The Effect of Past Performance Framing on Investors’ Belief Updating by SSRN
DG Value Surges On Recovery Plays
According to a copy of the firm's February investor update, Dov Gertzulin's DG Value Partners returned +4.48% net for the month of February, which ValueWalk has been able to review. Q4 2020 hedge fund letters, conferences and more Following this performance, the firm has returned +8.32% net for the year to the end of February. Read More
Maastricht University – School of Business and Economics – Department of Finance; Network for Studies on Pensions, Aging and Retirement (Netspar)
Maastricht University – School of Business and Economics – Department of Finance; Netspar
May 18, 2015
We analyze how past performance framing impacts investors’ belief updating. Over multiple rounds, we present 377 experimental subjects with past performance information and subsequently measure updates of their beliefs (return expectations and risk perceptions). We employ three different frames, varying the default information horizon (annual, monthly, daily). An important distinction compared to previous work is that subjects can easily and without costs opt out of the default and obtain past performance information on each of the three horizons. We find that, across all subjects, the default option does not have an impact on the magnitude of subjects’ belief updating. However, dividing the sample into subjects staying in versus opting out of the default uncovers an important pattern of results. While showing returns over a longer information horizon reduces the magnitude of belief updates for subjects staying in this default, we find the opposite effect for subjects opting out of this default.
The Effect Of Past Performance Framing On Investors’ Belief Updating- Introduction
Updates in individual investors’ beliefs, such as return expectations and risk perceptions, are important drivers of investment decisions (Hoffmann, Post, and Pennings 2013). Prior work shows that investors rely on naïve reinforcement learning when updating beliefs (Kaustia and Knüpfer 2008; Choi et al. 2009; Chiang et al. 2012). Accordingly, investors update beliefs by extrapolating individual past return experiences. Importantly, larger updates of beliefs lead to more active trading, hurting investment performance (Hoffmann and Post 2015). We examine how framing of past performance information, using different defaults regarding the information horizon, affects investors’ belief formation. In particular, we analyze whether presenting longer information horizons as a default option leads to smaller updates in beliefs. Doing so is important, as finding such an effect would constitute an easy-to-implement behavioral nudge that can positively affect individual investor welfare by a reduction in trading activity.
To answer our research question, we perform an experiment in which we place subjects in a situation closely resembling an online brokerage environment. We present subjects with a stock portfolio that they are asked to imagine to own, and assess their belief updates over time, based on the performance of this stock portfolio. Subjects are randomly assigned to three experimental conditions, which differ with regard to the default information horizon (annual, monthly, daily) at which portfolio performance is shown. Participants can easily and without costs opt out of the default and obtain past performance information on each of the three information horizons.
We find that across treatments, varying the default does, on average, not impact the magnitude of belief updating. However, an important pattern emerges when we divide the sample into subjects staying in the default vs. those opting out of the default. Over all rounds and treatments, more than half of the experimental subjects opt out of their default information horizon. Specifically, while showing returns over a longer information horizon compared to a shorter information horizon reduces the magnitude of belief updating for subjects who stay in this default condition, we find the opposite result for subjects who opt out of this default condition to look at portfolio performance over a shorter information horizon. In particular, for subjects originally assigned to the longer information horizon default, opting out confronts subjects with returns over a shorter horizon. This in turn leads to larger updates in their beliefs. In contrast, for subjects originally assigned to the shorter information horizon default, opting out confronts them with returns over a longer horizon, which leads to smaller updates in their beliefs.
Our paper is related to, but different from, previous work that examines how different evaluation horizons influence individual investors’ decision-making, such as Benartzi and Thaler (1995), Beshears et al. (2014), and Gneezy and Potters (1997). These studies typically recommend longer evaluation periods to improve individual investor decision-making (e.g., to increase their stock-market participation). However, the results of these previous studies are based on experiments which restrict access to information and make it cumbersome or even impossible for subjects to opt out of the default. In a setting in which subjects have immediate access to alternative information horizons and can easily opt of the default, we find that presenting returns over a longer information horizon is not necessary beneficial. That is, only for subjects that stay in their default information horizon, presenting portfolio performance over a longer information horizon has a mitigating effect on the magnitude of their belief updates.
Our results imply that showing a longer information horizon as a default does not help or harm investors on average. It is important to gather evidence what the fraction of investors will be that stick to the default in order to implement an effective simple nudge. In our sample, especially subjects with high financial literacy opt out of the default. This finding implies that if a particular population of investors is highly financially literate, a short default information horizon should be chosen. Otherwise, a long horizon information horizon default is preferable.
See full PDF below.