Good morning, this is Ed Blount and I am speaking to you from the Center for the Study of Financial Market Evolution here in Washington, D.C. I've been asked by my good friends at the Risk Management Association, RMA, just up the road in Philadelphia, to offer some thoughts on "how data-based models can be used to change the negative views of financial markets that are held by some bank customers and regulators, especially in the wake of the pandemic." So, that is an interesting question.
I'm going to approach the answer in two parts:
First, I'll talk about the nature of models that are increasing regulatory and litigation risk following the pandemic; and then,
Second, I'll talk about how to deal with those increased regulatory and litigation risks, based on modeling precedents using newer technologies.
As a preamble, and as the reason I've been asked for comment, I have been engaged as a testifying expert [1] over the last 10 years in more than a dozen class-action lawsuits or civil litigation matters; as well as regulatory enforcement actions for the Department of Labor and the Securities and Exchange Commission. In fact, I've testified before all three branches of the US federal government. So, that's the credentials part. Now, let's talk about models.
Deconstructing Negative "Views" to Create Rebuttal Models
Negative views are formed because someone has a belief that is based upon a structured proposition and a set of facts that are assumed to correspond with reality. Philosophers call this the "Correspondence Theory of Truth." There's a very good description available for free on the Stanford University website, in their "Encyclopedia of Philosophy."
I think it should be no surprise that, in our modern world, we call these theories of truth "models." And, if all financial modeling relies on pattern recognition, as we were reminded in a recent RMA podcast, then to change a negative view, we have to employ a model that corresponds to a truth that is derived from a positive pattern of propositions. In the world of financial litigation, we call that a "rebuttal."
POST-COVID REGULATORY AND LITIGATION RISKS
So, what is the greatest risk following the pandemic? I would say that it is the relative (a) ambiguity and (b) paucity of information available to defend banks against charges of "Breach of the Public Trust," as implied within the policy sphere that is popularly called ESG (environmental, social, and governmental policies).
We'll look at two proposed models that, in my opinion, will cause difficulty for banks and their customers over the next several years.
Negative ESG-derived Views of Securities Finance
I have had to deal with flawed academic papers in my role as a defense expert for banks or asset managers. Often enough, these papers are cited by the plaintiffs' attorneys who are filing charges of malfeasance [4] or, in some cases, market manipulation.
- 1. In the case of the negative academic paper [5] we're using as an example, charges were made that (a) banks were aiding index fund managers in neglecting their fiduciary duty to exercise their governance obligations and privileges as portfolio owners of many securities, by (b) lending those securities out in order to gain additional income and improve their tracking errors.
Now, the typical response to that charge is based on articulated voting policies included either in the documents creating the fund or in the trust that underlies it. Unfortunately, a lot of the available data that's being used for rebuttal is so limited that it lends itself to challenge by plaintiffs' attorneys.
- 2. The second critical model represents a significant regulatory risk, and is being used by ESMA to energize the European Parliament's call for a review of cross-border securities loans, that are being used (allegedly) to cheat the German and Danish governments out of an estimated 50 billion Euros in otherwise-entitled tax revenues.
MODELING ESG RISK FOR REGULATION AND LITIGATION
These are serious charges. Each represents a potential "Betrayal of the Public Trust," which goes to the S within ESG, as well as to the G.
There are strong challenges that can be brought against the structural propositions that are used in the (a) academic paper, as well as in the (b) basic model for ESMA's support of the intrusion into the investment process that would be represented by audits of cross-border securities loans.
Both of these negative models attack the field of Securities Finance, which is essential to the liquidity and price discovery functions of the global fixed income and equity markets, mostly through their support for short sales, as well as hedging by institutions.
So, how would a bank or an asset manager who's being challenged respond to those charges? Well, actually, we have a precedent to consider.
Spiking the Negative ESG Views
Several years ago, I was tasked with the response to a series of academic papers that alleged, in great detail and with great quantitative support, that (a) hedge funds were borrowing shares in the securities finance and lending markets in order to (b) gain control of the vote in an Annual General Meeting, determining (c) whether a proposed tender, merger or other corporate action, would be (d) approved by the shareholders. [8]
The argument was made by the academics that, based on the data they had collected from one custodian and one prime broker, that spikes across the proxy record date presented evidence of a theory that the hedge funds were moving before the record date in order to build up their positions. [9]
We at CSFME worked with RMA and the broker-dealer trade group, SIFMA, to collect and analyze more than 800 million loan records, and then to produce a white paper. In turn, the data was anonymized, encrypted and turned over to a team of academics that studied the data and concluded that, in fact, this was not evidence of vote manipulation, but rather it was further evidence that lender-customers of banks were recalling their loans. [10] So, it was a positive outcome from the spike. And that is the way that we responded. [11]
I believe that's the best way to respond to arguments that challenge the compliance of managers or banks with their stated policy goals or investment strategies.
Building DLT Models to Change Negative Views
Let's deal with the ESMA proposal to establish audits of cross-border loans by the European Union. The rebuttal opportunity there is to use the new technology that's already been developed -- distributed ledger technology -- to respond to charges of complicity in avoiding withholding taxes on cross-border dividend payments.
Unfortunately, the structure of the increased post-crisis regulatory disclosures that are available for rebuttal in securities finance is such that only positions are really being disclosed. It is true that loan data is being mined for those disclosures, but the objective of that data collection process is to identify excess leverage.
As a result, it is not easy to reconstruct that loan data into the end-to-end mapping that would be required to demonstrate that the lending spikes are indeed evidence of benign trading activity, not manipulation of the markets. In fact, it would be necessary to recreate what CSFME did before, that is, create a loan-level industry model to demonstrate that lenders, asset managers, agents, and borrowers (as represented by the prime brokers) are, in fact, complying with their disclosed policies about proxy voting.
That is potentially a big project. But the technology exists to get it done. [12]
In my opinion, that loan level model would not only be available for response to regulatory charges such as those being proposed by ESMA, but would also be available to courts in support of summary dismissal motions for any complaint brought that would be based, even partly, on the arguments embodied within the critical academic paper described above.
Hoisted on their own Petards (Spikes)
This is a different way to approach a model. It's not a credit or market risk model. It's a litigation or regulatory risk model. There are similarities in that their structures are both based on propositions and data. But the way to approach the rebuttal is not to try forecasting an outcome -- or to absolutely destroy an opponent's argument -- but rather to recast their own evidence in a positive view.
That's the way you change someone's opinion. You're not going to be able to argue methodology with them. You're not going to argue the facts because they're going to have a different interpretation of those same facts.
What you have to do is identify the common areas that you agree on and structure the rebuttal to reinterpret that set of facts.
In the cases that I mentioned before, we all agreed that there were spikes in the activity patterns being recognized by the models. But what we did at the time, and which should be done in the future, was to recast those spikes as evidence of a positive process: compliance with stated policies.
Thank you for listening and I hope everyone has a very good outcome from the end (hopefully) of the pandemic.
Originally posted here, as, "The Impact of Data-based Models on Financial Markets," by the Risk Management Association, June 7, 2021.
Notes:
1. Edmon W. Blount, Eric B. Poer, Tiko V. Shah, "Securities Finance Disputes," Chapter 30, The Litigation Services Handbook: The Role of the Financial Expert, 6th Edition, Wiley, 2017, pp 30.1 - 30.20, at https://www.amazon.com/Litigation-Services-Handbook-Financial-Expert/dp/1119166322
2. Stanford University Encyclopedia of Philosophy, https://plato.stanford.edu/entries/truth/#ReaTru
3. RMA Podcasts, “The Impact of Covid-19 on Modeling,” November 19, 2020, https://soundcloud.com/user-524270410/the-impact-of-covid-19-on-modeling.
4. Malfeasance, defined as "Evil-doing; the doing of that which ought not to be done; wrongful conduct, especially official misconduct; violation of a public trust or obligation; specifically, the doing of an act which is positively unlawful or wrongful, in contradistinction to misfeasance, or the doing of a lawful act in a wrongful manner." American Heritage Dictionary, at https://www.wordnik.com/words/malfeasance.
5. Hu, Edwin and Mitts, Joshua and Sylvester, Haley, The Index-Fund Dilemma: An Empirical Study of the Lending-Voting Tradeoff (December 22, 2020). NYU Law and Economics Research Paper No. 20-52 , Columbia Law and Economics Working Paper No. 647, Available at SSRN: https://ssrn.com/abstract=3673531 or http://dx.doi.org/10.2139/ssrn.3673531
6. ESMA, Final Report on Cum Ex and Other Multiple Withholding Tax Reclaim Schemes, Sept. 23, 2020. https://www.esma.europa.eu/document/final-report-cum-ex-and-other-multiple-withholding-tax-reclaim-schemes
7. Ibid.
8. Hu, Henry T. C. and Black, Bernard S., The New Vote Buying: Empty Voting and Hidden (Morphable) Ownership. As published in Southern California Law Review, Vol. 79, pp. 811-908, 2006, University of Texas Law, Law and Econ Research Paper No. 53, Available at SSRN: https://ssrn.com/abstract=904004
9. Christoffersen, Susan E. and Geczy, Christopher Charles and Reed, Adam V. and Musto, David K., Vote Trading and Information Aggregation (January 2007). AFA 2006 Boston Meetings Paper, Sixteenth Annual Utah Winter Finance Conference, ECGI - Finance Working Paper No. 141/2007, Available at SSRN: https://ssrn.com/abstract=686026 or http://dx.doi.org/10.2139/ssrn.686026
10. Moser, Shane and Van Ness, Bonnie F. and Van Ness, Robert A., Securities Lending Around Proxies: Is the Increase in Lending Due to Proxy Abuse, or a Result of Dividends? (December 6, 2011). Available at SSRN: https://ssrn.com/abstract=1969051 or http://dx.doi.org/10.2139/ssrn.1969051
11. CSFME, Borrowed Proxy Abuse: Real or Not, October 2010. https://www.sec.gov/comments/s7-14-10/s71410-202.pdf
12. CSFME, “Systems Experts Set the Bar for Blockchain in Securities Finance,” Feb. 19, 2019. https://csfme.org/Full_Article/category/all/systems-experts-set-the-bar-for-blockchain-in-securities-finance
13. See, note 4, supra.