Adapting financial rationality: Is a New Paradigm Emerging?
Mona Soufian1, William Forbes2, Robert Hudson3
1Newcastle
2School
Business School, Northumbria University
of Business and Economics, Loughborough University
3Newcastle
University Business School, Newcastle University
1
Abstract
We discuss the implications of an alternative to the efficient market hypothesis (EMH) the
adaptive market hypothesis (AMH). The AMH may give a theoretical basis for a new financial
paradigm which can better model such phenomena as the recent financial crisis. The AMH
regards the financial market order as evolving, tentative and defined by creative destruction
in which trading strategies are introduced, mutate to survive or face abandonment. The
concept of investor rationality is less helpful than the distinction between investment
strategies which are more or less well adapted to the prevailing market environment in
which they are deployed. We outline how a more systematic and grounded basis for
behavioural finance can be developed in line with the later approach. Based on this we
develop testable hypotheses allowing the AMH to be distinguished from the EMH. Finally we
discuss how the AMH can aid our understanding of important issues in finance. A crucial
feature is that in the survival of richest, as opposed to fittest, implied by the AMH there is
much room for misallocation of resources as price and value uncouple. In this evolution of
the financial market order the regulatory State features as a further market in which the
vote market verifies or disrupts settled market conditions.
Acknowledgements
We would like to acknowledge very insightful and helpful comments by the editor and three
anonymous referees on an earlier draft of this paper which have considerably improved the
work.
2
1. Introduction
The recent global financial crisis has dealt a huge and largely unanticipated shock to the
world economy. The Queen of England surely expressed the thoughts of much of the
world’s population in November 2008 when she asked in a visit to the London School of
Economics why no one had seen the crisis coming. We believe the fundamental answer to
her question lies in the dominance of the neoclassical financial paradigm. A few far-sighted
people had seen problems ahead but were largely ignoredi. The idea that markets rationally
price assets and risk was so entrenched amongst influential academics, practitioners,
regulators and politicians that dissenting views were completely marginalised.
Recent years have seen an almost continuous succession of financial crises including the
emerging markets crisis of the late 1990s, LTCM, the bursting of the ‘dot com’ bubble, the
accounting scandals at Enron and Worldcom, cumulating in the near collapse of the global
banking system in 2008 and on going problems with sovereign debt. As Hyman Minksy
(1986) has taught us financial crises are a recurring theme of economic history. What is so
disturbing is the escalating frequency and intensity of the crises we now observe. Galbraith
(1990, p viii) states the case thus “Recurrent speculative insanity and the associated financial
deprivation and larger devastation are, I am persuaded, inherent in the system. Perhaps it is
better this can be recognised and accepted.” Recently Ferguson (2012) has portrayed the
2008 Crisis as a critical point in a “great degeneration” of Western capitalist economises as
they enter a “stationary state” characterised by the rule of law being replaced by the rule of
lawyers and an intense rent-seeking amongst market participants for shares of a pie that has
ceased to grow or has even entered decline.
Mainstream finance theory has clearly failed to anticipate, or even convincingly explain, the
recent crises. Indeed to a large extent it might be considered to have caused them by giving
intellectual authority to the ideal of unrestrained financial markets and dogmatically
suppressing dissenting views. There seems a vital need to address this situation with new
research programmes to better model reality. In the terms of Kuhn’s seminal work on the
structure of scientific revolutions the crises are anomalies; that is a failure of the current
3
paradigm to take into account observed phenomena (Kuhn, 1962). An accumulation of
anomalies eventually leads to a change of paradigm. The current crisis may act as the
breakpoint enabling serious consideration of different theoretical approaches.
Perhaps we can finally now accept the cyclical nature of financial crises and move on to
explaining the function they serve. Periods of trauma and destruction may intensify the
speed at which differentiation based on evolutionary fitness proceeds. This suggests seeking
to build financial institutions that can withstand further crises may be a misplaced effort. It
may be better to try to put in place bank resolution procedures that are both easily triggered
and avoid capable of protecting the taxpayer’s purse. Taleb (2012) identifies a category of
trader/entrepreneurs who thrive of volatile, if not destructive times. Such innovators are
“antifragile” in the sense that they enter their own in periods when pressures to survive are
there most intense.
As one would expect a crisis of the magnitude we are experiencing has given rise to
enormous debate. There have been hundreds of popular and academic articles and books
on the subjectii. Different authors have emphasised different perspectives. Much of the
popular coverage has personalised the issues. Often individuals have become scapegoats for
behaviour they personify, for example Richard Fuld, Sean Fitzpatrick and Fred Goodwin, the
CEOs who presided over the demise of Lehman Brothers, Anglo-Irish, and the need for the
government rescue of RBS, respectively. A simple assertion that we have a flawed and
greedy banking culture is now commonplace as a result.
Whilst it is surely prudent to rapidly address particular flaws in financial practices and
regulations much of the post crisis response has been very piecemeal and ad-hoc in nature.
This type of response is inevitable given the evident lack of an appropriate and credible
theoretical basis to inform policy. The deficit in theory has been recognised even in some
essentially practical and hard-headed assessments of the crisis. This is one of the primary
insights of the UK's Turner Review into the failure of regulatory authorities to head off the
burgeoning securitised debt crisis (Turner, 2009). Turner concludes (Turner, 2009, pp 85)
“the conventional wisdom relating to the global financial system – that risks had been
diversified – was widely accepted and was wrong”. If the ability to diversify as a risk
reduction strategy now looks tarnished in the face of systemic risk then the very
4
fundamentals of received professional wisdom are in doubt. In a related review of the UK’s
equity market’s John Kay explicitly demures from a view (Kay, 2012), that “public policy
should proceed as if these ‘irrational’ behaviours did not exist: such an approach would not
be consistent with the fundamental goals of performing high performance companies.” This
insight echoes Jean-Claude Trichet’s (Governor of ECB) view that many of the economic
models used by advisors during the financial crisis (Trichet, 2010) “seemed incapable of
explaining what is happening in the economy in a convincing manner.” Thus the EMH seems
to have been discarded as the basis of sound public policy and the race is now on to gain
acceptance of a credible alternative. The AMH could become such an alternative source of
guidance.
In this paper we question the core assumption of the EMH which is the ‘rationality’ of
economic agentsiii. Following Gigerenzer et al (2003) we argue that true evolved rationality
emerges when the response when investors’ cognition is a good match to the demands of
the environment in which they find themselves trading. The distinction here is between the
investment strategy and its cognitive and external context, as opposed to a proposed
statistical property which is conjectured to prevail regardless of context or cognition. Based
on the above, in this paper we discuss directions for future research which offer some hope
to build a more persuasive and useful theorisation of financial decision-making. Initially we
propose replacement of the concept of the Efficient Markets Hypothesis (EMH) that financial
markets always act to set prices ‘rationally’ by an understanding that prices change as
investors’ constantly adapt their behaviour as markets evolve their own internal order. The
latter process is known as the Adaptive Markets Hypothesis (AMH) and was initially
proposed by Lo (2004, 2005). The AMH recognises the importance of behavioural finance
and was partly designed to offer a way to reconcile this emergent literature with the
mainstream.
Our second, and complimentary, proposal is to work towards a more
systematic and theoretically grounded basis for behavioural finance, building on the work of
Herbert Simon on bounded rationality, and the research programme of Gigerenzer on
heuristics. At the moment behavioural finance is somewhat fragmented from a theoretical
point of view and can be criticised as often being an arbitrary catalogue of observed
departures from rationality without a unifying theoretical vision to explain those anomalies.
5
Rebuilding financial theory is a huge task and it would be naive to over-commit to a specific
way forward at this stage. However, the research agenda we propose is far from exclusive in
scope. The AMH is much less theoretically restrictive than the existing paradigm in that it
recognises the possibility of a wide range of responses by players in the markets rather than
taking a particular view of investor rationality as axiomatic. An advantage of the approach is
that it may reconcile understandings from various research traditions, including neo-classical
economics, behavioural finance and psychology.
modelling approaches.
It can also accommodate different
This is important as the rapid development of computing is
introducing radically new research tools such as the direct modelling of economic agents
(agent-based modelling). These can be characterised as a ‘bottom-up’ approaches as they
focus on the behaviour of the individual agents in the market and then aggregate this
behaviour to deduce the implications for the overall market. This is, of course, a largely
meaningless activity within the neo-classical world as the agents are axiomatically assumed
to be both homogeneous and rational, according to the normative construction of the
“representative agent” model. In contrast to the agent-based approach most research to
date, both neo-classical and behavioural, takes a modelling approach that can be
characterised as being ‘top-down’ with an emphasis on observing movements in the prices
of financial instruments rather than the underlying behaviour of market participants. In
general, the motivation of the participants cannot be directly observed.
This creates
difficulties for practitioners of behavioural finance as motivations can only be deduced from
indirect evidence. From a methodological point of view it would be premature to definitely
favour the new agent-based approaches over much more established methods but equally it
would be foolish to reject them out of hand. They certainly permit the testing of hypotheses
that relate much more directly to the behaviour and motivations of individual market
participants.
While the AMH seems like a radical new departure to standard finance theorists
evolutionary theory has a history of use within transaction cost perspectives on both the
development of corporate organisational form and governance structures operating within
any chosen form (Nelson and Winter, 1982). This evolutionary theory of economic change
shifted the analytical focus from managers maximising a given objective of profits, or sales,
etc, to the selection of “routines” appropriate to a fluctuating and uncertain environment.
6
Rather than invoking calculus to solve for implicit maxima the evolutionary perspective
favours processes, such as markov-switching processes, for rotating across alternative
“routines”, selected to fit the trading environment the company faces. Such routines include
“well specified technical routines for producing things, … research and development or
advertising and business strategies about product diversification and overseas investment.”
(Nelson and Winter, 1982, p 14). In tranquil times old established routines are likely to
remain unchallenged and comfortably embedded. But in more turbulent periods, rapid
technological, or regulatory, change induces threats to survival requiring major revisions to
the set of routines adopted. In such critical periods, as Nelson and Winter (1982, p 58) state
“it is more natural to represent large scale motivational forces as a kind of persistent
pressure on decisions, a pressure to which the response is sluggish, halting and sometimes
inconsistent…. an … evolutionary purging of motives that diverge excessively from survival
requirements.”
However, within these bounds, imposed by the need to survive,
established, comfortable, if sub-optimal, routines abound. So while within the Nelson and
Winter (1982) schema environmental changes initiate switches in prevailing routines.
Realised learning (RL) methods characterise routine rotations as emerging from the diffusion
of individual successful adaptions through a broader investment community.
The evolutionary perspective on organisational form reflects a much older interest in
the analogy between economic and biological processes which dates back to Bernard
Madiville’s satirical sonnet “The Fable of the Bees” (1714) where the Bee hive mimics the
market by allowing the struggle for individual survival (or private vices) to produce a perfect,
if brutal, social order (of questionable public virtue). Indeed Charles Darwin was much
inspired in conceiving of “The Origin of Species” by Robert Malthus’s “Essay on the Principle
of Population” suggesting Economics and the life sciences shared a common analytical frame
in much the same way as Econophysics marries finance and the natural sciences now
(Ferguson, 2012 p 63).
In the next section of the paper, Section 2, we initially outline the background to current
mainstream theory with its emphasis on perfect rationality and follow this with a discussion
of an alternative approach which assumes that rationality is bounded and the resulting
implications of this for our proposed conceptual approach. The remainder of the paper is
7
structured as follows: Section 3 reviews the Adaptive Expectations Hypothesis (AMH)
followed by an examination of the role of bounded rationality in recent theoretical
developments and considers how the AMH may be tested using ‘top-down’ approaches.
Section 4 examines how a more systematic and theoretically grounded basis for behavioural
finance can be developed. The internal and external bounds on financial decision-making are
examined in sub-section 4.1. Heuristics and adaptive behaviour are examined in sub-section
4.2, followed by discussions on satisficing and investor heterogeneity in sub-section 4.3.
Finally, sub-section 4.4 outlines a tentative modelling strategy for an evolutionary
perspective on financial innovation based on agent-based modelling.
The various sub-
sections of Section 4 imply testable hypotheses, allowing us to distinguish the predictive
power of the AMH from the EMH using the ‘bottom-up’ methods associated with agentbased modelling. Section 5 considers some practical examples of how the AMH and EMH
differ by considering some current areas of finance research. The paper concludes with
overall remarks and suggestions for future work in Section 6.
2. Background
2.1 Mainstream Theory and Perfect Rationality
The classical assumptions of Finance theory are broadly that individuals are rational, seek to
maximise the expected utility of their wealth, are risk averse and follow the tenets of
subjective probability. Capital markets in turn are perfect and generate financial returns
which are not predictable. Despite broad critiques, not least in this journal, (see, for
example, Hudson et al, 1999; Keasey and Hudson, 2007; Hudson and Maioli, 2010; Shiller,
2000; Clarkeson, 2009; Krugman, 2009 and Akerlof and Shiller, 2009) this mainstream
approach has remained very dominant. There are reasons for this rigidity, all the elite
finance departments and academic journals are overwhelmingly dominated by scholars
steeped in the mainstream approach (see Whitley, 1986 and Fox, 2010 for accounts of the
rise of this dominance). In addition, the mainstream approach is very closely allied to the
philosophical belief that free markets are the best way to allocate resources. This was
almost a point of patriotic faith in the US during the Cold War era when mainstream finance
was developed and certainly an easy and powerful argument to make after the collapse of
the Soviet Union. However, it may be that, like reason itself, markets are a good servant but
8
a poor master. Below we discuss the concept of rationality as employed in mainstream
theory and then outline how the concept has become so associated with free market
policies.
The standard notion of rationality employed by economists is that of purposive rationality.
There are no value judgements of the desired ends simply whether the methods used to
achieve them are optimal. This notion lends itself to logical axioms which can be used to
determine rationality. For example, in his seminal multi-million copy selling text book, Paul
Samuelson, often considered the most influential post war economist, codified four
principles of rationality: completeness, transitivity, non-satiation and convexity (Samuelson,
1948, p45)iv. From this viewpoint the human being is seen as a utility maximising,
calculating machine (or “Laplacean Demon”) which raises obvious difficulties once we reflect
upon studies of actual human behaviour or merely our own social observation. Even
Kenneth Arrow, a Nobel Prize winning economist whose work is synonymous with the logical
analysis of economic issues, admitted that human beings could not be rational in this sense
(Arrow, 1986).
In practice the notion of rationality described above is often blurred in the literature around
financial markets with the prominence of the concept that financial markets are rational in
the sense that they produce the best social outcome. In much of the literature dealing with
finance the various notions of rationality are used in a rather cavalier and inter-changeable
manner often to make the rhetorical point that free markets are the best (and most rational)
way to allocate resources. For clarity in this paper we try to distinguish precisely what we
mean by rationality.
The academic theory most strongly justifying the approach of leaving allocative decisions to
the market is the EMH. The history of the development of the EMH is very revealing and can
be seen as not a disinterested scientific endeavour but one strongly influenced by
ideological considerations and the need to preserve the core methodological approaches of
neo-classical economics (see Fox, 2010, for an excellent account of its development). In
broad terms, by the 1950s mainstream economics had largely adopted the still dominant
neo-classical approach based on utility maximisation by rational agents. The methodology
of this approach draws very heavily on deterministic optimisation methods similar to those
9
used in classical (pre – quantum theory) natural sciences such as physics and chemistry. Paul
Samuelson in particular drew heavily on methods from thermodynamics in his writings (see
Samuelson, 1986). Securities markets were problematic within this framework. In popular
discourse their behaviour had long been a byword for irrationality and the random nature of
their behaviour was becoming mathematically formalised by the early 1960s (see, for
example, Kendall, 1953; Mandlebrot, 1963; Cootner, 1964). Samuelson rose to the challenge
and wrote an ingenious article seemingly squaring the circle by reconciling randomness with
rationality (Samuelson, 1965). The very title of the article ‘Proof that Properly Anticipated
Prices Fluctuate Randomly’ and the liberal use of mathematics in the article gave an air of
dispassionate scientific rigor to the exercise.
The reasoning in the article is in fact very straightforward and superficially persuasive. Since
it is hard to predict stock prices then surely they incorporate all available information and
therefore are rationally determined.
It is hard to overestimate the influence of this
argument on subsequent financial theory and practice. A very substantial body of academic
work was seen as supportive of Samuelson’s view. This work was built up by a number of
scholars through the 1960s. The general research pattern was to find empirical evidence that
prices were hard to predict by some metric and then tacitly or explicitly assume that this
showed that prices were properly anticipated. The scholar that is most closely associated
with this enterprise is Eugene Fama of the University of Chicago which personified fervent
support of free markets and positive economics.
Fama developed the term ‘efficient
markets’. In 1965 in an article in the Financial Analysts Journal he wrote “In an efficient
market, however, the actions of the many competing participants should cause the actual
price of a security to wander randomly about its intrinsic value” Fama (1965, p56).
But the meaning of randomness was left unclear and thus left open to opportunistic use by
EMH advocates in later discussion (Mlodinow, 2008, p 84-85). One interpretation of
randomness is that any of the conceivable outcomes are equally likely, rain or shine for
example. This meaning to the word is sometimes termed the frequency interpretation of
randomness. Another is that rain or shine are not equally likely but I cannot predict which
will occur with any reasonable degree of accuracy. This is sometimes termed the subjective
definition of randomness. The frequency interpretation judges the randomness of the
sample ex-post, by the frequency of outcomes. But the subjective interpretation of
10
randomness judges its presence ex-ante by simply asking whether I could know if it will rain
tomorrow or not. Flipping between these meanings gives EMH advocates considerable
scope for avoiding unwelcome refutations of their preferred theory.
Fama subsequently continued to refine the Efficient Markets Hypothesis to classify different
types of efficiency (Fama, 1970). The term ‘efficient market’ was a brilliantly persuasive
choice of words. Efficiency is one of the highest social values in the West and generally
assumed to be a good thing, albeit that in the context efficiency is given a meaning far
different from its normal usage. People who are not scholars of finance invariably and
understandably assume that an efficient market refers to the operational efficiency of
market trading, for example, how fast or how cheaply trades can be processed. The main
achievement, and perhaps purpose, of the efficient market project was, however, to show
that intrinsic value and market value of traded assets were one and the same with profound
implications. If market prices are a perfect guide to intrinsic value the market can then be
taken as an infallible guide to human affairs. This view of the world is aptly summed up by
mathematician David Orrell who describes the market under the EMH as “..some kind of
hyper-rational being that can outwit any speculator or government regulator” (Orrell, p
265).
Only a few mavericks have raised dissenting voices against the EMH project. Many years
before our present difficulties Robert Shiller pointed out a basic logical mistake in this
reasoning. Just because prices are difficult to predict it does not imply that they are rational
in the sense that they represent intrinsic value. Unpredictability is a necessary but not
sufficient condition for market prices to be rational. “This argument for the efficient markets
hypothesis represents one of the most remarkable errors in the history of economic
thought. It is remarkable in the immediacy of its logical error and in the sweep and
implications of its conclusion” (Shiller, 1984, p. 459). Shiller’s critique, however, has largely
fallen on deaf ears and the mainstream view remains dominant.
There has been much institutionalised inertia in the research approaches adopted within
Finance. Findlay and Williams (1985) argue that in conventional Finance the positivist
assertion that ‘‘assumptions do not matter if the model works’ has been subverted into the
notion that assumptions cannot be criticised so long as the model cannot be shown not to
11
work’’. In the relatively benign US financial environment between 1950 and the late 1990s,
the years of the “great moderation”, it was hard to show that the mainstream model was
clearly not working, despite the short sharp shock of the 1987 market crash. But in the
subsequent, more tumultuous, years its shortcomings have become transparent.
It is
untenable to argue that the stock market in the dot-com years, or the US housing and
financial market in the run up to the financial crisis in 2008, were acting in an way that
reflected underlying intrinsic values. Even Alan Greenspan, the well known advocate of free
markets and the Federal Reserve Chairman from 1987 to 2006, was forced to admit in
testimony to a House of Representative Committee that his pro-market ideology was flawed
(Greenspan, 2008).
The development of finance theory is certainly not merely an “academic” issue in the
pejorative sense that few outside the academy know or care about such developments. In a
now well developed body of work Donald MacKenzie has illustrated the “performative”
nature of modern finance theory and especially the Black-Scholes-Merton option pricing
theory and the portfolio insurance schemes that the formula underpinned (MacKenzie,
2004, 2006 (a) and (b)). While the Black-Scholes-Merton model was “performative” in the
sense of moving the quoted price of options to be more like those implied by the model it
also embedded the possibility of a “counter-performative” pricing distortions. This counterperformative feature of options pricing theory was perhaps most dramatically revealed by
the October 1987 Crash. Portfolio insurance, undertaken by creating “synthetic puts” in the
form of sell orders against the futures’ index contract, created positive feedback when sales
of the spot market contract became too heavy for the futures contract price to be speedily
constructed and quoted. Large sell orders were greatly amplified once effective hedging of
downside risk was nullified because market-makers were either unwilling or unable to
execute investors’ trades. Finance theory is rarely tested for validity upon data unaffected by
its own development. Rather the role of finance theory as an “engine not a camera”
(MacKenzie, 2004) means financial theories often create new market realities which the
theory was not initially designed to describe or understand.
2.2 Bounded Rationality and its Incorporation into a New Conceptual Framework
12
For some years there has been on going work into the implications of relaxing the
mainstream assumptions about the behaviour of individuals (see Daniel et al, 2002). Recent
events add impetus to these efforts and raise many more questions about the realism of
current mainstream approaches to finance. The recent global financial crisis has certainly
encouraged some scholars to re-examine the foundations upon which economics and
finance are based. Mallard (2011), among others, challenges the assumption of perfect
rationality in reviewing recent developments in modelling bounded rationality. As discussed
above, mainstream finance theory relies on the assumption of perfect rationality of market
participants and hence it is obvious that this is necessary for markets to be efficient.
However, the existence of efficient markets is not consistent with empirical evidence.
Evidence of departures from efficiency include the high volatility of asset prices and
apparent over-reaction and under reaction in the stock markets (see early studies by, for
example, Shiller, 1981 and De Bondt and Thaler, 1985).
In the growing behavioural finance literature departures from market efficiency are
generally attributed to behavioural biases amongst investors. Much of the mainstream
approach is tacitly retained in that investors are assumed to have purposive rationality and
departures from the fully rational behaviour in the mainstream models are due to the biases
and cognitive limitations of the individuals involved in executing their purpose. Thus most
work in behavioural finance has tended to focus on the bounded rationality of individuals
leading to departures from the optimum solution given by the mainstream model.
Other approaches to investigating bounded rationality offer more positive assessments of
alternatives to perfect rationality and its offspring the EMH. This has been reflected in the
recent development of the AMH combining Simon’s notion of bounded rationality (Simon,
1955) and evolutionary cognitive theories (see Lo, 2004, 2011).
The investigation of
bounded rationality is also the central theme of Gert Gigerenzer’s research programme
which has provided a better understanding of modelling heuristic decision-making.
Gigerenzer’s heuristic studies (1996, 2009, 2011) show that by employing specifically chosen
heuristics agents/investors can make accurate, fast decisions while expending little effort in
the decision-making process.
13
We advance a general conceptual framework that incorporates bounds on both an
individual's internal cognition and the external environment within which it is formed. Our
discussion considers how rationality emerges from the dynamic interplay between individual
internal cognition and its external environment. Environments are constructed and shaped
by the cognition of decision-makers who act within them. This process envisages an
evolutionary chain in which heuristics arise, mutate and thrive, or die out, according to the
progeny they yield. Successful heuristics/mental frames may become entrenched amongst a
wealthy trading elite, while others become marginalised to a dwindling impoverished few,
only to to be reborn in some new guise. A moment’s reflection on the history of
chartism/technical trading, now reborn in the guise of high-frequency trading, reflects this
evolutionary thread (Lo and Hasanhodzic, 2010). Similar evolutionary mechanisms to those
we observe in the natural world are also observed in our social and economic lives. But
crucially no one person observes all the changing heuristics and the role of the price system
is then to transmit a diverse range of value signals no single investor can possess. A primary
benefit of heuristic driven choices is that no one pretends they are either optimal or
anything other than expedient. Hence discarding them in the face of a preferred alternative
is fairly costless and unproblematic (Taleb, 2012, p 11).
We argue that a true evolved rationality emerges when the response from investors is a
good match to the demands of the environment in which they find themselves trading.
Heuristics aid increasingly rational/optimal choice by investors in a stable trading
environment. But financial markets rarely remain stable and fractures in the external
environment rarely leave existing mental-frames pre-eminent, or perhaps even prominent,
for long. Hence the norm for financial markets is the fluctuation of partially adopted failing,
or retrieved and recycled, mental frames which allow a partial rationalisation of the market
setting. In order for investors to adapt their decisions using heuristic devices it is crucial to
match the joint effect of the structure of their internal cognitions to their environment.
This alignment can be compared to two blades of a pair of scissors (Newall and Simon, 1972,
p 55). The first blade represents the agent’s internal cognition (the heuristic
mechanism/framing) while the second blade captures the market environment. An
appropriate conjunction of an investor's mental frame and the environment induces
boundedly rational behaviour. The constraints in our environment support the development
14
of simple strategies, using heuristics/frames that exploit the benefits/constraints on offer as
opportunities and guides. Such fast and frugal decision-making is suitable for an evolved and
still evolving market order. But this evolved order is rarely grasped by any single intelligence,
as Hayek states (1973, p 49-50)
“...The more complex the order aimed at, the greater will be that part of separate
actions which will have to be determined by circumstances not known to those who
direct the whole, and the more dependent control will be on rules rather than specific
commands.”
Hence investors must always choose an investment strategy based on partial, imperfect and
fragmented information recognising that competing traders, with their own flawed
cognition, do the same. A constantly revised, revolving, set of heuristics enable an organic
order to emerge but equally allow for it to dissemble until a new transient order prevails.
Even skills acquired by continuous interaction with others, with all its punishments and
rewards, are often wrongly rationalised as deriving from our individual skill or talent (Smith,
2008, p 8). If heuristics/frames can truly aid decision-making the question arises as to when
and why do some heuristics perform well for investors while others prove misleading or
even harmful? The evolution of a bounded rationality, selected for fitness of purpose in
some particular market context, is a focus of this paper. A conceptual framework is
developed that incorporates internal and external bounds upon rational choice, building
upon the key characteristics of bounded rationality. Three key characteristics are taken into
account in modelling bounded rationality. These are (a) the evolution of heuristics and
mental-frames by adaptive decision-making, (b) the impact of investor heterogeneity, and,
finally, (c) the concept of investor satisficing implied by bounded rationality. We show
bounded rationality provides a promising perspective for refining the foundations of the
AMH in an age when theory and practice have so wildly diverged and finance theory is
deemed unhelpful for evaluating policy reform.
Financial markets are subject to uncertainty, risk and ever changing events; these aspects
make such markets a particularly interesting field to examine market participants’ behaviour
and decision-making processes. Financial innovation has often increased such uncertainty in
financial markets. The multiple triggers for recapitalisation of Special Purpose Entities
(SPE’s), which housed securitised debt, or its division into tranches, illustrate this trend.
Price/quantity choices are made by investors within a trading environment itself in a
15
constant state of flux. A theory of expectation formation is a crucial part of any theory of
investment and asset pricing. Most theories in finance rely on the assumption that investors
form rational expectations. Professors Christopher Sims and Thomas Sargent were awarded
the Nobel Prize in 2011 for contributions to this work. In this area rationality is usually held
to imply an evaluation of risky outcomes in conformity with the von Neumann-Morgenstern
axioms. These axioms imply agents consistently rank alternative risk outcomes/gambles
based on an accurate projection of probability weighted future cash-flows. These axioms
are themselves largely a normative theory of how decisions should be made in uncertain
environments rather than providing a positive theory of how they are made. To quote
Vernon Smith, the 2002 Nobel Prize winner (2008, p. xv)
“Practitioners are into problem solving and do not relate naturally to discussions
driven by economic theory...but they can appreciate working models when they see
and experience them.”
As discussed above, despite the clear behavioural biases of investors standard finance
commonly invokes perfect “rationality” in the sense described by the von NeumannMorgenstern axioms.
3.
The AMH as an alternative framework to perfect rationality, and its
implementation
This section reviews a recent alternative to the EMH’s assumption of perfect individual
rationality and the role of bounded rationality in the development of this theory and its
implementation.
Behavioural finance attempts to provide evidence that the market is not efficient and that
market participants are not rational. Many empirical findings indeed jar with standard
Finance theory. Investors tend to overreact to private information signals and under-react to
public information signals, such as earnings announcements (Bernard and Thomas, 1990).
Stock market overreaction and consequent longer-run mean-reversion are foundational
findings for the behavioural perspective on investment. Further, the equity premium puzzle,
of equity's relatively high return as an asset class, remains puzzling even after the on going
financial crisis has partially dented its edge (Mehra and Prescott, 2008).
16
To date, however, no integrated theory of investment behaviour to rival standard theory has
yet emerged. As critics state behavioural finance can often seem a rag bag of ad hoc models
to explain individual anomalies with no overarching structure (see, for example, Fama,
1998).
One attempt to develop such an integrating framework is the AMH (Lo, 2004). While such
models are currently in their infancy, they provide hope of salvaging the best of the efficient
markets theorem while excising some of that theory’s worst follies. AMH explains loss
aversion, overreaction and other behavioural biases by the fact that investors react to a
changing market environment by invoking new heuristics/mental frames. The processes of
learning, heuristics management and adaptive decision-making are central to AMH.
Accordingly individual agents are regarded as not perfectly, but rather boundedly, rational or
“satisficers”. Hence a rational decision is always evolving, in construction and contention as
mental schemata are invoked, adapted for purpose, and discarded as market conditions
change. Hence we cannot understand market outcomes through the eyes of one
representative investor. Rather it is the very exchange, rejection and promulgation of mental
frames upon valuation that evolves the market condition we observe at any given point (Lo,
2004, p. 21). Investors satisfy, as opposed to optimise, by truncating search at varying points,
trading and observing the consequences.
Great, or perhaps just lucky, traders choose good truncation points and come to dominate
the market, poor traders truncate their search too early or too late and are eased out of the
market. But this difference can only be observed after the event with successful traders’
confidence being fuelled by the downfall of their peers. Trading strategies, high-frequency,
contrarian, etc. can be thought of as mutating species within an overall market ecology. The
growing weight of money flowing towards the favoured mental frame earns a declining rate
of return restarting the cycle of mental frame selection, via the search for yield in a
Darwinian selection process in which trading species grow, mutate and thrive or face
extinction.
In this process there are no fixed valuation rules, but rather an array of
constantly recycled and sometimes outdated market plays of temporary opportunistic use.
Market plays are here neither rational nor irrational, but rather adapted or maladapted to
the context within which they are deployed. So trading profits guide the evolving chain of
market strategies, “By viewing economic profits as the ultimate food source on which
17
market participants depend for their survival, the dynamics of market interactions and
financial innovation can be readily derived.”(Lo, 2004, p. 23). This “survival of the richest”
rather than the fittest allows for substantial noise to surround the signals about asset value
conveyed by prices. Further successful actions speak louder than disastrous omissions, as
one feted movie executive put it “If I had said yes to all the projects I turned down and no to
all the other ones I took, it would of worked out much the same” (Molodinow, 2009, p 12).
Availability bias means few traders learn from even their most damaging omissions. So
learning is very history dependent with little inference being made from unused, but
potentially profitable, alternative trading strategies.
Scholars of psychology and behavioural finance are aware that beyond a certain level of
complexity human cognition is limited, the question is how bounded rationality can be more
realistically modelled in finance. There are two main approaches to modelling bounded
rationality (a) optimising agents who face constraints-with the focus placed upon external
bounds, and (b) satisficing decision-making agents- which focus upon internal bounds to
cognition.
Heuristics work by using small, but highly relevant, sets of information to resolve immediate
problems quickly. Investors apply heuristics to adapt to new information, for example as
market conditions change. New opportunities arise from the old heuristics. Think of the
“weightless economy” of the late 90's that made reported losses almost seem virtuous as
internet players struggled to “dominate the space” at all costs (Brynjolfsson and Kahn, 2000).
Since market conditions are subject to continuous and abrupt changes, investors will
continue to adjust and adapt, as the evolutionary ebb and flow of markets progresses (Lo,
2004).
According to the AMH arbitrage opportunities appear and disappear as a result of adaptive
responses to changing economic conditions; as they do so investors adapt new heuristics to
match new challenges. The most basic requirement of any trading strategy is its survival
value, its profitability and risk reduction value come after this most fundamental market
test.
Strategies that “blow up” their adopters are unlikely to live long enough to
18
appropriately adapt to the market environment in which they are deployed. So where do
new market strategies or ways of seeing the market come from?
De Bondt (2004) interviewed over 500 European investors with between Euro 100,000 and
Euro 1 million to invest and concluded “individual clusters of attitudes and belief, often
associated with national character, gender, age and religion influence portfolio choice.
[Hence] Culture matters.” As Polyani pointed out (1944) financial markets are embedded
within a broader social and political settlement whose transformation (in a revolutionary
episode) can breach the control of such heuristic tools control. So mental frames influence
investment strategies and are shared and communally formed. Such cognitive schema are
inherited, adapted and eventually abandoned, cannibalised or recycled in the face of new
realities. Crucially our financial and personal lives are integrated and mutually enforcing and
not sealed off purely economic calculations
But how does this happen? What do we know about the evolution of the behavioural norms
we all share? One norm we all rely on is some, perhaps minimal, degree of co-operation
from others. One setting in which this collaboration is sorely needed is within the context of
a prisoner's dilemma where a failure to co-operate induces losses for all. Axelrod (1984)
reports the results of a competition to devise a computer program capable of optimally
solving the prisoner's dilemma problem. The winning program always initially uses a cooperative strategy until the opposing player defects and then retaliates until the opponent
co-operates again. This ‘tit-for-tat’ strategy turns out to be optimal almost whatever the
opponent does. If my opponent is continually uncooperative he gets one period of grace
before my retaliation kicks in. If my opponent always cooperates so will I. Interestingly the
strategy that performed worst is the most seemingly “sophisticated” invoking learning and
probability distributions to constantly update behaviour. Within financial markets the
problem of counter-party risk, so vividly displayed during the recent sub-prime debt crisis
reflects such a choice (Khanani and Lo, 2007 and Lo, 2009). Heiner (1983, p563) notes a
rather similar pattern of development in the evolution of blackjack play amongst
professional players. Card counting techniques, requiring immense memory skills, have given
way to very simple structured response strategies. It appears simple structured strategies
dominate more explicit “maximisation” strategies in many competitive environments.
19
The evolving order of the AMH contrasts with the EMH for which efficiency is seen as a static
concept. It is important, however, to consider how the two hypotheses can be tested
empirically. Direct tests of the AMH are quite easy to conceptualise using the ‘top-down’
approach. A large number of tests have been designed over the years to test for the EMH
and usually these involve some sort of quantification of efficiency. In broad terms the AMH
can be tested by investigating whether the level of efficiency in a market varies significantly
over time. An increasing number of papers report results consistent with the AMH. Neely et
al (2009) report regularity in appearing and disappearing profit opportunities in the foreign
exchange market in the 1970's through early 1990's. Lo (2004) and Kim et al (2011) report
fluctuating levels of market efficiency over very long historical periods. In addition, quite a
number of studies report fluctuations and even reversals in well known stock market
anomalies (see, for example, Brusa et al 2005; Chong et al, 2005; Marquering et al, 2006 and
Moller and Zilca, 2008). Yet the AMH must presently be regarded more as a way of seeing,
rather than a fully specified alternative model to the EMH. In Section 5 we adopt that way of
seeing to re-evaluate some classic problems in Finance.
The following section of the paper outlines conceptual frameworks incorporating both
internal and external bounds. It reviews the inclusion of internal and external bounds
separately in the modelling of bounded rationality, highlighting the importance of taking the
two sets of constraints upon effective decision-making into account when modelling the
limitations upon rationality. These conceptual frameworks allow direct testing of the AMH
using ‘bottom-up’ approaches.
4. A more systematic and theoretically grounded basis for behavioural finance and its links
to AMH
In this section we examine how a more systematic and theoretically grounded basis for
behavioural finance can be developed. The internal and external bounds on financial
decision-making are examined in sub-section 4.1. Heuristics and adaptive behaviour are
examined in sub-section 4.2, followed by discussions of satisficing and investor
heterogeneity in sub-section 4.3. We show how each section implies testable hypotheses
allowing us to distinguish the predictive power of the AMH from the EMH using ‘bottom-up’
methods which is discussed in sub-section 4.4.
20
4.1. Internal and external bounds upon investment decisions
Both environmental context and cognitive boundaries are crucial factors when modelling
decision-making under uncertainty. There are two types of model of bounded rationality.
One type of model focuses on the external bounds on the investor. These models are based
upon investor optimisation under externally imposed constraints (Lee, 2011, p 514). Here
investors are assumed to be rational in terms of maximising some objective, whereas
satisficing is interpreted as non-optimal and simply accepting some, at least tolerable,
outcome given the external constraints. Following this external constraint approach, some
studies use an optimal search model framework, where investors do not possess full
information about the choices set they face. But Weitzman's (1979) allows investors to
invoke heuristics in deciding when to stop their search. Asymmetric information is often
invoked as investors are often not equally informed about the distribution and location of
the prices on offer.
The second type of model of investor satisficing is based on internal cognitive limitations
which tend to induce errors in the judgements of boundedly rational agents (Salop and
Stiglitz, 1977, Bekiros, 2010). The focus of this type of model is therefore on the internal
cognitive bounds upon the investor.
Following this approach, such models imbue
consumers/investors with heterogeneous response functions recognising that, for example,
investor's cognition may be reduced through habitual behaviour such as dividing price
quotes into, high, medium and low bands. In this example, traders with low cognitive ability
can only form a small number of partitions (for example the single partition into low and
high prices), whereas the more cognitively gifted traders are able to form a greater number
of partitions (say low, medium and high) and are thus able to more astutely assess the price
distribution.
Heiner (1983) portrays the origins of predictable investor behaviour as deriving from the
need of investors to assess the reliability of information before it is acted upon. Actions in
this view are only undertaken if they cross a threshold for the reliability of the information
eliciting the possible action. More frequently deployed actions are by their nature more
reliable, if only because their effect has been played out more. The reliability of information
is a function of both an investor’s perception and environment. For a given environment less
21
perceptive investors will be able to assess the reliability of a smaller range of actions,
because of their inability to recall them being used. This makes sense when we compare the
variety of behaviour we observe in humans compared to animals, or cats compared to
baboons. The more developed the mind the less repetitive and predictable the agent’s
behaviour. So in Heiner’s (1983) framework task uncertainty reduces the scope of actions
adopted by challenging investors’ to assess the reliability of possible actions. It is this very
contraction of the feasible action set, induced by the need to select a reliable action, that
produces predictability in observed investor behaviour. So when reliability of actions is an
issue uncertainty may be a boon for the social scientist/researcher, by rendering investor
behaviour predictable.
The above two models of bounded rationality present an incomplete picture of the notion of
bounded rationality since they only apply one type of bound. In a more complete version of
bounded rationality, both internal and external bounds intertwine in determining the
agents’ decision-making. When only one set of bounds is taken into account, bounded
rationality is seen as a barrier to an optimal solution whereas if the two sets of bounds are
taken into account, as Simon’s scissors metaphor shows, a richer concept of rational action is
born. This fit between internal and external bounds highlights the evolving nature of limits
upon investor rationality.
According to this holistic view of decision-making our environment contains both potential
problems and solutions. This is because environments are constructed by decision-makers’
behaviour and the emerging structure of our environment interacts with the heuristics each
trader adopts in order to manage and remould that environment. Therefore, the structure of
the environment maps onto and is redrawn by to the structure of individuals' cognition. This
implies the following hypothesis:
Hypothesis 1: The structures for both internal cognition and the external environment
within which it arises define rationality for investors within an evolving market order.
The inclusion of both sorts of bounds allows an evolved form of rationality to emerge
amongst investors. Successful investors respond to their changing external environment
adaptively and appropriately. This matching process of environment to its cognition is
22
undertaken by the construction and adaption of heuristics/mental-frames aimed at limiting
cognitive errors without incurring unacceptably high costs in the struggle.
4.2
Heuristic and adaptive behaviour
The use of decision-making heuristics and the construction of mental frames is perhaps the
most influential departure from standard expected utility theory, it plays an important
function in prospect theory (Kahnneman and Tversky, 1979, 1992), the theory of mental
accounting Thaler (1985, 1999) and asset pricing Bernartzi and Thaler (1995).
Early studies of heuristics were associated with error-prone intuitions or apparent
irrationality (Kehneman and Tversky (1992). In the literature, heuristics are also viewed
positively as a natural feature of decision choice requiring a combination of learning and
adaptation (Gigerenzer and Brighton, 2009). Heuristics that are not rational if cognitive
processing was limitless may be understood as a form of constrained rationality given our
true mental abilities.
Recent studies attempt to model heuristics using an adaptive tool box built on three decision
blocks of (a) searching, (b) stopping rules and (c) final decision criterion. These studies show
that employing specific heuristics in an adaptive way helps to make accurate, fast and frugal
decisions. The frugality derives from adapting least cost filters to screen out clearly inferior
options and settle upon good enough outcomes. In this view, heuristics are the internal
mechanisms that guide search and determine when it should end. Using heuristics in this
adaptive way means that the two sources of bounds upon rationality, from internal and
external sources, nest neatly. So Gigerenzer et al (1999, p 13) state
“A heuristic is ecologically rational to the degree that it is adapted to the structure of
the environment.”
Here “'ecological' is just another word for the occurrence of a rule-governed, self-organized
order” (Smith 2008, p. 6). Decision-makers can make good decisions, by using their mind to
best infer future value given by trading in their environment. If external bounds are fairly
immutable from the investors’ standpoint, then internal bounds can be evolved to take
advantage of the structure of the external environment. A relatively stable environment
combined with a constantly adapting investor understanding of that environment makes for
better decision-making.
23
Heuristic search involves identifying a search procedure requiring limited computational
capacity to render complex decisions solvable. This involves an “editing” of decisions to
leave a feasible choice set capable of fairly simple, rule-based, evaluation. So, for example,
we sort stocks into value or glamour, small or big, bins prior to final selection. Heuristic
rules/mental frames are used to manage the consequences of cognitive limitations and
especially so in situations where there are constraints in time, cost and skills or when
information is limited or ambiguous. There are two reasons for rationality to break down:
firstly, beyond a certain level of complexity our logic fails us. Secondly, in the cut and thrust
of market trading investors cannot rely on the rationality of other agents, so they are forced
to make guesses regarding the behaviour of others.
Consider the game of chess. On the first move each player can move any of their 8 pawns or
either of their 2 Knights. This implies 40 moves are available to play in total at the opening
move. This fact implies at the end of first move 400 board configurations are possible. On
the second move this expands to 71,582 possible board configurations (Shenk, 2006, p. 69).
So good Chess players, like great traders, must make fast and frugal decisions focussing
directly on likely evolutionary paths of play and resulting threats and opportunities. Similarly
Rubic’s cube has 43 trillion possible initial starting points (Heiner, 1983, p 563). So recalling
the best way to unscramble the cube from each of them is impossible. So good players
simply sequentially adopt a few basic unscrambling strategies pursuing them until they
prove fruitless. While this involves far more manipulations of the cube than a
perfect/optimal strategy, for any given starting point, it is hugely superior to any random
search strategy.
This results in subjective beliefs and the assumption of managed risks.
Given the possible outcomes of play are so numerous co-ordinating the editing of possible
outcomes, prior to decision-making, becomes impossible even if players/traders were to
attempt it. Hence investors are almost certain to adopt very different mental frames for
trading purposes and it is only the evolutionary process of trading itself that can remove the
most unhelpful frames and promote the power of the most successful frames. In this way
each investor “promote(s) an end which is no part of his intentions.” (Smith, 1776, p. 421).
This is achieved by unconsciously sifting and perfecting new, more appropriate, mental
schemata thrown into prominence by changing market conditions.
24
Adaptive agents use specific heuristics that seem appropriate to the changing environment.
In this context adaptive decision-making is rational. Successful heuristics rotate over time,
with no fool-proof trading rule being on offer. As Lo (2004) points out rationality in the
context of EMH and irrationality in the context of behavioural finance are two sides of the
same coin i.e. agents adjust and adapt to new information. This is why Lo favours replacing
the polarities of rationality and irrationality with a more nuanced concept of adaptive
behaviour in complex, evolving, markets. This is consistent with a definition of an evolved
rationality, where investors try out, adapt and reject heuristics as they assess new market
opportunities and withdraw from past loss-making positions. From the above the following
hypothesis is derived:
Hypothesis 2: Adaptive decision-making enhances alignment between the structures
of internal cognition and the external environment.
Here the reliability of a strategy, and hence its practical use, depends on both the
decision-makers environment and their perception of it. Heiner (1983) reminds us that the
most sophisticated agent’s may seek to change their environment as well as improve their
ability to perceive it. Of course such possibilities are limited, which may explain the
extinction of many animals with poorer perceptual abilities than humans.
More appropriate/adaptive responses to the trading environment enhances the match
between internal and external bounds. This hypothesis accords with the fact that cognitive
limits are not necessarily a curse, but may rather be a blessing in guiding investors to fast,
frugal, yet profitable, trading decisions in a world where time is money and the sources of
money change rapidly.
4.3.
Satisficing and heterogeneity
Satisficing, as opposed to optimising, might be thought of as the first fruit of bounded
rationality. Satisficing embeds the art of the possible in a constrained environment with
sparse information and tight deadlines on decision-making.
Satisficing fulfils feasible
aspirations while accepting money will always be left on the table. Often we must choose
the least bad option. But investors’ ability to conceive of and rank options will differ. For
some investors a tech stock is a tech stock, but some prescient investors, like John Doerr of
25
Kleiner Peabody Caulfield Byers realise Amazon, Netscape and Google are just that little bit
special. It is from such refined mental processes that fortunes are made.
Satisfaction of investors’ desires is limited by their cognitive abilities and powers of selfcontrol and these will differ across investors. Investors differ in their psychological makeup
and cognitive brain function, as well as their risk-profiles and investment horizons. From
the above, the following hypothesis is derived:
Hypothesis 3: Optimising agents who utilise their best internal capacity to adapt to
their observed environmental structure are satisficers.
Investors optimise their investment portfolios subject to their personal cognitive bounds and
ability to match their cognition to objective market conditions. Putting this interpretation of
satisficing into a financial context implies that investors differ in their trading strategies in
addition to setting different goals according to their skills and knowledge.
From the above discussion we identify four main features of bounded rationality for
incorporation into a successful modelling framework for investment. These features, listed
below, will be discussed within this section and applied in Section 6.
1) Investors’ cognitions are bounded in different ways and to different degrees.
2) The goal of a satisficer is to achieve an acceptable solution by using their current
cognition, taking the heterogeneity of other agents into account.
3) Through adapting appropriate heuristics, the investor's own cognitive bounds and the
bounds set by their trading environment are matched.
4) Agents’ satisfy their goals through adaptive behaviour and the adoption, recasting and
rejection of decision heuristics.
The interaction between investors as decision-makers within financial markets is illustrated
in Figure 1. Here markets are in the short run “voting machines” not weighing machines as
the famous value investor Benjamin Graham pointed out (Graham and Dodd, 1934, p 23).
What matters is not what a trader believes but what traders think others believe. The
market reflects the instability of Keynes' “beauty contest” which yields an unstable
equilibrium in investors' expectations. Expectations and decisions by investors shape market
26
outcomes, the resulting structure of the market order is then reflected in individual
investor’s assessments of the effectiveness of their chosen trading strategy. This drives
iterative rounds of adoption, adjustment and rejection of mental frames by investors. This
adaptive process can be both costly and slow as a new order tentatively emerges.
Figure 1 about here
4.4.
Implementation of Artificial Intelligent Agents
This sub-section outlines how artificial Intelligence models can be constructed in order to
test the hypotheses of the prior subsections.
In response to the highly centralised, top-down and deductive approach that characterises
the mainstream neoclassical economic approach, agent-based modelling using artificial
intelligent agents is decentralised, bottom-up, and inductive, within dynamic environments
Agent-based modelling was developed in the early 1980s and is now a rapidly growing
approach to modelling economic behaviour (Duffy, 2006). While looking at the advantages
and disadvantages of this computational approach to economic behaviour Duffy (2006),
shows that these models often provide a better fit to experimental data and he suggests the
possibility of combining an agent-based methodology productively with human
subjects/experimental research methods.
Agent-based models have the considerable advantage that controlled experiments can be
run with the attributes of both the population of agents and market being capable of being
adjusted to see the effects on the resulting market dynamics (see, for example, Cincotti et
al., 2003; Yeh and Yang, 2010; Tseng et al., 2010)). Thus it is possible to directly test
hypotheses such as those developed in the three subsections above. An interesting sub set of
this work is that looking at ‘zero-intelligence’ agents who have no trading skill at all to see
how they fare in markets sometime in interactions with more skilled or faster learning
agents (see, Yeh, 2008). In these exercises the roles of skill and chance can be explicitly
investigated in a way that is very difficult in the traditional paradigm.
27
Machine learning refers to systems capable of independent learning; that is integrating and
acquiring knowledge. Researchers in artificial intelligence are actively engaged in developing
learning algorithms applicable in a real world complex system context. Modelling a financial
market is an ideal application for this field because financial markets are huge and complex.
Investors’ behaviour is complex and conflicting and so far standard linear asset pricing
models have struggled to describe it adequately. Therefore, the methodology and modelling
that allows agents to learn and adapt may become a crucial part of modelling financial
market behaviour.
Reinforcement Learning is a sub-field of machine learning and now commonly used in
artificial intelligence originated from psychology. Investor behaviour is heuristically
determined and depends on agents’ cognitive ability to learn and adapt, therefore RL
methodology appears to provide an ideal approach to model agents’ learning, experience,
intuition, knowledge and beliefs.
RL provides a formal framework for defining the interactions between goal-directed learning
agents within a dynamic environment in terms of states, actions and reward (Sutton and
Barto, 1998). Sutton and Barto (1998, p.3) describe RL as learning how to map actions into
the environment in order to maximise reward. They continue their definition as follow: “The
learner is not told which action to take, as in most forms of machine learning, but instead
must discover which actions yield the most reward by trying them out. In the most
interesting and challenging cases, actions may affect not only the immediate reward faced
but also the next situation encountered and, through that, all subsequent rewards. These
two characteristics-trial and error search and delayed reward- are the most important
distinguishing features of RL”.
Moreover, the dynamics of RL can be reduced to four main elements: a policy, a reward
function, a value function, and a model of the environment. An agent searches for an
optimal policy which maximises total reward. The agent is continuously analysing the
consequences of their actions through trial and error interactions with the environment. In
more recent RL frameworks agents simultaneously learn by trial and error a model of the
environment and, the use of that model for planning (Sutton and Barto, 1998).
28
RL consists of a set of environmental states, S, a set of agent actions, A, and a scalar
reward/reinforcement function, R, mapping environments and actions into pay-offs. The
environment is typically modelled as a finite-state Markov decision process (MDP), this make
RL algorithms highly related to dynamic programming techniques.
Greif and Laitlin (2004) show reinforced learning within institutions can make the difference
between continued survival and collapse. Comparing the dominant post-war socialist parties
in Italy and Sweden and the medieval cities of Genoa and Venice they trace how an inability
to learn to adapt to a radically changing environment lead to the demise of the Italian left
and the Genoese podesteria. These authors show how reinforcement learning can be built
into a simple repeated-game structure. They conclude, after considering a variety of casestudies of institutional reform, that while initially institutions reinforce their structure over
time exogenous shocks tend to undermine this trend by posing an existential threat to the
increasingly atrophied institution. According to a survey done by Kaelbling et al. (1996) RL
has been successfully applied to complex problem-solving and decision making.
RL can be applied for the purpose of getting an optimal policy that resolves control problems
as well as prediction problem. Different versions of RL have been used in the context of
financial markets. For example, Lee (2001) adopts a RL algorithm which learns only from
past data for stock market prediction using the Korean stock market. He finds RL provides a
more useful indicator than conventional indicators of stock trading and he recommends
extending this approach by applying other RL algorithms where the definition of reward on
offer is slightly different from that in his paper.
An adaptive fuzzy RL approach, which combines the adaptive action selection policy of RL
and fuzzy logic, is employed by Bekiros (2010). In this approach agents beliefs are
represented by fuzzy inference rules to advance the literature on heterogeneous agents
based market. He finds that the predictive ability of the adaptive fuzzy RL is significantly
higher than other competing models.
5.
EMH or AMH? - Some applications.
29
In this section we try to illustrate how a shift to viewing financial markets through the lens of
the AMH, as opposed to the traditional EMH, might aid our understanding and improve
research methods in finance. The greatest effect of such a shift would be to broaden the
type of questions that can legitimately be asked in finance, a prospect we discuss in our
conclusion. However, an AMH focus would illuminate many existing issues and the specific
ways they are conceptualised and addressed.
To illustrate the AMH approach we discuss three specific and important issues out of many
in the existing literature that can be illuminated by the AMH approach. We revisit these
standard research questions through the lens of our three proposed research hypotheses
developed above. We ask how those hypotheses might be used to enhance current model
building and financial decision-making. We later show the AMH helps us to understand State
intervention in financial markets not as a constraint on the operation of those markets but
as part of the evolution of financial institutions’ market power into full expression in the
political arena.
5.1 Three examples of applying the AMH
The first example we deal with concerns the random walk.
Perhaps the most basic
implication of the EMH is that market prices follow a random walk. Tomorrow’s price will
only differ from today’s price in random unpredictable ways. As stated above, an array of
empirical studies have rejected this claim in favour of under-reaction/momentum at
relatively short intervals of a year or so and longer-term correction/reversals implying earlier
market overreaction. Such refutations have emerged as a set of “anomalies” which form
much of the grist to the current mill of behavioural finance studies.
i) Momentum
Recently Barberis, Sheifer and Vishny (1998) have advanced a theoretical model in which
earnings (and hence stock values) always follow a random-walk, as the EMH implies, but
investors perceive them to either revert to trend or exhibit momentum with some
predetermined probability. The transition between an investor focus upon momentum and
reversion regimes with a corresponding effect on prices, in clear denial of the random-walk
property of observed prices, is an obvious anomaly/irrationality within the EMH framework.
30
These agents fail to comply with the requirement of Hypothesis 3 above in matching their
perceptions of value creation to the observed environment; thus they fail the test
hypothesis 3 poses for boundededly rational agents. The distinction here, as in Hypothesis 1
above, is between the investment strategy and its cognitive and external context, as
opposed to a proposed statistical property, which is conjectured to prevail regardless of
context or cognition. The AMH cautions us that it is not primarily value characteristics
themselves that drive price movements but rather the mental frames and heuristics through
which value is perceived. Hence the irrationality of investors, as suggested by the EMH, in
the Barberis et al (1998) framework is simply their adaptability, or evolving rationality, to
those adopting the AMH lens on the same phenomena.
The success of momentum based trading strategies is known to be a function of the
extent of informational asymmetry prevailing in the financial market (Easley et al, 2002),
suggesting the need to adapt trading strategies according to the environment into which
they are deployed. Neither momentum nor reversion/correction based strategies are
dominant universally but rather are revealed to be so by deployment into a favourable
trading environment.
So either the transition probability, underpinning the markov-
switching process between regimes in the Barberis Shleifer and Vishy model tracks changes
in the regime probabilities (as investors flip between momentum/reversion “routines”) or
the choices made by competing autonomous agents are allowed to aggregate to determine
overall market asset demand and supply schedules.
Seru et al (2010) show market learning consists of two types of correction. Firstly
investors discover their true market ability in the process of trading. Secondly those who
realise they have little aptitude for trading, having been punished with losses, exit the
market in order to passively hold a diversified portfolio. Those who so exit may be largely
individual investors whom Hur et al (2010) identify as being particularly associated with the
presence of profits to following momentum based strategies.
ii) The disposition effect
It is now well established that investors seem to sell winners (stock that have had recent
rises in price) too quickly and hold losers (stocks with recent price declines) too long. In any
market with transaction costs longer trades are, on average, more profitable trades. The
costs of trading opens up a range of prices between which arbitraging differences in prices
31
of assets of equivalent value makes no sense (Constantinides, 1983). In their seminal study
Schlarbaum et al (1978) document this fact. Returns to short holding periods are highest
amongst their sample of individual investors, although surprisingly good more generally. So
why do the returns of individual investors decline in the holding period over which they
trade?
Shrefrin and Statman (1985) argue that this occurs because such investors have a disposition
to sell winning stocks, whose price has just risen, too quickly and hold on to losing stocks,
whose price has just fallen, too slowly. They do this to avoid the “regret” of realising paper
losses or missing out on easily envisaged gains. This pattern of trading is all the more difficult
to understand given tax incentives to sell stocks whose value has fallen to enable the
realised loss to be used as a tax shelterv (Constantinides, 1984) . This is because the tax rate
levy/relief on capital gains/losses are equal so it makes sense to realise losses in order to
help harvest longer-term post-tax returns on your winning stocks. It appears individual
investors at least do not seem to trade to maximize risk-adjusted returns, leaving tax
subsidies from the government on the table. In contradiction of hypothesis 3 they fail to
match their cognitive frame to the tax environment they find themselves in, suggesting they
are not rational in this bounded sense.
What about more sophisticated traders, such as hedge funds and investment banks? Hur et
al (2010) confirm that the presence of both momentum and the disposition effect is directly
proportional to the percentage of individual traders present in each market segment. As
with momentum strategies the intensity of a heuristic and its explanatory power varies with
the environment into which it is deployed. While institutional investor dominated markets
offer little scope for arbitraging the disposition effect more “noisy” markets, in which
individual investors have a greater share of trades, offer greater opportunities for counterdisposition effect arbitrage profits to be reaped. Hence the disposition effect flourishes in
individual investor dominated markets but is arbitraged away elsewhere.
Yet it is not only the traders’ characteristics that determine the intensity of the observed
disposition effect. Kumar (2009) shows firstly that the intensity of the disposition effect is
strongly related to various proxies for the uncertainty involved in valuing the stock and that
there is also evidence informed traders recognize such pockets of “hard to value” stocks and
that they proactively trade to arbitrage out such mispricing. Using standard metrics of the
intensity of informed tradingvi Kumar (2009) reports evidence that informed traders actively
32
predate upon the relatively uniformed who misguidedly trade in hard to value stocks. This
evidence brings us closer to direct observation of the elimination of noise-traders in a
process of Darwinian competition to implement profitable investment strategies.
iii) corporate governance and organisational form.
A tragic example of this process of a mismatch of cognition within a changing environment is
the way in which British Petroleum’s “upstream” exploration and extraction unit, BPX, was
focussed, by its then Head John Browne, on identifying very large hydrocarbon deposits of
oil and natural gas; so called “elephants” in the industry (Roberts, 2007). In the 90’s BP had
lost ground to smaller, niche player, independent explorers/extractors. So Browne cut
competing “asset managers” loose within BPX to compete as they saw fit. This replaced
centralised control with information-sharing amongst “peer group” autonomous asset
managers within BPX (Roberts, 2007, p 25-27). The Deepwater Horizon rig explosion (in the
Gulf of Mexico on April 20th 2010) and the subsequent inability of BP to cap the
underground well until mid September that year exposed the dangers of such intense
internal rivalry. Safety thresholds were pushed to breaking point under such intense
pressure. Following the spill survival required hyper-vigilence with regard to rig safety as
politicians made clear their willingness, if not eagerness, to react to public fury. Hence
corporate history matters, especially in times of economic crisis by determining the nature
and scope of corporate routines available to enable survival.
This “path dependence”, deriving from the effect of history of the array of corporate
routines is most starkly illustrated by the claim that shareholder rights are better protected
in common law jurisdictions (largely the UK and its former colonies, US, India, Malaysia, etc)
than in civil law ones (France, Germany or their former colonies) (La Porta, et al, 1998,
2000). Yet, as other others have pointed out in critiquing the “path dependency” story of
national shareholder protection arrangements, the civil law retains much of the plasticity
the common law offers by differential enforcement of a unified set of codified company laws
(Pistor et al, 2002, Pistor and Xu, 2003). The AMH allows us to understand both the “path
dependency” of governance arrangements and how legal enforcement of shareholder’s
rights operates as an evolutionary selection strategy within seemingly divergent governance
paths.
33
The examples given above have been drawn directly from the existing finance and
accounting literature and the insights gained can be fitted reasonably easily into the
standard methodological approach of creating analytically tractable models. However, the
traditional analytical methods are very limited in their scope to model interactions between
large numbers of heterogeneous agents when decision systems evolve though time. In so
far as papers in the literature deal with these issues, most either use either single
representative agent models with non-constant belief systems (see, for example, Barberis
and Huang, 2008) or models with two types of agent, one rational and the other nonrational (see the literature on noise trader models, for example, De Long et al , 1990).
Models with much more complexity than this quickly become intractable analytically. It is
possible, however, to use computer simulations to directly model the interactions of large
numbers of heterogeneous agents with varying degrees of intelligence.
This use of
intelligent agent models is a very fast growing field (see, the special issue of the Journal of
Economic Surveys in 2011 devoted to this and closely related topics) but is currently largely
disconnected from mainstream academic finance. Work of this nature is often done by
computer scientists, mathematicians and physicists and receives very little coverage in the
most prestigious mainstream finance journals. Given that the assumptions adopted in this
field are evidently a much closer depiction of reality this field of research looks very
promising and its dismissal by the mainstream all the more troubling.
5.2 The AMH and financial reform
The AMH allows us to see the relationship between the markets and the State in a new,
perhaps more helpful, way. A keystone of the EMH is the belief that markets know best and
the State can at best only re-allocate wealth produced by the gains from the trade that
characterise free-markets. The recent financial crisis makes a mockery of such claims. In
reality the primary role of the TARP, UK bailout of RBS, EU bailout of Dexia, Bankia, etc, has
been to recapitalise bankrupt private sector institutions at the tax-payers expense, in a
“socialism for the rich” parody of the welfare state. In reality two parallel, and often
complementary, markets explain these facts. The market for votes is itself as much central to
effective financial hegemony as any financial market exchange. Increasingly political and
financial power has been fused in a way that makes the prudential role of banking
34
authorities much in doubt (Johnson and Kwak, 2010). The central figures in the management
of the financial crisis, Hank Paulson, Timothy Geithner and Larry Summers all had previous
lives in investment banking or headed straight to such an institution slightly after their
tenure in office. The close integration between the State and financial power suggests a
symbiotic, and perhaps even mutually parasitic, existence. The AMH recognises the need of
investment institutions to adapt to pressures from vote markets within democratic polities.
Political preferment is thus just another survival/advancement strategy of those using vote
markets to buttress their position within allegedly “efficient” financial markets. State
intervention is not necessarily a constraint on market forces but rather one further
manifestation of market forces power.
5.3 The AMH and nature of Darwinian economic competition.
The AMH with its acknowledgement of evolutionary behaviour makes clear some of
the problems of equating competition with automatic improvements in social welfare. Much
economic competition is truly Darwinian in that it is being ranked first as an individual, not
increasing the overall social product, that counts. Frank (2011, page 7) states the problem
thus
“Charles Darwin was one of the first to perceive the problem clearly. One of his
central insights was that natural selection favours traits and behaviours primarily according
to their effect on individual organisms not larger groups.”
A tension arises between the interest of individuals and the broader social group because
natural selection implies traits which confer individual fitness thrive regardless of their
consequence for the broader species. Frank (2011, p12) gives the example of drug cheats in
the world of professional sports. All cyclists may agree that a “clean” race is the only race
that counts. But each may feel forced to use drugs in a World where “everyone is doing it”.
The need to stretch to reach the highest possible rank means we may achieve the worst of
all possible worlds, with no worthy Champions to hail, rather than the best.
The proposed financial transaction/Tobin tax (FTT) 0.1% tax on all bond and equity
transactions now adopted by ten EU member states might be seen as a helpful attempt to
bring about a community-wide benefit in the face of individual nation states incentives to
act as a home to “flight capital”. While access to such capital clearly benefits whichever State
in which it chooses to domicile the need to appease it may serve to reduce community-wide
35
wealth and social cohesion. Such a “public good” might be seen as sufficient justification for
State intervention to dampen speculative trades. Indeed the costs of an individual State
adopting a FTT is already well known from Swedish experiments in the 1990’s. So it is clear
any benefits from the FTT can only be reaped by transnational co-operation rather than
competition. An evolutionary perspective helps us understand that the problem is not that
financial markets are insufficiently competitive, but rather that the intensity of competition
we observe is damaging to broader social objectives of stability. Hence the State has a
legitimate role in constraining individual competitive effort which damages broader social
policy
This symbiotic relationship between the State and the market has been known at
least since Karl Polyani, explained how the “laissez-faire” system of free-market capitalism
was planned by England’s political ruling elite. In a similar way today intense and resource
intensive “rent seeking” is deployed by the wealthy to divert resources from the middleclass to themselves (Stiglitz, 2010). The predatory lending practices and credit card services
marketing of financial institutions are examples of legal, but clearly immoral, transfers from
the poor to the wealthy. But in this the banks follow a well recognised path in procurement,
privatisation and regulatory practice which facilitates State mandated transfers from the
middle–class to a wealthy elite. The impact of such transfers is already very clear in the data.
In 2007, just before the Crisis hit America, the top 0.1% of American households had an
income 220 times higher than the average for the remaining 90% of American households
(Stiglitz, 2012, p 2)
More recently Rajan (2010) has traced the roots of the sub-prime crisis to the
increasing impoverishment of the uneducated and unskilled, often black, citizenry in the US.
From 2001 onwards economic recovery, even when it came, often was of the jobless
recovery form, where new jobs went to the young, better educated, more flexible elements
of the workforce leaving behind a surplus to requirement residue. The absence of any longterm social welfare provision in the US means such victims of past restructuring enter a
lower paid, more transient, workforce more exposed to the vagaries of the market.
The obvious way of alleviating the most pressing needs of the swollen ranks of the
working poor, short of State provision, was easier access to credit. While this stoked up
problems for the future it did at least ameliorate the anxiety of current poverty. Rajan (2010,
p 35) finds no shortage of evidence that US government policy promoted the indebtedness
36
of the working poor. For example the Federal Housing Enterprise and Housing Enterprise
Safety and Soundness Act of 1992 instructed the Department of Housing and Urban
Development (HUD) to develop affordable housing goals specifically targeting ethic
minorities and those living in zip/post code areas rarely attracting mortgages. This initial
encouragement to the development of the sub-prime lending market was reinforced later by
President George W. Bush as part of his vision of expanding the American dream via a
homeowners’ democracy.
Recently Ferguson (2012) has diagnosed Western capitalism as entering an era of a
“great degeneration” as market participants are driven to use rent-seeking to derive benefits
from the State over claims to a constant, or even declining, national product. This descent
into “stationary state” allows the rule of law to be replaced by the rule of lawyers as
distributional conflicts dominate attempts to create a more efficient, productive, society.
Within the modern “beehive” of financial networks extreme complexity allows small
perturbations within the system to swiftly gather pace with disastrous consequences. While
such network externalities are now being addressed by finance theorists
(Easley and
Kleinberg, 2010) and implemented tentatively by regulatory authorities (Gai et al, 2011) they
remain irreconcilable with the standard finance theory of price-taking trades amongst
anonymous atomistic agents.
Financial institutions are the embodiment of the “rules of the game” of financial
markets (Heiner, 1983) and it is these very rules which have now clearly broken down. So
much of the post crisis challenge lies in appropriate post-crisis institutional reform. While
many commentators portray almost any institutional reform as an attempt to traduce
market allocations this ignores the role of financial institutions in maintaining the integrity of
financial markets and hence securing gains from trade. A competitive strategy which gives
an individual an edge ahead may nudge the society he lives in backwards. If free-market
order takes much planning, as Polansky suggests (1944), it is financial institutions which
must both bring a free market into existence and maintain its integrity once it exists. As Greif
and Laitlin (2004) show such institutions are limited in their capabilities of internal renewal
and so likely to being overwhelmed by historical events and hence collapse. This seems to
have been the fate of many of the existing regulatory authorities discredited by the crisis
(the UK’s FSA being perhaps the most obvious victim). Such bodies are unable to survive the
37
competitive process due to an inability to adapt to the greatly intensified rent-seeking
behaviour the financial crisis brought.
6.
Conclusion
In the light of recent events which have exposed the shortcomings of the current financial
paradigm this paper discusses the implications of moving to an approach based on the AMH.
The AMH is much less theoretically restrictive than the existing paradigm as it does not
assume that market participants uniformly act in accordance with the rationality axioms of
neo-classical economics. Instead more realistic notions of bounded rationality are totally
consistent with the AMH approach.
We present the main features of bounded rationality suggesting three testable hypotheses
to determine the degree to which observed trading behaviour conforms to the tenets of
bounded rationality. This reflects an evolutionary concept of bounded rationality that seems
more psychologically plausible than that commonly invoked by Finance researchers in its
EMH form.
As we have seen in the previous section the move to this approach would enable many
specific issues in the discipline to be addressed in a more constructive and meaningful way.
Many empirically established facts would cease to be troublesome anomalies to be ignored
or marginalised but phenomena to be integrated into the mainstream of the subject.
Yet it is in forming policy regarding financial markets that the transition from an EMH to an
AMH paradigm may yet have it biggest impact. Possibly the greatest effect might be at a
conceptual level by making acceptable discussion about and research into issues that would
previously have been regarded as resolved. It is clear that more questioning of this nature
might well have helped to avoid or alleviate the present crisis.
The AMH is still a tentative and untested competitor to its EMH rival. But we have shown
some reasons to be excited about its sphere of application both in theory testing and policy
discussion. The evolving financial crisis may provide the context for such alternatives to the
prevailing orthodoxy to at last be taken seriously.
38
References:
Akerlof, G, Shiller, R.Animal Spirits: How human psychology drives the economy and why it
matters for global capitalism:Princeton University Press New Jersey, USA: 2009.
Armstrong, M, Vickers, J Zhou, J. Prominence and Consumer Search. RAND Journal of
Economics2009:209-233.
Arrow, K.J. Rationality of Self and Others in an Economic System. The Journal of Business.
1986: Vol. 59, No. 4: 385-399.
Arthur, W. Inductive Reasoning and Bounded Rationality. The American Economic Review,
1994 No. 2 (Papers and Proceedings of the Hundred and Sixth Annual Meeting of the
American Economic Association 1994:(May) 406-411.
Axelrod, R.The evolution of cooperation: Basic Books Cambridge, Massachussetts, USA:
1984.
Barberis, N, Shleifer, A,Vishny, R.A Model of Investor sentiment, Journal of Financial
Economics 1998: 49(December) 307-343.
Basov, S. Bounded rationality: static versus dynamic approaches. Economic Theory
2005:25(4), 871-886. Available at: http://works.bepress.com/suren_basov/3.
Bekiros, S. Heterogeneous trading strategies with adaptive fuzzy Actor-Critic reinforcement
learning: A behavioral approach. Journal of Economic Dynamics and Control2010:34(6)
1153-1170.
Benartzi, S, Thaler, R.Myopic loss aversion and the equity premium puzzle. The Quarterly
Journal of Economics1995: 110, 73-92.
Bernard, V. and Thomas, J. Evidence that stock prices do not fully reflect the implications of
current earnings for future earnings. Journal of Accounting and Economics:1990 13: 305340.
British Academy, Open letter to Her Majesty the Queen. 2009:
http://www.britac.ac.uk/events/archive/forum-economy.cfm
Brock, W,Hommes, C.A rational route to randomness. Econometrica: Journal of the
Econometric Society1997:651059-1095.
Brown, A, Chua, Z,Camerer, C. Learning and visceral temptation in dynamic saving
experiments.Quarterly Journal of Economics2009: 124(1) 197-232.
Brusa, J, Pu, L, Schulman, C. Weekend Effect, ‘Reverse’ Weekend Effect, and Investor Trading
Activities.Journal of Business Finance &Accounting2005: 32(7/8) 1495-1517.
39
Chang, E, Lewellen, W.Market Timing and Mutual Fund Performance. Journal of
Business1984: 57 57-72.
Cincotti, S. ,Focardi S. M., Marchesi M., Raberto M., Who wins? Study of long-run trader
survival in an artificial stock market. Physica A 2003: 324 1–2 227–233.
Chong, R, Hudson, R, Keasey, K, Littler, K.Pre-Holiday Effects: International Evidence on the
Decline and Reversal of a Stock Market Anomaly. Journal of International Money and
Finance 2005: 24 1226-1236.
Clarkeson, R. Actuarial insights into the global banking catastrophe. Journal of Financial
Regulation and Compliance 2009:17(4) 381-397.
Cootner, P. (Ed) The Random Character of Stock Market Prices. Cambridge MA: MIT Press,
1964, 297-337.Critical Perspectives on International Business, Special Issue on the Financial
Crisis 2009: 5(1/2).
Critical Perspectives on International Business, Special Issue on the Financial Crisis 2009:
5(1/2).
Daniel, K, Hirshleifer, D,Teoh, S. Investor psychology in capital markets: Evidence and policy
implications, Journal of Monetary Economics2002: 49(1) 139-209.
Daniel, K, Hirshleifer, D, Subrahmanyam, A. Investor psychology and security market underand overreactions. Journal of Finance1998: 53(6) 1839-1885.
De Bondt, W, Thaler, R. Does the stock market overreact? Journal of Finance1985:40(3) 793-805.
De Bondt, W,Thaler, R. Further evidence on investor overreaction and stock market
seasonality. Journal of Finance1987:42(3), 557-581.
De Bondt, W. A portrait of the individual Investor, European Economic Review 1998:42 831844.
De Bondt, W. The value and beliefs of European investors. In Knorr Cetina, K, Preda A., eds.
The Sociology of Financial Markets, Oxford University Press: 2005.
De Long, B., Shleifer, A. Summers L. Waldman R. Noise trader risk in financial markets.
Journal of Political Economy, 1990: 98(4): 703-738.
Easley, D., S. Hvidkjaer, O’Hara, M. Is information risk a determinant of asset returns?
Journal of Finance 2002:57(5): 2185-2221.
Easley, D. and J. Kleinberg. Networks, Crowds and Markets: Reasoning about a highly
connected world. New York, USA, Cambridge University Press. 2012
40
Fama, E. Random Walks in Stock Market Prices. Financial Analysts Journal. Sept-Oct 1965.
55-59.Fama, E. Efficient Capital Markets: A Review of Theory and Empirical Work. Journal of
Finance, 1970:25(2) 383–417.
Fama, E.Market efficiency, long term returns and behavioral finance. The Journal of Financial
Economics 1998:49 283-306.
Ferguson, N. The Great Degeneration: How institutions decay and economies die. London,
England, Allen Lane. 2012.
Findlay, M, Williams, E. A Post Keynesian View of Modern Financial Economics: In Search of
Alternative Paradigms, Journal of Business Finance and Accounting1985:12(1) Spring.
Fox, J.The Myth of the Rational Market:Harriman House. Petersfield, UK: 2010.
Frank, R., The Darwin Economy: Liberty competition and the public good. Princeton,
Princeton University Press. New Jersey, USA, 2011
Gai, P., Haldane, A. Kapedia, S. Complexity, concentration and contagion. Journal of
Monetary Economics 58:July 453-470. 2011
Gabaix, X,Laibson, D. A boundedly rational decision algorithm. The American Economic
Review2000:90(2) 433-438.
Gigerenzer, G, Brighton, H. Homo heuristicus: Why biased minds make better inferences.
Topics in Cognitive Science2009:1(1) 107-143.
Gigerenzer, G, Gaissmaier, W. Heuristic Decision Making. Annual Review of Psychology,
2011:62 451–482.
Gigerenzer, G, Goldstein, D. Reasoning the fast and frugal way: models of bounded
rationality. Psychological Review 1996:103(4), pp. 650-669.
Gigerenzer, G, Selten, R. Bounded Rationality: The Adaptive Toolbox: MA MIT
PressCambridge : 2001.
Gigerenzer, G, Todd, P, Group, A. Simple Heuristics That Make Us Smart: Evolution and
Cognition Series Oxford University Press: 1999.
Graham, B, Dodd, D. Security Analysis: McGraw-Hill New York, USA: 1937.
Greenspan, A. Testimony to House of Representatives Committee on Government Oversight
and Reform Oct 23: 2008.
Greif, A. and D. Laitlin, A Theory of Endogenous political change. American Political Science
Review 2004:98(4): 633-652.
41
Heiner, R. The Origins of predictable behavior. American Economic Review 73(4): 560-595.
1983.
Hao, S, Jin, Q, Zhang, G. Investment Growth and the relation between equity value, earnings
and equity book value. Accounting Review2011:86(2) 605-635.
Hommes, C. Financial markets as nonlinear adaptive evolutionary systems. Quantitative
Finance2001:1(1) 149-167.
Hudson, R,Keasey, K, Littler, K, Dempsey, D. Time diversification: an essay on the need to
revisit finance theory. Critical Perspectives on Accounting 1999: 10501–19.
Hudson, R, Maioli, S. A response to a ‘Reflections on a global financial crisis’. Critical
Perspectives on International Business 2010:6, 1 53-71.
Hulbert, M.Timing and selection within the bond market: A look at the record, American
Association of Individual Investors(AAII) Journal, 1996:(November).
Hur, J., M. Pritamani, et al. Momentum and the disposition effect: The role of individual
investors. Financial Management 2010:39(3): 1155-1175.
Journal of Economic Surveys, Special Issue on Nonlinearity, Randomness and Complexity,
2011:25(3).
Journal of Financial Regulation and Compliance, 2009, Special Issue on the Financial Crisis,
Vol 17, Issue 4.
Johnson, S, Kwak, J.13 Bankers: The Wall Street takeover and the next financial meltdown:
Pantheon Books, New York, USA: 2010.
Kahneman, D, Slovic, P,Tversky, A. Judgment under uncertainty: Heuristics and
biases:Cambridge University Press: 1982.
Kahneman, D, Tversky, A. Prospect theory: An analysis of decision under risk.
Econometrica1979:47 263–291.
Kay, J. Review of UK equity markets and long-term decision-making. London, England,
Department of Business Innovation and Science. 2012.
Keasey, K, Hudson, R. Finance theory a house without windows. Critical Perspectives on
Accounting2007:18 932-951.
Kendall, M. The Analysis of Economic Time Series – Part I: Prices. Journal of the Royal
Statistical Society, Series A (General) 116, no. 1 11-34.
42
Kim, J,Shamsuddin, A, Lim K. Stock return predictability and the adaptive markets
hypothesis: Evidence from century-long U.S. data. Journal of Empirical Finance 2011:
forthcoming.
Khandani, A, Lo, A. What happened to the quants in August 2007? Journal of Investment
Management2007:5 5-54.
Krugman, P.How Did Economists Get It So Wrong? The New York Times, September 2nd,
2009:
http://www.nytimes.com/2009/09/06/magazine/06Economict.html?pagewanted=1&_r=2.
Kuhn, T. The Structure of Scientific Revolutions. 1st. ed., Chicago: Univ. of Chicago Press:
1962.
Kumar, A. Hard to value stocks, behavioural biases and informed trading. Journal of
Financial and Quantitative Analysis 2009:44(6): 1375-1401.
Lee, C. Bounded Rationality and The Emergence of Simplicity Amidst Complexity. Journal of
Economic Surveys2011: 25507-526.
Lo, A. The Adaptive Markets Hypothesis: Market Efficiency from an Evolutionary Perspective.
Journal of Portfolio Management,2004: 30th Anniversary Issue 15-29.
Lo, A. Reconciling efficient markets with behavioural finance: the adaptive markets
hypothesis. Journal of Investment Consulting2005: 7 (2) 21-44.
Lo, A. Regulatory reform in the wake of the financial crisis of 2007-2008. Journal of Financial
Economic Policy2009:1(1) 4-43.
Lo, A, Hasanhodzic, J. The evolution of technical analysis: financial prediction from
Babylonian tablets to Bloomberg terminals: Bloomberg Press, Hoboken, New Jersey, USA:
2010.
Lo, A, Fear, Greed and financial crises: A cognitive neurosciences perspective, MIT Working
Paper, https://mitsloan.mit.edu/finance/pdf/Lo-20120109d.pdf, 2011.
Mandeville, B. The Fable of the Bees: Or private vices and public virtues. London, England,
Penquin Books. 1714/1970.
Mandelbrot, B. The Variation of Certain Speculative Prices. The Journal of Business 1963 36
(4) 394-419
Mallard, G. Modelling cognitively bounded rationality: an evaluative taxonomy. Journal of
Economic Surveys,2011: forthcoming.
43
Marquering, W, Nisser, J, Valla, T. Disappearing anomalies: a dynamic analysis of the
persistence of anomalies, Applied Financial Economics 2006: 16(4) 291-302.
McLean, B, Nocera, J. All the devils are here: The hidden history of the financial crisis:
Portfolio-Penguin, London, England: 2010.
Mehra, R, Prescott, E. The equity premium: A puzzle. Journal of Monetary
Economics1985:15(2), 145-161.
Mehra, R, Prescott, E. The equity premium: ABC’s. Handbook of the Equity RiskPremium:North Holland, Elsevier, Amsterdam, The Netherlands: 2008.
Merton, R. On Market Timing and Investment Performance, I. An Equilibrium Theory of the
Value for Market Forecasts. Journal of Business1981:54 263-290.
Minsky, H. Stabilising an unstable economy: McGraw Hill Professional New York USA: 1986.
Mlodinow, L. The drunkard's walk: How randomness rules our lives: First Vintage Books
New York, USA:, 2009
Moller, N, Zilca, S. The evolution of the January effect. Journal of Banking & Finance2008:32
(3), 447-457.
Neely, C, Weller, P, Ulrich, J. The Adaptive Markets Hypothesis: Evidence from the Foreign
Exchange Market. Journal of Financial and Quantitative Analysis2009:44 467-488.
Newell, A,Simon, H. Human Problem Solving: Englewood Cliffs, NJ: PrenticeHall: 1972.
Polyani, K. The great transformation: The political and economic origins of our time Beacon
Press. Boston Massachusetts, USA 1944.
Rajan, R., Fault lines: How Hidden fractures still threaten the world economy. Princeton,
New Jersey, USA. Princeton University Press. 2010
Rogers, R,Sahakian, B, Hodges, J, Polkey, C, Kennard C, Robbins,T. Dissociating executive
mechanisms of tasks control following frontal lobe damage and Parkinson's disease.
Brain.1998:121 815-842.
Rubinstein, A.On Price Recognition and Computational Complexity in a Monopolistic. Journal
of Political Economy1993:39 83-96.
Rubinstein, A.On price recognition and computational complexity in a monopolistic
model.Journal of Political Economy, 1993:101(3) 473-484.
Salop, S, Stiglitz, J.Bargains and Ripoffs: A Model of Monopolistically Competitive Price
Dispersion. The Review of Economic Studies1977:44 493-510.
44
Samuelson, P. Economics. 1948. New York. McGraw-Hill.
Samuelson, P. Proof that Properly Anticipated Prices Fluctuate Randomly. Industrial
Management Review 1965 6:2 41-49.Sapra, S, Zak, P.Neurofinance: Bridging Psychology,
Neurology, and Investor Behavior. 2008:Working Paper Series, Issue Available at SSRN:
http://ssrn.com/abstract=1323051.
Shenk, D.The immortal game:Anchor Books, New York, USA: 2006.
Seru, A., T. Shumway, Stoffman N. Learning by Trading. Review of Financial Studies
2010:(23): 705-739.
Shiller, R.Do Stock Prices Move Too Much to be Justified by Subsequent Changes in
Dividends?American Economic Review 1981: 71(3): 421–436.
Shiller, R.Stock Prices and Social Dynamics.Brookings Papers on Economic Activity, 1984:(2)
457-510.
Shiller, R.Irrational Exuberance.2nd Edition: Princeton University Press, New Jersey, USA:
2000.
Shleifer, A, and Summers, L.The Noise Trader Approach to Finance.Journal of Economic
Perspectives 1990:4(2) 19-33.
Simon, H.A behavioral model of rational choice.The Quarterly Journal of Economics, 1955:1
99-118.
Simon, H. Studying human intelligence by creating artificial intelligence.American
Scientist1981:69(3) 300-309.
Simon, H.A mechanism for social selection and successful altruism.Science, 1990:250 16651668.
Simon, H.Invariants of human behavior. Annual Review of Psychology, 1990:41(1), 1-20.
Sorkin, R.Too big to fail:Penguin, London, England: 2009.
Smets, F, Wouters, R. An estimated dynamic stochastic general equilibrium model of the
euro area.Journal of the European Economic Association2003:1(5), 1123-1175.
Smith, V. Rationality in Economics: Constructivist and Ecological forms. Cambridge University
Press, New York, USA: 2008.
Stiglitz, J., The Price of Inequality. London, England, Allen Lane, 2012.
Taleb, N., Antifragile: How to live in a world we don't understand. London, England, Allen Lane.
2012.
45
Thaler, R. Mental Accounting and Customer Choice. Marketing Science1985:4(3), 199-21.
Thaler, R,Mental Accounting Matters.Journal of Behavioral Decision Making.1999: 12(3) 183206.
Todd, P.Fast and frugal heuristics for environmentally bounded minds. In: G. Gigerenzer & R.
Selten, eds. Bounded rationality: The adaptive toolbox (Dahlem Workshop
Report):Cambridge, MA: MIT Press, 51-70: 2001.
Todd, P,Gigerenzer, G. Bounding rationality to the world. Journal of Economic
Psychology2003:24(2) 143-165.
Trichet, J.-C. Reflections on the nature of monetary policy non-standard measures and
finance theory., European Central Bank Press Release.
http://www.ecb.int/press/key/date/2010/html/sp101118.en.html, 2010
Tseng, J., Lin C., Lin C., Wang S., Li S., Statistical properties of agent-based models in markets
with continuous double auction mechanism. Physica A 2010: 389 8 1699–1707.
Tseng, K.Behavioral finance, bounded rationality, neuro-finance, and traditional finance.
Investment Management and Financial Innovations, 3(4) 7-18.
Tversky, A, Kahneman, D.Advances in prospect theory: Cumulative representation of
uncertainty. Journal of Risk and uncertainty1992:5(4) 297-323.
Turner, A.The Turner Review: A regulatory response to the global banking crisis: Financial
Services Authority: 2009.
Veit, E, Cheney, J.Are mutual funds market timers? Journal of Portfolio Management,
1982:8(Winter) 35-42.
Von Neumann, J., and Morgenstern, O. Theory of Games and Economic Behavior, Princeton,
NJ, Princeton University Press, second ed. 1947.
Weitzman, M.Optimal Search for the Best Alternative.Econometrica1979: 47, 641-654.
Whitley, R.The transformation of business finance into financial economics: the roles of
academic expansion and changes in U.S. capital markets. Accounting, Organisations and
Society 1986: 11 171–92.
Yeh, C. The Effects of Intelligence on Price Discovery and Market Efficiency. Journal of
Economic Behaviour & Organization, 2008: 68.613-625.
Yeh, C. and Yang, C. Examining the Effectiveness of Price Limits in an Artificial Stock Market.
Journal of Economic Dynamics and Control, 2010: 34.pp.2089-2108.
46
Zhang, G.Accounting information, capital investment decisions and equity valuation: Theory
and empirical implications, Journal of Accounting Research 2000:Vol 38 271-295.
47
Figure 1
48
Figure 2
49
Figure 3
i
A good example of this is Rajan, 2005. This paper was delivered in August 2005 to a
gathering of top economists at Jackson Hole which that year was honouring the retirement
of Alan Greenspan as Chairman of the Federal Reserve. In hindsight this paper seems
remarkably prescient but it was attacked quite violently at the time. Lawrence Summers, for
example, told the audience he found "the basic, slightly lead-eyed premise of (Rajan's) paper
to be misguided." (Wall Street Journal, 2009). A paper by Bezemer, 2011 outlines a number
of research studies that did forecast a crisis and concludes that they commonly adopted
accounting or flow-of-funds models rather than neo-classical equilibrium models.
ii
Books by McLean and Nocera, 2010 and Sorkin, 2009, are popular examples of the genre.
One of the better high level overviews is a letter to the Queen seeking to answer her
question written by members of the British Academy (British Academy, 2009). For diverse
collections of academic papers on the crisis see special issues of the Journal of Financial
Regulation and Compliance, 2009 and Critical Perspectives on International Business, 2009
iii
Rationality is a rather ill-defined and potentially loaded concept as we discuss later in the
paper.
iv
Very similar axioms of rationality were set out by von-Neumann and Morgenstern in their
work on the theory of games and economic behaviour.(von-Neumann and Morgenstern,
1947).
vRepurchasing
after 30 days if the asset is expected to rally, to avoid the sale being classified
as a “wash sale” and discarded for tax purposes.
vi
The PIN metric of Easley et al (2002).
50