridgecrest in 2013. Ridgecrest's participation in the rare-earth concentrates market is more limited than its presence in feldspar and talc, reflecting the geographic concentration of its mining assets.

Pinnacle Earth Resources Ltd. ("Pinnacle"). Pinnacle is an Ontario corporation with its U.S. operations headquartered in Salt Lake City, Utah. Pinnacle is a top-five North American producer of rare-earth concentrates and specialty industrial minerals, with mining operations in Utah, Idaho, and British Columbia, and processing facilities in Utah and Quebec. For FY 2021, Pinnacle reported revenue of approximately CAD 1.45 billion (approximately USD 1.15 billion at the average 2021 exchange rate). Pinnacle's primary competitive overlap with TMC is in the rare-earth concentrates and industrial talc markets; its presence in the high-purity feldspar market is comparatively smaller but not negligible.

North American Specialty Minerals Council (NASMC). NASMC is an industry trade association headquartered in Arlington, Virginia. NASMC publishes quarterly pricing benchmarks for the specialty-minerals industry, organizes quarterly conferences attended by industry participants, and engages in advocacy on regulatory and trade matters affecting the industry. The quarterly conferences, which rotate among venues across the United States, are attended by executives, sales managers, and technical personnel from the member companies. The DOJ's investigation has alleged that side meetings at NASMC conferences served as one of the venues for the alleged coordination among TMC, Ridgecrest, and Pinnacle representatives.

Collectively, TMC, Ridgecrest, and Pinnacle account for a significant share of U.S. supply for each of the three product categories, though the precise market share figures are subject to confidentiality restrictions and vary by product category and year. Based on publicly available information and the data assembled by Hargrove, the three companies collectively account for an estimated 55–70% of U.S. supply for high-purity feldspar, 45–60% for industrial talc, and 50–65% for rare-earth concentrates during the relevant period. This level of concentration is consistent with the conditions under which coordinated pricing behavior is both theoretically more likely to emerge and more likely to be sustained.

B. Product Descriptions

High-Purity Feldspar. High-purity feldspar is a mineral used primarily in the manufacturing of ceramics, glass, and specialty coatings. It serves as a flux in glass and ceramic production, lowering the melting temperature of the batch and improving the durability and finish of the final product. The pricing of high-purity feldspar is influenced by several factors, including quality grade (particularly the level of iron and other impurities), silica content, particle size distribution, and regional transportation costs, which are significant given the bulk nature of the product and the geographic dispersion of mining operations and end-use customers. Pricing is typically quoted on a per-ton basis, with premiums for higher-purity grades. The market is characterized by a mix of long-term supply contracts with major industrial customers (ceramics manufacturers, glass producers) and spot sales to smaller purchasers.

Industrial Talc. Industrial talc is a hydrated magnesium silicate mineral used in a wide range of industrial applications, including paper manufacturing (as a filler and coating), plastics (as a reinforcing filler), paint and coatings (as an extender pigment), and personal-care products (as a base for talcum powder and cosmetics). Pricing is influenced by particle size, mineral purity, brightness, and end-use specifications, with different grades commanding different price points depending on the application. The industrial talc market is characterized by a moderate degree of product differentiation based on these quality attributes, though the underlying mineral is relatively homogeneous. Transportation costs are again a significant pricing factor.

Rare-Earth Concentrates. Rare-earth concentrates encompass a group of 17 chemically similar elements used in high-technology applications, including electronics, permanent magnets (for wind turbines and electric vehicle motors), catalysts (for petroleum refining), and defense applications (guidance systems, satellite components). Rare-earth pricing is inherently more volatile than feldspar or talc pricing due to the influence of geopolitical factors — particularly the dominant role of Chinese production in global rare-earth supply — supply constraints from environmental and permitting challenges, and the relatively small size and illiquidity of the spot market. Processing costs, which include complex chemical separation procedures, are also a significant pricing factor. The North American rare-earth market has grown in strategic importance during the period under analysis, driven by concerns about supply-chain security and the growing demand for permanent magnets in electric vehicles and renewable energy systems.

C. DOJ Alleged Affected Commerce

The DOJ Antitrust Division has estimated the total affected commerce across all three participants — TMC, Ridgecrest, and Pinnacle — at approximately USD 1.34 billion over the alleged conspiracy period (January 2017 – August 2021). This figure represents the aggregate U.S. sales of high-purity feldspar, industrial talc, and rare-earth concentrates by the three companies during the approximately four years and eight months of the alleged conspiracy.

TMC's share of affected commerce, based on the DOJ's preliminary estimates and Hargrove's independent analysis of TMC sales data, is approximately USD 478 million, representing approximately 35.7% of total affected commerce ($478M / $1,340M ≈ 35.7%). This figure encompasses all U.S. sales of the three product categories by TMC's U.S. Specialty Minerals Division during the conspiracy period, including both spot sales and sales under long-term contracts.

These affected-commerce figures are preliminary and subject to refinement in the formal affected-commerce analysis to be prepared by Hargrove in connection with penalty negotiations. The figures are relevant to this memorandum because they provide context for the magnitude of the pricing conduct under analysis and will form the basis for the DOJ's penalty calculation, as discussed in Section IX.

IV. Pricing Analysis Results

This section presents the detailed results of the pricing analysis for each of the three product categories. For each product, results are reported for each of the three analytical windows: the pre-conspiracy baseline (January 2015 – December 2016), the alleged conspiracy period (January 2017 – August 2021), and the post-conspiracy period (September 2021 – September 2022). Detailed data tables are presented in Appendix 1, and full correlation matrices are presented in Appendix 2.

A. High-Purity Feldspar

1. Pre-Conspiracy Period (January 2015 – December 2016)

During the pre-conspiracy baseline period, TMC and Ridgecrest feldspar prices exhibited a moderate positive correlation, with a Pearson correlation coefficient of approximately r ≈ 0.52. This level of correlation is consistent with the two firms operating in the same concentrated market, responding to common input cost fluctuations (particularly energy and transportation costs), and adjusting to similar demand conditions in the downstream ceramics and glass manufacturing sectors. The Spearman rank correlation for this pair was approximately ρ ≈ 0.49, confirming the moderate positive association observed in the Pearson measure.

The correlation between TMC and Pinnacle feldspar prices was weaker during the baseline period, with a Pearson r ≈ 0.38 and a Spearman ρ ≈ 0.35. This lower correlation reflects Pinnacle's comparatively smaller feldspar market share and the geographic concentration of Pinnacle's feldspar operations, which exposes Pinnacle to somewhat different regional cost and demand dynamics than TMC and Ridgecrest.

During this period, TMC's average feldspar price adjustments occurred approximately every four to six months, consistent with TMC's standard practice of semi-annual price reviews. Ridgecrest's price adjustments followed a broadly similar cadence — also approximately semi-annual — but the timing of adjustments appeared to be independently determined. There was no statistically significant lead-lag relationship between TMC and competitor price changes at weekly or monthly frequency during this period; that is, neither TMC's price changes reliably preceded Ridgecrest's nor vice versa, suggesting that the observed moderate correlation was driven by common market factors rather than price signaling or following behavior.

2. Alleged Conspiracy Period (January 2017 – August 2021)

The pricing relationship between TMC and Ridgecrest feldspar changed markedly during the alleged conspiracy period. The Pearson correlation coefficient between TMC and Ridgecrest feldspar prices increased from approximately 0.52 in the baseline period to approximately 0.91 during the conspiracy period, a difference that is statistically significant at the p < 0.001 level using Fisher's z-transformation test. The Spearman rank correlation increased correspondingly, from approximately 0.49 to approximately 0.88. This level of correlation is exceptionally high and substantially exceeds what would be expected based on common market factors alone, as confirmed by the regression analysis discussed in Section VII.

The TMC-Pinnacle feldspar correlation also increased during the conspiracy period, from a Pearson r ≈ 0.38 to approximately r ≈ 0.74 (p < 0.01). While this increase is less dramatic than the TMC-Ridgecrest increase, it is nonetheless statistically significant and reflects the broadening of the alleged coordination to include Pinnacle, particularly with respect to pricing in the western U.S. market where Pinnacle has a larger feldspar presence.

Beyond the magnitude of the correlation, the timing of price adjustments became highly synchronized during the conspiracy period. Hargrove identified 32 discrete price-change events for TMC feldspar during the conspiracy period (defined as published price adjustments of at least 1% from the prior posted price). In 28 of these 32 events, Ridgecrest adjusted its feldspar prices within a 14-day window of the TMC adjustment. This frequency of near-simultaneous adjustment is statistically anomalous when compared to the pre-conspiracy period, during which such tight synchronization occurred in only approximately 40% of price-change events. Moreover, in 19 of the 28 synchronized events, the magnitudes of the TMC and Ridgecrest price adjustments were within ±1.5 percentage points of each other, further suggesting coordinated rather than independently determined pricing.

Structural break analysis (Chow test) identifies a statistically significant break in the TMC-Ridgecrest feldspar price relationship occurring in the first quarter of 2017, consistent with the alleged conspiracy start date. Critically, this structural break aligns temporally with the first documented VaultChat exchange between Gavin Holt and Carl Edgerton, which occurred on March 14, 2017 — a communication in which, according to the VaultChat Message Analysis Memorandum, Holt and Edgerton discussed an upcoming feldspar price increase and the timing of its announcement. The structural break results are discussed in further detail in Section V.

After controlling for common cost drivers — energy costs, transportation costs, and raw material prices — through the OLS regression models described in Section VII, the residual correlation between TMC and Ridgecrest feldspar prices remains elevated, with a partial correlation of approximately 0.78. This finding indicates that common cost factors, while explaining a meaningful portion of the price co-movement, account for only a fraction of the observed synchronization during the conspiracy period. The remaining co-movement is attributable to factors not captured by the cost controls — a category that includes, but is not limited to, coordinated pricing behavior.

3. Post-Conspiracy Period (September 2021 – September 2022)

Following Gavin Holt's departure from TMC, effective September 30, 2021, and the approximate cessation of the alleged conspiracy in August 2021, the pricing relationship between TMC and Ridgecrest feldspar reverted toward the pre-conspiracy baseline. The Pearson correlation coefficient declined from approximately 0.91 during the conspiracy period to approximately 0.47 during the post-conspiracy period — a level closely approximating the pre-conspiracy baseline of 0.52. While the post-conspiracy period is relatively short (approximately 13 months) and thus the statistical precision of the correlation estimate is lower, the direction and magnitude of the decline are consistent with the cessation of coordinated pricing behavior.

The timing of price adjustments also diverged in the post-conspiracy period. In 5 of the 8 post-conspiracy feldspar price-change events identified for TMC, TMC and Ridgecrest announced price changes more than 30 days apart — a pattern more consistent with independent pricing decisions than with the tight synchronization observed during the conspiracy period.

Structural break analysis confirms a second statistically significant break in the TMC-Ridgecrest feldspar price series in Q3–Q4 2021, aligning with the end of the alleged conspiracy period and Holt's departure. This second break is discussed in further detail in Section V.

B. Industrial Talc

1. Pre-Conspiracy Period (January 2015 – December 2016)

The TMC-Ridgecrest price correlation for industrial talc during the pre-conspiracy baseline period was somewhat higher than for feldspar, with a Pearson r ≈ 0.58 and a Spearman ρ ≈ 0.55. This higher baseline correlation reflects several features of the industrial talc market: the number of significant market participants is smaller than in the feldspar market, the product is more homogeneous across suppliers (with fewer quality-based pricing differentials), and common cost factors — particularly energy and transportation — exert a proportionally greater influence on talc pricing. These structural characteristics mean that a higher degree of price co-movement is expected even in the absence of coordination.

The TMC-Pinnacle talc price correlation during the baseline period was Pearson r ≈ 0.44, reflecting Pinnacle's smaller but meaningful participation in the talc market, concentrated primarily in the western U.S.

2. Alleged Conspiracy Period (January 2017 – August 2021)

During the alleged conspiracy period, the TMC-Ridgecrest talc price correlation increased from the baseline level of approximately 0.58 to approximately 0.88 (p < 0.001). The TMC-Pinnacle talc price correlation increased from approximately 0.44 to approximately 0.79 (p < 0.001). Both increases are statistically significant and economically substantial, indicating a marked shift in the degree of price co-movement during the conspiracy period.

The industrial talc category is notable for the intensity of the documentary evidence linking pricing changes to competitor communications. Of the 37 documented VaultChat exchanges between Holt and Edgerton or Tsang, 15 specifically referenced talc pricing — making talc the most frequently discussed product category in the recovered communications. These 15 exchanges addressed a range of topics, including planned price increases for specific talc grades, the timing of price announcements to key customers, and the allocation of certain large industrial accounts between TMC and competitors.

The event study analysis (Section VI) reveals particularly strong temporal alignment between VaultChat exchanges and TMC pricing events for talc. On 9 of the 12 occasions on which Sandra Chu, TMC's Director of Pricing Analytics, received pricing-adjustment instructions from Gavin Holt (as identified through Chu's witness interview and confirmed by review of her calendar entries and email records), the timing of the instruction correlated within one to five business days with a VaultChat exchange involving Holt and a competitor counterpart. This pattern is consistent with a sequence in which Holt communicated with a competitor counterpart to agree upon a pricing action and then directed Chu to implement the agreed-upon change through TMC's pricing systems.

In addition to the pricing correlation findings, Hargrove's analysis of TMC transaction data revealed customer allocation patterns during the conspiracy period that are consistent with market allocation. Specifically, certain large industrial customers — including two Fortune 500 chemical manufacturers — exhibited unexplained shifts in purchasing patterns during the conspiracy period, with a reduction in competitive bidding and an increase in sole-source purchasing from TMC (or, in other cases, a shift away from TMC to Ridgecrest or Pinnacle), that reversed following the end of the alleged conspiracy. While the customer allocation analysis is not the primary focus of this memorandum, it is noted here because it provides additional corroborative context for the pricing findings.

3. Post-Conspiracy Period (September 2021 – September 2022)

The TMC-Ridgecrest talc price correlation declined to approximately Pearson r ≈ 0.50 during the post-conspiracy period, approaching the pre-conspiracy baseline of approximately 0.58. The TMC-Pinnacle talc price correlation declined to approximately Pearson r ≈ 0.41, closely approximating its pre-conspiracy level of 0.44. As with the feldspar analysis, the direction and magnitude of these declines are consistent with the cessation of coordinated pricing behavior and the reversion to competitive market dynamics.

C. Rare-Earth Concentrates

1. Pre-Conspiracy Period (January 2015 – December 2016)

Rare-earth concentrate pricing is inherently more volatile than feldspar or talc pricing due to the influence of geopolitical factors, global supply-chain dynamics (particularly the dominance of Chinese production), and the relatively small size of the North American spot market. This heightened volatility has implications for the interpretation of correlation coefficients, as higher price volatility can either increase or decrease measured correlations depending on the nature of the underlying shocks.

During the pre-conspiracy baseline period, the TMC-Pinnacle rare-earth price correlation was Pearson r ≈ 0.61 — the highest baseline correlation observed among any of the product-competitor pairs. This reflects the fact that Pinnacle is TMC's primary North American competitor in rare-earth concentrates, and the two companies' pricing is significantly influenced by the same global benchmark prices and supply-chain factors. The Spearman rank correlation for this pair was ρ ≈ 0.58.

The TMC-Ridgecrest rare-earth correlation during the baseline period was substantially lower, at Pearson r ≈ 0.33, reflecting Ridgecrest's comparatively limited presence in the rare-earth concentrates market. Ridgecrest's rare-earth operations are a smaller component of its overall business, and its pricing may be influenced by factors specific to its particular product grades and customer base.

An important data limitation applies to the rare-earth analysis: Pinnacle's rare-earth concentrate pricing data prior to mid-2016 is incomplete in the available data sources. As a result, the baseline analysis for the TMC-Pinnacle rare-earth pair effectively begins from July 2016, providing only approximately six months of pre-conspiracy data rather than the full two years available for other product-competitor pairs. This shorter baseline period reduces the precision of the pre-conspiracy correlation estimate and should be considered in interpreting the magnitude of the change between periods.

2. Alleged Conspiracy Period (January 2017 – August 2021)

During the alleged conspiracy period, the TMC-Pinnacle rare-earth price correlation increased from the baseline level of approximately 0.61 to approximately 0.85 (p < 0.001). The TMC-Ridgecrest rare-earth correlation increased from approximately 0.33 to approximately 0.62 (p < 0.05), a more modest increase consistent with Ridgecrest's smaller role in the rare-earth market.

The most pronounced coordination in rare-earth pricing appears to have occurred during the 2019–2020 sub-period, when global rare-earth supply disruptions — including Chinese export restrictions, environmental enforcement actions that curtailed production at several Chinese mines, and pandemic-related logistics challenges — created a pricing environment characterized by elevated uncertainty and rapid price movements. Hargrove's analysis indicates that the TMC-Pinnacle rare-earth correlation was particularly high during this sub-period (Pearson r ≈ 0.92 for 2019–2020 alone), suggesting that the alleged coordination may have been especially active during periods of market disruption, when participants had both greater incentive and greater opportunity to coordinate pricing responses.

After controlling for global rare-earth commodity indices — specifically, the global rare-earth price index and the London Metal Exchange rare-earth reference prices — the residual TMC-Pinnacle rare-earth correlation remains statistically significant, with a partial correlation of approximately 0.54. This indicates that the elevated correlation during the conspiracy period is not fully explained by common responses to global rare-earth market conditions, though global factors do account for a meaningful portion of the co-movement. The residual correlation is somewhat lower than for feldspar and talc, which is consistent with the greater influence of exogenous global factors on rare-earth pricing and with the less frequent VaultChat communications specifically addressing rare-earth pricing.

3. Post-Conspiracy Period (September 2021 – September 2022)

During the post-conspiracy period, the TMC-Pinnacle rare-earth correlation declined to approximately Pearson r ≈ 0.55, representing a partial but meaningful reversion toward the pre-conspiracy baseline of approximately 0.61. The fact that the post-conspiracy correlation (0.55) is slightly below the pre-conspiracy baseline (0.61) may reflect ongoing market turbulence in the rare-earth sector during late 2021 and 2022, including supply-chain adjustments and shifting demand patterns that have introduced new sources of idiosyncratic pricing variation.

The TMC-Ridgecrest rare-earth correlation declined to approximately Pearson r ≈ 0.29, closely approximating its pre-conspiracy baseline of approximately 0.33 and confirming the reversion of Ridgecrest's rare-earth pricing to a largely independent trajectory.

V. Structural Break Analysis — Detailed Results

The structural break analysis was designed to test whether the pricing dynamics between TMC and its competitors changed at specific points in time, and whether those change points align with the alleged conspiracy start and end dates. Two complementary methodologies were employed: the Chow test, which tests for a structural break at a specified date, and the Bai-Perron multiple structural break test, which endogenously identifies the number and timing of structural breaks without requiring the analyst to specify break dates in advance.

Chow Test Results. The Chow test was applied to each TMC-competitor price series pair, with hypothesized break dates at Q1 2017 (alleged conspiracy start) and Q3 2021 (alleged conspiracy end). The results are summarized as follows:

• TMC-Ridgecrest Feldspar: Statistically significant structural breaks identified at both Q1 2017 and Q3 2021 (p < 0.01 at both break points). The F-statistics for both breaks substantially exceed the critical values, indicating that the pricing relationship shifted materially at both the beginning and end of the alleged conspiracy period.

• TMC-Ridgecrest Talc: Statistically significant structural breaks identified at both Q1 2017 and Q4 2021 (p < 0.01 at both break points). The slight delay in the talc break relative to the feldspar break at the conspiracy end is consistent with the observation that certain long-term talc supply contracts may have delayed the reversion to competitive pricing following the cessation of coordination.

• TMC-Pinnacle Rare-Earth Concentrates: Statistically significant structural breaks identified at Q1 2017 (p < 0.05) and Q3 2021 (p < 0.05). The somewhat lower significance levels for the rare-earth series reflect the higher volatility of rare-earth pricing and the shorter effective baseline period, as discussed in Section IV.C.1.

• TMC-Pinnacle Feldspar and Talc: Structural breaks at the conspiracy start date are marginally significant (p ≈ 0.06–0.08) for both product categories, reflecting Pinnacle's smaller market share in these categories and the resulting weaker statistical signal.

• TMC-Ridgecrest Rare-Earth Concentrates: A structural break at Q1 2017 is marginally significant (p ≈ 0.09), consistent with Ridgecrest's limited rare-earth market presence.

Bai-Perron Test Results. The Bai-Perron multiple structural break test was applied to the full January 2015 through September 2022 TMC-Ridgecrest price series for both feldspar and talc. The results confirm the existence of two structural breaks in each series:

• Feldspar: The Bai-Perron test identifies break dates of approximately February 2017 (±2 months) and approximately October 2021 (±2 months). These endogenously determined break dates closely align with the conspiracy period boundaries derived from the documentary evidence.

• Talc: The Bai-Perron test identifies break dates of approximately March 2017 (±2 months) and approximately November 2021 (±3 months). Again, these dates are consistent with the conspiracy timeline, with the slightly later second break date for talc paralleling the Chow test finding discussed above.

The Bai-Perron test was also applied to the TMC-Pinnacle rare-earth series, but the results were less definitive, identifying one clear break in early 2017 but providing an ambiguous result for the conspiracy end date, likely due to the continued volatility in rare-earth markets during the post-conspiracy period.

Implication. The structural break analysis provides strong statistical evidence that the pricing dynamics between TMC and its principal competitors changed materially at the beginning and end of the alleged conspiracy period. The alignment between the endogenously determined break dates and the conspiracy timeline derived from documentary and testimonial evidence is striking and supports the inference that the elevated pricing correlations observed during the conspiracy period were not the result of a gradual or trend-driven change in market structure but rather reflected a discrete shift in pricing behavior — a pattern consistent with the commencement and cessation of coordinated pricing conduct.

Full statistical output for the Chow and Bai-Perron tests, including F-statistics, critical values, confidence intervals for break dates, and residual diagnostics, is presented in Appendix 3.

VI. Event Study — VaultChat Communications and Price Changes

A. Methodology

The event study analysis was designed to examine whether TMC pricing events — specifically, the dates on which TMC published changes to its posted prices for high-purity feldspar, industrial talc, or rare-earth concentrates — were temporally associated with documented VaultChat communications between Gavin Holt and Carl Edgerton (Ridgecrest) or Kevin Tsang (Pinnacle).

The analysis cross-referenced the dates of the 37 documented VaultChat exchanges identified in the VaultChat Message Analysis Memorandum dated September 15, 2022, prepared by Calderwood & Harkness, with TMC price-change announcement dates and competitor price-change announcement dates. The event window was defined as ±15 calendar days from the VaultChat exchange date — that is, a 31-day window centered on each documented communication. Within this window, Hargrove examined (a) whether TMC announced a price change, (b) whether any competitor announced a price change, (c) whether the price changes were in the same direction (i.e., both increases or both decreases), and (d) whether the magnitudes of the changes were similar (defined as within ±2 percentage points of each other).

In addition, the analysis examined the temporal relationship between the 12 pricing-adjustment instructions from Holt to Sandra Chu, as identified through Chu's witness interview and confirmed by review of her calendar entries and email records, and the dates of VaultChat exchanges.

B. Findings

The event study produced the following principal findings:

Of the 37 documented VaultChat exchanges between Holt and Edgerton or Tsang, spanning the period from March 14, 2017, through July 9, 2021, 29 occurred within 15 calendar days before or after a TMC or competitor price adjustment. The remaining 8 exchanges either fell outside the ±15-day event window or could not be reliably matched to a specific pricing event due to data limitations. The frequency with which VaultChat communications cluster around pricing events (29 of 37, or approximately 78%) is substantially higher than would be expected if the communications and pricing events were independently timed. Under a null hypothesis of random timing, given the average frequency of price-change events during the conspiracy period, the expected proportion of VaultChat exchanges falling within a ±15-day window of a pricing event is approximately 35–45%, depending on the product category.

In 22 of the 29 instances where a VaultChat exchange coincided with a pricing event, TMC and at least one competitor made pricing changes in the same direction (both increases or both decreases) and within a similar magnitude range (±2 percentage points) within the 15-day event window. This pattern of same-direction, similar-magnitude adjustments clustered around communication dates is consistent with the use of the VaultChat platform to coordinate the timing and magnitude of price changes.

The first documented VaultChat exchange, dated March 14, 2017, between Holt and Edgerton, provides a particularly instructive example. According to the VaultChat Message Analysis Memorandum, this exchange involved a discussion of an upcoming feldspar price increase and the proposed timing of its announcement. Within 10 days of this communication, both TMC and Ridgecrest announced feldspar price increases of approximately 4%, effective within days of each other. This first documented exchange and its associated pricing event are temporally consistent with the structural break identified by the Chow and Bai-Perron tests in Q1 2017.

The analysis of the 12 pricing-adjustment instructions from Holt to Chu shows statistically significant clustering around VaultChat exchange dates. Specifically, 9 of the 12 instructions fell within 5 business days of a documented VaultChat communication. The temporal sequence in most of these instances is consistent with the following pattern: Holt communicated with a competitor counterpart via VaultChat, and within 1–5 business days thereafter, Holt directed Chu to implement a specific pricing adjustment — often specifying both the magnitude and timing of the change — without reference to the normal cost-based pricing models that Chu's Pricing Analytics group would typically employ to justify pricing decisions. Chu noted in her interview that this departure from standard pricing-review procedures struck her as unusual, though she did not raise the matter with anyone other than Holt at the time.

C. NASMC Conference Correlation

In addition to the VaultChat communications, Hargrove examined whether TMC pricing events were temporally associated with NASMC quarterly conferences, which the DOJ alleges served as venues for in-person coordination among TMC, Ridgecrest, and Pinnacle representatives.

Thomas Eriksen, TMC's Regional Sales Manager for the Western Region, attended four NASMC quarterly conferences during the conspiracy period. Conference venues included Arlington, VA (NASMC's headquarters city, where several conferences were held); Scottsdale, AZ; and Orlando, FL. Gavin Holt attended additional NASMC conferences during the conspiracy period, as confirmed by TMC travel and expense records reviewed by Calderwood & Harkness. The total number of NASMC conferences attended by TMC personnel during the conspiracy period has not been fully determined, as TMC's travel records are still being analyzed, but at least six conferences are confirmed to have been attended by Holt or other TMC personnel.

The event study reveals elevated clustering of price-change events in the two to four-week periods following NASMC quarterly conferences. Specifically, during the conspiracy period (January 2017 – August 2021), 14 of 32 feldspar and talc price-change events (approximately 44%) occurred within two to four weeks of an NASMC conference. By comparison, during the pre-conspiracy baseline period (January 2015 – December 2016), only 3 of 14 price-change events (approximately 21%) occurred within this window relative to NASMC conference dates. The difference in these proportions is suggestive, though not by itself statistically significant at the 95% confidence level given the relatively small sample sizes, particularly in the pre-conspiracy period.

While NASMC conference attendance alone is not evidence of coordination — trade association conferences are a legitimate and commonplace feature of the specialty-minerals industry — the statistical pattern of post-conference price changes, combined with Eriksen's testimony regarding Holt's meetings with competitor representatives at conferences and the timing of subsequent pricing guidance changes, is consistent with the DOJ's allegation that side meetings at NASMC conferences served as one of the mechanisms through which pricing coordination was arranged and maintained.

VII. Regression Analysis — Controlling for Market Factors

A. Model Specification

To distinguish between pricing correlation attributable to common market factors and pricing correlation potentially attributable to coordinated behavior, Hargrove estimated a series of OLS regression models for each product category. The models were specified as follows:

Dependent Variable: Monthly average TMC price for the relevant product category (high-purity feldspar, industrial talc, or rare-earth concentrates), expressed in U.S. dollars per metric ton.

Independent Variables:

(1) Competitor Monthly Average Price: The monthly average price for the same product category for the relevant competitor (Ridgecrest or Pinnacle), used to measure the degree of price co-movement.

(2) Energy Cost Index: The U.S. Energy Information Administration's composite energy cost index for industrial users, reflecting the cost of natural gas and electricity used in mining and processing operations.

(3) Transportation Cost Index: The Bureau of Labor Statistics' Producer Price Index for rail and truck transportation of minerals, reflecting the cost of delivering product from mine or processing facility to customer.

(4) Raw Material/Mining Input Index: A composite index constructed from PPI sub-indices for industrial chemicals, explosives, and mining equipment, reflecting the cost of key mining and processing inputs.

(5) Downstream Demand Indicators: The Federal Reserve's industrial production index for the relevant downstream sector(s) — chemical manufacturing (NAICS 325) for talc and rare-earth concentrates, and glass and glass products (NAICS 3272) for feldspar — used as a proxy for end-use demand conditions.

Dummy Variables and Interaction Terms:

• A binary dummy variable for the conspiracy period (equal to 1 for January 2017 – August 2021 and 0 otherwise), capturing any level shift in the TMC-competitor pricing relationship during the conspiracy period.

• An interaction term between the conspiracy-period dummy and the competitor price variable, capturing any change in the degree of price co-movement (i.e., the slope of the TMC-competitor pricing relationship) during the conspiracy period.

Estimation Method: All models were estimated by ordinary least squares with heteroskedasticity-robust standard errors (White standard errors). The Durbin-Watson statistic was computed for each model to test for serial correlation in the residuals. Where serial correlation was detected, the models were re-estimated with Newey-West standard errors to ensure valid

PRIVILEGED & CONFIDENTIAL ATTORNEY-CLIENT PRIVILEGED / WORK PRODUCT DOCTRINE NOT FOR DISTRIBUTION WITHOUT PRIOR WRITTEN AUTHORIZATION OF CALDERWOOD & HARKNESS LLP
PRICING DATA ANALYSIS MEMORANDUM
PREPARED AT THE DIRECTION OF COUNSEL — CALDERWOOD & HARKNESS LLP
November 1, 2022
TO: Elaine Forsythe, Partner, Calderwood & Harkness LLP
FROM: Dr. Vanessa Tremblay, Engagement Partner, Hargrove Forensic Advisory LLC Aaron Liu, Lead Associate, Calderwood & Harkness LLP Monica Reyes, Working Associate, Calderwood & Harkness LLP
RE: Pricing Data Analysis — Torchlight Mining Corp. U.S. Specialty Minerals Division Price Movements vs. Competitor Price Movements (2015–2022)
MATTER: DOJ Antitrust Division Grand Jury Investigation — Grand Jury No. GJ-22-0194 (N.D. Cal.); Calderwood & Harkness Matter No. 1039-00001
PRIVILEGE DESIGNATION: This memorandum and its attachments constitute attorney work product and attorney-client privileged communication prepared in anticipation of litigation and at the direction of counsel. This memorandum was prepared under Kovel protections pursuant to the engagement of Hargrove Forensic Advisory LLC by Calderwood & Harkness LLP on July 1, 2022. Disclosure of this memorandum or any portion thereof to any person or entity other than the client, Torchlight Mining Corp., or its authorized representatives, without the prior written consent of Calderwood & Harkness LLP, may constitute a waiver of the attorney-client privilege and/or work product protection. All recipients are instructed to treat this memorandum as strictly confidential and to limit further distribution to those individuals with a need to know in connection with the above-referenced matter.
Right-click to update Table of Contents
I. Executive Summary
This memorandum presents the findings of a joint analysis conducted by Calderwood & Harkness LLP and Hargrove Forensic Advisory LLC examining the pricing behavior of Torchlight Mining Corp. ("TMC") in the U.S. specialty-minerals market relative to the pricing behavior of its two principal competitors, Ridgecrest Minerals Inc. ("Ridgecrest") and Pinnacle Earth Resources Ltd. ("Pinnacle"), over the period January 2015 through September 2022. The analysis was undertaken at the direction of counsel as part of the internal investigation into the conduct forming the basis of the DOJ Antitrust Division's ongoing grand jury investigation, Grand Jury No. GJ-22-0194 (N.D. Cal.).
The three product categories at the center of the DOJ investigation — high-purity feldspar, industrial talc, and rare-earth concentrates — were analyzed across the full time period. These three product categories represent the principal product lines of TMC's U.S. Specialty Minerals Division and, according to the DOJ's theory of the case, the subjects of alleged price-fixing activity among TMC, Ridgecrest, and Pinnacle.
To facilitate meaningful comparison, the analytical period was divided into three discrete windows:
(a) Pre-Conspiracy Baseline (January 2015 – December 2016): The period preceding the alleged commencement of coordinated pricing behavior, used as a benchmark for normal pricing correlations attributable to common market factors.
(b) Alleged Conspiracy Period (January 2017 – August 2021): The period during which, according to the DOJ's investigation and the internal investigation findings, coordinated pricing activity occurred among the three companies, principally through VaultChat encrypted communications between Gavin Holt (TMC Vice President, Pricing Strategy) and Carl Edgerton (Ridgecrest) and Kevin Tsang (Pinnacle), as well as through side meetings at NASMC quarterly conferences.
(c) Post-Conspiracy Period (September 2021 – September 2022): The period following the departure of Gavin Holt from TMC (effective September 30, 2021) and the approximate cessation of the alleged conspiracy, used to evaluate whether pricing dynamics reverted to pre-conspiracy patterns.
Key Finding. The analysis identifies statistically significant pricing correlations between TMC and its competitors during the alleged conspiracy period (January 2017 – August 2021) that are materially elevated compared to both the pre-conspiracy baseline and the post-conspiracy period. Across all three product categories, Pearson correlation coefficients between TMC and competitor price series increased substantially during the conspiracy period — in some instances more than doubling — and declined to levels approximating the pre-conspiracy baseline following the end of the alleged conspiracy. These patterns are consistent across multiple analytical approaches, including time-series correlation analysis, structural break testing, regression analysis controlling for common cost drivers, and an event study linking pricing changes to documented VaultChat communications.
It is important to note, however, that the analysis identifies pricing patterns consistent with coordinated behavior during the conspiracy period. Price correlation alone is not dispositive of conspiracy; parallel pricing can also result from legitimate market dynamics, particularly in concentrated industries with relatively homogeneous products and common input cost structures. To address this, the analysis employed regression models controlling for common cost drivers — including energy costs, transportation indices, raw material prices, and downstream demand indicators — and found that the elevated correlations during the conspiracy period persist even after accounting for these market-level explanations. Nevertheless, the statistical evidence should be considered in conjunction with the documentary evidence (VaultChat communications), witness testimony (Sandra Chu, Thomas Eriksen), and other investigative findings before drawing conclusions regarding the existence or scope of the alleged conspiracy.
This memorandum supports the internal investigation findings to be presented to the TMC Board of Directors on January 20, 2023, and will inform the cooperation proffer and penalty negotiation strategy. The analysis will be refined in a formal Hargrove expert report addressing affected-commerce calculations and penalty analysis, which is scheduled for delivery in connection with the anticipated penalty negotiations with the DOJ Antitrust Division.
II. Scope and Methodology
A. Engagement Background
Hargrove Forensic Advisory LLC ("Hargrove") was retained by Calderwood & Harkness LLP on July 1, 2022, under Kovel protections, to provide forensic economic advisory services in connection with the DOJ Antitrust Division's grand jury investigation of pricing practices in the U.S. specialty-minerals market. The engagement was authorized by TMC's General Counsel, Rebecca Aldridge, and the Audit and Compliance Committee of the TMC Board of Directors, pursuant to the Board's resolution of June 15, 2022, authorizing the internal investigation.
The engagement partner at Hargrove is Dr. Vanessa Tremblay, who maintains her principal office at 233 South Wacker Drive, Suite 4800, Chicago, IL 60606. Dr. Tremblay holds a Ph.D. in Economics from the University of Chicago and has over twenty years of experience in antitrust economics, including extensive experience with price-fixing investigations and the economic analysis of coordinated conduct in concentrated industries. The Hargrove team also includes three senior economists and two data analysts, all of whom have been identified on the privilege log maintained by Calderwood & Harkness.
Hargrove was tasked with three principal assignments in connection with this engagement:
(1) Economic analysis of pricing patterns in the specialty-minerals market, including time-series analysis, correlation analysis, structural break analysis, regression analysis, and event studies linking pricing events to documentary evidence of coordination;
(2) Affected-commerce volume assessment, including the calculation of TMC's affected commerce during the alleged conspiracy period for purposes of penalty determination under the United States Sentencing Guidelines and the DOJ Antitrust Division's penalty guidelines; and
(3) Support for NPA penalty negotiations, including the preparation of economic analyses and presentations for use in negotiations with the DOJ regarding the appropriate monetary penalty, taking into account cooperation credits, remediation efforts, and mitigating factors.
This memorandum addresses task (1). Tasks (2) and (3) will be addressed in subsequent work product, as described in Section IX below.
B. Data Sources
The analysis draws upon five principal categories of data, each of which is described below. The data collection, processing, and quality-assurance procedures are described in greater detail in Appendix 6.
TMC Internal Data. The primary source of TMC pricing data consists of published price lists, transaction-level sales data, customer contract pricing (including both spot and long-term contract prices), and internal pricing memoranda from TMC's U.S. Specialty Minerals Division covering the period January 2015 through September 2022. These data were sourced from two principal repositories: (a) the approximately 487,000 documents produced to date by TMC in response to Grand Jury Subpoena No. GJ-22-0194-SD-003, which requested all documents relating to pricing decisions, competitor contacts, and trade association activities for the U.S. Specialty Minerals Division; and (b) TMC's internal enterprise resource planning (ERP) system (Harborline Software) and customer relationship management (CRM) system (Pinnacle Health IT), from which transactional data were extracted and processed. The collection and forensic processing of TMC's electronic data was performed by Redstone Digital Forensics Inc. (Neil Ashworth, Project Director), which was separately retained by Calderwood & Harkness under Kovel protections for digital forensics support.
Competitor Pricing Data. Because Hargrove does not have direct access to the internal pricing records of Ridgecrest or Pinnacle, competitor pricing data were assembled from multiple sources. The primary source of competitor benchmark pricing is the quarterly pricing surveys published by the North American Specialty Minerals Council (NASMC), headquartered in Arlington, VA. NASMC publishes average transaction prices for high-purity feldspar, industrial talc, and rare-earth concentrates on a quarterly basis, disaggregated by grade and region. These benchmarks provide a reliable, if somewhat coarse, measure of market-level pricing trends. Supplemental competitor pricing data were obtained from industry trade publications, including Mineral Markets Weekly, Industrial Minerals Review, and Specialty Minerals Quarterly. In addition, TMC's own records contained competitor price sheets, bid tabulation documents, and competitor pricing intelligence collected by TMC's sales personnel in the ordinary course of business, which provided more granular, transaction-level insight into Ridgecrest and Pinnacle pricing for specific products and time periods.
Market Data. To control for common cost drivers and market-level factors that could independently produce pricing correlation, Hargrove assembled a comprehensive set of market and cost data, including: (a) energy cost indices (U.S. Energy Information Administration natural gas and electricity price indices); (b) transportation cost indices (Bureau of Labor Statistics Producer Price Index for rail and truck transportation of minerals); (c) mining consumables and raw material input cost indices (PPI for industrial chemicals, explosives, and mining equipment); and (d) downstream demand indicators, including the Federal Reserve's industrial production index for chemical manufacturing (NAICS 325) and for glass and glass products (NAICS 3272), which serve as proxies for demand in the end-use markets for the three product categories. These data were sourced from publicly available federal databases and supplemented by industry reports from NASMC and the U.S. Geological Survey.
VaultChat Communications. A critical corroborative data source consists of the pricing-related communications exchanged between Gavin Holt (TMC) and Carl Edgerton (Ridgecrest) and Kevin Tsang (Pinnacle) via the VaultChat encrypted messaging application. These communications were identified, recovered (to the extent technologically possible), and catalogued in the VaultChat Message Analysis Memorandum dated September 15, 2022, prepared by Calderwood & Harkness as internal work product. That memorandum identified at least 37 documented exchanges between March 14, 2017, and July 9, 2021, in which pricing information — including specific prices, planned price-change timing, and customer allocation — was shared between Holt and one or both competitor counterparts. The event study component of the present analysis cross-references the dates and content of these VaultChat exchanges with TMC and competitor pricing events.
Witness Interview Data. The analysis also draws upon information obtained through Calderwood & Harkness's internal investigation witness interviews, conducted under Upjohn warnings. Two witnesses provided information particularly relevant to the pricing analysis:
• Sandra Chu (Director, Pricing Analytics, TMC U.S. Specialty Minerals Division): Chu stated during her interview on August 22, 2022, that she received pricing-adjustment instructions from Gavin Holt on at least 12 occasions between approximately mid-2017 and mid-2021. Chu described these instructions as unusual because they directed specific price levels or price-change timing without reference to the normal cost-based pricing models employed by the Pricing Analytics group. The dates of these instructions, as identified from Chu's calendar entries and email records, were provided to Hargrove for use in the event study analysis.
• Thomas Eriksen (Regional Sales Manager, Western Region, TMC U.S. Specialty Minerals Division): Eriksen stated during his interview on September 2, 2022, that he attended four NASMC quarterly conferences during the conspiracy period and observed Gavin Holt meeting privately with individuals he identified as representatives of Ridgecrest and Pinnacle at conference social events. Eriksen could not provide details of what was discussed in these meetings but noted that pricing guidance from Holt often changed in the days following NASMC conferences.
C. Analytical Methodology
Hargrove employed five principal analytical approaches, each designed to examine a different dimension of the pricing relationship between TMC and its competitors. The methodologies are summarized below; additional technical detail is provided in Appendix 6.
Time-Series Analysis. Monthly average prices for each product category (high-purity feldspar, industrial talc, and rare-earth concentrates) were computed for TMC and each identified competitor (Ridgecrest and Pinnacle) over the full January 2015 through September 2022 period. Monthly averages were computed from transaction-level data for TMC and from the best available data for competitors, as described in Section II.B above. The resulting time series were examined visually and statistically for trends, seasonality, and co-movement patterns across the three analytical windows.
Correlation Analysis. Pearson product-moment correlation coefficients and Spearman rank correlation coefficients were calculated for each TMC-competitor price series pair within each of the three analytical windows (pre-conspiracy, conspiracy period, and post-conspiracy). Pearson correlations measure the linear association between two price series, while Spearman correlations measure the monotonic association without assuming linearity. Both measures were employed to ensure robustness. The statistical significance of each correlation coefficient was assessed using standard t-tests, and Fisher's z-transformation was used to test whether the correlation coefficients differed significantly across time periods.
Structural Break Analysis. Chow tests were applied to the TMC-competitor price series to determine whether statistically significant shifts in the pricing relationship occurred at the hypothesized break dates of January 2017 (alleged conspiracy start) and September 2021 (alleged conspiracy end). In addition, Bai-Perron multiple structural break tests were applied without imposing a priori break dates to determine whether the data independently identify structural breaks consistent with the conspiracy timeline. The Bai-Perron methodology allows for the endogenous determination of break dates and the number of breaks, providing an important check against the risk of confirmation bias in the choice of break dates.
Regression Analysis. Ordinary least squares (OLS) regression models were specified for each product category to control for common cost drivers and market-level factors that could independently explain pricing correlations. The dependent variable in each model is the monthly average TMC price for the relevant product category. Independent variables include the competitor's monthly average price, energy cost indices, transportation cost indices, raw material and mining input cost indices, and downstream demand indicators. Conspiracy-period dummy variables and interaction terms were included to test whether the degree of price co-movement changed during the alleged conspiracy period after controlling for common market factors. Heteroskedasticity-robust standard errors (White standard errors) were employed, and models were tested for serial correlation using the Durbin-Watson statistic.
Event Study. An event study analysis was conducted to examine the temporal relationship between specific pricing events — dates on which TMC published price changes — and three categories of potentially relevant events: (a) documented VaultChat communications between Holt and competitor counterparts, (b) NASMC quarterly conference dates, and (c) competitor price-change announcements. The event window was defined as ±15 calendar days from the TMC price-change announcement date. The analysis examined the frequency with which VaultChat communications, NASMC conferences, and competitor price changes fell within the event window and compared this frequency to what would be expected under a null hypothesis of random timing.
Significance Levels. All statistical tests were conducted at the 95% confidence level (α = 0.05), with robustness checks at the 99% confidence level (α = 0.01). Results are reported with exact p-values where available to permit assessment of the strength of the evidence at multiple significance thresholds.
D. Limitations and Caveats
The analysis is subject to the following limitations, which should be considered in evaluating the findings and their implications.
First, the analysis is based on data available as of November 1, 2022. At the time of this analysis, the DOJ had not yet issued the anticipated second subpoena (GJ-22-0194-SD-007), which is expected to request additional categories of documents, including communications between TMC and Ridgecrest/Pinnacle outside the VaultChat platform. Additional data produced in response to that subpoena or obtained through other investigative channels may affect the findings presented herein.
Second, competitor pricing data is incomplete for certain periods and products. Most significantly, Pinnacle's rare-earth concentrate pricing data prior to mid-2016 is unavailable from the sources described in Section II.B, which limits the effectiveness of the pre-conspiracy baseline analysis for the TMC-Pinnacle rare-earth pair. In addition, the competitor pricing data assembled from public and semi-public sources is generally less granular than the TMC transaction-level data, which may introduce measurement error that could attenuate estimated correlations.
Third, the VaultChat communications were only partially recovered due to the application's self-destructing message features and the deletion of message histories from certain devices. The 37 documented exchanges identified in the VaultChat Message Analysis Memorandum may understate the total number of pricing-related communications between Holt and his competitor counterparts. Consequently, the event study results may undercount the instances of temporal alignment between communications and pricing events.
Fourth, and most fundamentally, correlation does not establish causation. Parallel pricing can result from legitimate market dynamics, particularly in concentrated industries with relatively homogeneous products and common cost structures. While the regression analysis controls for observable common cost drivers, it is possible that unobserved factors — such as common strategic responses to demand shocks, simultaneous receipt of market intelligence, or oligopolistic interdependence — could contribute to elevated correlations without any agreement or coordination. The statistical evidence presented in this memorandum should therefore be evaluated in conjunction with the documentary and testimonial evidence of the alleged conspiracy, not in isolation.
III. Market Overview
A. Industry Structure
The North American specialty-minerals market for high-purity feldspar, industrial talc, and rare-earth concentrates is a highly concentrated market in which a small number of producers account for a substantial share of total U.S. supply. This concentrated structure is relevant to the pricing analysis because, as noted in Section II.D, some degree of pricing correlation is expected in concentrated markets even in the absence of coordination, due to oligopolistic interdependence, common cost structures, and the transparency of pricing in markets with few participants.
Torchlight Mining Corp. ("TMC"). TMC is a publicly traded Canadian mining and minerals company with its corporate headquarters in Toronto, Ontario, and its U.S. operational headquarters in Phoenix, Arizona. TMC's U.S. Specialty Minerals Division operates mines and six downstream chemical-processing facilities across the United States, with significant operations in Nevada, Utah, Georgia, and North Carolina. For FY 2021 (TMC's fiscal year ending December 31, 2021), the U.S. Specialty Minerals Division reported revenue of approximately USD 687 million, representing approximately 20.3% of TMC's consolidated revenue of approximately USD 3.38 billion (CAD 4.27 billion at the average 2021 exchange rate). TMC is the largest North American producer of high-purity feldspar and among the top three North American producers of industrial talc and rare-earth concentrates.
Ridgecrest Minerals Inc. ("Ridgecrest"). Ridgecrest is a Delaware corporation headquartered in Denver, Colorado. Ridgecrest is the second-largest North American producer of high-purity feldspar and industrial talc, with mining operations primarily in Colorado, Wyoming, and Montana, and processing facilities in Colorado and Texas. Ridgecrest reported FY 2021 revenue of approximately USD 1.92 billion across all business lines, though its specialty-minerals operations account for a smaller share of total revenue than TMC's corresponding division. Ridgecrest is a portfolio company of Marlowe Asset Management Peak Capital Partners LP, a Denver-based private equity firm that acquired a controlling interest in Ridgecrest in 2013. Ridgecrest's participation in the rare-earth concentrates market is more limited than its presence in feldspar and talc, reflecting the geographic concentration of its mining assets.
Pinnacle Earth Resources Ltd. ("Pinnacle"). Pinnacle is an Ontario corporation with its U.S. operations headquartered in Salt Lake City, Utah. Pinnacle is a top-five North American producer of rare-earth concentrates and specialty industrial minerals, with mining operations in Utah, Idaho, and British Columbia, and processing facilities in Utah and Quebec. For FY 2021, Pinnacle reported revenue of approximately CAD 1.45 billion (approximately USD 1.15 billion at the average 2021 exchange rate). Pinnacle's primary competitive overlap with TMC is in the rare-earth concentrates and industrial talc markets; its presence in the high-purity feldspar market is comparatively smaller but not negligible.
North American Specialty Minerals Council (NASMC). NASMC is an industry trade association headquartered in Arlington, Virginia. NASMC publishes quarterly pricing benchmarks for the specialty-minerals industry, organizes quarterly conferences attended by industry participants, and engages in advocacy on regulatory and trade matters affecting the industry. The quarterly conferences, which rotate among venues across the United States, are attended by executives, sales managers, and technical personnel from the member companies. The DOJ's investigation has alleged that side meetings at NASMC conferences served as one of the venues for the alleged coordination among TMC, Ridgecrest, and Pinnacle representatives.
Collectively, TMC, Ridgecrest, and Pinnacle account for a significant share of U.S. supply for each of the three product categories, though the precise market share figures are subject to confidentiality restrictions and vary by product category and year. Based on publicly available information and the data assembled by Hargrove, the three companies collectively account for an estimated 55–70% of U.S. supply for high-purity feldspar, 45–60% for industrial talc, and 50–65% for rare-earth concentrates during the relevant period. This level of concentration is consistent with the conditions under which coordinated pricing behavior is both theoretically more likely to emerge and more likely to be sustained.
B. Product Descriptions
High-Purity Feldspar. High-purity feldspar is a mineral used primarily in the manufacturing of ceramics, glass, and specialty coatings. It serves as a flux in glass and ceramic production, lowering the melting temperature of the batch and improving the durability and finish of the final product. The pricing of high-purity feldspar is influenced by several factors, including quality grade (particularly the level of iron and other impurities), silica content, particle size distribution, and regional transportation costs, which are significant given the bulk nature of the product and the geographic dispersion of mining operations and end-use customers. Pricing is typically quoted on a per-ton basis, with premiums for higher-purity grades. The market is characterized by a mix of long-term supply contracts with major industrial customers (ceramics manufacturers, glass producers) and spot sales to smaller purchasers.
Industrial Talc. Industrial talc is a hydrated magnesium silicate mineral used in a wide range of industrial applications, including paper manufacturing (as a filler and coating), plastics (as a reinforcing filler), paint and coatings (as an extender pigment), and personal-care products (as a base for talcum powder and cosmetics). Pricing is influenced by particle size, mineral purity, brightness, and end-use specifications, with different grades commanding different price points depending on the application. The industrial talc market is characterized by a moderate degree of product differentiation based on these quality attributes, though the underlying mineral is relatively homogeneous. Transportation costs are again a significant pricing factor.
Rare-Earth Concentrates. Rare-earth concentrates encompass a group of 17 chemically similar elements used in high-technology applications, including electronics, permanent magnets (for wind turbines and electric vehicle motors), catalysts (for petroleum refining), and defense applications (guidance systems, satellite components). Rare-earth pricing is inherently more volatile than feldspar or talc pricing due to the influence of geopolitical factors — particularly the dominant role of Chinese production in global rare-earth supply — supply constraints from environmental and permitting challenges, and the relatively small size and illiquidity of the spot market. Processing costs, which include complex chemical separation procedures, are also a significant pricing factor. The North American rare-earth market has grown in strategic importance during the period under analysis, driven by concerns about supply-chain security and the growing demand for permanent magnets in electric vehicles and renewable energy systems.
C. DOJ Alleged Affected Commerce
The DOJ Antitrust Division has estimated the total affected commerce across all three participants — TMC, Ridgecrest, and Pinnacle — at approximately USD 1.34 billion over the alleged conspiracy period (January 2017 – August 2021). This figure represents the aggregate U.S. sales of high-purity feldspar, industrial talc, and rare-earth concentrates by the three companies during the approximately four years and eight months of the alleged conspiracy.
TMC's share of affected commerce, based on the DOJ's preliminary estimates and Hargrove's independent analysis of TMC sales data, is approximately USD 478 million, representing approximately 35.7% of total affected commerce ($478M / $1,340M ≈ 35.7%). This figure encompasses all U.S. sales of the three product categories by TMC's U.S. Specialty Minerals Division during the conspiracy period, including both spot sales and sales under long-term contracts.
These affected-commerce figures are preliminary and subject to refinement in the formal affected-commerce analysis to be prepared by Hargrove in connection with penalty negotiations. The figures are relevant to this memorandum because they provide context for the magnitude of the pricing conduct under analysis and will form the basis for the DOJ's penalty calculation, as discussed in Section IX.
IV. Pricing Analysis Results
This section presents the detailed results of the pricing analysis for each of the three product categories. For each product, results are reported for each of the three analytical windows: the pre-conspiracy baseline (January 2015 – December 2016), the alleged conspiracy period (January 2017 – August 2021), and the post-conspiracy period (September 2021 – September 2022). Detailed data tables are presented in Appendix 1, and full correlation matrices are presented in Appendix 2.
A. High-Purity Feldspar
1. Pre-Conspiracy Period (January 2015 – December 2016)
During the pre-conspiracy baseline period, TMC and Ridgecrest feldspar prices exhibited a moderate positive correlation, with a Pearson correlation coefficient of approximately r ≈ 0.52. This level of correlation is consistent with the two firms operating in the same concentrated market, responding to common input cost fluctuations (particularly energy and transportation costs), and adjusting to similar demand conditions in the downstream ceramics and glass manufacturing sectors. The Spearman rank correlation for this pair was approximately ρ ≈ 0.49, confirming the moderate positive association observed in the Pearson measure.
The correlation between TMC and Pinnacle feldspar prices was weaker during the baseline period, with a Pearson r ≈ 0.38 and a Spearman ρ ≈ 0.35. This lower correlation reflects Pinnacle's comparatively smaller feldspar market share and the geographic concentration of Pinnacle's feldspar operations, which exposes Pinnacle to somewhat different regional cost and demand dynamics than TMC and Ridgecrest.
During this period, TMC's average feldspar price adjustments occurred approximately every four to six months, consistent with TMC's standard practice of semi-annual price reviews. Ridgecrest's price adjustments followed a broadly similar cadence — also approximately semi-annual — but the timing of adjustments appeared to be independently determined. There was no statistically significant lead-lag relationship between TMC and competitor price changes at weekly or monthly frequency during this period; that is, neither TMC's price changes reliably preceded Ridgecrest's nor vice versa, suggesting that the observed moderate correlation was driven by common market factors rather than price signaling or following behavior.
2. Alleged Conspiracy Period (January 2017 – August 2021)
The pricing relationship between TMC and Ridgecrest feldspar changed markedly during the alleged conspiracy period. The Pearson correlation coefficient between TMC and Ridgecrest feldspar prices increased from approximately 0.52 in the baseline period to approximately 0.91 during the conspiracy period, a difference that is statistically significant at the p < 0.001 level using Fisher's z-transformation test. The Spearman rank correlation increased correspondingly, from approximately 0.49 to approximately 0.88. This level of correlation is exceptionally high and substantially exceeds what would be expected based on common market factors alone, as confirmed by the regression analysis discussed in Section VII.
The TMC-Pinnacle feldspar correlation also increased during the conspiracy period, from a Pearson r ≈ 0.38 to approximately r ≈ 0.74 (p < 0.01). While this increase is less dramatic than the TMC-Ridgecrest increase, it is nonetheless statistically significant and reflects the broadening of the alleged coordination to include Pinnacle, particularly with respect to pricing in the western U.S. market where Pinnacle has a larger feldspar presence.
Beyond the magnitude of the correlation, the timing of price adjustments became highly synchronized during the conspiracy period. Hargrove identified 32 discrete price-change events for TMC feldspar during the conspiracy period (defined as published price adjustments of at least 1% from the prior posted price). In 28 of these 32 events, Ridgecrest adjusted its feldspar prices within a 14-day window of the TMC adjustment. This frequency of near-simultaneous adjustment is statistically anomalous when compared to the pre-conspiracy period, during which such tight synchronization occurred in only approximately 40% of price-change events. Moreover, in 19 of the 28 synchronized events, the magnitudes of the TMC and Ridgecrest price adjustments were within ±1.5 percentage points of each other, further suggesting coordinated rather than independently determined pricing.
Structural break analysis (Chow test) identifies a statistically significant break in the TMC-Ridgecrest feldspar price relationship occurring in the first quarter of 2017, consistent with the alleged conspiracy start date. Critically, this structural break aligns temporally with the first documented VaultChat exchange between Gavin Holt and Carl Edgerton, which occurred on March 14, 2017 — a communication in which, according to the VaultChat Message Analysis Memorandum, Holt and Edgerton discussed an upcoming feldspar price increase and the timing of its announcement. The structural break results are discussed in further detail in Section V.
After controlling for common cost drivers — energy costs, transportation costs, and raw material prices — through the OLS regression models described in Section VII, the residual correlation between TMC and Ridgecrest feldspar prices remains elevated, with a partial correlation of approximately 0.78. This finding indicates that common cost factors, while explaining a meaningful portion of the price co-movement, account for only a fraction of the observed synchronization during the conspiracy period. The remaining co-movement is attributable to factors not captured by the cost controls — a category that includes, but is not limited to, coordinated pricing behavior.
3. Post-Conspiracy Period (September 2021 – September 2022)
Following Gavin Holt's departure from TMC, effective September 30, 2021, and the approximate cessation of the alleged conspiracy in August 2021, the pricing relationship between TMC and Ridgecrest feldspar reverted toward the pre-conspiracy baseline. The Pearson correlation coefficient declined from approximately 0.91 during the conspiracy period to approximately 0.47 during the post-conspiracy period — a level closely approximating the pre-conspiracy baseline of 0.52. While the post-conspiracy period is relatively short (approximately 13 months) and thus the statistical precision of the correlation estimate is lower, the direction and magnitude of the decline are consistent with the cessation of coordinated pricing behavior.
The timing of price adjustments also diverged in the post-conspiracy period. In 5 of the 8 post-conspiracy feldspar price-change events identified for TMC, TMC and Ridgecrest announced price changes more than 30 days apart — a pattern more consistent with independent pricing decisions than with the tight synchronization observed during the conspiracy period.
Structural break analysis confirms a second statistically significant break in the TMC-Ridgecrest feldspar price series in Q3–Q4 2021, aligning with the end of the alleged conspiracy period and Holt's departure. This second break is discussed in further detail in Section V.
B. Industrial Talc
1. Pre-Conspiracy Period (January 2015 – December 2016)
The TMC-Ridgecrest price correlation for industrial talc during the pre-conspiracy baseline period was somewhat higher than for feldspar, with a Pearson r ≈ 0.58 and a Spearman ρ ≈ 0.55. This higher baseline correlation reflects several features of the industrial talc market: the number of significant market participants is smaller than in the feldspar market, the product is more homogeneous across suppliers (with fewer quality-based pricing differentials), and common cost factors — particularly energy and transportation — exert a proportionally greater influence on talc pricing. These structural characteristics mean that a higher degree of price co-movement is expected even in the absence of coordination.
The TMC-Pinnacle talc price correlation during the baseline period was Pearson r ≈ 0.44, reflecting Pinnacle's smaller but meaningful participation in the talc market, concentrated primarily in the western U.S.
2. Alleged Conspiracy Period (January 2017 – August 2021)
During the alleged conspiracy period, the TMC-Ridgecrest talc price correlation increased from the baseline level of approximately 0.58 to approximately 0.88 (p < 0.001). The TMC-Pinnacle talc price correlation increased from approximately 0.44 to approximately 0.79 (p < 0.001). Both increases are statistically significant and economically substantial, indicating a marked shift in the degree of price co-movement during the conspiracy period.
The industrial talc category is notable for the intensity of the documentary evidence linking pricing changes to competitor communications. Of the 37 documented VaultChat exchanges between Holt and Edgerton or Tsang, 15 specifically referenced talc pricing — making talc the most frequently discussed product category in the recovered communications. These 15 exchanges addressed a range of topics, including planned price increases for specific talc grades, the timing of price announcements to key customers, and the allocation of certain large industrial accounts between TMC and competitors.
The event study analysis (Section VI) reveals particularly strong temporal alignment between VaultChat exchanges and TMC pricing events for talc. On 9 of the 12 occasions on which Sandra Chu, TMC's Director of Pricing Analytics, received pricing-adjustment instructions from Gavin Holt (as identified through Chu's witness interview and confirmed by review of her calendar entries and email records), the timing of the instruction correlated within one to five business days with a VaultChat exchange involving Holt and a competitor counterpart. This pattern is consistent with a sequence in which Holt communicated with a competitor counterpart to agree upon a pricing action and then directed Chu to implement the agreed-upon change through TMC's pricing systems.
In addition to the pricing correlation findings, Hargrove's analysis of TMC transaction data revealed customer allocation patterns during the conspiracy period that are consistent with market allocation. Specifically, certain large industrial customers — including two Fortune 500 chemical manufacturers — exhibited unexplained shifts in purchasing patterns during the conspiracy period, with a reduction in competitive bidding and an increase in sole-source purchasing from TMC (or, in other cases, a shift away from TMC to Ridgecrest or Pinnacle), that reversed following the end of the alleged conspiracy. While the customer allocation analysis is not the primary focus of this memorandum, it is noted here because it provides additional corroborative context for the pricing findings.
3. Post-Conspiracy Period (September 2021 – September 2022)
The TMC-Ridgecrest talc price correlation declined to approximately Pearson r ≈ 0.50 during the post-conspiracy period, approaching the pre-conspiracy baseline of approximately 0.58. The TMC-Pinnacle talc price correlation declined to approximately Pearson r ≈ 0.41, closely approximating its pre-conspiracy level of 0.44. As with the feldspar analysis, the direction and magnitude of these declines are consistent with the cessation of coordinated pricing behavior and the reversion to competitive market dynamics.
C. Rare-Earth Concentrates
1. Pre-Conspiracy Period (January 2015 – December 2016)
Rare-earth concentrate pricing is inherently more volatile than feldspar or talc pricing due to the influence of geopolitical factors, global supply-chain dynamics (particularly the dominance of Chinese production), and the relatively small size of the North American spot market. This heightened volatility has implications for the interpretation of correlation coefficients, as higher price volatility can either increase or decrease measured correlations depending on the nature of the underlying shocks.
During the pre-conspiracy baseline period, the TMC-Pinnacle rare-earth price correlation was Pearson r ≈ 0.61 — the highest baseline correlation observed among any of the product-competitor pairs. This reflects the fact that Pinnacle is TMC's primary North American competitor in rare-earth concentrates, and the two companies' pricing is significantly influenced by the same global benchmark prices and supply-chain factors. The Spearman rank correlation for this pair was ρ ≈ 0.58.
The TMC-Ridgecrest rare-earth correlation during the baseline period was substantially lower, at Pearson r ≈ 0.33, reflecting Ridgecrest's comparatively limited presence in the rare-earth concentrates market. Ridgecrest's rare-earth operations are a smaller component of its overall business, and its pricing may be influenced by factors specific to its particular product grades and customer base.
An important data limitation applies to the rare-earth analysis: Pinnacle's rare-earth concentrate pricing data prior to mid-2016 is incomplete in the available data sources. As a result, the baseline analysis for the TMC-Pinnacle rare-earth pair effectively begins from July 2016, providing only approximately six months of pre-conspiracy data rather than the full two years available for other product-competitor pairs. This shorter baseline period reduces the precision of the pre-conspiracy correlation estimate and should be considered in interpreting the magnitude of the change between periods.
2. Alleged Conspiracy Period (January 2017 – August 2021)
During the alleged conspiracy period, the TMC-Pinnacle rare-earth price correlation increased from the baseline level of approximately 0.61 to approximately 0.85 (p < 0.001). The TMC-Ridgecrest rare-earth correlation increased from approximately 0.33 to approximately 0.62 (p < 0.05), a more modest increase consistent with Ridgecrest's smaller role in the rare-earth market.
The most pronounced coordination in rare-earth pricing appears to have occurred during the 2019–2020 sub-period, when global rare-earth supply disruptions — including Chinese export restrictions, environmental enforcement actions that curtailed production at several Chinese mines, and pandemic-related logistics challenges — created a pricing environment characterized by elevated uncertainty and rapid price movements. Hargrove's analysis indicates that the TMC-Pinnacle rare-earth correlation was particularly high during this sub-period (Pearson r ≈ 0.92 for 2019–2020 alone), suggesting that the alleged coordination may have been especially active during periods of market disruption, when participants had both greater incentive and greater opportunity to coordinate pricing responses.
After controlling for global rare-earth commodity indices — specifically, the global rare-earth price index and the London Metal Exchange rare-earth reference prices — the residual TMC-Pinnacle rare-earth correlation remains statistically significant, with a partial correlation of approximately 0.54. This indicates that the elevated correlation during the conspiracy period is not fully explained by common responses to global rare-earth market conditions, though global factors do account for a meaningful portion of the co-movement. The residual correlation is somewhat lower than for feldspar and talc, which is consistent with the greater influence of exogenous global factors on rare-earth pricing and with the less frequent VaultChat communications specifically addressing rare-earth pricing.
3. Post-Conspiracy Period (September 2021 – September 2022)
During the post-conspiracy period, the TMC-Pinnacle rare-earth correlation declined to approximately Pearson r ≈ 0.55, representing a partial but meaningful reversion toward the pre-conspiracy baseline of approximately 0.61. The fact that the post-conspiracy correlation (0.55) is slightly below the pre-conspiracy baseline (0.61) may reflect ongoing market turbulence in the rare-earth sector during late 2021 and 2022, including supply-chain adjustments and shifting demand patterns that have introduced new sources of idiosyncratic pricing variation.
The TMC-Ridgecrest rare-earth correlation declined to approximately Pearson r ≈ 0.29, closely approximating its pre-conspiracy baseline of approximately 0.33 and confirming the reversion of Ridgecrest's rare-earth pricing to a largely independent trajectory.
V. Structural Break Analysis — Detailed Results
The structural break analysis was designed to test whether the pricing dynamics between TMC and its competitors changed at specific points in time, and whether those change points align with the alleged conspiracy start and end dates. Two complementary methodologies were employed: the Chow test, which tests for a structural break at a specified date, and the Bai-Perron multiple structural break test, which endogenously identifies the number and timing of structural breaks without requiring the analyst to specify break dates in advance.
Chow Test Results. The Chow test was applied to each TMC-competitor price series pair, with hypothesized break dates at Q1 2017 (alleged conspiracy start) and Q3 2021 (alleged conspiracy end). The results are summarized as follows:
• TMC-Ridgecrest Feldspar: Statistically significant structural breaks identified at both Q1 2017 and Q3 2021 (p < 0.01 at both break points). The F-statistics for both breaks substantially exceed the critical values, indicating that the pricing relationship shifted materially at both the beginning and end of the alleged conspiracy period.
• TMC-Ridgecrest Talc: Statistically significant structural breaks identified at both Q1 2017 and Q4 2021 (p < 0.01 at both break points). The slight delay in the talc break relative to the feldspar break at the conspiracy end is consistent with the observation that certain long-term talc supply contracts may have delayed the reversion to competitive pricing following the cessation of coordination.
• TMC-Pinnacle Rare-Earth Concentrates: Statistically significant structural breaks identified at Q1 2017 (p < 0.05) and Q3 2021 (p < 0.05). The somewhat lower significance levels for the rare-earth series reflect the higher volatility of rare-earth pricing and the shorter effective baseline period, as discussed in Section IV.C.1.
• TMC-Pinnacle Feldspar and Talc: Structural breaks at the conspiracy start date are marginally significant (p ≈ 0.06–0.08) for both product categories, reflecting Pinnacle's smaller market share in these categories and the resulting weaker statistical signal.
• TMC-Ridgecrest Rare-Earth Concentrates: A structural break at Q1 2017 is marginally significant (p ≈ 0.09), consistent with Ridgecrest's limited rare-earth market presence.
Bai-Perron Test Results. The Bai-Perron multiple structural break test was applied to the full January 2015 through September 2022 TMC-Ridgecrest price series for both feldspar and talc. The results confirm the existence of two structural breaks in each series:
• Feldspar: The Bai-Perron test identifies break dates of approximately February 2017 (±2 months) and approximately October 2021 (±2 months). These endogenously determined break dates closely align with the conspiracy period boundaries derived from the documentary evidence.
• Talc: The Bai-Perron test identifies break dates of approximately March 2017 (±2 months) and approximately November 2021 (±3 months). Again, these dates are consistent with the conspiracy timeline, with the slightly later second break date for talc paralleling the Chow test finding discussed above.
The Bai-Perron test was also applied to the TMC-Pinnacle rare-earth series, but the results were less definitive, identifying one clear break in early 2017 but providing an ambiguous result for the conspiracy end date, likely due to the continued volatility in rare-earth markets during the post-conspiracy period.
Implication. The structural break analysis provides strong statistical evidence that the pricing dynamics between TMC and its principal competitors changed materially at the beginning and end of the alleged conspiracy period. The alignment between the endogenously determined break dates and the conspiracy timeline derived from documentary and testimonial evidence is striking and supports the inference that the elevated pricing correlations observed during the conspiracy period were not the result of a gradual or trend-driven change in market structure but rather reflected a discrete shift in pricing behavior — a pattern consistent with the commencement and cessation of coordinated pricing conduct.
Full statistical output for the Chow and Bai-Perron tests, including F-statistics, critical values, confidence intervals for break dates, and residual diagnostics, is presented in Appendix 3.
VI. Event Study — VaultChat Communications and Price Changes
A. Methodology
The event study analysis was designed to examine whether TMC pricing events — specifically, the dates on which TMC published changes to its posted prices for high-purity feldspar, industrial talc, or rare-earth concentrates — were temporally associated with documented VaultChat communications between Gavin Holt and Carl Edgerton (Ridgecrest) or Kevin Tsang (Pinnacle).
The analysis cross-referenced the dates of the 37 documented VaultChat exchanges identified in the VaultChat Message Analysis Memorandum dated September 15, 2022, prepared by Calderwood & Harkness, with TMC price-change announcement dates and competitor price-change announcement dates. The event window was defined as ±15 calendar days from the VaultChat exchange date — that is, a 31-day window centered on each documented communication. Within this window, Hargrove examined (a) whether TMC announced a price change, (b) whether any competitor announced a price change, (c) whether the price changes were in the same direction (i.e., both increases or both decreases), and (d) whether the magnitudes of the changes were similar (defined as within ±2 percentage points of each other).
In addition, the analysis examined the temporal relationship between the 12 pricing-adjustment instructions from Holt to Sandra Chu, as identified through Chu's witness interview and confirmed by review of her calendar entries and email records, and the dates of VaultChat exchanges.
B. Findings
The event study produced the following principal findings:
Of the 37 documented VaultChat exchanges between Holt and Edgerton or Tsang, spanning the period from March 14, 2017, through July 9, 2021, 29 occurred within 15 calendar days before or after a TMC or competitor price adjustment. The remaining 8 exchanges either fell outside the ±15-day event window or could not be reliably matched to a specific pricing event due to data limitations. The frequency with which VaultChat communications cluster around pricing events (29 of 37, or approximately 78%) is substantially higher than would be expected if the communications and pricing events were independently timed. Under a null hypothesis of random timing, given the average frequency of price-change events during the conspiracy period, the expected proportion of VaultChat exchanges falling within a ±15-day window of a pricing event is approximately 35–45%, depending on the product category.
In 22 of the 29 instances where a VaultChat exchange coincided with a pricing event, TMC and at least one competitor made pricing changes in the same direction (both increases or both decreases) and within a similar magnitude range (±2 percentage points) within the 15-day event window. This pattern of same-direction, similar-magnitude adjustments clustered around communication dates is consistent with the use of the VaultChat platform to coordinate the timing and magnitude of price changes.
The first documented VaultChat exchange, dated March 14, 2017, between Holt and Edgerton, provides a particularly instructive example. According to the VaultChat Message Analysis Memorandum, this exchange involved a discussion of an upcoming feldspar price increase and the proposed timing of its announcement. Within 10 days of this communication, both TMC and Ridgecrest announced feldspar price increases of approximately 4%, effective within days of each other. This first documented exchange and its associated pricing event are temporally consistent with the structural break identified by the Chow and Bai-Perron tests in Q1 2017.
The analysis of the 12 pricing-adjustment instructions from Holt to Chu shows statistically significant clustering around VaultChat exchange dates. Specifically, 9 of the 12 instructions fell within 5 business days of a documented VaultChat communication. The temporal sequence in most of these instances is consistent with the following pattern: Holt communicated with a competitor counterpart via VaultChat, and within 1–5 business days thereafter, Holt directed Chu to implement a specific pricing adjustment — often specifying both the magnitude and timing of the change — without reference to the normal cost-based pricing models that Chu's Pricing Analytics group would typically employ to justify pricing decisions. Chu noted in her interview that this departure from standard pricing-review procedures struck her as unusual, though she did not raise the matter with anyone other than Holt at the time.
C. NASMC Conference Correlation
In addition to the VaultChat communications, Hargrove examined whether TMC pricing events were temporally associated with NASMC quarterly conferences, which the DOJ alleges served as venues for in-person coordination among TMC, Ridgecrest, and Pinnacle representatives.
Thomas Eriksen, TMC's Regional Sales Manager for the Western Region, attended four NASMC quarterly conferences during the conspiracy period. Conference venues included Arlington, VA (NASMC's headquarters city, where several conferences were held); Scottsdale, AZ; and Orlando, FL. Gavin Holt attended additional NASMC conferences during the conspiracy period, as confirmed by TMC travel and expense records reviewed by Calderwood & Harkness. The total number of NASMC conferences attended by TMC personnel during the conspiracy period has not been fully determined, as TMC's travel records are still being analyzed, but at least six conferences are confirmed to have been attended by Holt or other TMC personnel.
The event study reveals elevated clustering of price-change events in the two to four-week periods following NASMC quarterly conferences. Specifically, during the conspiracy period (January 2017 – August 2021), 14 of 32 feldspar and talc price-change events (approximately 44%) occurred within two to four weeks of an NASMC conference. By comparison, during the pre-conspiracy baseline period (January 2015 – December 2016), only 3 of 14 price-change events (approximately 21%) occurred within this window relative to NASMC conference dates. The difference in these proportions is suggestive, though not by itself statistically significant at the 95% confidence level given the relatively small sample sizes, particularly in the pre-conspiracy period.
While NASMC conference attendance alone is not evidence of coordination — trade association conferences are a legitimate and commonplace feature of the specialty-minerals industry — the statistical pattern of post-conference price changes, combined with Eriksen's testimony regarding Holt's meetings with competitor representatives at conferences and the timing of subsequent pricing guidance changes, is consistent with the DOJ's allegation that side meetings at NASMC conferences served as one of the mechanisms through which pricing coordination was arranged and maintained.
VII. Regression Analysis — Controlling for Market Factors
A. Model Specification
To distinguish between pricing correlation attributable to common market factors and pricing correlation potentially attributable to coordinated behavior, Hargrove estimated a series of OLS regression models for each product category. The models were specified as follows:
Dependent Variable: Monthly average TMC price for the relevant product category (high-purity feldspar, industrial talc, or rare-earth concentrates), expressed in U.S. dollars per metric ton.
Independent Variables:
(1) Competitor Monthly Average Price: The monthly average price for the same product category for the relevant competitor (Ridgecrest or Pinnacle), used to measure the degree of price co-movement.
(2) Energy Cost Index: The U.S. Energy Information Administration's composite energy cost index for industrial users, reflecting the cost of natural gas and electricity used in mining and processing operations.
(3) Transportation Cost Index: The Bureau of Labor Statistics' Producer Price Index for rail and truck transportation of minerals, reflecting the cost of delivering product from mine or processing facility to customer.
(4) Raw Material/Mining Input Index: A composite index constructed from PPI sub-indices for industrial chemicals, explosives, and mining equipment, reflecting the cost of key mining and processing inputs.
(5) Downstream Demand Indicators: The Federal Reserve's industrial production index for the relevant downstream sector(s) — chemical manufacturing (NAICS 325) for talc and rare-earth concentrates, and glass and glass products (NAICS 3272) for feldspar — used as a proxy for end-use demand conditions.
Dummy Variables and Interaction Terms:
• A binary dummy variable for the conspiracy period (equal to 1 for January 2017 – August 2021 and 0 otherwise), capturing any level shift in the TMC-competitor pricing relationship during the conspiracy period.
• An interaction term between the conspiracy-period dummy and the competitor price variable, capturing any change in the degree of price co-movement (i.e., the slope of the TMC-competitor pricing relationship) during the conspiracy period.
Estimation Method: All models were estimated by ordinary least squares with heteroskedasticity-robust standard errors (White standard errors). The Durbin-Watson statistic was computed for each model to test for serial correlation in the residuals. Where serial correlation was detected, the models were re-estimated with Newey-West standard errors to ensure valid

PRICING DATA ANALYSIS MEMORANDUM

Record

Parties

Dates

Amounts