Human Relations - Semantic Scholar

Human Relations http://hum.sagepub.com

Knowing the risks: Theory and practice in financial market trading Paul Willman, Mark P Fenton O'Creevy, Nigel Nicholson and Emma Soane Human Relations 2001; 54; 887 The online version of this article can be found at: http://hum.sagepub.com/cgi/content/abstract/54/7/887

Published by: http://www.sagepublications.com

On behalf of:

The Tavistock Institute

Additional services and information for Human Relations can be found at: Email Alerts: http://hum.sagepub.com/cgi/alerts Subscriptions: http://hum.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav

Downloaded from http://hum.sagepub.com at PENNSYLVANIA STATE UNIV on April 12, 2008 © 2001 The Tavistock Institute. All rights reserved. Not for commercial use or unauthorized distribution.

05willman (ds)

30/5/01

12:18 pm

Page 887

Human Relations [0018-7267(200107)54:7] Volume 54(7): 887–910: 017839 Copyright © 2001 The Tavistock Institute ® SAGE Publications London, Thousand Oaks CA, New Delhi

Knowing the risks: Theory and practice in financial market trading Paul Willman, Mark P. Fenton O’Creevy, Nigel Nicholson and Emma Soane

A B S T R AC T

Theories about trading in financial markets are well developed and practically influential. However, traders’ behaviour within these markets appears to deviate substantially from that predicted by theory. Using data from a study of traders within financial markets in London, this article seeks to document this apparent paradox and assess its implications. General theories about how the financial world works are distinct from, but compatible with, more instrumental behavioural rules about how to work in the financial world. The latter may be seen as internally consistent recipes for action which require concurrent belief in both the validity of the general theories – for example about the relationship between risk and return – and in the ability of individual agency to secure outcomes which, in terms of the general theory, have low probability.

KEYWORDS

financial markets qualitative analysis risk theories in use traders

1. Introduction Financial markets are heavily theorized domains. Academic finance theory uses some of the most powerful analytical tools available in social science to model the operation of such markets, usually on the basis of strong assumptions about the rationality of actors and the availability of information. There 887


05willman (ds)

888

30/5/01

12:18 pm

Page 888

Human Relations 54(7)

are also practitioner theories, which often put less emphasis on rationality; books by and about traders describe trading strategies that may generate large rewards. Trading in financial markets sustains heroes and myths; it sustains movies and television shows. People who trade in financial markets are generally aware of both sets of theories. They are trained to understand the academic theories of the market, but also immersed in the folklore of the practitioner theorists. These two sets provide, by the nature of the markets they govern, two rather different things. The first is a theory of how the world of financial markets works. The second is a theory or set of theories which act as guidelines for successful action within markets – theories about how to work the world. As Giddens (1990) notes, financial markets are very specific types of domain. They are institutionally structured risk environments. Risk is not incidental to their activities; rather, the activities themselves involve the measured pursuit of risk. They are also domains of sophisticated reflexivity in which behaviour is influenced by the type of theory preferred by the actor. In such markets, traders trade according to one or more theories, knowing that others act similarly. In this article, we are primarily concerned with the theories that guide trader behaviour. Specifically, we explore the relationship between the general theories of the market and the specific theories that traders use. Section 2 reviews the main theories about financial markets. It looks in particular at the ways in which economic theories try to explain both the behaviour of individuals in the market and the aggregate effects of that behaviour. Section 3 looks at some essential properties of individual theories in action in this context. Section 4 describes the data and methods from an empirical study of trading behaviour in the London financial markets. Section 5 uses the data to explore traders’ individual theories; three features are selected for closer examination – intuition (‘flair’), the role of reflexivity and the emergence of contrarian beliefs. Section 6 looks at the relationship between formal theories and theories in action, attempting an assessment of performance implications.

2. Academic work on financial markets As Coase (1988: 9) notes, financial markets are often cited as examples of perfect competition. As such, they have three features. First, they are assumed to be efficient; they have prices that fully reflect available information (Fama, 1970). Second, they are characterized by perfect information, such that all actors have the information necessary to trade both costlessly and immediately;


05willman (ds)

30/5/01

12:18 pm

Page 889

Willman et al.

Knowing the risks

transaction costs are low. Third, they are characterized by perfect competition, such that multiple sellers cannot fix prices (Swedberg, 1994: 274). Paradoxically, as Coase (1988) also notes, these conditions require an authority structure to secure the market ‘involving an intricate system of rules and regulations to prevent malfeasance’ (p. 7). Swedberg (1994) and Coase (1988) also argue that, in financial economics, markets are often assumed to exist rather than be empirically analysed. However, this is a rather incomplete picture. Financial economics does tend to have a rather rigid neo-classical structure in which individuals are seen to maximize utility under the assumption of a close correlation between risk and return (Bernstein, 1996; Brealey & Myers, 1988; Markowitz, 1952). In the aggregate, these actions determine asset prices. To this extent, market efficiency stems from joint hypotheses about individual behaviour and market structure.1 But financial economists do not generally assert that financial markets are perfect and informationally efficient. If this were the case, then there would be no profit opportunities (defined as a return in excess of the risk). As Bernstein puts it (1996): ‘at any level of risk, all investors would earn the same rate of return’ (p. 297). In fact, much work on market microstructure has been concerned to understand how prices are set and how opportunities for profit may arise. Two broad conventional approaches can be identified. The first focuses on inventory, looking at the flow of trades and how temporal imbalances between supply and demand may arise. The second looks at informational differences between traders. Together, they imply that profits may emerge from the existence of transaction costs and private information (for a review, see O’Hara, 1995). There is empirical work on financial markets in support of both approaches (Ito et al., 1998; Lyons, 1998) which remains fully within the neo-classical tradition in assuming rational, utility maximizing actors. Although risk-return relationships emerge in the aggregate in financial markets (Brealey & Myers, 1988),2 it is not necessary to assume that all individuals have the same risk-return preferences. The behaviour of individual actors in relation to risk is determined by the asymmetry of their utility functions above and below current wealth, which may be variable. Interpersonal variation aggregates into a demand curve for any given level of risk over a range of returns accruing to such risk. In combination with the aggregate supply curve for a given level of risk, this produces a market price for risk. It is thus possible to distinguish the market price for risk from the risk preferences of individuals in the market. Expected utility theorists can acknowledge evidence, for example, from experimental economics, that individuals do not always behave according to the predictions of expected utility theory, relying on the weaker proposition that, on average, investors’


889

05willman (ds)

890

30/5/01

12:18 pm

Page 890


behaviour is consistent with the prediction. Expected utility theory can be shown to be consistent with most aggregate market behaviour most of the time (Bernstein, 1996). However, some aggregate market phenomena are difficult to explain. Consider ‘noise’ trading. As Dow and Gorton (1997) note, ‘there appears to be a consensus that trading volume or turnover (trading volume as a fraction of total market value) is inexplicably high’ (p. 1025); Black (1986) originated the term ‘noise’ to describe this excess. Noise traders may be acting rationally for liquidity or hedging reasons, or they may be acting rationally as agents in ways that differ systematically from the behaviour of those acting in markets as principals. For example, Dow and Gorton (1997) argue that noise trading exists because investors (principals) force traders (agents) to trade rather than be idle. Other approaches take a more radical view. De Long et al. (1990) argue, first, that noise traders are irrational, having ‘erroneous stochastic beliefs’. Second, irrationality helps, in that it may generate higher expected returns than those accruing to rational traders. Third, the rational traders’ reaction to irrationality may cause prices to diverge from fundamental values. Close attention to irrationality, the possible benefits from irrational action and its aggregate consequences are central to the project of ‘behavioural finance’, which seeks not only to account for phenomena such as noise trading but also for investor behaviour, stock market overreactions and bubbles (Shiller, 2000; Thaler, 1991). In practice, the focus is on ‘quasirationality’ (Thaler, 1991), which Bernstein (1996) describes as analysing how market actors ‘struggle to find their way between the give and take between risk and return, one moment engaging in cool calculation and the next yielding to emotional impulses’ (p. 287). One theory, prospect theory, has become particularly important in behavioural finance because of its implications for expected utility theory. The latter posits that individuals are motivated to maximize not expected financial returns but rather the expected utility of their actions. Utility functions possess a general form following from the proposal that the disutility arising from a fall in wealth is greater than the utility arising from an increase in wealth of the same size. This accounts for risk aversion; an individual will require a risk premium to engage in a trade with an element of risk in the return. A monetary value can be put on risk aversion by examining how much greater than the stake the winnings need to be to encourage individuals into a fifty-fifty bet (Bernstein, 1996; Markowitz, 1952). Prospect theory (Kahneman & Tversky, 1979; Thaler 1991; Tversky & Kahneman, 1986) argues that individuals treat gains and losses differently; this is depicted in Figure 1. The risk neutrality line is contrasted with the


05willman (ds)

30/5/01

12:18 pm

Page 891

Willman et al.

Knowing the risks

Figure 1 Prospect theory framing effects Source: Bazerman (1998: 49)

framing effects of prospect theory depicted in the value line. Framing effects depict risk aversion in the domain of gains and loss aversion – risky behaviour seeking to avoid loss altogether – in the domain of losses. Individual reference points dividing gains from losses may vary, depending on performance targets and past history (Wiseman & Gomez-Meija, 1998). Despite their importance, framing effects are just one example of a set of heuristics and biases which affect decision making under uncertainty (Kahnemann et al., 1982); many have been used to explain deviations from rationality in financial markets (Shefrin, 2000; Thaler, 1993). An important point to make for the subsequent argument of this article is that within the behavioural finance approach, rational behaviour and heuristics coexist but the former is theoretically dominant. A comprehensive list of decision heuristics does not sum to a theory of individual market behaviour, rather it sums to a theory about deviations from the neo-classical model (Willman, 2000). In explaining how individuals behave in financial markets, the final set of work to consider concerns social influences. Sociological work on financial markets has tended either to examine the empirical pattern of market


891

05willman (ds)

892

30/5/01

12:18 pm

Page 892


transactions or to examine social and non-rational influences on decision making. The former is the concern of the micro structural work of Baker and colleagues (Baker, 1984a, 1984b; Baker & Iyer, 1992). Baker contests the market model implicit in financial economics by describing highly differentiated exchange networks in securities markets characterized by interpersonal differences in network centrality. He identifies systematic relationships between network size and density on the one hand and price volatility on the other. His central finding that increased network size is associated with increased volatility challenges economic assumptions about perfect markets. Keynes (1936) described stock market behaviour as ‘anticipating what average opinion expects average opinion to be’ (p. 156), manifesting the concern with reflexivity which is characteristic of many who focus on the social construction of financial markets. A key strand of work is that on herding, where mutually aware actors engage in mimetic, irrational behaviour (Adler & Adler, 1984). Abolafia and Kilduff (1988), following Kindleberger (1978) have described panics in which investment manias followed by panic selling imply that market actors both create and are influenced by their trading environments (see also Warner & Molotch, 1993). Abolafia (1996) combines a social constructionist approach with detailed ethnographic work in a variety of Wall Street markets. These markets are seen, not as neo-classical phenomena but as socially produced and reproduced by market makers. In particular, specific market conditions are linked to bounded rationality phenomena, such that different behaviours may be understood as responses to different market circumstances. For example, bond traders, subject to continued electronic information flow, use information editing heuristics foreign to pit traders in commodity markets where the main information sources are interpersonal. These theories have different implications for traders in markets. A peculiar feature of the efficient markets hypothesis underpinning financial economics is that efficiency stems from joint hypotheses about individual behaviour and market structure. Other approaches differ. In behavioural finance, market outcomes are influenced by deviations from rationality whose origins do not lie within the market. In sociological approaches, there is variance in the primacy accorded to structure and action but in all cases individual action occurs within a socially produced market context. We argue that all approaches imply that, in addition to an understanding of financial theory, individuals in the market need a set of theories and heuristics to guide their own profit seeking behaviour. We also argue that these individual theories have social dimensions. We detail the argument in the next section.


05willman (ds)

30/5/01

12:18 pm

Page 893

Willman et al.

Knowing the risks

3. Individual and general theories The key distinction is between general theories of how the market works and the personal theories guiding specific individuals working in the market. Markets may be perfect, rational and governed by expected utility theory, but individuals trade within differentiated networks, more frequently than necessary and with variable risk-return trade-offs. Their behaviour is reflexive in two senses. First, they learn idiosyncratic trading strategies based on their market experience. Second, they factor predictions of others’ behaviour into their own trading strategies. The view offered here is that this difference is systematic. In the following section, we examine the following propositions: Proposition (i): Traders believe that the market as a whole works in accordance with general finance theory – this is ‘how the world works’. Proposition (ii): Theories that guide individual action in the market involve a belief in the existence of exceptions to the general theories – this is ‘how to work the world’. Consider Figure 2. The figure is a simple representation of an aggregate risk return relationship in the market. Traders in the market wish to succeed, i.e. they wish to trade in the area above the trend line, but without moving the trend line up. A market in which there is perfect information about risk-return relationships will, other things being equal, offer few such opportunities.3 The strategy of the successful trader is thus, from one perspective, an attempt to identify or create and then subsequently use and protect the market imperfections that allow the majority of trades to occur above the line. The theory of how to work the world differs from the prevailing theory of how the world works in the following respects. 1. 2.

3. 4.

It seeks anomalies in the operation of the general theory; it is often contrarian in nature. However, it depends logically on a belief in the general validity of the theory of the world; it involves plays around the trend line which require a general belief in its existence. It deploys this knowledge consciously and instrumentally; this may be to the advantage of the principal, or the trading agent, or both. There are, ex ante, clear incentives to render the individual theory in use inappropriable both by others operating competitively in the market and by principals.


893

05willman (ds)

894

30/5/01

12:18 pm

Page 894


Figure 2 The relationship between risk and return

5.

Interpersonal variation in the content of such theories is likely.

Theories of how to work the world generate variance around the trend line. It is not assumed at this point that they are rational. Such a theory targeted at the domain of gains may simply be optimistic bias or illusion of control, and may be associated with poor trading performance (Fenton O’Creevy et al., 1999). They may involve beliefs about fundamental properties of the stock or instrument being traded. They may involve beliefs about patterned movements in the market, accessed by analysis of historic data. They may involve more intuitive responses, based on ‘feel’ for the market. They may thus also vary in their degrees of formality. Such theories are instrumental recipes for action predicated on the need to generate purposive action in a world governed by probabilistic rules; they function to generate a sense of control over the environment. They are reflexive in the sense that individuals are aware of the development of their own theories. They are heuristic in the sense that they reduce complicated tasks on the assessment of


05willman (ds)

30/5/01

12:18 pm

Page 895

Willman et al.

Knowing the risks

probability and value to simple judgement operations. (Kahnemann et al., 1982). They have symbolic functions in at least two – slightly contradictory – senses. First, they are used to render actions accountable; they must fit within a broader vocabulary of trading motives (Wright Mills, 1963) to assist presentation of self. Second, they may be used to articulate a distinctive trading strategy and identity; they may function to generate self esteem. As we discuss below, these two senses may relate to accountability in the domains of loss and gain respectively. Work-the-world theories are thus similar in many respects to lay theories, particularly in terms of their functions (Hewstone, 1983). However, they differ from ‘pure’ lay theories in several respects. They are the product of experience in a specific expert context, not pure ‘common sense’ (Furnham, 1988). They are theories of process rather than content, often deductive rather than inductive, situational rather than individualistic (Furnham, 1988). However, this is not to say that such theories about the market do not contain pure ‘lay’ elements in their characterization of the psychology of other traders. The difference from, and relationship to, general knowledge is necessary rather than contingent. If trading behaviour is routinized application of general rules, then – as many have argued – it can be automated (Wilhelm, 1999). Routines imply fatalistic trading behaviour which can, incidentally, be exploited by others in the market. By contrast, popular accounts stress the existence of ‘wizards’ and ‘heroes’ who ‘beat the market’ (Insana, 1996); our own data show not fatalism but unrealistic optimism and illusion of control among traders (Fenton O’Creevy et al., 1999). From one perspective, traders simply avoid what might be termed the ‘actuarial fallacy’, defined as the individual presumption that aggregate population probabilities apply to individual action. To say that, for example, 50 percent of trades lead to losses is not to say that each trade has a 50 percent probability of failure (Nicholson et al., 1999). Traders routinely display optimistic bias in assigning low probabilities of loss to trades based on their work-the-world theories. They interpret the aggregate theories as defining an action space rather than providing behavioural rules; we explore the implications of this in more detail below.

4. Data and methods The data reported here are part of a larger study of individual and contextual influences on the risk-taking behaviour of traders in financial markets. Respondents came from the London operations of four large investment banks. They were engaged in trading fixed income and equity products. Both


895

05willman (ds)

896

30/5/01

12:18 pm

Page 896


market making and proprietary trading are included. No open outcry markets were covered, since the majority of London trading is now electronic. The total sample size is 118, of whom 57 were traders, 46 trader/managers, 15 desk managers and 10 senior managers who did not trade. The data collection process had three stages. First, participants completed a short questionnaire covering demographic, tenure, career history and earnings data. Second, semi-structured interviews were conducted with all respondents. Interviews addressed a range of issues, including risk, emotions, trading strategies, the experience of gain and loss and questions about organizational culture. Interviews took between 35 minutes and 1.5 hours. The key variable influencing the length of interviews was the amount of time participants could be away from their desks. Each interview was taped and verbatim transcripts were produced. Third, participants completed a set of self-ratings concerned with four aspects of performance and a questionnaire covering attitudes towards compensation. The qualitative data presented here are based on analysis of the content of the interview transcripts. Table 1 illustrates, by underlining, the main interview categories used. The approach to analysis was phenomenological, as in several other studies of this type (Furnham, 1988; Heath, 1999) although reference is made to whether a type of comment reflected a broadly shared opinion or not. Interview data were coded for analysis using the QSR NUD*IST Vivo (NVivo) program. The program enables sections of transcribed interview text to be coded and categorized. The analysis took place in two stages. First, every reference to trading style or strategy was coded. A new document comprising each strategy-related quotation was created. Second, the quotations in this document were sub-coded. The new set of codes was derived from the data. This method of categorization, rather than theoretically based coding, was used because it fitted with the exploratory nature of the article. Each of the authors read the document containing all references to strategies. The key data sub-categories were reached through discussion of the data. The second set of codes comprised categories agreed upon by the authors. The NVivo program was again used to code the interview data into categories. The categories of interest were concerned with traders’ beliefs about how to trade, beliefs about the trading world, tacit knowledge, contrarian beliefs and boredom trades. The quotations presented in this article have been selected as representative examples of the material from each category. Although the relevant data are reported elsewhere, it is appropriate, in light of the previous discussion, to note the evidence on the existence of nonrational behaviour in the sample. These include pervasive loss aversion (Willman et al., 1999), illusion of control (Fenton O’Creevy et al., 1999) and emotional influences on trading behaviour (Soane et al., 1999). In addition,


05willman (ds)

30/5/01

12:18 pm

Page 897

Willman et al.

Knowing the risks

Table 1 Main interview categories Trader interview items 1. The trader’s job 2. Traders’ decision making processes 3. Emotions associated with trading 4. Learning and strategy development 5. The use of trading limits 6. The impact of loss and gain 7. Team work 8. Management style 9. The compensation process 10. Organizational culture Manager interview items 1. Traders’ use of heuristics 2. The effects of reward systems on traders’ behaviour 3. Stress management 4. Trader development 5. Matching of traders to different types of markets 6. Traders’ use of information 7. Teamwork 8. Performance evaluation and management 9. Error management 10. Risk management

there were dispositional differences in risk propensity (Nicholson et al., 1999). These traders used heuristics and biases in their own recipes for action.

5. Working the world In this empirical section, we explore the content of working-the-world theories. Three elements are important. The first is a belief in the importance of flair and intuition; traders believe technical knowledge is not enough. We argue that this belief sustains the generation of tacit and inappropriable knowledge. The second is a belief in the importance of reflexivity; we argue that this belief sustains learning by doing which in turn may generate noise trading. The third involves contrarian strategies. These are, in Giddens’s (1990) sense, counterintuitive beliefs that generate divergent behaviour. We


897

05willman (ds)

898

30/5/01

12:18 pm

Page 898


argue that these generate intractable management problems. They cannot easily be articulated and their pursuit often involves dissimulation.

(i) Flair and intuition The following exchange took place between the first-named author and a senior trading manager on an equities desk. What happens to new entrants? People join as juniors, answering phones and monitoring positions . . . they sit with traders and move around the desks . . . if they show flair, they try out in trading. We are recruiting MBAs but you need flair; flair plus a good technical base. What is flair? Flair is anticipating the market, showing intuition, having a contrary, different view of events; not going with the herd, not following the market trend. Traders also had views on flair: Having a feeling is not the same as experience, it’s like having whiskers, like being a deer . . . you need a certain type of intelligence, but it’s more about intuition. Knowledge and experience do count for a lot; but there are some people you could never teach trading to in your life. Some people are just too academic. There is an instinct that you have and you build on that. It was seen by many to be something to do with learning from experience, and with learning about information processing: Unless you’ve been in a black hole yourself, you can’t explain it to anyone; you see someone just staring at the screen and they can’t get out. When you have a position like that, you just look at it. You need someone to come along and say ‘OUT’. So traders have to be approachable.


05willman (ds)

30/5/01

12:18 pm

Page 899

Willman et al.

Knowing the risks

The subjective part of trading strategy is the assessment of the quality of information and how prepared you are to rely on it to assess your risk . . . you have to tell the difference between probabilities and possibilities. No-one has told me how to trade since I have been here. I sit next to the guy that runs the desk, so they’re always asking ‘what’s the P&L, how’s the desk doing, what’s the desk doing?’ Intuition is particularly difficult to justify ex ante and particularly glamorous ex post. Trader interviews nonetheless abounded with assertions of the importance of intuition and, conversely, the failure of purely technical ability to explain success. This differed by market, being more likely in ‘vanilla’ equities trading than in more complicated derivatives trading; however, it was not absent in the latter: It is very difficult in my experience to articulate why you get into a trade and what you do when you’re in a trade. The importance of intuition was expressed in a variety of contexts. Most generally, managers would note that some people ‘get it’ and some do not, referring to the performance of trainees. Those who did not ‘get it’ went to sales or research. At the most fundamental level, then, displaying some form of intuition was a ticket of entry to the trading floor. Different trading styles were noted, some focusing on the fundamentals of the stock or instrument, some looking at past trading data (‘technical’ traders) and others using intuition. The combination of these three different forms of approach offers managers a portfolio of trading styles that could be used to diversify risk. Finally, there is intrapersonal variance. Traders remarked on times when they were ‘on a roll’, ‘seeing the ball really well’, ‘going for it’, when the intuitive feel for the market had to be exploited to the full. They reported being more likely to trade frequently in such circumstances. Use of intuition may be associated with noise trading and, conceivably, with risk taking: I think one of the keys to successful trading is to know when you’re hot and there’s no better feeling than when you’re trading from strength. Once this tacit skill was developed, not only was it difficult to articulate, there were few incentives for many to do so:


899

05willman (ds)

900

30/5/01

12:18 pm

Page 900


Everyone has the right to know arbitrage information, but people do not necessarily think it is their duty to share the information. This business is not about team spirit. We should be better at trading as a group . . . but in reality people get very parochial and very protective of what they’re doing. If I didn’t know something and went and asked someone, this gives them bargaining power. Systems are not set up to be able to see what other people are doing. Information is not shared enough and there is too much Chinese whispers about position taking. In short, traders believe both that they have private information and that they develop tacit knowledge.

(ii) Reflexivity Financial markets are highly reflexive domains in which traders ‘colonise the future’ in terms of models from the past (Giddens, 1990). There is awareness of the general theories, awareness of one’s personal theory in use and awareness of the existence of variance between one’s own and others’ theories. I think it’s the blending of the quantitative skills with an understanding of how people react. Let’s face it . . . what is the market? It’s you and me and a hundred other people sitting around and playing poker through screens; you are trading emotion, but you do have quantitative value running through it. You have to build a framework on how you believe the world is working . . . you have your mental picture of what’s going on. When you are making money, this mental picture is being reinforced. Some traders reflected on the contradictions involved: When we make a lot of money, I tend to worry about it. I think we can’t be making all that money, something is wrong. Our systems are not all that good. Once you’ve entered a position, you realize no rule book prepares you


05willman (ds)

30/5/01

12:18 pm

Page 901

Willman et al.

Knowing the risks

for it. And we’ve all read the world’s best trader books. Every person has a different approach. One way of approaching this issue is to examine three of the theories about loss and gain which pass the membership test (i.e. traders talk in these terms). They are depicted in Figure 3. Figure 3a depicts risk neutrality in the face of loss and gain, which is generally assumed in finance theory. Whatever the wealth or performance position of the trader, each trade is dealt with according to its own risk level, analytically and rationally. In interviews, traders would acknowledge the need to behave in this way but would also point to the experience of losses, the progress towards bonus targets and their ‘mood’ as influences on trading behaviour. Figure 3b presents the prospecttheoretic pattern of risk aversion in the domain of gains and loss aversion in the domain of losses. This was to be avoided; several traders quoted the maxim ‘you don’t cut profits and you don’t chase losses’. Figure 3c illustrates the ideal; chase profits, cut losses. Figure 3a represents what traders know they should do; Figure 3b what they should not do – but apparently, on occasion, find themselves doing (Willman et al., 1999); and Figure 3c what they would like to have done ex post. They are simultaneously aware of the rules, the need to break the rules and the risks involved in so doing. They know the risks in a fairly sophisticated way, but not in the way predicted by finance theorists. Finding out about the risks involved learning by doing. Experience was seen to be important: To trade anything well, you need at least a year’s experience of trading that stock. The year is a continuous progression of trading experiences. You can never know enough and you can never learn too much. Traders were more likely to report trading more frequently when making money and less frequently when losing: When I make money I think it shows I’m doing something right. If I’m right, I will try and do more of them, to increase the size of my position to make more money. I think if you are on a roll, that is when you are prepared to put more money at risk. When you are not sure what is going on and you have a few losses, that is when you pull back.


901

05willman (ds)

902

30/5/01

12:18 pm

Page 902


(a)

(b)

(c)

Figure 3 The theories about loss and gain that pass the membership test: (a) risk neutrality; (b) prospect-theoretic prediction; and (c) ideal trader

I think it is a good idea when you lose money, especially a lot of money, to sit back, take it easy and maybe stop trading for a while. On average, people will trade more often when they are making money compared with when they are losing money because their risk aversion and loss tolerance change. Other random effects may be evident: You do boredom trades because you can be sitting up there doing nothing and you think, well I’ll do that because it gives me something to do. The next thing you know you are wrong and you’ve lost money on it. Or you’re right and you’re inclined to do it again in bigger volumes.


05willman (ds)

30/5/01

12:18 pm

Page 903

Willman et al.

Knowing the risks

These quotes imply a rather different view of noise trading. If trading is seen as a rational activity affecting the portfolios of investors, the volume of trading in markets is high. If trading is seen as the generation of expertise through learning by doing, this high level of trading can be explained in rational terms as leading to the acquisition of tacit knowledge by traders in the market.

(iii) Contrarian beliefs Behaviour manifesting flair was often seen as the opposite of herding. Good traders must be better than average: A lot of people in the market are trend followers and that’s where opportunities are created. You need to believe in your own ability because when things go wrong, how will you cope? You have to think you are better than average in the market. Every year my filtering processes have improved, so now I can beat the market. I think that to get to the top you would have to adopt a higher risk strategy because otherwise it is a risk-return trade off. If you’re happy to take bigger risks then you might end up getting fired or getting to the top. Contrarian beliefs use the language of the general theory to articulate vocabularies of motive. The general laws of the market are interpreted in idiosyncratic ways as part of the vocabularies of motive used to justify situated action by traders (Wright Mills, 1963). Consider the following, from a trading manager in equities: Six weeks ago [i.e. mid-October 1998] I woke up on Sunday, read the papers and decided that the market was going to turn. On the Monday, I told the traders ‘you can be level or long but don’t come back short’. We made a lot of money . . . it’s been the best six months in a while . . . when everyone tells you something, don’t be with the crowd. It was a bad market . . . we decided it was better to do something than do nothing.


903

05willman (ds)

904

30/5/01

12:18 pm

Page 904


Another trader manager remarked: I ask my team what they are doing and then I do the opposite. Of course, by definition, everybody cannot be contrarian. A trading cliché is ‘the trend is my friend’. However, individuals ‘on a roll’ may have the confidence to buck the trend. Managers may use contrarians or deliberately encourage a contrarian position on their desks in order to hedge against the trend. If all traders on a desk are trading with the herd, the manager may see benefit in having some contrarian positions which ‘question existing wisdom’. We asked traders about their best experiences and about their best trades. Several responded by talking about the times they had made a great deal of money, or made a big trade. However, several responded to this openended question by talking in explicitly contrarian terms; the best trade was seen as the one that proved their theory right, in the face of arguments to the contrary, irrespective of profitability: Gut feeling is my sixth sense. I would rather go with it and be wrong than not go with it and be right because if my gut feeling tells me to do something and I don’t and it happens then I am doubly annoyed. The context in which these beliefs are expressed is important. Traders often have to justify their trading positions to managers. Trader managers in this sample were predominantly ex-traders who continued to run their own books. They had fairly wide spans of control and could not monitor the trading of individuals directly; they simply knew inputs (i.e. trading strategies) and outcomes (profit or loss). Overwhelmingly, they focused on ensuring traders avoid losses rather than encouraging them to maximize gains (Willman et al., 1999). Because they do not wish to rely solely on retrospective outcome data, they frequently resort to monitoring inputs, i.e. the logic of the trading position. In one interview, a manager reported firing a trader because of deficiencies in his explanation of his position, rather than on the basis of outcome data. This episode epitomizes the social aspects of ‘work-the-world’ theories. There may be strong incentives for traders to offer ‘herding’ representations of contrarian positions, or simply to conceal such positions from managers ex ante, until the hypothesis prompted by the theory in use has been tested. Ex post, a successful contrarian strategy may be publicized as the basis for reputation building. Consequently, the contrarian views expressed in the interviews involve


05willman (ds)

30/5/01

12:18 pm

Page 905

Willman et al.

Knowing the risks

retrospection and are thus likely at best to be a biased sample of the contrarian positions actually taken. The failed ones will not form part of the folklore of trading, while the successful ones achieve a cachet perhaps out of proportion to their significance. Successes from herding or simply from luck may be retrofitted to a contrarian argument. Traders who strongly believe in their own efficacy may be subject to fundamental attribution error. In short, there may be strong incentives for traders ex post to emphasise the use of flair for the purposes of reputation building. As Frankfurter and McGoin (1999) note, those who work in financial markets may be predisposed to interpret beneficial outcomes which are the result of chance events as the result of deployment of tacit skill. We concur to the extent that we found traders had unrealistic beliefs in their ability to control events in this sample.

6. Knowing the risks Our central argument has been that the knowledge deployed by traders in financial markets consists of general, quasi-scientific knowledge about markets and tradable instruments on the one hand and, on the other, specific and idiosyncratic recipes for trading success. Perfect markets and rational decision making in the aggregate models may be set against the interest in the generation of imperfections, the use of heuristics and learning by doing at the individual level. It is beyond the scope of this article to argue that these differences between, on the one hand, aggregate theories and, on the other, individual ones serve to undermine the former.4 It may be that, as in the philosophy of science literature, one regards individual decision making by traders as involving the critical and reflexive process of ‘depending on ideas while assessing their dependability’ (Sabel, 1994). From this perspective, overcoming what we have termed the actuarial fallacy, (i.e. believing aggregate probabilities apply to specific actions) is similar to maintaining Feyerabend’s (1970) notion of the ‘principle of tenacity’ under which scientists maintain a belief despite acknowledging its infirmity. The fundamental belief here is that one can beat the market and its infirmity lies in the fact that many traders hold it and they cannot all be right. We conclude by examining three issues. The first is the pursuit of the perfect market, particularly by regulators. The second is the pursuit of learning by doing by traders. The third is the generation of contrarian strategies by traders. All three have both theoretical and policy implications. Financial markets are not perfect and traders do not have perfect information. Moreover, their actions keep things that way. Imperfections are


905

05willman (ds)

906

30/5/01

12:18 pm

Page 906


essential to making money but are intrinsically unknowable to the majority of traders at any time. The markets are in fact a complicated mixture of formal, publicly available knowledge and knowledge which is tacit in one of two senses (Willman, 1997). 1.

2.

It may be necessarily tacit, based on intuition about market change. This knowledge is typically generated in learning-focused noise trading which results in experientially grounded ‘feelings’ about the market which provide a basis for risk exposure. It may be conceptualized as a deeply embedded heuristic which traders describe as flair, the precise content of which cannot easily be articulated. We refer to this as a Type 1 inappropriability. It may be contingently tacit – kept so for profit by the originating trader or traders in order to be locally rather than generally appropriable. This knowledge may contain facts about arbitrage possibilities or market maker reactions that are in principle easily articulated but which will be protected by individuals as the basis for continued trading success. This is Type 2 inappropriability.

As Hayek (1945) points out, in information terms, markets are highly decentralized. Relevant knowledge ‘never exists in concentrated or integrated form, but solely as the dispersed bits of incomplete and frequently contradictory knowledge which all separate individuals possess’ (p. 519). Winning strategies in such markets may not last long and individuals have incentives to prevent others from smothering the advantage their strategy creates (Bernstein, 1996). Both forms of knowledge are used by traders to target the domain of gains outlined in Figure 1. Use of know-how which is Type1 inappropriable is not fully rational, but it may generate profit. Use of know-how which is Type 2 inappropriable is rational for traders but may cause problems for managers or regulators; it will generally be covert. Ex ante, traders may misrepresent or dissimulate the bases of their trades in both cases. In the first case, because of the irrationality; it would be very difficult to explain to managers and peers. In the second, because of appropriability issues; if everybody follows the strategy, it may yield no advantages. The vocabulary of motives that traders use to describe their trades to managers or other position monitors is that underlying Figure 3c. This relates to both noise trading and contrarian strategies. Our approach to noise trading is that a significant amount of it consists of learning by doing. Traders are experimenting with their work-the-world theories. Since these trades may not be condoned by managers, often the


05willman (ds)

30/5/01

12:18 pm

Page 907

Willman et al.

Knowing the risks

basis of the trade will be couched in language which conceals intent. Similar considerations affect contrarian strategies, but to a rather more compounded extent. Ex ante I will wish to conceal my contrarian strategy not only, first, because my manager will not necessarily endorse a riskseeking counter-intuitive trade but also, second, because it can only succeed to the extent that it does not immediately induce mimetic herding. Ex post, it may be the basis for reputation building. Not only do these motives underpin the dual symbolic role for work-the-world theories outlined above, they also suggest a link between trader learning, contrarian strategies and noise trading. In short, the structure of the knowledge base which sustains trader success generates covert behaviour which also sustains the occasions of trader malfeasance which attract great publicity; the most famous, that of Leeson in Barings, contains intuition, contrarian beliefs and the attempted exploitation of a market imperfection (Bank of England, 1997). Heath (1999) argues that extrinsic incentives bias – assuming others are more motivated by extrinsics than intrinsics – is endemic in principal–agent relationships. A priori, there is a strong case for suspecting this exists in financial markets where traders are rewarded predominantly by bonus payments. Heath argues that this bias emerges because principals use inappropriate lay psychological theories to structure incentives for agents. Our analysis of the practical theories trader–agents themselves use may offer hints for the design of incentives and constraints in financial markets which tap into the intrinsic elements of motivation that underpin such theories. Our focus here has been on knowledge intensive work rather than knowledge intensive firms. All our respondents worked within large investment banks but we have chosen the individual and the market as our units of analysis. Knowledge in markets is fragmented. Work may be knowledge intensive but the knowledge used and generated in work cannot easily be codified. It thus illustrates a particular form of knowledge management problem, but the appropriability considerations may be relevant to a variety of contexts.

Acknowledgements We are grateful for comments on this article from attendees at the Oxford Knowledge Management Conference, as well as from anonymous reviewers of the journal. We are also grateful for the co-operation of the four institutions that supported the field research, who wish to remain anonymous. This research was sponsored by ESRC grant No. L211252056.


907

05willman (ds)

908

30/5/01

12:18 pm

Page 908


Notes 1 We are grateful to an anonymous reviewer for this point. 2 This is not necessarily the case for other markets; see Wiseman and Bromiley (1991), McNamara and Bromiley (1999). 3 Traders can of course profit simply by making markets using liquidity to benefit from bid-offer spreads. 4 For a view of financial economics as wholly ideological, see Frankfurter and McGoin (1999).

References Abolafia, M.Y. Making markets: Opportunism and restraint on Wall St. Cambridge, MA: Harvard University Press, 1996. Abolafia, M.Y. & Kilduff, M.K. Enacting market crisis: The social construction of a speculative bubble. Administrative Science Quarterly, 1988, 33, 173–93. Adler, P.A. & Adler, P. The social dynamics of financial markets. Greenwich, CT: JAI Press, 1984. Baker, W. The social structure of a national securities market. American Journal of Sociology, 1984a, 89(4), 775–811. Baker, W. Floor trading and crowd dynamics. In P.A. Adler and P. Adler (Eds), The social dynamics of financial markets. Greenwich, CT: JAI Press, 1984b, pp. 107–27. Baker, W. & Iyer, A. Information networks and market behaviour. Journal of Mathematical Sociology, 1992, 16, 305–32. Bank of England Report to the Board of Banking Supervision: Inquiry into the circumstances of the collapse of Barings. London: HMSO, 1997. Bazerman, M. Judgement in managerial decision making. New York: Wiley, 1998. Bernstein, P.L. Against the gods: The remarkable story of risk. New York: Wiley, 1996. Black, F. Noise. Journal of Finance, 1986, 41, 529–43. Brealey, R. & Myers, S. Principles of corporate finance. New York: McGraw-Hill, 1988. Coase, R. The firm, the market and the law. Chicago, IL: University of Chicago Press, 1988. Delong, J.B., Schleifer, A., Summers, L. & Woldermann, R.J. Noise traders and risk in financial markets. Journal of Political Economy, 1990, 98(4), 703–38. Dow, J. & Gorton, G. Noise trading, delegated portfolio management and economic welfare. Journal of Political Economy, 1997, 105, 1024–50. Fama, E. Efficient capital markets: A review of theory and evidence. Journal of Finance, 1970, 25, 34–105. Fenton O’Creevy, M., Nicholson, N., Soane, E. & Willman, P. Trading on illusions. Academy of Management Conference, Chicago, 1999. Feyerabend P. Consolations for the specialist. In I. Lakatos and A. Musgrave (Eds), Criticism and the growth of knowledge. Cambridge: Cambridge University Press, 1970, pp. 197–230. Frankfurter, G.M. & McGoin, E.G. Ideology and the theory of financial economics. Journal of Economic Behaviour and Organisation, 1999, 39, 159–77. Furnham, A. Lay theories. Oxford: Pergamon, 1988. Giddens, A. The consequences of modernity. Oxford: Polity, 1990. Giddens, A. Modernity and self-identity. Oxford: Polity, 1991. Hayek, F. von The use of knowledge in society. American Economic Review, 1945, 35, 519–30. Heath, C. On the social psychology of agency relationships: Lay theories of motivation overemphasize extrinsic incentives. Organizational Behavior and Human Decision Processes, 1999, 76(1), 25–62.


05willman (ds)

30/5/01

12:18 pm

Page 909

Willman et al.

Knowing the risks

Hewstone, M. (Ed.) Attribution theory: Social and functional extensions. Oxford: Blackwell, 1983. Insana, R. Traders’ tales. New York: Wiley, 1996. Ito, T., Lyons, R.K. & Melvin, M.T. Is there private information in the FX market? The Tokyo experiment. Journal of Finance, 1998, 1111–30. Kahnemann, D. & Tversky, A. Prospect theory: An analysis of decisions under risk. Econometrica, 1979, 47, 262–91. Kahnemann, D., Slovic, P. & Tversky, A. (Eds) Judgement under uncertainty: Heuristics and biases. Cambridge: Cambridge University Press, 1982. Keynes, J.M. The general theory of employment, interest and money. London: Macmillan, 1936. Kindleberger, C.P. Manias, panics and crashes: A history of financial crises. New York: Basic Books, 1978. Lyons, R.K. Profits and position control: A week of FX dealing. Journal of International Money and Finance, 1998, 97–115. McNamara & Bromiley, P. Risk and return in organisational decision making. Academy of Management Journal, 1999, 42(3), 330–9. Markowitz, H.M. (1952) The utility of wealth. Journal of Political Economy, 1952, 60, 151–8. Nicholson, N., Fenton O’Creevy, M., Soane, E. & Willman, P. Individual dispositions and risk propensity. LBS COR Working Paper, 1999. O’Hara, M. Market microstructure theory. Oxford: Blackwell, 1995. Sabel, C. Learning by monitoring: The institutions of economic development. In R. Swedberg and N. Smelser (Ed.), Handbook of Economic Sociology. Princeton, NJ: Princteon University Press, 1994, pp. 137–66. Schiller, R. Irrational exuberance. Princeton, NJ: Princeton University Press, 2000. Shefrin, H. Beyond greed and fear: Understanding behavioral finance and the psychology of investing. Cambridge, MA: Harvard Business School Press. Soane, E., Nicholson, N., Fenton O’Creevy, M. & Willman, P. Trading on emotions. Academy of Management Conference, Chicago, 1999. Swedberg, R. (1994) Markets as social structures. In R. Swedberg and N. Smelser (Eds), Handbook of economic sociology. Princeton, NJ: Princeton University Press, 1994, 255–83. Thaler, R. Towards a positive theory of consumer choice. Journal of Economic Behaviour and Organisation, 1980, 1, 19–80. Thaler, R. Quasi-rational economics. New York: Russell Sage Foundation, 1991. Thaler, R. (Ed.) Advances in behavioral finance. New York: Russell Sage Foundation, 1993. Tversky, A. & Kahnemann, D. Rational choice and the framing of decisions. Journal of Business, 1986, 59, 251–78. Warner, K. & Molotch, H. Information in the marketplace: Media explanations of the 1987 crash. Social Problems, 1993, 40, 167–88. Wilhelm, W.J. Internet investment banking: The impact of information technology on relationship banking. Journal of Applied Corporate Finance, 1999. Willman, P. Appropriability of technology and internal organisation. In I. McLoughlin and M.J. Harris (Eds), Innovation, organisational change and technology. London: ITB Books, 1997. Willman, P., Fenton O’Creevy, M., Nicholson, N. & Soane, E. Traders’ tales: Agency, managerial behaviour and loss aversion in investment banking. LBS COR Working Paper, 1999. Willman, P. Risk, greed, sound and fury. Business Strategy Review, 2000, 11(2), 72–4. Wiseman, R.M. & Bromiley, P. Risk return associations: Paradox or artifact? An empirically tested explanation. Strategic Management Journal, 1991, 12, 231–42. Wiseman, R.M. & Gomez-Meija, L.R. A behavioural agency model of managerial risk taking. Academy of Management Review, 1998, 23(1), 133–53. Wright Mills, C. Situated actions and vocabularies of motive. In D. Horowitz (Ed.), Power politics and people. Oxford: Oxford University Press, 1963.


909

05willman (ds)

910

30/5/01

12:18 pm

Page 910


Paul Willman is Professor of Management Studies at University of Oxford. He holds degrees from the Universities of Cambridge and Oxford and has held teaching posts at Imperial College, London, Cranfield School of Management and London Business School. He is a research fellow at the Centre for Economic Performance, LSE. He has published several books and numerous articles in the fields of labour relations and human resource management and is currently researching perceptions of risk and their relationship to risk behaviour. [E-mail: [email protected]] Mark Fenton O’Creevy is a Lecturer at the Open University. [E-mail: [email protected]] Nigel Nicholson is Professor of Organizational Behaviour at London Business School. [E-mail: [email protected]] Emma Soane is a Research Officer at London Business School [E-mail: [email protected]]
