Financial Professionals' Overconfidence: Is It Experience, Function, or Attitude?

• Published date: 01 March 2014
• DOI: http://doi.org/10.1111/j.1468-036X.2011.00636.x
• Date: 01 March 2014

European Financial Management, Vol. 20, No. 2, 2014, 236–269

doi: 10.1111/j.1468-036X.2011.00636.x

Financial Professionals’

Overconfidence: Is It Experience,

Function, or Attitude?

Oliver Gloede

Department of Economics, Leibniz Universit¨

at Hannover, K ¨

onigsworther Platz 1, 30167 Hannover,

Germany

E-mail: gloede@gif.uni-hannover.de

Lukas Menkhoff

Department of Economics, Leibniz Universit¨

at Hannover, K ¨

onigsworther Platz 1, 30167 Hannover,

Germany, and Centre forEuropean Economic Research (ZEW) in Mannheim, Germany

E-mail: menkhoff@gif.uni-hannover.de

Abstract

This paper examines financial professionals’ overconfidence in their forecasting

performance. We compare individuals’ self-rating of performance with the true

performance, both measured relative to the same peer group. The forecasters

in our sample show overconfidence on average, although to a moderate degree,

including many cases of underconfidence. In analysing this, we find that working

experience is accompanied by less overconfidence. Function is also related to less

overconfidence, such as being a fund manager and using fundamental analysis.

The same effect is found for the attitude to herd, whereas recent success appears

with more overconfident professionals.

Keywords: better-than-average,self-rating,forecasting,foreign exchange,

performance

JEL classification: G1, D84, F31

We thank participants at the annual conferences of the European Financial Management

Association, German Economic Association, German Finance Association, Money Macro

Finance Research Group, and the RoyalEconomic Society, as well as the audiences of several

seminars, in particular Michael Br¨

auninger, Jerry Coakley, Michael Frenkel,Markus Glaser,

Alan Kirman, Thomas Lux, Christian Pfeifer, Winfried Pohlmeier, Michael Schr ¨

oder and

Philipp Sibbertsen for their helpful comments and discussion. We gratefully acknowledge

very constructive comments by the editor and two anonymousreferees. In addition, we thank

the Centre for European Economic Research (ZEW) in Mannheim, Germany, for providing

the data set. Oliver Gloede gratefully acknowledges a scholarship of the Foundation of

German Business (Stiftung der Deutschen Wirtschaft).

C

2011 Blackwell Publishing Ltd

Financial Professionals’ Overconfi dence 237

1. Introduction

We know from a series of, by now, ‘classical’ studies that most people are overconf ident

regarding their own abilities, such as their driving performance (Svenson, 1981).

Subsequent research has shown that overconfidence is not an invariant characteristic

but also depends on circumstances. What is important in this respect is the feedback

that people receive, which seems helpful in adjusting one’s own perception. Financial

markets provide frequent and precise feedback to their participants, so that overconfident

behaviour may be less expected here. In addition, financial markets punish overcon-

fidence, as overconfidence reduces an investor’s performance (Odean, 1998). Given

the lack of research in this realm, exploring the extent of overconfidence of financial

professionals and its potential determinants seems warranted.

Our study contributes to this issue and is the first one to examine determinants

of the better-than-average (BTA) variant of overconfidence in the case of financial

professionals. Accordingly, we need two ingredients for this study, i.e., a BTA-measure

of overconfidence and the linkage of this measure to potential determinants. First, the

BTA-measure of overconf idence is the difference between a professional’s self-rated

relative performance with respect to a reference group and the same person’s true

relative performance within the same reference group. This precise measure is usually

not available. Second, we analyse relationships between individual forecasters’ overcon-

fidence and their demographic, (job) function and forecasting characteristics. Findings

reported here enrich our understanding about the determinants of overconfidence among

financial professionals, perhaps facilitating the introduction of measures for reducing

overconfidence and its costly consequences.

Our research uses a sample of 105 professional forecasters who are regularly

contributing to an established financial market survey in Germany. From this survey,

we gather monthly forecasts for the USD/DM and (later) USD/EUR exchange rate

over several years, so that we can calculate a meaningful forecasting performance. We

complement this performance information with data from additional surveys conducted

at the same time as the regular surveys. These supplementary surveys reveal the

professionals’ self-rating and further characteristics about them.

The financial professionals in this study are experienced, educated and hold senior

positions in the financial industry so that the sample seems to be relevant for our

purpose. Based on this sample, the BTA-measure of overconfidence shows reasonable

attributes (which will be related to the literature below). First, these professionals regard

themselves mostly better in their self-rating than their fellow forecasters and thus show

overconfidence on average. Second, the average degree of overconfidence is moderate,

which may be expected for professionals whoreceive frequent feedback and in a market,

foreign exchange, where forecasting is particularly difficult. Third, the combination

of moderate average overconfidence with substantial individual heterogeneity implies

that many professionals show underconfidence. Fourth, the self-rating of financial

professionals is not significantly related to the same person’s performance, indicating

that a correct self-rating is not easy. Fifth, our BTA-measure of overconfidence has the

expected correlations to alternative measures of overconfidence and thus seems reliable.

Below we document the following determinants of overconf idence. The working

experience of a professional is tentatively related to less overconf idence, suggesting that

experience helps in assessing one’s true performance. Also, two function-related criteria

are related to less overconfidence: first, being a fund manager, which may be due to

the immediate feedback that these professionals receive; and second, a heavy reliance

C

2011 Blackwell Publishing Ltd

Oliver Gloede and Lukas Menkhoff

238

on fundamental analysis, which may indicate these professionals’ remarkable effor ts

in forecasting. Finally, a professional’s tendency to herd in his forecasts is associated

with less overconfidence. Thus, herding may be interpreted as a sign of less confidence

and possibly as a rational reaction to inferior information or ability. In contrast to the

relations just discussed,recent forecasting success is positively related to overconfidence,

with success inducing higher self-evaluation. These patterns provide obviouslessons for

financial professionals and their superiors.

This paper is structured as follows: Section 2 reviews the previous literature, showing

that our approach is original. Section 3 introduces the data used. Section 4 details

relationships between self-rating, overconfidence and performance. Section 5 provides

the main findings from regression analyses. And Section 6 concludes.

2. Literature

The aim of this study is to analyse overconfidence of professionals in financial markets.

Even though financial professionals are not the main subjects in empirical studies, we can

learn about determinants of overconfidence from the finance and psychology literature

which deals with overconfidence and BTA in general. A survey of the BTA literature is

provided by Alicke and Govorun (2005).

Overconfidence biases are expected to ease when tasks involve frequent feedback,

as is the case for financial markets (Lichtenstein and Fischhoff, 1980). The effect of

training could be dampened if these biases are deeply rooted in personality (see Preston

and Harris, 1965; Brehmer, 1980; Menkhoff and Nikiforow, 2009). Consequently, one

should expect a lower degree of overconf idence here than elsewhere. However, the

impact of feedback on behaviour requires that it is asked for and understood (Kruger

and Dunning, 1999). In this respect, there may be differences between various groups

in financial markets: professionals in particular might use and be able to interpret the

feedback they get. In contrast, non-professional participants, such as individual investors,

may have lower incentives to use feedback and may have less ability to interpret it.1

Available studies do support the notion that there is also overconfidence in f inancial

markets, but this evidence refers mainly to non-professionals or to financial markets

in general, where non-professionals are included. Specifically, there are three kinds of

studies. First, one stream of the literature relies on the theoretically-derived finding

that overconfidence of financial market participants can be detected by their increased

trading activity (Odean, 1998). There is ample evidence of ‘too much trading volume’

in financial markets in general (e.g., Statman et al., 2006). Second, other studies use

information about individual investors which is collected from their trading accounts

(among others are Odean, 1999; Glaser and Weber, 2007a; Grinblatt and Keloharju,

2009). Third, overconf idence is shown in experimentssimulating f inancial markets (e.g.,

Biais et al., 2005; Deaves et al., 2009). Thus, we know from these various perspectives

that overconfident behaviour appears in financial markets, but we know little about

whether financial market professionals also show this overconfident behaviour.

As professionals trade the largest volumes and have the best information among

financial market participants, evidence on their behaviour is of particular interest.

1In fact, professionals might even need a reasonable level of overconf idence to sustain their

optimistic and risk-loving attitude after failure, which is apparentlya warranted characteristic

of the financial industry (Taylor and Brown, 1988; Oberlechner and Osler, forthcoming).

C

2011 Blackwell Publishin

g

Ltd