New York – The SEC recently released an order relating to three AXA Rosenberg entities that has interesting implications for quantitative researchers:
The SEC’s order instituting administrative proceedings against the firms found that senior management at BRRC and ARG learned in June 2009 of a material error in the model’s code that disabled one of the key components for managing risk. Instead of disclosing and fixing the error immediately, a senior ARG and BRRC official directed others to keep quiet about the error and declined to fix the error at that time.
“To protect trade secrets, quantitative investment managers often isolate their complex computer models from the firm’s compliance and risk management functions and leave oversight to a few sophisticated programmers,” said Robert Khuzami, Director of the SEC’s Division of Enforcement. “The secretive structure and lack of oversight of quantitative investment models, as this case demonstrates, cannot be used to conceal errors and betray investors.”
The SEC additionally charged BRRC with failing to adopt and implement compliance policies and procedures to ensure that the model would work as intended.
The SEC found that the error, which was introduced into the model in April 2007, was eventually fixed for all portfolios. However, knowledge of the error was kept from ARG’s Global CEO until November 2009. ARG then conducted an internal investigation and disclosed the error to SEC examination staff in late March 2010 after being informed of an impending SEC examination of ARIM and BRRC. ARG disclosed the error to clients on April 15.
The SEC’s order further found that ARG, BRRC, and ARIM made material misrepresentations and omissions about the error to ARIM’s clients. The firms failed to disclose the error and its impact on client performance, attributed the model’s underperformance to market volatility rather than the error, and misrepresented the model’s ability to control risks. BRRC did not have reasonable compliance procedures in place to ensure that the model would assess certain risk factors as intended. The coding process for the model represented a serious compliance risk for BRRC and its clients because accurate coding is required for the model to function properly and in the manner represented to clients.
A recent memo on this order from the law firm Cadwalader outlines exactly how the violations took place in this case:
The Adviser developed computer code for a quantitative investment model (the Model) used by
its affiliated advisers to manage client portfolios. In June 2009, an employee of the Adviser
discovered an error in the Models computer code. This code had been put into service in April
2007 and the discovered error was characterized by the Order as having, in effect, eliminated one
of the key components in the Model for managing risk. See paragraph 2 of the Order. The
employee brought this error to the attention of a person who was a senior official at both the
Adviser and the Advisers parent. In response, the employee who had discovered the error was
directed to keep quiet about it and to not fix it at that time. The error was fixed eventually in the fall
of 2009 and, at about that same time, the error was disclosed to the parents CEO by an employee
of the Adviser who felt compelled to do so. Id.
After learning of the error, the parent company conducted an internal investigation that concluded in
mid-March 2010. The Adviser also obtained advice of external legal counsel concerning its
obligation to disclose the error. Id. A few weeks after the conclusion of the internal investigation,
the parent company disclosed the error to SEC examination staff. Significantly, this disclosure
came after the parent was informed by the SECs Office of Compliance Inspections and
Examinations of an impending examination of the Adviser and other affiliated entities.
Clients were informed of the error a few weeks later on April 15, 2010. Prior to such disclosure,
the Adviser had misinformed clients about the ability of the Model to control risk. Indeed,
according to the Order, even after the error was discovered, clients were falsely informed that
underperformance of client portfolios had resulted from market volatility and other factors unrelated
to the error.
The most interesting implication of the SEC order comes from two quotes given by SEC officers:
“Quant managers must be fully forthcoming about the risks of their model-driven strategies, especially when errors occur and the models don’t work as predicted,” said Bruce Karpati, Co-Chief of the Asset Management Unit in the SEC’s Division of Enforcement.
Rosalind R. Tyson, Director of the SEC’s Los Angeles Regional Office, added, “Quant managers need to ensure that their compliance policies and procedures are tailored to the risks of their model’s strategies, and that compliance personnel are integrated into the development and maintenance of their investment models.”
These quotes imply two major conclusions for the way in which quantitative funds operate: that their marketing and client communication languages be circumspect about the actual risks of their model-driven strategies, and that compliance personnel oversee the development and maintenance of models to ensure that risks are not hidden from clients.
The latter implication may seem at first glance an unrealistic expectation – compliance personnel typically do not have the quantitative background that would be required to be intimately involved with model development. It seems that the real challenge for compliance officers will be to put in place a risk identification process that enables them to detect risks within quantitative models that the developers or portfolio managers directly involved with the model would prefer to hide. Some might have thought that quantitative research would be relatively unaffected by the compliance concerns (unrelated to this case) currently affecting expert networks and channel checkers. However, this order shows that quantitative firms may well have to re-assess their internal compliance policies and procedures.
Links of interest: