‘But-For’ Modeling in the Stock Lending Class Action Settlements

In an earlier piece about the upcoming claims administration for the $581M+ partial settlement of the antitrust class action involving stock lending, I suggested the significant time lag we’re seeing between preliminary approvals and the still-to-be-announced date for claims submissions may relate to complexities in formulating a distribution plan.[1]

This article elaborates on these challenges, discussing how the distribution plan will likely have a ‘but for’ recognized loss (RL) formula, and why this methodology will cause a lengthy claims administration process and delay distribution payments.



The stock lending market is largely over the counter (OTC).  There is no central market for participants to engage in direct exchanges or to obtain real-time pricing information.  The decentralized nature of this market results in high ‘search’ costs, creating profit opportunities for the defendant banks acting as agents and intermediaries for borrowers and lenders. The complaint alleges these banks resisted new entrants to the market and the adoption of technologies that would have reduced search costs and did so to protect their spreads or mark-ups on stock lending transactions.


For class certification, plaintiffs’ experts modeled class-wide harm.

In support of their proposed class for litigation (the Litigation Class), the plaintiffs engaged experts to opine on how the introduction of competition to the stock lending market – including a transition to direct electronic exchanges and data companies offering greater pricing transparency – would have benefited share borrowers and lenders.

In expert jargon, absent alleged anti-competitive conduct by the defendant banks, the market would have shifted from a bilateral trading system, where share borrowers and lenders trade through agents or intermediaries, to a multilateral trading system in which they simultaneously obtain competing bids or offers, often anonymously.

They concluded that by January 1, 2012, at the latest, the emergence of competitors offering electronic trading platforms, central clearing services, and post-trade pricing data would have successfully facilitated direct dealings between stock borrowers and lenders and significantly reduced intermediary profits for the defendant banks.

In short, stock borrowers would have paid less, and lenders would have received more, as the banks were either forced to reduce their spreads or suffer disintermediation.  Plaintiffs argued that class certification was appropriate because damages could be determined on a class-wide basis, using common formulas and information from the defendant banks and markets, rather than by individualized determinations based on facts specific to each participant.


The distribution plan formulas will require similar ‘but for’ modeling.

The settlements will put the plaintiffs’ damages theories to practice.  We expect the distribution plan will use a ‘but for’ approach for loss estimation, with RL formulas estimating inflated spreads over time. To estimate the latter, experts will have to make assumptions about the when and to what extent direct trading and data transparency would have been adopted in the market absent the alleged anti-competitive conduct, thereby reducing borrowing costs and lender returns.

At its core, the damage formulas will compare real-world pricing against estimated pricing in this hypothetical ‘but-for’ world.  For example, short seller damages at any given point during the class period will likely be calculated as the loan prices the Prime Broker defendants charged borrowers less the modeled ‘but for’ prices they would have paid absent the alleged misconduct.  There will be similar formulas for share lenders, with damages equaling the difference between their actual compensation from the banks and the ‘but for’ compensation they would have received if competition existed.

Presumably, claimant losses will be greater later in the class period.  At that point, absent cartel behavior, competition would have been more firmly entrenched, reducing search costs more.  Estimated compensation to lenders will also vary depending on the types of shares involved.  Anti-competitive forces would have had a greater impact on harder-to-borrow shares than easier-to-locate stocks.

Modeling the years from 2017 to 2023 will be particularly challenging.  As noted in my earlier piece (footnote 1), the parties did not exchange discovery on events after 2017, making it difficult for the experts to determine the actual price, quantity, and other terms the defendant banks charged.

Finally, the experts will have to assess whether absent the banks’ alleged misconduct, the spreads would have declined over time anyway, as stock lending participants sought ways to trade on better negotiated terms or via alternative methods.  They will also have to account for the broader economic forces that drive overall volumes of stock lending activity.

In short, to ensure claimants receive fair compensation, the experts will have to construct a complex, hypothetical world, with many assumptions about what could or would have been.  Creating such ‘but for’ models take time and resources, explaining the late start to this administration.



Challenges around the ‘but for’ modeling of class damages will also extend the time necessary to complete claims processing and make payments.  Like the earlier Forex settlement, which also had broad eligibility, we expect this administration will take more time than the typical antitrust case.  Assuming class notices are sent in the first quarter of 2024, distribution payments will likely not occur until the end of 2025 at the earliest, and more likely not until 2026.

[1] Michael Lange, Esq., Stock Lending Settlements: If “Interested”, Capitalize Now for Returns Later (October 30, 2023)


Related Content