Journal of Economic Issues, Vol. 51 Issue No. 4, December 2017
IMPERFECT INFORMATION AND OPPORTUNISM
Information is a scarce resource. It is inherently available only in limited form to decision
makers. Limited or imperfect information is caused by uncertainty – both ontologic and
epistemic, limitations in cognitive capabilities or bounded rationality, hidden information and
information asymmetries. This has fundamental implications for the manner in which the self-
interested behaviour of agents will manifest itself. The paper argues that in the context of
imperfect information, self-interest can function in a manner quite different from what standard
approaches assume. This has been shown by the recent financial crisis. However, there has
been limited consideration in mainstream models, both of the neoclassical and institutional type,
as to what the exact nature of self-interestedness is, and how this affects the market behaviour
of agents. The nature of self-interest therefore needs to be modeled explicitly, to improve the
explanatory power of economic theories.
JEL: B40, B52, D03, D80, D82
Key Words: Information, uncertainty, opportunism, bounded rationality, self-interest
IMPERFECT INFORMATION AND OPPORTUNISM
Information is a key input into the decision making processes whereby resources are allocated.
In the same way that all resources are scarce, the nature of reality is such that information is
inherently scarce, as well. The three reasons why information is available only in imperfect form
to decision makers are the presence of uncertainty; complexity and the limited cognitive
capabilities of human agents; and finally, the existence of hidden or asymmetric information. In
contrast to this, the standard neoclassical model in economics has traditionally treated
information as being available fully, and to an unlimited extent depending on the requirements of
optimizing agents (Arrow 1974). Even where the initial information is lacking, the world is taken
to having ontological characteristics which enable full information to become available to
decision-makers over time (Savage 1954). This unrealistic view of information and its availability
needs to be done away with. Fortunately, the economic science is moving away from building
models based on unrealistic assumptions, and as Stiglitz (2000: 1441) notes, there is now
“recognition that information is imperfect, that obtaining information can be costly, that there are
important asymmetries of information ……. (and this has) a profound effect on how we think
about economics today.”
A major consequence of economic agents having to make judgments and act upon imperfect
information is that the content of their self-interestedness, and the manner in which it manifests
itself in market behaviour, becomes variable. If information were perfect, self-interested
behaviour would conform to straight-forward maximization subject to constraints, and any
deviant behaviour would be automatically arbitraged out of the market. However, the recent
sub-prime related financial crisis has shown that information is imperfect, that there are serious
information asymmetries, and these result in self-interest functioning in a manner which is quite different from that assumed in standard approaches. The U.S. Securities and Exchange
Commission reports that as of September 2014, 175 entities and individuals were charged with
misconduct that led to or arose from the financial crisis, with fines and penalties totaling $3.59
billion. The behaviour included outright fraud, misrepresentations and false statements, faulty
and selective disclosure, improper auditing and accounting, and in general undertaking
schemes to deliberately mislead investors. Such activities were not one off, they continued over
a period of time, and were exhibited almost across the entire financial sector, covering reputable
and well-established financial entities such as Citigroup, Goldman Sachs, J.P. Morgan, Merrill
Lynch, Bank of America, Fannie Mae and Freddie Mac, Jeffries & Co., Bear Stearns, UBS
Securities, Credit Suisse, and KPMG, amongst others (http://www.sec.gov./spotlight/enf-
actions-fc.shtml). The persistence of such modes of behaviour, which vary from the
assumptions of standard rational agent theory, indicates that there are alternative modes of self-
interest, and that the precise nature of self-interest needs to be explicitly modeled if our intention
is to provide better explanations for agent behaviour and the actual functioning of markets.
Both the standard neoclassical model (Friedman 1953, Savage 1954, Arrow 1974) and the
mainstream new institutional economics/transactions cost approach (Coase 1937, Coase 1960,
Alchian 1969, Furubotn and Richter 2005), give limited consideration to what the exact nature of
self-interestedness is, as empirically exhibited in the market behaviour of agents. In the new
institutional economics (NIE) literature a few authors do consider this issue (Klein 1980, Alchian
and Woodward 1988). But the only significant contributor is Williamson (1985, 1993). The paper
will therefore, first elaborate the reasons why information is inherently available only in imperfect
form to decision makers. It will then construct a taxonomy of self-interest. Based on this, the
adequacy of the alternative models mentioned above will be considered by assessing how they
treat decision making under imperfect information, in circumstances where self-interest can
function in different modes.
Uncertainty: Ontological and Epistemic
Uncertainty is the first and foremost reason for the existence of imperfect information. There are
two ways of viewing uncertainty: ontological, where it is associated with the nature of reality;
and, epistemic, where it relates to a lack of knowledge. Ontological uncertainty relates to the
unknowability of the future and of events that are distant in space. It is a property of the real
world, and belongs to the nature of the environment (Bianchi 1990). Uncertainty of this kind has
been characterized in two ways in the literature. The first view is that uncertainty can be
modeled and quantified through a probabilistic mathematical calculus (Savage 1954). The
second view is that expounded by Knight (1921), and others. In this, the ontology of the system
is open, and since the future is yet to be created, the set of outcomes cannot be fully identified.
Therefore, no probabilistic quantification is possible and fundamental uncertainty exists. Let us
discuss these two views.
Agreeing with the first view of uncertainty, Samuelson (1969) states that, the claim of economics
to be a science rests on its acceptance of the ergodic hypothesis. In this view, reality can be
analyzed in terms of stochastic processes. Explaining this hypothesis, Dunn (2001) indicates
that all stochastic processes yield time series data for which statistical measures such as
averages and standard deviations can be calculated. These measures consist of our empirical
knowledge of past and future relationships. As the number of observations increase, the
measures coincide such that averages calculated from past data eventually collapse into an
objective probability distribution that describes all possible past, present and future realizations.
If the relationship between economic variables is ergodic, the future is merely a statistical
reflection of the past. Uncertainty viewed in this form can therefore be subjected to a
quantification using the theory of stochastic processes. Random shocks can also be
accommodated within this system by introducing an uncorrelated error term which captures the
impact of casual exogenous events.
The alternative view is that propounded by Knight (1921). He argues that the legitimacy of a
probabilistic quantification is limited to planned or naturally occurring situations that are
repeated indefinitely under identical conditions. In such circumstances uncertainty can be
characterized as risk and actuarially measured. As Radner (2000) points out, for such a calculus
the decision-makers’ underlying model of the world must be fixed so that the probability space
remains unchanged. However, as argued by Keynes, Schumpeter, and others, this is not a
proper characterization of a real economy. Under capitalism there are endogenous pressures to
innovate in order to make more profits. Innovation creates new states of the world and leads to
fundamental uncertainty. Thus Minsky (1996:8) states that “uncertainty … is a deep property of
decentralized systems in which a myriad of independent agents make decisions whose impacts
are aggregated into outcomes that emerge over a range of tomorrows”. In situations where
there is creativity and structural change, the future is yet to be created, and so some relevant
information cannot be known, even in principle. This leads to a situation in which a reliable
probabilistic estimate cannot be made. Fundamental uncertainty implies that an objectively
defined optimal solution does not exist ex ante (Dequech 2001).
Elaborating on this view, Dunn (2001) indicates that fundamental uncertainty relates to a world
not characterized by stochastic processes or governed by ergodic laws. In a non-ergodic world,
even if agents can collect all the required information about the past and process it, this will not
provide a reliable basis for forecasting the future, and expectations based on past probabilities
will differ persistently from the time averages that emerge as the future unfolds. Although
ontological in character, this view also underlies the epistemic concept of uncertainty
emphasized by Keynes (1921). In the Keynesian view, agents are not sure about the validity of
the model that guides their actions. Uncertainty arises therefore from the way one thinks about
the world. It is epistemic because it relates to the beliefs of agents, rather than ontological,
where it is a property of the real world. In the words of Minsky (1996), the elements determining
long term expectations change so often that what happens in the economy at any date is
contaminated by market conditions that reflect actions determined by mental models that differ
from the model that now guides expectations. In such a world, agents are truly uncertain in that
there are no deterministic laws to be discovered, and they have to create the future themselves
by their own actions. Crucial decisions have to be taken which are non-repeatable, which are of
necessity creative, and which change the economic landscape forever. Dunn (2001) points out
that the presence of such crucial decisions is a sufficient condition for the existence of non-
deterministic processes, and consequently fundamental uncertainty.
Complexity and Human Cognitive Capabilities
The second area underlying the presence of imperfect information is that of the complexity of
information and limitations in human cognitive capabilities. Even if complete information exists in
the system, in the real world decisions are likely to be based on limited or imperfect information
because of the innate or behavioural characteristics of the decision-maker. Orthodox price
theory deals with simple market exchanges. In this model, the price system is the mechanism
for communicating all the information necessary for parties to conduct the exchange. Changes
in market opportunities are fully reflected in changes in relative prices, and this provides an
adequate basis for economic actors to adjust spontaneously (Hayek 1945). However, as Arrow
(1974) has pointed out, for rational decision-making, even relating to simple exchanges, a more
complex process of optimization has to be followed. Information on future markets, commodities
and prices are needed. Where forward contracts providing for such information needs are not
available, the only theoretical device to restore the functioning of markets, are contingent
markets. If uncertainty prevents most of the above from existing, the only solution possible then
is for there to be an estimation of expected utility based on some set of subjective probabilities.
In these circumstances, even if all the information needed is potentially available in the system,
its complexity is likely to be such that in practice only a limited set of imperfect information is
available to the average economic actor. This is equally true even where there are no search or
In any event, most transactions in the modern world are not simple rather they are characterized
by complex market exchanges. In such exchanges, mutuality of advantage requires continuity of
the exchange relationship (Commons 1932). Such continuity indicates that there is a need for
contracts to facilitate exchange. Further, where transactions require complex contracts, this
makes the information requirements for the conclusion of such contracts even more complex. In
such a situation, any process of rational decision-making will place significant demands on the
cognitive capabilities of the economic actor. As Simon (1972) notes, complexity in the
environment could be so great that it prevents the actor from calculating the best course of
action. Using the example of chess, Simon indicates that the chess player’s difficulty in
behaving rationally i.e. choosing the optimum strategy, has nothing to do with uncertainty – in
the sense of statistical decision theory. As von Neumann and Morgenstern (1953) observe,
chess is a game of perfect information. Theoretically, the full tree of possible games can be
known. Considering expert chess, on average this amounts to 10 120 possible games (Shannon
1950). Given the impracticality of such a calculation, even though the best chess players store
up to 300,000 chunks i.e. common chess position patterns, in their memories (Gobet and Simon
2000), they do not consider more than 15 to 20 moves ahead from a given position. Imperfect
information thus becomes a basis for decision making even in a perfectly certain environment
(in a game theoretic sense) because of the computational inability to ascertain the structure of
the environment. Complexity is therefore an important reason for the use of imperfect
information by market participants.
The discussion above indicates that rational decision-making in the context of complex market
exchange requires that human actors have the cognitive capabilities to process the necessary
information. The problem of complexity can only be addressed if we assume that economic
agents are hyper-rational or nearly omniscient. However, given natural human cognitive
limitations, we are led to the concept of bounded rationality (Simon 1957). Rational behaviour in
the real world is as much determined by the “inner environment” of people’s minds, as by the
“outer environment” of the world on which they act. Bounded rationality is the idea that the
choices people make are determined not only by some overall goal, but by the knowledge that
decision makers do and don’t have of the world, and their ability or inability to evoke that
knowledge when it is relevant (Simon 2000). Bounded rationality recognizes that there are
constraints on the information processing and computing capabilities of human beings.
Consequently as Simon (1957: xxiv) puts it, human actors are “intendedly rational but only
As a consequence of mounting evidence from the behavioural economics literature (Kahneman,
Slovic and Tversky 1982, Kahneman 2012), there seems to be increasing agreement that
bounded rationality is the appropriate cognitive assumption for economic actors (Williamson
2010). In fact Selten (1999) indicates that even in situations involving no risk and uncertainty,
people do not follow optimization procedures to maximize utility. Given the limits of a decision-
makers mental capacity to consider all alternatives and their consequences, it is inevitable that
he must apply some heuristic procedure to arrive at the best decision. This of course implies
that limited or imperfect information will be the basis for decision-making. Conlisk (1996) states
that, over the past 30 years hundreds of studies have been done by psychologists and
experimental economists showing that human beings make economic decisions based on
bounded rationality. Reviewing the major studies in this area, he argues that the literature on
‘heuristics and biases’ conclusively shows that subjects make systematic errors by using
decision heuristics or rules of thumb, which fail to accommodate the full logic of a decision.
Unbounded rationality not only has a weak empirical basis, but to be the logical basis for any
optimization or maximization process, requires the additional assumption that deliberation costs
are zero. If deliberation costs are positive, the decision maker has to find the optimal
deliberation effort to devote to the choice. This leads to infinite regress. Conlisk (1996) and
Pingle (1992) point out that when such costs are included in the optimization process, it means
that the augmented optimization problem will itself be costly to analyze, and the new
deliberation cost will be neglected. As Johansen (1977: 144) puts it, “The question of how far to
go ….. is in itself an optimization problem,” consequently, “At some point a decision must be
taken on intuitive grounds.” The infinite regress problem indicates that optimization, or
maximization of some objective function, based on unbounded rationality cannot be the ultimate
logical basis for behaviour. Rather a more fruitful approach is to accept bounded rationality and
consequent deliberation costs, which permit a truncated decision-making process to occur
based on imperfect information. In short, the idea that rational choice is a universally used
decision making system has to be rejected (Furubotn and Richter 2005: 523).
Hidden Information and Information Asymmetries
The final reason why decision-making in the real world is based on imperfect information is the
presence of hidden information. Hidden information occurs when the agents who are
participating in a transaction fail to self-disclose either the true attributes of a good or service ex
ante, or fail to self-disclose true performance ex post (Williamson 1993). The standard
neoclassical model accepts that in a decentralized economic system knowledge is inherently
localized and therefore the information available to an agent is necessarily incomplete. But as
Hayek (1937, 1945) argued the price mechanism can provide the necessary coordination,
whereby through an adaptive process such limited information can become the complete
information necessary for efficient solutions to emerge at the aggregate market level. However,
as Williamson (1991) argues, for the adaptive process in such a model to work, all the parties to
a transaction must behave honestly and candidly, and tell the whole truth and nothing but the
truth. If they do not do so, a situation of asymmetric information between the parties develops,
and this encourages human actors to pursue individual sub-goals rather than solutions which
serve the common interest.
The problem of hidden information and hidden action has best been elaborated in the context of
the principal-agent relationship (Arrow 1984). Arrow states that although the standard economic
model postulates an arm’s length relationship between agents, it is now recognized that the
agency relationship is a pervasive fact of economic life, and such interactions are virtually
universal in an economy. Hidden information about characteristics leads to the adverse
selection problem. The classical example of adverse selection in the goods market is the
‘lemons’ problem analyzed in the context of the used car market by Akerlof (1970). In this,
buyers unable to assess the exact quality of a used car will offer prices which result in all the
good quality cars withdrawing from the market. Consequently, there will be market failure, with
little or no trade occurring in the resultant ‘pooling’ equilibrium in which cars of all qualities
remain pooled in the same market. The more common case of the adverse selection problem,
and one that has been widely analyzed in the literature, is that of the insurance market. The
insured (the agent) has better knowledge about his/her health, probability of falling sick, or
dying, but the principal (the insurance company) cannot check this information in order to
ascertain whether this has been used in a way that best serves the company’s interests. Since
the population of those being insured is heterogeneous and the insurance company cannot
differentiate between the various categories, a common premium will be charged to all. This will
result in high risk individuals purchasing more insurance and low risk individuals less, leading to
an inefficient market solution (Arrow 1984, Rothschild and Stiglitz 1976).
Information asymmetries about performance, typically hidden action relating to the effort of an
agent leads to the moral hazard problem. Common agency relationships in this area include that
of the patient (principal) and physician (agent); stockholders (principal) and management
(agent); and, landlord (principal) and tenant (agent). As Arrow (1984) points out these
relationships are a significant departure from the arm’s length fixed price relation amongst
economic agents postulated in economic theory. The principal does not buy agent services at a
fixed price set by the market, nor does the principal simply buy output from the agent. Rather
the principal observes the outcome but cannot analyze its two components – the impact of the
agent’s action and exogenous uncertainty. Consequently, the variable to be determined is not a
price but a complex functional relationship.
If information were perfect, then potential hidden action could be resolved through contracts
which would stipulate all the actions that may need to be taken to bring about an optimal
solution. However, where information asymmetries exist, a mutually beneficial outcome is
unlikely to occur because the incentives of one or both parties in a transaction will be to pursue
individual sub-goals. This has been termed as the incentive alignment problem. Thus if
individuals are insured against risk, they have inadequate incentives to take actions that avoid
that risk. In a firm, managers have special expertise and inside information that they have
acquired over time about the organization. This creates an information asymmetry between
management and shareholders, thus creating opportunities for the former to use their
managerial discretion to obtain “rents” i.e. payments in excess of opportunity cost. Managers
will consequently pursue objectives other than maximizing shareholder value (Edlin and Stiglitz
1995). As Arrow (1984: 10) states: “In general, there does not exist a Walrasian (or Nash)
equilibrium with the property that no principal has an incentive to introduce new profitable
alternatives.” Stiglitz (2000) further points out, that dealing with the moral hazard problem
through incentive contracts and pay also presumes that there are no transaction costs. For
instance, if managers’ differential information is a major source of the rents they receive, they
will take action to increase the information asymmetries through obfuscation and creating noise.
In the presence of search costs therefore, market forces will not result in arbitrage, rather
incentives are created to bring about non-optimal outcomes (Edlin and Stiglitz 1995).
The literature on information economics shows that hidden information and hidden action, which
lead to the adverse selection and moral hazard problems, are pervasive. Such phenomena are
ubiquitous and affect the behaviour of individuals and firms (Stiglitz 1985; Arnott and Stiglitz
1991; Edlin and Stiglitz 1995). There are therefore, little grounds for believing in the Pareto
efficiency or inherent equilibrating capability of a market economy. On the contrary, in the
presence of information asymmetries, markets will be prone to serious fluctuations,
disequilibrium and crises. In Mishkin (1990) we find an analysis of how adverse selection and
moral hazard can cause financial crises. Considering evidence from financial crises that
occurred between 1850 and 1940, Mishkin concludes that the observed relationship between
banking panics, and the timing of increased interest rate spreads between low and high quality
borrowers, supports the asymmetric information view of the causes of these crises. He also
indicates that many of the other features of these crises, in particular the consequences on the
real economy, are hard to explain using any other theoretical point of view. Kirabaeva (2010)
and Stiglitz (2008) come to the same conclusion about the causes of the 2007-2008 sub-prime
related financial crisis.
Nature of Self-Interest
The foregoing discussion indicates that information is inherently available only in imperfect form.
Before considering how this affects the content of self-interest, in this section we elaborate on
what the potential modes of self-interested behaviour are. Self-interest can be defined in the
neoclassical manner, as proposed by rational choice theory. In this the content of self-interest is
that economic actors undertake utility or wealth maximizing behaviour subject to some
constraints. In the formal model, the only additional requirements are that decision making be
undertaken based on certain rules of choice, which include transitivity, invariance, dominance,
etc. The content of self-interest then consists only of maximizing behaviour, and no other
behavioural attributes are required to generate the standard neoclassical model. Self-interest in
this sense combined with market forces, learning processes, and evolution render all other
behavioural attributes or human qualities irrelevant. The logic of this model is that actors who
fail to make rational self-interested decisions, or adapt through a process of learning, will not
survive in a competitive world.
Disagreeing with this view, Mullainathan and Thaler (2000) point out that even at a theoretical
level market forces and evolution cannot be relied upon to make economic agents behave in a
rational manner as presumed by the neoclassical model. Competition works through arbitrage,
but there are limits on shorting and arbitraguers can be wiped out of the market before their bets
are proven right. Evolution does not necessarily weed out those who fail to maximize. For
instance, evolutionary arguments can be used to explain ‘over-confidence’: seeing ‘over-
confidence’, a ‘rational’ opponent may choose to back down. Finally, learning itself may not
necessarily lead to a change in strategy when there are high costs associated with
experimentation and learning. In such circumstances a ‘rational’ agent is likely to stick to a non-
In the classical system, although self-interest and competitive forces were seen as being central
to achieving optimal solutions, the view of self-interest was much more nuanced than in
neoclassical theory. Adam Smith’s original exposition on this matter is found in his Theory of
Moral Sentiments (1759). In this book, Smith compares market competition to a race and states
that for the race to be successfully run there must be no violation of fair play (quoted from his
Theory of Moral Sentiments in Wilson 1976:83). This assumes that agents interact on the basis
of a harmony of interests, and consequently their self-seeking behaviour is governed, at least to
some degree, by notions of fair play, common understanding and trust. As Arrow (1974:23)
states, “Trust is an important lubricant of the social system ……. Trust and similar values,
loyalty or truth telling ……… increase the efficiency of the system.” In other words, for efficient
solutions to be achieved, the self-interested utility or profit maximizing behaviour of homo
oeconomicus needs to be bound by some set of informal rules which include fair play and some
degree of trust. This is very different from the neoclassical view, which as stated above rejects
the need for any additional behavioural attributes beyond simple maximization.
Elaborating on this, Williamson (1985:47) states that for the neoclassical system to work, when
agents confront self-interested others across markets, bargains must be struck in a rule-bound
manner on the basis of fully and candidly disclosed initial positions. The model treats individuals
like playing a game with fixed rules that they all obey (Williamson 1993). There can be no
problematic behaviour attributable to rule deviance amongst human actors. Williamson (1985)
terms self-interested behaviour of this kind as simple self-seeking. However, this does not
necessarily have to be the case, and human action may well be guided by self-interest seeking
behaviour which is of a strategic kind. Williamson (1985: 47) defines such behaviour as
“opportunism” or “self-interest seeking with guile”. Opportunistic self-seeking behaviour includes
subtle forms of deceit, such as attempts to mislead, disguise, distort or confuse, but also more
blatant efforts such as lying, cheating, embezzling or stealing. Furthermore, it can be ex ante or
ex post, i.e. it can occur before or after a transaction has been completed. Williamson thus
distinguishes between opportunistic self- interested behaviour or opportunism, and simple self-
interest seeking behaviour. Neoclassical economics assumes that agent behaviour is dominated
by simple self-seeking, and excludes the possibility that opportunism can occur in a manner
which is significant enough to affect market outcomes.
Yet another type of self-interested behaviour emerges when we drop the assumption that actors
will necessarily limit their choices to a basket of opportunities that lie within their own budgets
and capabilities. Instead they could well try to take away the endowments of others by choosing
to rob, steal or act in some other predatory manner (Bardhan 2001, Chakravarti 2012). Such
behaviour, which could be termed as being predatory self-interest, is precluded in the classical
or neoclassical view of self-interest. However, this is not a realistic presumption because the
scope for predation by private parties on each other is created by the inherent inequality
between individuals in terms of physical strength, wealth, or other means that give one person
greater power than another. Veblen (1949) had hypothesized such behaviour as being
characteristic of the leisure class, and suggested that predatory habits were an essential part of
an individual’s struggle for existence. In principle therefore, there is no reason why self-interest
in the form of predatory behaviour should not be generalized as a potential characteristic of all
economic agents. In our analysis however, we will not take this type of self-interest into
consideration because self-interest of this kind operates through non-market means, and our
primary focus in this paper is to consider the effect of imperfect information on behavioural
attributes in the context of market exchange.
Let us now discuss how the neoclassical and mainstream NIE-related approaches consider the
manner in which self-interest needs to function in their core models, and whether the presence
of imperfect information is considered to have any influence on the content of this self-
interestedness or not. In Arrow (1974) we find a succinct exposition of the key elements of the neoclassical model.
Arrow indicates that one of the central concepts of the model is that the behaviour of agents is
governed by the criterion of optimization under constraints. For the purposes of optimization the
individual agents need not know his utility function or production possibility set because
“markets have supplied the information economized on, in the form of prices” (Arrow 1974:4).
However, rational decisions about consumption and production must be made with reference to
the future as well as the present, and information about future commodities must include their
prices. These prices can only be found in a suitable market in which future supply and demand
are equated. Unfortunately, such markets do not yet exist. The non-existence of futures goods
markets implies that the information needed by the optimizer is not provided by existing markets
and consequently he faces a world of uncertainty. Hence the optimizer must replace market
commitments to buy and sell, by expectations. Expected prices and utilities, of course, require
some probabilities. This logic leads to the conventional dominant paradigm of neoclassical
theory, the expected utility hypothesis.
Radner (2000) provides us with an outline of the subjective expected utility hypothesis first
expounded by Savage (1954). Savage’s hypothesis is a theory of unbounded rationality in
decision making in the face of uncertainty. The focus is on how a rational decision maker
revises initially subjective beliefs in the light of new observations about the world. The theory
consists of: a set of alternative states of the world, alternative actions, and alternative
consequences. States can be ascribed probabilities, consequences can be given utilities, and
using a scale of probabilities and utilities, an expected utility can be calculated for each act. The
updating of beliefs based on new information is done through a process of Bayesian learning
i.e. governed by the laws of conditional probability. As Dunn (2001) notes, in this model, the
long run is an asymptotic end state of a process of learning. Subjective expected utility is based
on the ergodic axiom that in the long run all choices made, based on Bayesian learning processes, will collapse into the substantive rationality of pre-programmed choices. We are
therefore dealing with an immutable system in which omniscience is possible in the long run.
In the neoclassical framework therefore, the inherent presence of imperfect information caused
by uncertainty and complexity in the real world are dealt with through: the ontological
assumption that the world is ruled by ergodic laws and that uncertainty can be quantified
through the theory of stochastic processes; and, that human beings have adequate cognitive
powers, or unbounded rationality, to modify their behaviour based on observation and learning
procedures which follow the laws of conditional probability. As Schlicht (1990) indicates,
obviously the world being described in this model is remote from any real world economic
processes. Consequently, as a gesture to realism, the neoclassical school has proceeded to
defend instrumental or unbounded rationality through what is called the “as if” defense originally
propounded by Vaihinger (1968). In the modern interpretation of the “as if” defense it is
admitted that the motivation and mental processes underlying economic behaviour may be
diverse. However, it is argued that this nevertheless leads to choices which approximate that
which would have been obtained by assuming unbounded rational behaviour i.e. “as if”
optimization is being undertaken by the agent with a utility function being maximized.
The classic defense using the “as if” argument is provided by Friedman and Savage (1948),
using the expert billiard player example. The hypothesis is that an expert billiard player makes
his shots “as if” he knew all the complicated mathematical formulae of angles, etc. that are
required to make the perfect shot. Friedman (1953) argues that confidence in the hypothesis is
not based on the belief that players can or do go through such processes, rather “it derives from
the belief that unless in some way or another they were capable of reaching essentially the
same result, they would not in fact be expert billiard players” (Friedman 1953:21). This
argument is based on Friedman’s “predictionism”, which is that assumptions are justified so long
as they lead to correct predictions, and no explanation of the mechanism linking the variables is
required. However, since this hypothesis was proffered, a considerable literature has emerged
which suggests that heuristics, and learning theory based on this, provide a better explanation
for the billiard players expertise. The manner in which expert chess players think has already
been referred to above (Simon 1972). As von Neumann and Morgenstern observe, since
computing the optimal strategies is practically impossible, it “necessitates the use of those
incomplete, heuristic methods of playing, which constitute good chess” (von Neumann and
The use of heuristics rather than optimal strategies is now well established as the basis on
which expert chess players play the game. Thus the limited rules that govern chess tactics
include: a calculation of points gained or lost in an exchange; whether a combination of some
moves results in a better position such as control of the centre, or control of a diagonal or file,
etc. These rules are also the basis on which chess playing computer programs are designed.
Even modern computer based chess programs therefore do not use optimization strategies
because they are computationally intractable. Gigerenzer (2004) provides a further example of
what cricket or baseball players actually do in order to catch a ball in flight, which shows that the
“as if” approach to optimization is inaccurate and has no value. Experiments show that players
do not make any calculations or measurements as the optimization approach suggests. Rather
they use what is termed a “gaze heuristic”. This involves running towards the ball while keeping
the angle of the gaze between the eye and the ball constant. Successful catching has therefore
nothing to do with optimization. Further, in comparison to the “as if” approach, heuristics provide
a useable explanation which can be used to train inexperienced players.
Beyond this discussion about the use heuristics, we have the vast literature on cognitive
illusions, anomalies and biases (Kahneman and Tversky 1979; Kahneman, Slovic and Tversky
1982; Rabin 1998), which shows that human behaviour systematically deviates from the
rationality assumptions of the neoclassical model. Instrumental, abstract or unbounded
rationality, and it’s “as if” justification therefore has little empirical validity, and is an unsafe
hypothesis from an explanatory or predictive point of view. Such assumptions may be of value
in formulating an elegant theoretical framework, but do not provide a basis for explaining the
behaviour of individuals or firms when faced with uncertainty and complexity in the real world.
Let us now consider the manner in which the neoclassical model deals with the third source of
imperfect information: hidden information and its consequences. We have already noted that the
model treats individuals as if they are playing a game with fixed rules that they all obey. It is
expected that bargains are struck in a rule-bound manner on the basis of fully and candidly
disclosed initial positions (Williamson 1985:47, Williamson 1993). Such a rule-based
interpretation of self-interest implies that symmetrical information is available to all actors in a
bargain. Consequently, hidden information and all the problems of adverse selection and moral
hazard are assumed away. Thus by assuming that the self-interested behaviour of agents
consists of simple self-seeking, and that there is no problematic behaviour attributable to rule
deviance (Williamson 1985), the problems surrounding imperfect information and its potential
influence on actual modes of self-interest are ignored in the neoclassical model. Clearly this is
not a satisfactory solution. As we have discussed earlier, asymmetric information is a very real
phenomenon, it is pervasive, and has a major impact on agent behaviour and market
New Institutional Economics
While the neoclassical model sees agents as optimizers, the NIE approach starts with a
behavioural model of man that is based on cognitive psychology. It accepts Simon’s (1957) view
that economic agents have limited computational and cognitive capabilities, and therefore
“bounded rationality” is a more appropriate concept to describe decision making processes.
Bounded or limited rationality implies that all decision making will be made on the basis of
imperfect information. Consequently, agents will inevitably incur search, bargaining or more
generally “transaction costs” in the process of making a decision. As Furubotn and Richter
(2005) indicate, “Transaction costs are encountered universally because of the character of the
individuals who make decisions”. The impact of bounded rationality and transaction costs on
market exchanges is that all complex contracts will unavoidably remain incomplete (Williamson
1991). Market outcomes, in this view, then become a function of the nature and extent of
transaction costs and the measures that agents take to mitigate them in order to achieve the
best possible outcomes.
The second major departure that the NIE approach has from the neoclassical model is its
proposition that self-interest should not necessarily be taken to presume a harmony of interests
between agents, or be viewed as being benign in character. As we noted earlier, self-interested
behaviour need not necessarily be in the nature of simple self-seeking, but can be self-
interested opportunism (Williamson 1985). Perfect information implies that opportunism, either
exante or expost, cannot persist as a mode of self-interested behaviour because agents who
behave in this manner will be weeded out of the market by competitive forces and will not
survive. Perfect information also implies that uncertainty, limitations of cognitive capabilities and
hidden information, have been overcome in some way, or eliminated by assumption, and are
therefore no longer relevant factors in the decision making process. In a world where there is
perfect information, there are no transaction costs. Agents have all the information they need to
costlessly conclude optimal contracts.
Transaction costs therefore, can only occur if information is imperfect. But are transaction costs
and imperfect information two sides of the same coin, such that the presence of one
automatically implies the presence of the other? Further, is the market outcome primarily
determined by the nature and extent of the transaction costs present, as held by the NIE
literature, irrespective of the precise mode of self-interest affecting agent behaviour? These
questions are considered below in the context of potential agent behaviour to achieve greater
contractual completeness and more profitable outcomes when faced by imperfect information.
In particular, given the possibility of opportunism, the central position that has been given to
transaction costs in influencing market outcomes is re-assessed using the lens of information
that has been elaborated above.
First we look at the transaction cost implications of uncertainty – both fundamental and that
which can be characterized as risk. Arrow (1974) indicates that when fundamental uncertainty
exists, firstly, it becomes evident that any variable that improves predictability will have
economic value. Agents will therefore seek to acquire additional information. But information is
costly. The fact of information being imperfect, and that transaction costs are involved in
obtaining additional information, then becomes an integral factor that will influence the
functioning of markets. Efficiency can no longer be consider purely in terms of productive
efficiency, but will have to include efficiency in information gathering and processing. Secondly,
Arrow indicates that fundamental uncertainty amongst buyers and sellers about prices and
quantities will result in the non-existence of markets for future goods. Theoretically, this can be
addressed by proposing contingent markets. However, in the real world transactions involving
forward contracts for future goods are found to have very limited coverage. This is so because
contracts are not enforceable without costs and forward contracts are even more costly to
enforce than contemporaneous contracts. In the presence of fundamental uncertainty therefore,
optimal long term contracts cannot be concluded and agents will have an incentive to incur
transaction costs and resort to or establish non-market mechanisms which can mitigate the lack
What if we assume an ergodic world in which uncertainty can be analysed in terms of stochastic
processes and a risk based quantification undertaken? Dunn (2001) argues that in such an
environment through observation and a learning process governed by Darwinian natural
selection agents can arrive at the objective probabilities. But this can only happen in the long
run, and consequently in the transitional period agents will have to incur transaction costs to
support the learning processes they use to overcome the imperfect information they face. Given
the initial imperfect information therefore, as we noted earlier, optimal long term contracts
cannot be concluded exante. However, could rational expectations be formed and ex ante
maximization replace sub-optimal solutions if bargaining were based on complete sequential
short term contracting? Williamson (1993) does not agree. He indicates that to prevent ex post
maladaptation problems from arising, which will lead to contractual difficulties, it is necessary for
rule-based self-interest to guide the decision making of agents. Parties at the outset must self-
disclose all relevant information candidly, and continue to do so at every contract renewal
interval so that the promises made by all parties become self-enforcing.
Even for complete sequential contracting to occur therefore, whether it is an ergodic or non-
ergodic world, information must be candidly and fully self-disclosed by the parties at each
contracting interval in the bargaining process. If this occurred, efficiency related adjustments
could be made from one period to the next, and forward contracting is not likely to be
problematic. Enforcement would also become a simple matter. Such a situation could then
approximate one which is characterized by perfect information, with learning processes
providing the adjustment mechanism from one period to the next so that efficient solutions could
be achieved in the long run. What is evident from this discussion, is that the nature of the
market solution arrived at depends primarily on whether opportunism is present or not. If one or
both parties behave opportunistically and create information asymmetries with a view to
generating rents or gaining a disproportionate share of the gains from exchange, then non-
optimal solutions including market failure can occur. Moreover, if opportunism is present, even
the incentive of agents to incur transaction costs will depend entirely on their assessment of
whether expending resources on such costs is likely to mitigate their information disadvantage
or not. If the costs are assessed to be high with no guarantee that the required information could
be acquired, a judgment or decision could well be made on some intuitive principle or heuristic
basis without any transaction costs, except for some minimal deliberation costs, being incurred.
The same argument applies to imperfect information in the form of hidden action or hidden
information. Hidden information about characteristics, or hidden action, creates the possibility of
generating rents based on the superior information possessed. As observed above, this
encourages one of the parties to pursue individual sub-goals and behave opportunistically,
leading to the moral hazard and adverse selection problems. In these circumstances again,
expending resources on searching for information or bargaining will not necessarily lead to a
preferred market solution. As Edlin and Stiglitz (1995) indicate, the incentives on one side of the
bargain may well be not just to not reveal information, but in fact to increase the information
asymmetry through greater obfuscation and creating noise. Since the imperfection in this case
exists in principle e.g. where behavior is unobservable, and is generated by the inherent
characteristics or potential action of one party in the transaction, the other party may not be able
to eliminate the asymmetry through incurring expenditure on search, processing or other costs.
The commonly used transaction cost view in the literature, that information gaps can be reduced
or eliminated through the expenditure of resources on transaction costs, is therefore an
incomplete view of the real nature of the bargaining process.
The fundamental problem here relates to the nature of the incentives generated by the
information asymmetry. These could either encourage outright opportunistic behaviour, or if
there is some potential beneficial mechanism through which an alignment of incentives can
occur, could also result in behaviour that leads to a Pareto superior incentive compatible market
solution. In the context of the latter case, some of the mechanisms considered in the literature
which could bring about an alignment of incentives between parties, include the reputation effect
(Shapiro 1983), signaling devices such as warranties (Varian 1992), advertising (Klein and
Leffler 1981), and educational qualifications (Spence 1974). The use of such mechanisms to
arrive at a mutually beneficial solution may require transaction costs to be expended by one or
both parties, but this would only be a by-product of the bargaining process and not a
determinant of the solution.
The discussion above indicates that the relationship between imperfect information and
transaction costs is not that of a one to one equivalence. An information asymmetry emerges
from the fundamental character of the bargaining relationship between two parties e.g. in the
principal-agent case, which can be exploited by one of the parties opportunistically or not; or, it
could even be created by one of the parties in the trade because of a perceived greater benefit
from behaving opportunistically e.g. the recent VW emissions scandal case (see below).
Elimination of the imperfection therefore depends on the presence or absence of opportunism,
and whether the potential gains from behaving honestly and resorting to some incentive
alignment mechanism are assessed to be greater than those which may result from behaving
It could be argued that an intractable information asymmetry exists because of prohibitive
transaction costs. If transaction costs were low, agents would have an incentive to eliminate the
asymmetry. Therefore distinguishing between these two concepts is incorrect. However, as we
have noted above such a view is a misunderstanding of the nature of the bargaining relationship
which generates such an asymmetry. For instance, where behaviour is unobservable, as noted,
the incentive one one-side of the bargain may well be to increase the information imperfection.
In other words, the inherent nature of an asymmetry could be such that expending resources on
transaction costs by parties in the bargain cannot eliminate it. Therefore, achieving a mutually
beneficial or improved outcome has nothing to do with the extent of the transaction costs
involved and whether they are considered prohibitive or not. The market outcome will be
determined by the pay-off matrix contained in the information asymmetry itself, with the role and
relevance of transaction costs becoming uncertain.
In these circumstances to argue that an intractable information asymmetry is due to prohibitive
transaction costs is similar to Friedman’s (1953) ‘as if’ argument. There is no way of proving or
disproving it, and we have no idea of how to assess the quantum of high or prohibitive
transaction costs in such a case. However, what we do know from the discussion above is that
an asymmetry is aggravated by opportunistic behaviour. Therefore, a better approach would be
one which directly addresses the factors underlying the misalignment of incentives, and alters
the pay-off matrix which is causing such behaviour. For instance, there is general consensus
today that de-regulation and too much reliance on free market forces were key factors
underlying the recent financial crisis, and that the prevention of banking and financial crises
require more stringent non-market interventions such as better regulations, enforcement and
harsher punishment (Chakravarti 2012: Chapter 9). Measures of this type, which are intended to
directly address the incentives underlying opportunistic behaviour are therefore more likely to be
effective in achieving Pareto superior market outcomes. Such non-market interventions do
involve costs, and such costs could even be termed as transaction costs, but they are third party
costs which are required to improve the functioning of the market (Arrow 1969:48) and not direct
costs on increasing information, communications, contractual agreements, etc., which need to
be borne by the transacting parties themselves. Therefore, mitigating such costs will not be a
relevant factor that influences agent behaviour in arriving at a particular market outcome.
The foregoing discussion can be summarized in the following remarks.
Remark 1: Where perfect information exists, there are no transaction costs and opportunism is
not a viable mode of self-interest.
Remark 2: Where imperfect information exists, self-interest in the mode of simple self-seeking
can lead to near optimal or even optimal solutions. Transaction costs will be incurred initially,
but bargaining can result in vanishing transaction costs and optimal solutions through repeated
Remark 3: Where imperfect information exists and there is self-interest in the mode of
opportunism, sub-optimal solutions including market failure can occur. In the presence of
opportunism, the link between imperfect information and transaction costs is uncertain and
depends on the nature of the information asymmetry and incentives in the bargain.
Game theoretic solutions
The above discussion on alternative market outcomes in the context of different modes of self-
interest can be formally presented in a game theoretic framework. In game theory, a situation
characterized by opportunism can be modeled as a non-cooperative game in which the players
adopt individual maximization strategies in open conflict with all other players. Non-cooperative
games where players interact only once produce Pareto-inefficient outcomes e.g. the standard
one-shot Prisoners Dilemma. Our findings above that opportunism in a bargain results in sub-
optimal outcomes is consistent with this formal result from game theory. However, market actors
whether individuals, firms or institutions do commonly interact repeatedly and over long periods
of time. The question therefore arises as to whether repeated interaction, or repeated games,
can result in some form of implicit collusion which overcomes opportunistic behaviour based on
a non-cooperative attitude, and brings about an outcome which replicates explicitly cooperative
Cooperative solutions are efficient by definition (Nash 1950). If bargaining is assumed to be
cooperative, then efficient solutions automatically result. Given imperfect information however,
in general, ex post efficiency cannot be achieved if information is incomplete on both sides of
the bargain (Myerson and Satterthwaite 1983), or even if it is one-sided incomplete information
(Schweizer 1988). Incentive compatible direct mechanisms do exist to resolve the problem of
incomplete information, but this requires that the mechanism be freely accepted by both parties,
and that voluntary bargaining occurs based on individual rationality. Schweizer (1988) further
indicates that a mechanism can be considered to be incentive compatible only if it never pays
not to tell the truth. Based on this framework, we can consider how effective the mechanisms
mentioned above can be in generating incentives that encourage actors to become cooperative.
Our first solution to the problem of imperfect information relates to Remark 2, where it is
assumed that there is rule-based simple self-seeking, or truthfulness and trust amongst agents.
In this situation, bargaining based on Individual rationality i.e. the expectation that the benefits
from the transaction will be non-negative, will encourage signaling or some other mechanism to
be proposed by the parties so that their true type is revealed. By giving information on type, the
incomplete information will be corrected and bargaining will bring about a Pareto improvement
over the existing situation. Additionally, as noted earlier, with simple self-seeking, transaction
costs are likely to be low. This will permit reasonable communications to proceed and a
cooperative solution to be achieved, with the possibility that transaction costs will vanish over
time as the agreements become self-enforcing (Schweizer 1988).
The second situation relates to Remark 3, where there is the presence of opportunism. One of
the solutions proposed in the game theory literature for this problem is repetition. It is argued
that repetition can alleviate the phenomena of adverse selection in the first period and moral
hazard thereafter. Repetition permits priors to be revised, and instead of a ‘pooling’ equilibrium
which leads to market failure, opens the possibility of a ‘separating’ equilibrium in which the
market clears (Nechyba 2010: Chapter 22). After all, it is argued, one cannot expect to benefit
beyond one round of cheating. Additionally, the seller may wish to build up his reputation in the
market. Factors such as gaining a reputation or earning trust influence long term pay-offs, and
should therefore provide incentives to overcome short-term strategic behaviour (Shapiro 1983).
Based on this it is argued that repeated games can bring about optimal, or at least superior
outcomes, compared with non-cooperative bargaining solutions. If this were so, then
opportunistic behaviour would only be of concern in one-off interactions, and would not be very
relevant when considering more continuous and regular market transactions between parties.
This issue is considered below using a few examples from standard game theory (Lambertini
Our first example is a supergame characterized by complete, symmetric but imperfect
information, with a finite time horizon, in which each stage is a Prisoners Dilemma. When this
game is solved using backward induction, it is found that at each stage the Prisoners Dilemma
is reproduced and the players do not generate a Pareto-efficient solution. Repetition in this case
does not result in an efficient outcome. If the game with the same characteristics is infinitely
repeated over time however, the players can achieve a Pareto improvement over the Nash
equilibrium of the constituent game. But this depends on the weight given to future periods and
pay-offs. If agents are myopic and severely discount future pay-offs, no improved solution can
be achieved. Achieving the Pareto-efficient result also requires that the players play a collusive
strategy (Friedman 1971). If they do not collude, but play a tit-for-tat strategy for instance, as
proposed by Axelrod (1981), then the result is again sensitive to the time discounting factor
relative to the size of the pay-offs. Repetition by itself therefore does not guarantee that
strategic behaviour will be eliminated and optimal outcomes arrived at. Many additional
assumptions about agent behaviour are required for such a result to be achieved.
The second example relates more directly to whether the reputation effect provides a strong
incentive for sellers to maintain their quality and act honestly or not (Lambertini 2011:143). The
question is whether repetition can alleviate exante opportunism or adverse selection, through
the reputation that the seller may wish to build up. This issue has been considered in the
context of a firm selling experience or durable goods where the quality can be modified from
one period to the next. Lambertini shows through a formal two-period model that the
reputational effect will provide an incentive to the seller to maintain quality and be honest only if
the immediate cost advantage is greater than the discounted value of the maximum losses the
seller expects to suffer if the consumer buys less or not any more. If this is not the case then
cheating on product quality will be the dominant strategy for the seller.
That cheating can be a dominant strategy, and that repetition and the reputation effect do not
necessarily mitigate opportunistic behaviour is starkly brought out by the recent VW emissions
scandal. In 2008, VW announced that it had achieved a new level of Clean Diesels technology
which met stringent European Union and other emission standards. VW diesel car sales in the
U.S. rebounded. Between 2009 and 2015 almost 11 million clean diesels were sold, of which
500,000 were in the U.S. The new technology won several environmental awards and received
tax breaks. In 2014 studies came out which found that there were discrepancies between VW’s
diesel vehicle test and on-road emissions. These caught the attention of the U.S. Environmental
Protection Agency (EPA), resulting in its threat in September 2015 to not certify 2016 VW
diesels. VW responded by admitting that its diesels had been installed with an emissions
compliance “defeat device” which was designed to cheat testing. The device consisted of
software in the diesel engine which could detect when it was being tested and could change its
performance to improve the emission results.
Subsequent to the scandal breaking, VW’s share price fell by a third, and it has had to make a
provision of EUR 16 billion towards the cost of schemes to refit the defective cars, compensate
owners, and law suits. What all this may eventually cost as against the profit gained from
increased sales during 2009-2015 is difficult to assess. However, the company does not seem
to have suffered long term damage. During 2016 it achieved the position of being the largest
automobile manufacturer in the world, and polls and surveys in the U.S. and Germany even
show that its reputation is relatively intact, with most consumers indicating that they have a
positive impression of VW cars. Reports on the internet
(https://en.wikipedia.org/wiki/Volkswagen_emissions_scandal), in fact indicate that defeat
devices of one type or another have been used in the past by several other automobile
companies including Chrysler, Ford, General Motors, and several heavy truck manufacturers
such as Caterpillar, Mack and Volvo. As in the case of the financial sector prior to and during
the sub-prime crisis, the automobile industry also shows that opportunistic behaviour can be
widespread, and the theoretical presumption that repeated interaction or reputational effects will
mitigate such behaviour or naturally weed out agents who behave in this manner is incorrect.
Unlike the neoclassical model, NIE-related approaches accept that uncertainty, complexity, and
information asymmetries result in decision makers facing imperfect information. Further these
factors are not inconvenient variables which can be assumed away, but are essential elements
of the world we live in, and consequently need to be taken account of in any model that tries to
explain the functioning of the economic system. In the context of information asymmetries, to
explain the fact that human actors are given to sub-goal pursuits, the presence or absence of
opportunism as a behavioural assumption then becomes a central postulate. If there were
candid self-disclosure, clearly information asymmetries can no longer exist, and Akerlof’s (1970)
“lemons” problem would disappear. As it happens, adverse selection, moral hazard, shirking by
employees, and managerial discretion, all of which fall under the broad category of opportunism
are widely observed phenomena (Arnott & Stiglitz 1991, Williamson 1993). Moreover, an
increasing number of studies now use these forms of opportunism as a basis to explain a range
of phenomena that neoclassical theory cannot explain. This includes the functioning of
insurance markets; the lending behaviour of banks and financial institutions; managerial
incentives and rent-seeking; and, the causes of the recent sub-prime related financial crisis.
At the behavioural level, NIE-related approaches explicitly state that bounded rationality is the
appropriate cognitive assumption for human actors. This automatically rules out unrealistic
explanations based on unbounded or instrumental rationality and “as if” type of justifications.
However, for bounded rationality to provide a good basis to explain many market outcomes and
real world phenomena that arise due to imperfect information, additional assumptions about
opportunism are necessary. Bounded rationality leads to imperfect information. However, the
market outcome which emerges out a situation of imperfect information depends on the
potential pay-offs contained in the imperfection and whether agents decide to exploit these
opportunistically or not. The market outcome then becomes a function of the presence or
absence of opportunism rather than the transaction costs associated with various mechanisms
intended to correct or mitigate the lack of knowledge. In the literature, only Williamson (1991,
1993) has continued to give equal importance to bounded rationality and opportunism as key
factors that should be considered in any theory of economic organization.
Although information asymmetries have become an important basis in the recent literature to
explain a range of economic phenomena, without the behavioural assumption of opportunism,
asymmetries do not necessarily translate into adverse selection or moral hazard. In spite of this,
most studies have not put opportunism centre stage. Others, such as Klein (2006), hold that
opportunism is just self-interest extended over a larger set of margins. This is a total
misunderstanding of the meaning and impact of opportunism. As we have argued in this paper,
self-interest in the form of opportunism has implications fundamentally different from the
Smithian view of simple self-interest, and needs to be a key assumption of any theory that
attempts to explain various key economic phenomena observed, including the performance of
the economic system itself.
Alchian, A. “Information Costs, Pricing, and Resource Unemployment” Western Economic
Journal No. 7, 1969: 109-128.
Alchian, A. & Woodward, S. “The Firm is Dead: Long Live the Firm: A Review of Oliver E.
Williamson’s The Economic Institutions of Capitalism” Journal of Economic Literature No. 26,
Axelrod, R. “The Emergence of Cooperation Among Egoists” American Political Science Review
No. 75, 1981: 306-318.
Akerlof, G. A. “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism”
Quarterly Journal of Economics Vol. LXXXIV, 1970: 488-500.
Arnott, R. & Stiglitz, J. E. “Moral Hazard & Nonmarket Institutions: Dysfunctional Crowding Out
of Peer Monitoring?” American Economic Review Vol. 81 No.1, 1991: 179-190.
Arrow, K.J. “The Organization of Economic Activity: Issues Pertinent to the Choice of Market
versus Non-Market Allocations” in The Analysis and Evaluation of Public Expenditures: The
PBB System, Joint Economic Committee, 91 st Congress, Vol.1, Washington D.C.
Arrow, K. J. “Limited Knowledge and Economic Analysis” American Economic Review Vol. 64
No. 1, 1974: xiii-xiv + 1-10.
Arrow, K. J. “The Economics of Agency” Technical Report No.451, 1984. Center for Research
on Organizational Efficiency, Stanford University, California.
Bardhan, P. “Distributive conflicts, collective action, and institutional economics”, in G.M. Meier
and J. Stiglitz (eds), Frontiers of Development Economics. Oxford: Oxford University Press.
Bianchi, M. “The unsatisfactoriness of satisficing behaviour: from bounded rationality to
innovative rationality”, Review of Political Economy Vol. 2 No. 2, 1991: 149-167.
Chakravarti, A. Institutions, Economic Performance and the Visible Hand: Theory and Evidence.
Cheltenham: Edward Elgar. 2012.
Coase, R.H. “The Nature of the Firm” Economica, No. 4, 1937:386-405.
Coase, R. H. “The Problem of Social Cost”, Journal of Law and Economics, Vol. 3, 1960: 1-44.
Commons, J. R. Institutional Economics. New York: Macmillan. 1932.
Conlisk, J. “Why Bounded Rationality?” Journal of Economic Literature Vol. 34 No. 2, 1996:
Dequech, D. “Bounded Rationality, Institutions and Uncertainty”, Journal of Economic Issues
Vol. XXXV No. 4, 2001: 911-929.
Dunn, S. P. “Bounded Rationality is not fundamental uncertainty: A post Keynesian
perspective”, Journal of Post Keynesian Economics Vol. 23 No. 4, 2001: 567-587.
Edlin, A. S. & Stiglitz, J. E. “Managerial Rent Seeking and Economic Inefficiencies”, American
Economic Review Vol. 85 No. 5, 1995: 1301-1312.
Friedman, J. “A Non-Cooperative Equilibrium for Supergames” Review of Economic Studies No.
38, 1971: 1-12.
Friedman, M. & Savage, l. “The Utility Analysis of Choices Involving Risk”, Journal of Political
Economy Vol. 56, 1948: 279-304.
Friedman, M. Essays in Positive Economics. Chicago: University of Chicago Press. 1953.
Furubotn, E. G. & Richter, R. Institutions and Economic Theory, 2 nd edition. Ann Arbor:
University of Michigan Press. 2005.
Gigerenzer, G. “Striking a Blow for Sanity in Theories of Rationality”, in M. Augier & J. March
(eds), Models of Man: Essays in memory of Herbert A. Simon. Cambridge MA: MIT Press.
Gobet, F. & Simon, H. A. “Five Seconds or Sixty? Presentation time in expert memory”,
Cognitive Science No. 24, 2000:651-682.
Hayek, F. A. “Economics and knowledge”, Economica Vol.4, 1937: 33-54.
Hayek, F. A. “The uses of knowledge in society”, American Economic Review Vol.35, 1945:
Johansen, L. Lectures on Macroeconomic Planning, Part I: General Aspects. Amsterdam: North
Kahneman, D. Thinking, Fast and Slow. UK: Penguin Random House. 2012.
Kahneman, D. & Tversky, A. “Prospect Theory: An Analysis of Decisions Under Risk”,
Econometrica Vol. 47 No. 2, 1979: 263-291.
Kahneman, D., Slovic, P. and Tversky, A., (eds) Judgment under Uncertainty: Heuristics and
Biases. New York: Cambridge University Press. 1982.
Keynes, J.M. Treatise on Probability. London: Macmillan & Co. 1921.
Kirabaeva, K. “Adverse Selection and Financial Crisis”, Bank of Canada Review Winter
2010/11, 2010: 11-19.
Klein, B. “Transaction Cost Determinants of “Unfair” Contractual Arrangements” American
Economic Review Papers and Proceedings, 1980, No. 70: 356-362.
Klein, P. “Does Transaction Cost Economics Need Opportunism?” Paper available on:
http://www.organizationandmarkets.com 6 October 2006.
Klein, B. & Leffler, K. B. “The Role of Market Forces in Assuring Contractual Performance”,
Journal of Political Economy Vol. 89, 1981: 615-641.
Knight, F. Risk, Uncertainty and Profit. New York: Houghton Mifflin. 1921.
Lambertini, L. Game Theory in the Social Sciences. Oxford: Routledge. 2011.
Minsky, H. “Uncertainty and the Institutional Structure of Capitalist Economies”, Working Paper
No. 155, 1996. The Jerome Levy Economics Institute of Bard College.
Mishkin, F. S. “Asymmetric Information and Financial Crisis: A Historical Perspective”, NBER
Working Paper No. 3400, 1990.
Mullainathan, S. & Thaler, R. “Behavioural Economics“ NBER Working Paper No. 7948, 2000.
Myerson, R. & Satterthwaite, M. “Efficient Mechanisms for Bilateral Trading“ Journal of
Economic Theory, 1983: 265-281.
Nash, J. “The Bargaining Problem“ Econometrica No. 21, 1950: 128-140.
Neumann, J. von & Morgenstern, O. Theory of Games and Economic Behaviour (3 rd edition).
Princeton: Princeton University Press. 1953.
Nechyba, T. Macroeconomics: An Intuitive Approach with Calculus. Maison, USA: South
Western Cengage Learning. 2010.
Pingle, M. “Costly Optimization: An Experiment” Journal of Economic Behaviour and
Organisation No.17, 1992: 3-30.
Rabin, M. “Psychology and Economics”, Journal of Economic Literature Vol. 36 No.1, 1998:
Radner, R. “Costly and Bounded Rationality in Individual and Team Decision-making”, Industrial
and Corporate Change Vol. 9 No. 4, 2000: 623-658.
Rothschild, M. & Stiglitz, J. E. “Equilibrium in Markets: An Essay on the Economics of Imperfect
Information”, Quarterly Journal of Economics Vol. 90 No. 4, 1976:629-649.
Samuelson, P. A. “Classical and Neo-Classical Monetary Theory”, in R. Clower (ed) Monetary
Theory. Harmondsworth: Penguin. 1969:170-190.
Savage, L. The Foundations of Statistics. New York, Wiley. 1954.
Schlicht, E. “Rationality, Bounded or Not, and Institutional Analysis”, Journal of Institutional and
Theoretical Economics Vol. 146, 1990:703-719.
Schweizer, U. “Externalities and the Coase Theorem: Hypothesis or Result? Journal of
Institutional & Theoretical Economics No.150, 1988: 245-266.
Selten, R. “What is Bounded Rationality”, SFB Discussion Paper B-454. Prepared for Dahlen
Shannon, C. “Programming a Computer for Playing Chess”, Philosophical Magazine Ser. 7 Vol.
41 No. 314, 1950.
Shapiro, C. “Premiums for High Quality Products as Returns to Reputations” Quarterly Journal
of Economics No. 97, 1983: 659-679.
Simon, H. A. Models of Man: Social and Rational. New York: Wiley. 1957.
Simon, H. A. “Theories of Bounded Rationality”, in C.B. McGuire & R. Radner (eds), Decision
and Organization. North Holland Publishing Company. 1972.
Simon, H. A. “Bounded Rationality in Social Science: Today and Tomorrow”, Mind & Society
Vol. 1, 2000: 25-39.
Smith, A. Theory of Moral Sentiments. London: Ward Lock & Co. 1759.
Spence, A. M. Market Signalling: Information Transfer in Hiring and Related Screening
Processes. Cambridge: Harvard University Press. 1974.
Stiglitz, J. E. “Information and Economic Analysis: A Perspective”, Economic Journal Vol.95
Supplement, 1985: 21-41.
Stiglitz, J. E. “The Contributions of the Economics of Information to the Twentieth Century
Economics”, Quarterly Journal of Economics Vol. 115 No. 4, 2000: 1441-1478.
Stiglitz, J. E. “The Financial Crisis of 2007/2008 and its Macroeconomic Consequences”, 2008.
Working Paper available at http://www0.gsb.columbia.edu/faculty/jstiglitz/download/papers
Vaihinger, H. The Philosophy of “As If”: A system of the Theoretical, Practical and Religious
Fictions of Mankind. Translated by C.K. Ogden. New York: Barnes and Noble. 1968.
Varian, H. R. Microeconomic Analysis, 3 rd edition. New York: Wiley. 1992.
Veblen, T. The Theory of the Leisure Class. London: George Allen and Unwin. 1949.
Wilson, T. “Sympathy and self-interest”, in T. Wilson & A. Skinner (eds), The Market and the
State: Essays in Honour of Adam Smith. Oxford: Clarendon Press. 1976: 73-99.
Williamson, O. E. The Economic Institutions of Capitalism. New York: Free Press. 1985.