Contemporary Challenges in Risk Management

Dealing with Risk, Uncertainty and the Unknown

T. Andersen, T. Andersen, Kenneth A. Loparo

• Publisher: Palgrave Macmillan
• Year: 2014
• ISBN:
  9781137447623
• Topic: Business
• Subtopic: Financial Risk Management
• Index: Business

This book focuses on two central aspects of the risk managing process, namely 1. how managers (can and do) assess developments in the external risk environment and deal with them, and 2. analysing the effects of risk management and different managerial approaches. The articles represent state of the art academic analyses and research contributions.

1

Distinguishing Rationality and Bias in Prices: Implications from Judgments of Risk and Expected Return

Hersh Shefrin

Introduction

There is a gulf between what theory and practice tell us about how risk premiums reward investors for bearing risk. An elegant theory relates expected return to both mean-variance efficient portfolios and to the covariance between returns and a pricing kernel. However, this theory has not proved to be especially valuable in empirical work, where risk premiums are instead explained using simple factor models involving size and book-to-market equity (B/M), for which there is little theoretical justification.

There is also a lack of consensus on the root cause of the factor structure associated with the cross-section of stock returns. One possibility is that the factor structure reflects fully rational prices, while another is that the factor structure reflects investors’ behavioral biases. Fama and French (2004) argue that it is not possible to distinguish between the two possibilities empirically, and have maintained this position even with the emergence of new results about sentiment-based predictability in returns (e.g. Baker and Wurgler, 2006, 2007). This lack of agreement among scholars is partly due to a reliance on realized returns in discriminating between various explanations (Black, 1993). In this regard, the moments of realized returns are ex post variables that cannot be automatically equated with investors’ ex ante judgments of risk and expected returns.

To shed light on whether prices are fully rational, or instead reflect behavioral bias, I introduce new data consisting of judgments by professional investors about the risk and returns of holding different stocks. These data, collected over a 15-year period, paint a clear, consistent picture of the cross-section of investors’ judgments of stock market risk and return. My findings indicate that investors’ collective judgments about risk and expected return display some of the rational pricing features emphasized by Fama and French (2004) and some of the behavioral features emphasized by Baker and Wurgler (2006, 2007).

With respect to Fama and French, I find strong and consistent evidence that investors’ judgments about risk are negatively correlated with size, and positively correlated with B/M. This finding accords with the Fama-French view, even though Fama and French admit that they have no compelling explanation for why size and B/M should underlie systematic risk. Nevertheless, my data show that investors do indeed judge large cap stocks to be safer than small cap stocks and growth stocks to be safer than value stocks.

With respect to Baker and Wurgler, I find that investors’ collective judgments about expected return are significantly related to the sentiment variable (SENT) (Baker and Wurgler, 2006). Notably, sentiment modulates the relationship between both size and realized returns, and B/M and realized returns. My findings show that sentiment modulates investors’ judgments about these relationships as well.

With respect to bias, I find that investors’ collective judgments about the cross-section of expected returns are consistently at odds with the cross-section of realized returns. My results indicate that the majority of investors expect higher returns from large cap stocks than from small cap stocks, and higher returns from growth stocks than from value stocks. In other words, investors act as if they attempt to implement a Fama-French factor model, but in the course of doing so, reverse the signs of the coefficients.

My data suggest that the bridge between the rational price view of risk and the behavioral view of expected return involves investors’ judgments of how risk and expected return are related. Ganzach (2000) reports evidence that investors perceive risk and expected return to be negatively related. Finucane (2002) discusses why judgments based on effect generally induce people to believe that risk and benefits are negatively related. In this regard, stocks are just one example. My data indicate that the majority of investors form judgments of risk and expected return as if they believe that the capital market line is negatively sloped and the security market line is negatively sloped. Therefore, even if they form appropriate judgments about a security risk, most form biased judgments about the associated expected return.

At the same time, my data make clear that not all investors are alike. There is substantial heterogeneity in investors’ judgments of risk and return. Roughly 20% of the investors in my sample make judgments in line with the Fama-French view. I became aware of this heterogeneity in 1999 when I ran an in-company workshop for a US hedge fund specializing in value investing. My analysis showed that along almost every dimension, the judgments of the fund’s director of research and chief investment officer (CIO) were in line with the Fama-French view. However, my analysis also showed that less than 15% of the portfolio managers and analysts reporting to the CIO formed like-minded judgments. Instead, most expected higher returns from larger cap stocks than from smaller cap stocks, they expected higher returns from growth stocks than from value stocks, and judged the relationship between risk and return to be negative.

I suggest that taken together, the three following elements combine to make the case that prices are not fully rational, and instead reflect behavioral bias. First, Baker and Wurgler (2006, 2007) document return predictability based on sentiment. Second, the relationship between investors’ judgments of expected return and Baker-Wurgler sentiment is positive and statistically significant. Third, the judgments about risk and expected return in my data feature biases that remain strong and consistent over the 15 years of my sample.

There is a long tradition in finance about the difficulty of using realized returns alone to identify the degree to which prices are fully rational. Black (1993) suggests that the connection of realized returns to size and B/M most likely stems from data mining, and it is with this in mind that Shefrin and Statman (1995, 2003) suggest analyzing whether size and B/M drive investors’ judgments, instead of focusing exclusively on realized returns. Doing so avoids the data mining quandary. To this end, they use data involving judgments about stocks’ value as a long-term investment (VLTI) from Fortune magazine’s annual corporate reputation survey. Notably, they find that judgments about VLTI strongly and consistently reflect size and B/M over time.¹

Shefrin and Statman report that VLTI is positively related to size and negatively related to B/M, which are opposite in sign to those for realized returns. They argue that this pattern suggests that prices reflect bias, and therefore are not fully rational. This line of argument appears to have had a limited impact among those debating whether or not prices are rational or behavioral (Fama and French, 2004). There are at least two possible reasons for this limited impact: first, because VLTI, unlike expected return, has no clear or precise definition; and second, demonstrating irrationality on the part of some investors, even many investors, does not necessarily imply that these irrational elements are manifest in market prices.

In Shefrin (2001), I reported the results from data based on workshops given in 1999 and 2000. My new data strongly reinforce the original findings from Shefrin (2001), and provide additional insight into the arguments advanced in Shefrin and Statman (1995, 2003). In my sample, VLTI is nearly always positively related to size, and negatively related to B/M. Moreover, with few exceptions, the size and B/M sign patterns for my data coincide with those from the Fortune data, for the years in which I have access to data from both. These findings provide further indications about the associations involving VLTI, size and B/M.

At the same time, my data suggest that VLTI is not a perfect proxy for judgments of expected return. I find that although VLTI is positively correlated with judgments of expected return, at times it is also negatively correlated with perceived risk. Moreover, the strengths of the correlations vary over time. From 2005 on, perceived risk impacts VLTI as strongly as expected return. Indeed, in both 2009 and 2012, perceived risk is statistically significant, but expected return is not. These results make clear that treating VLTI as a perfect proxy for expected return can be problematic.² The results also highlight the importance of having data directly measuring expected returns.

The remainder of this chapter is organized as follows: the next sections review the current thinking about the nature of risk and expected return in the asset pricing literature, describe the data, and present the cross-sectional properties of perceived risk. The subsequent sections discuss the cross-sectional properties of expected returns and analyze the behavioral features underlying the relationship between perceived risk and expected return. Then the strength of the relationship between expected return and the Baker-Wurgler sentiment index is described, the cross-sectional properties of the expected return series derived from analysts’ target prices is discussed, and the findings are related to earlier work based on the annual Fortune magazine reputation survey as a robustness test. Finally, the main issues are recapitulated and some basic conclusions are drawn from the study. Appendices 1–5 contain supporting details and provides a Bayesian perspective to interpret results.

Current thinking about the nature of risk and return

Fama and French (2004) survey the theory and evidence associated with the capital asset pricing model (CAPM). They point out that the CAPM provides a theoretical definition for total risk, and a decomposition of total risk into the sum of systematic and non-systematic components. In this section, I summarize key features of the Fama-French survey, and also quote extensively from it in order to capture the nuance and flavor of Fama and French’s language.

One of the strengths of the CAPM is in identifying systematic risk as the sole basis for risk premiums. At the same time, the CAPM is plagued by a variety of weaknesses, both theoretically, in how to define the market portfolio, and empirically, in whether beta explains realized returns. In addressing these weaknesses, Fama and French state: “In the end, we argue that whether the model’s problems reflect weaknesses in the theory or in its empirical implementation, the failure of the CAPM in empirical tests implies that most applications of the model are invalid” (p. 26).

In addition, Fama and French discuss the intertemporal capital asset pricing model (ICAPM) developed by Merton (1973), which extends the CAPM by drawing attention to the importance of state variables associated with future consumption and investment. Examples of state variables are labor income and prices of consumption goods. The state variable approach is centr...

Table of contents

• Cover
• Title
• Introduction: Contemporary Challenges in Risk Management
• 1  Distinguishing Rationality and Bias in Prices: Implications from Judgments of Risk and Expected Return
• 2  Looking under the Lamppost? A Research Agenda for Increasing Enterprise Risk Management’s Usefulness to Practitioners
• 3  The Risk-Return Outcomes of Strategic Responsiveness
• 4  Exploring the Effect of Effective Risk Management Capabilities
• 5  The “Soft” Side of Strategic Risk Management: How Top Managers’ Leadership Style Affects Volatility in Performance
• 6  Mixed Risk Management Practices: Insights from Management Accounting and What It Means for Strategic Risk Management
• 7  Subjective Beliefs and Statistical Forecasts of Financial Risks: The Chief Risk Officer Project
• 8  Defaults and Returns in the High-Yield Bond and Distressed Debt Markets: Review and Outlook
• Index