The recent enforcement action by the Securities and Exchange Commission (“SEC”) against Aegon USA Investment Management (“AUIM”) and three affiliated companies provides helpful guidance to investment advisers who employ quantitative models to manage their funds1  Specifically, the order highlights the need for investment advisers utilizing quantitative models to have the proper compliance policies in place to ensure the models are working as intended and to make sure that they are properly disclosing the risks of using such models in their marketing. 

Background2
According to the SEC’s Cease and Desist Order, AUIM’s quantitative models were developed by an inexperienced analyst in 2010 (a fact which was never disclosed to investors). Neither AUIM nor the analyst took any action to ensure the accuracy of the models before they were implemented - no formal models were employed by the analyst to confirm the accuracy of the models, and AUIM did not provide any oversight during development to confirm they worked as intended. Although AUIM identified potential risks with using the models by 2011, including via an internal audit highlighting the lack of controls and testing around the model, it went ahead with launching products employing the model. While AUIM began to take steps to adopt and implement a formal validation process, it continued to launch new products using the unvalidated models. It took until July 2013 for AUIM to create and implement a written model-validation policy. Ultimately, when the models were tested in the summer of 2013, more than 50 errors were discovered. 

AUIM and the other advisers named in the Order marketed the products and their use of the quantitative models without disclosing the risks associated with using such models to investors. The respondents marketed the available products by underscoring the benefits of using such a model, highlighting the “emotionless” and “model-driven” investment process used to manage the products. Such marketing implied to investors that the models worked as intended, despite respondents’ lack of reasonable belief that this was the case. After learning of the errors in the models in March 2014, the advisers failed to disclose this to investors. Instead, they stopped using the models and revised the prospectus and other marketing materials to say that the products “may” use a quantitative model or that they use “a combination of qualitative and quantitative factors.”

As a result of this conduct, the respondents were found to have violated the general fraud provisions of the Investment Adviser Act, as well as Rule 206(4)-1(a)(5) for publishing advertisements with untrue or misleading statements and Rule 206(4)-7 for failing to adopt and implement written policies and procedures.  To resolve the matter, respondents paid about $61 million in disgorgement and pre-judgment interest and another $36 million in civil penalties.

Lessons Learned
Investment advisers who develop their own quantitative model or employ those developed by others need to ensure they take certain steps to keep out of the SEC’s crosshairs.

1. Have formal controls and procedures to confirm models function as intended.  

This involves maintaining a written validation policy that includes conducting an independent validation of modeling results before launching products as well as periodically thereafter. Implement procedures regarding testing, approval and documentation of changes to models to ensure model development is controlled,3  and follow up on internal audits by adopting or updating policies and procedures reasonably designed to address identified risks.

2. Develop a formal policy regarding when the portfolio manager can trade on a discretionary basis outside the model results.  

The policy should require that a portfolio manager’s discretion to depart from model results must be defined, monitored and documented. If the portfolio manager can depart from the model results, do not represent in disclosures that the product is “emotionless” or “free from manager bias” or that “decisions will be made as dictated by the model.”

3. Inexperienced personnel must be provided guidance, training, and/or oversight while developing models and confirming they work as intended. No one should be an island.  

There should also be a business continuity plan in place to ensure back-up personnel with sufficient knowledge and involvement are available to adequately administer the model when key personnel are unavailable.

4. Implement procedures to ensure necessary disclosure.  

Disclosures should include (i) the person(s) involved in the day-to-day management of the model (and if such person(s) is inexperienced, disclose that fact as well); (ii) that the success of the model depends in part on the portfolio manager; (iii) risks associated with the use of models; and (iv) material changes to the model, discovered errors and the discontinued use of models that don’t function. 

5. Finally, do what you say will do.  

Do not be asleep at the switch and ignore your responsibilities, especially those you have expressly agreed to carry out.

Additional Resources
For a more in depth conversation on the SEC’s recent focus on internal controls and disclosures under the Investment Advisers Act, we invite you to join Dorsey & Whitney LLP’s Private Funds Symposium on September 26, 2018.  One of our six discussion panels will focus on Private Fund Investment Adviser Regulation and Enforcement. For more information, please contact Genna Garver and Kimberly Frumkin.


1 In the Matter of Aegon USA Investment Management, LLC, Admin. Proc. File No. 3-18681 (Aug. 27, 2018). 
2 This represents a summary of the charges and findings most pertinent to the lessons learned for use of quantitative models.
3 See also In the Matter of AXA Rosenberg Group, LLC, Admin. Proc. File No. 3-14224 (Feb. 3, 2011) (finding a violation of Rule 206-4(7) in part because the adviser “failed to conduct sufficient quality control over the coding process before putting that model into production”).