Debiasing in action

With increasing pressure on the active fund management industry to deliver returns for its clients, more work needs to be done to understand where fund managers add value and where they can improve. One way to address this issue is to understand the decision-making biases that affect investors, the value such biases destroy, and what to do to limit the effect they have on investment decisions.

Many asset management firms are already becoming alert to the impact of decision-making biases on fund performance. Working with analytics experts and behavioral scientists, some firms are applying machine-learning algorithms to their historical data to uncover clusters of suboptimal investment decisions across all asset classes. Upon close examination, many suboptimal decisions turn out to be the results of consistent bias.

Debiasing in action

One asset management firm examined two of its funds, analyzing trades and processes—as well as the emotions that the fund managers reported experiencing at the time of the suboptimal decisions—for signs of bias. They found that more than 35 percent of fund managers’ selling timing decisions were influenced by biases, resulting in significant value being left on the table. Our own experience and academic research show that the endowment effect—which leads fund managers to sell too late—destroys 150 bps per year for investors.

Selling too late

Having gone through the debiasing process, the first fund manager found it “helped by analyzing historic investing activity to enable me to understand where I and the team had done well and where we hadn’t. However, the project went beyond just the data and started to help us understand the potential behavioral drivers behind our decisions.” This enabled the fund to adjust and improve their process to produce better results for their clients.

The results showed that the fund manager had held on to 20 percent of positions for too long under the influence of anchoring and risk aversion. Both these biases are examples of stability biases, which relate to the desire to stick to what is already there, or known.

When considering which debiasing methodologies to use to reduce the effect biases have on performance, investors need to take their personalities, team culture, and existing decision-making processes into account. Once the fund manager and his team took this step, they implemented the following methodologies: premortem analysis into the decision-making process, which would be triggered if the stock price fell by 30 percent over a period of three months, or if it had a negative attribution to fund performance of more than 50 bps on a rolling yearly basis. The three key topics for concern, following the premortem, were then brought into a red team-blue team discussion, for more objective input. This enabled the team to overcome instances where they “were struggling rather than adjusting to the new reality,” bringing in new monitoring and review processes to reflect on their conviction in early assumptions.

The introduction of a more objective and robust decision-making process led to more than 150 bps of improvement in performance over six months, resulting in the creation of £60 million for unit holders.

Selling too early

Having identified areas for improvement, the fund manager running the second fund the firm examined focused on reducing the effect of loss aversion—caused primarily by the psychological impact associated with profits and losses—on his performance. The pain of losing money is psychologically roughly twice as great as the pleasure derived from gaining money. To avoid this pain, investors become averse to loss, often selling positions early based on impatience and fear.

In this case, the fund manager reported “real progress” in counteracting the influence of loss aversion. “The study clearly showed a desire to lock in profits too early and not let a full rerating occur when a company grows into its valuation. The last two years have been the perfect environment for this kind of investment and so that characteristic trait has been tested consistently. The ‘old me’ would have succumbed to the temptation to sell time and time again as these investments climbed to new highs on a consistent basis.”

Having instituted processes to counter his tendencies—including bringing in colleagues to challenge his decisions—the fund manager has avoided the aggressive trimming he would have been likely to execute before addressing the bias. “It has helped me stay in the game this year in what is a notoriously difficult market environment for my strategy.” 

When considering selling stocks that have reached his original fair value, but where he sees continued internal improvement in line with the underlying change story, the fund manager now applies two debiasing methodologies:

  • Clean-sheet-redesign—reviewing fair value assumptions from scratch, and asking whether one would buy this stock today
  • Devil’s advocate—If the clean-sheet leads to a change in decision, a devil’s advocate is introduced—an independent colleague arguing for a counter approach that challenges the fund manager’s new view

Using these debiasing methodologies enabled the fund manager to capture more than 200 bps of improvement in performance over twelve months, creating more than £200 million for unit holders.

In an industry that is continuously looking to improve the value created for their investors, identifying and capturing the value of behavioral change, using a combination of advanced analytics and behavioral science, will be a key differentiator.