The Economy as a Complex Adaptive System
A Review of
Eric D. Beinhocker,
The Origins of Wealth: Evolution, Complexity,
and the Radical Remaking of Economics
Herbert Gintis∗
October 31, 2006
In this broad-ranging book, Eric Beinhocker defends a vision of the economy as
a complex adaptive system. The theory that explains the operation of the economic
system he calls Complexity Economics. The Origin of Wealth is a frontal attack on
Neoclassical economic theory. Beinhocker recognizes the successes of this theory,
but locates them in the past. The combined aridity of classical game theory, in which
there have been no new insights for almost twenty years, and general equilibrium
theory, which has produced nothing of general interest since the consolidation of
existence theorems in the 1950’s, accounts for the current eclipse in the status of pure
theory in the eyes of many contemporary economists. For such-minded individuals,
Beinhocker has a message. “Economics can do better,” he says, “it’s time to move
on.” (p. 23)
Some will dismiss this book out of hand because Beinhocker does not provide
a workable analytical alternative to the Neoclassical model. He should not be
dismissed, since the market economy is, in fact, a complex adaptive system, a fact
that materially alters the analytical tools best deployed to model economic behavior.
Indeed, some of the appropriate tools remain to be invented.
From its characterization as a complex adaptive system, it follows that the market
economy follows an evolutionary dynamic. It is well known in evolutionary biology that one can treat genetic and cultural evolution using the same analytical tools
(including the same set of differential equations, the so-called replicator equations),
∗ Santa Fe Institute and Central European University. I would like to thank Samuel Bowles and
Doyne Farmer for helpful comments, and the John D. and Catherine T. MacArthur Foundation for
financial support.
1
and even combine them creatively in gene-culture coevolution (Cavalli-Sforza and
Feldman 1973, Richerson and Boyd 2004). Beinhocker adds that institutions, including firms, also evolve, as does technology and business culture.
1
Complexity Economics
Complexity economics is in a sense a mirror inversion of Neoclassical theory. Beinhocker favorably quotes Axel Leijonhufvud, who remarks that Neoclassical theory
models “smart people in unbelievably simple situations,” while the real world involves ”simple people [coping] with incredibly complex situations.” (p. 52)
What is the Complex Economy? For one thing, the complex economy is never in
equilibrium, but is constantly subjected to shocks, both exogenous and endogenous,
that affect its short-term movements. There are frequent local nonlinear resonances
that lead to significant deviations of economic variables (prices, quantities, wages,
asset prices) from their equilibrium values even in the absence of strong or systematic
perturbations to the system. We see such deviations in many economic time series,
which often have the “fat tails” characteristics of the power laws of complex systems,
as opposed to the Gaussian distributions of Neoclassical theory (Farmer and Lillo
2004).
Beinhocker stresses a second characteristic of the complex economy: the Law
of One Price fails (pp. 61-62). For instance, Beinhocker notes that in the European
Union, the standard deviation of prices rose from 12.3% in 1998 to 13.8% in 2003,
despite the extensive dropping of trade barriers and movement to a common currency
over this period.
A third characteristic of complex adaptive systems is that they rarely, if ever,
achieve the sort of optimality that can be attained in simple engineered systems.
For instance, since economies are rarely in equilibrium, most production, trade,
and consumption takes place out of equilibrium, and hence is Pareto-suboptimal, at
least when measured against a complete information Walrasian economy that has
somehow attained equilibrium.
There have been notable contributions to complexity economics in other areas in recent years. These include Brian Arthur’s work on increasing returns
(Arthur 1994), Peyton Young and Mary Burke’s analysis of crop sharing (Young
and Burke 2001), evolutionary models inspired by Nelson and Winter (1982) and
Hodgson (1998), William Brock and Stephen Durlauf’s study of social interaction
(Brock and Durlauf 2001), Edward Glaeser, Bruce Sacerdote and Jose Scheinkman’s
treatment of crime (Glaeser, Sacerdote and Scheinkman 1996), Samuel Bowles’
treatment of institutional evolution (Bowles 2004), Robert Axtell’s study of firm
size (Axtell 2001), Alan Kirman and his colleagues models of financial markets
2
(Kirman, Foellmer and Horst 2005), and models of the evolution of other-regarding
preferences (Gintis 2000b, Bowles, Choi and Hopfensitz 2003) and the agent-based
simulation of general equilibrium and barter exchange (Gintis 2006a, Gintis 2006b).
Tesfatsion and Judd (2006) is a comprehensive overview of computational methods
in complexity economics.
2
Five Big Ideas in Complexity Economics
Beinhocker offers the following useful summary of the differences between Neoclassical theory and Complexity Economics (p. 97).
a. Dynamics: The complex economy is thermodynamically open, dynamic, nonlinear, and generally far from equilibrium, whereas the Walrasian economy is
thermodynamically closed, static, and linear in the sense that it can be understood using algebraic geometry and manifold theory.
b. Agents: In the complex economy, agents have limited information and face
high costs of information processing. However, under appropriate conditions,
they evolve non-optimal but highly effective heuristics for operating in complex
environments. There is no assurance that when faced with novel environments,
individuals will shift efficiently to new heuristics. In the Neoclassical economy,
by contrast, agents have perfect information and can costlessly optimize.
c. Networks: Agents in the complex economy participate in sophisticated overlapping networks that allow them to compensate for having limited information
and facing formidable information processing costs. In the Walrasian economy, agents do not interact at all. Rather, each agent faces an impersonal price
structure.
d. Emergence: In the complex economy, macroeconomic patterns are emergent
properties of micro-level interactions and behaviors, in the same sense as the
chemical properties of a complex molecule, such as carbon, is an emergent
property of its nuclear and electronic structure, or that thermodynamics is an
emergent property of many-particle systems. In such cases we cannot analytically derive the properties of the macro system from those of its component parts,
although we can apply novel mathematical techniques to model the behavior
of the emergent properties. In the case of the complex economy, these higher
level modeling constructs are currently largely absent, although agent-based
modeling may provide the data needed to develop the appropriate mathematical tools. By contrast, the Walrasian economy has no macro properties that
cannot be derived from its micro properties (for instance, the First and Second
Welfare Theorems).
3
e. Evolution: In the complex economy, the evolutionary process of differentiation,
selection, and amplification provides the system with novelty and is responsible
for the growth in order and complexity. In the Walrasian economy there is no
mechanism for creating novelty or growth in complexity.
3
Evolutionary Economics
Schumpeter (1934) and Nelson and Winter (1982) are among the several leading
economists who have proposed an evolutionary theory of institutional development. Such approaches have been slighted by Neoclassical theorists. Beinhocker
stresses that in Complexity Economics the treatment of organizations, markets, and
economies as subject to an evolutionary process is not meant to be an analogy with
biology, but rather a literal description of dynamical processes in the economy (p.
187). “Evolution,” he concludes, is “a process of sifting from an enormous space of
possibilities…There is no foresight, no planning, no rationality, and no conscious
design. There is just the mindless grinding out of the algorithm.” (p. 198)
This description applies well to institutional and organizational development,
cultural change, and even scientific discovery. The evidence for this view is that
almost all attempts at technical or institutional innovation fail, and few individuals
are responsible for more than one successful innovation. How, then, are great
thinkers possible? When asked how he was able to make so many discoveries,
Linus Pauling replied: “You must have lots of ideas and just throw away the bad
ones.” Great thinkers for the most part simply are attuned to generating mutant
ideas, they evaluate more effectively the prospects for a new mutant idea, and
discard more rapidly the defective mutations. A similar argument likely obtains for
technical change, institutional innovation, and product innovation.
4
Imitation and Learning
Imitation is at the very epicenter of biological evolution, since offspring inherit
the genes of their parents. Imitation is also at the very heart of cultural evolution.
Indeed, the replicator equations for cultural evolution—the very same equations
that hold for genetic evolution—are derived by assuming individuals with lowpayoff strategies switch with some positive probability to the strategies of more
successful agents, with a probability that is increasing in the difference of the payoffs (Gintis 2000a). While there have been a few contributions to the economic
literature on behavioral change through imitation (Conlisk 1988, Bikhchandani,
Hirshleifer and Welsh 1992), and technological diffusion is included in macroeconomic modeling, Neoclassical theorists considers the notion non-standard. Rather,
4
economics has traditionally treated learning as process of individual data gathering and experimentation. The most plausible explanation of this bizarre prejudice
is simply disciplinary cultural inertia. In fact, the human capacity to imitate is
one of the most important and virtually unique aspects of human cognitive ability
(Tomasello, Carpenter, Call, Behne and Moll 2005).
Imitation acquires its importance from the fact that in an evolutionary context
most mutations are deleterious. An agent is thus more likely to improve his position by imitating a successful other, rather than experimenting personally. “Technological evolution,” says Beinhocker, “is not a mere metaphor. It is the result
of humandkind’s deductive-tinkering search through the near infinite possibilities
of Physical Technology space. The nature of the process of differentiation, selection, and replication in this substrate is different from that of biology, but is an
evolutionary process nonetheless.” (p. 259)
5
A New View of Markets
Neoclassical theory has a genius for hijacking terms from everyday discourse and
turning them analytically into their virtual opposites. Consider, for instance, the
term competition. The American Heritage dictionary give several meanings to
the term: the act of competing, as for profit or a prize; a test of skill or ability;
rivalry between two or more businesses striving for the same customer or market;
the simultaneous demand by two or more organisms for limited environmental
resources, such as nutrients, living space, or light. In the perfectly competitive
Walrasian model all agents are price-takers, so there is no competition at all in
the Walrasian economy in any of the above senses. Indeed, perfect competition is
precisely a situation in which individuals have no effect on market outcomes.
Similarly, a market in the Walrasian model is precisely not a market in the
ordinary sense of an institution mediating the exchange of goods. Rather, the
Walrasian market consists of a centralized pricing agent (the “auctioneer”) who
calls out prices, determines how much of each good would be forthcoming given
these prices, and adjusts the prices until excess supply in all markets is zero. Only at
this point does exchange occur, and it apparently takes the form (I say “apparently”
because the process of reallocating goods once prices are set is never described)
of agents dumping their supply goods a certain collection points and picking up
their demand goods at other collection points. In no sense, then is the “market”
of general equilibrium theory akin to the markets in which agents engage in real
competition and exchange.
One of the ironies of history is that if the Walrasian model were plausible, there
would be no need for real markets, real competition, or even capitalism itself. Social-
5
ism, consisting of a bureau of technocrats implementing the Walrasian auctioneer,
could harness the general equilibrium system to a system of public ownership of
wealth. This aspect of the general equilibrium model was clearly understood by
Oskar Lange, F. M. Taylor, and Enrico Barone in their famous defense of market
socialism (Barone 1935, Lange and Taylor 1938). This defense was so successful that it induced Josef Schumpeter to predict the imminent demise of capitalism
(Schumpeter 1942), and led Friedrich von Hayek to rethink, and finally abandon,
his commitment to neoclassical theory (Hayek 1945).
Ironically, however, Neoclassical theory has been an unrelenting defender of
capitalism, and by casting its lot with real-world “competition” and real-world
“markets”, it has thereby made a strategic choice that ensured its victory over the
Socialists, the Syndicalists, the Institutionalists, the Populists, the Anarchists, the
Communalists, and the other various movements that proposed alternatives to capitalism. Nevertheless, neoclassical theory is quite incapable of explaining what role
“competition” and “markets” in fact play in a successful economy, since the terms
refer to completely different concepts in Walrasian theory and in economic reality.
Game theoretic modeling of market competition comes close to explaining why
this economic institution is so central to the success of an economy. “Competition
among agents,” noted Bengt Holmström (1982), “has merit solely as a device to
extract information optimally. Competition per se is worthless.” In other words, it
is only when we have incomplete information that real markets are valuable; in the
complete information context of the general equilibrium model, they are otiose.
Beinhocker does not make the game theoretic argument, but it is implicit in
his evolutionary defense of market competition. “An evolutionary view of the
economy leads one to agree…that markets are good, but for some very different
reasons [from those of Walrasian Economics].” (p. 294) Markets are, in his view,
“an evolutionary search mechanism. Markets provide incentives for the deductivetinkering process of differentiation. They then critically provide a fitness function
and selection process that represent the broad needs of the population…Finally,
they provide a means of shifting resources toward fit modules and away from unfit
ones, thus amplifying the fit modules’ influence.” He explicitly recognizes that the
differentia specifica of markets cannot be captured by neoclassical theory: “Markets
win over command and control, not because of their efficiency at resource allocation
in equilibrium, but because their effectiveness at innovation in disequilibrium.” (p.
294)
6
6
Politics and Policy
Throughout much of the world, the pitched ideological and political battles between
Right and Left over economic policy are a thing of the past. On the Left, there is no
serious movement for the abolition of private property or even the nationalization
of basic industries. Unionionism is increasingly confined to the government sector,
and socialist policy is espoused only in a couple of oil-rich countries. Liberal
Keynesianism is dead in an era where conservatives are prone to run deficits and
liberals to piously admonish such policy as reckless and short-sighted. On the
Right, the notion that the state should be limited to protecting private property is
maintained, against all evidence, only by a radical fringe. Complexity Economics is
not responsible for defusing Right-Left polarities in policy circles. Ineffectiveness
was probably the real culprit.
Nevertheless, as Beinhocker stresses (p. 416), there remain deep ideological
divides of a Right-Left nature that are affected by the considerations analyzed in
The Origin of Wealth. One concerns human nature. “If one digs deeply into the
Left-Right divide…one finds two conflicting views of human nature. On the Left
is the view that human beings are inherently altruistic; that greed and selfishness
stem…from the construction of the social order; and that humans can be made
better through a more just society.…On the Right is the view that human beings
are inherently self-regarding and that the pursuit of self-interest is an inalienable
right. The most effective system of government is one that accommodates rather
than attempts to change this aspect of human nature.” (p. 418) Beinhocker asserts
that behavioral game theoretic findings defuse this polarity by showing that both
sides are wrong, and give us the materials for a far more nuanced and effective set
of policy options than are envisioned in the Left-Right dialogue.
The second Left-Right divide is over the proper weight to be afforded to markets
versus state intervention, the Left stressing market failure and favoring widespread
state intervention, the Right stressing state failure and favoring strict constraints
on intervention. The historical fact that strong states and strong market economies
have coevolved suggests that state intervention is an aspect of the economy as a
complex adaptive system, and the idea of a minimal state is simply a conservative
fantasy. Conversely, the notion that the state can successfully supplant the market is
incompatible with the importance of competition in the economy’s evolutionary dynamic. However, a proper understanding of economic dynamics and human nature
hold out the possibility of rendering state interventions, for instance in addressing
poverty and environmental problems, considerably more effective and politically
popular than hitherto possible.
7
7
Human Nature and Strong Reciprocity
Beinhocker stresses that behavioral economic research, in work pioneered by Fehr
and Gächter (1998), extended by Henrich, Boyd, Bowles, Camerer, Fehr and Gintis
(2004), and summarized in Gintis, Bowles, Boyd and Fehr (2005) and Henrich,
Boyd, Bowles, Camerer, Fehr and Gintis (2005), supports the notion that most
individuals are not purely self-regarding in their social interactions within a group,
but rather are a combination of conditional cooperators, who prefer to sacrifice
personally on behalf of other group members, and altruistic punishers, who prefer
to punish other group members who violate the group’s cooperative norms. The case
of both cooperation and punishment, individuals are willing to so behave even when
there is no possibility of future material reward stemming from their actions—as
is the case, for instance, in one-shot anonymous games in the laboratory and in the
field. This has come to be known as strong reciprocity (Gintis 2000b), the adjective
indicating that this goes beyond the sorts of mutual reciprocity based on enlightened
long-term self-interest stressed in the theory of repeated games (Fudenberg and
Maskin 1986).
The prevalence of strong reciprocity effectively undermines the Hobbesian view
of human nature so dear to conservatives. But it conflicts with the notion of unconditional altruism favored by liberals on two counts. First, strong reciprocators are
conditional cooperators who, in repeated but anonymous interactions, will generally withdraw cooperation if it appears that others are not reciprocating. Second,
strong reciprocators will generally revert to self-interested behavior if the cost of
cooperation becomes sufficiently high (Andreoni and Miller 2002, Gneezy 2005).
Moreover, both Left and Right have generally ignored the altruistic punishment
aspect of strong reciprocity. Indeed, only carefully controlled laboratory studies
have succeeded in isolating altruistic punishment as a “pure” motive rather than
simply a statistical aberration or an attempt to establish a reputation as a hard bargainer. These studies suggest, by contrast with traditional political philosophies of
both Right and Left that altruistic punishment may be the most potent enforcer of
prosocial norms, both in the evolution of our species and in contemporary social
life. Informal, decentralized sanction will normally be more effective in motivating
prosocial behavior than bureaucratic juridical sanctions. “Leges,” wrote Horace in
the Third Ode, “sine moribus vanae” (laws without morality are useless).
8
Strong Reciprocity and Social Welfare Policy
(Gintis et al. 2005) recently edited a volume the first part of which established
the nature of and modeled plausible evolutionary origins of strong reciprocity, and
the second applied this model of human strategic interaction to several spheres of
8
economic policy, including public support for the welfare state (Fong, Bowles and
Gintis 2005). Beinhocker draws upon this example to illustrate the possible uses of
strong reciprocity preferences in framing social policy.
The most widely accepted model of the demand for redistribution in Neoclassical economics is the median voter model, which holds that each voter chooses a
redistribution schema that maximizes his personal wealth (Roberts 1977). It follows
that a the redistribution implemented by majority-rule voting is that preferred by the
median-income voter. Because the distribution of income is generally skewed to the
right, voters and will therefore demand a positive level of redistribution. By contrast, a strong reciprocity voter will choose to redistribute towards individuals who
are needy and deserving, and away from individuals who are rich or non-deserving.
Fong et al. (2005) provide evidence strongly supporting the latter model of voter
behavior.
Beinhocker concludes that to mobilize rather than offend “reciprocal values,”
redistributive policies should be contingent upon the moral behavior of the moral
standing of the recipients. For instance, to the extent that persistent poverty is the
result of low wages and lack of educational opportunity, or is due to mental or
physical defect beyond the control of the individual, a large majority of voters in
the United States are prepared to redistribute in their favor. Similarly, a majority
of individuals approve of policies that insure individuals against bad luck but not
against the consequences of their own actions.
Many traditional projects of egalitarians, such as land reform and employee ownership of their workplaces are strongly consistent with reciprocity norms, as they
make people the owners not only of the fruits of their labors, but more broadly of the
consequences of their actions (Bowles and Gintis, 1998, 1999 provide overviews
based on contemporary principal-agent models). The same may be said of more
conventional initiatives such as improved educational opportunity and policies to
support home ownership. There is good evidence, for example, that home ownership
promotes active participation in local politics and a willingness to discipline personally those engaging in antisocial behaviors in the neighborhood (Sampson, Raudenbush and Earls 1997). An expansion of subsidies designed to promote employment
and increase earnings among the poor, suggested by Edmund Phelps (1997), would
tap powerful reciprocity motives. Similarly, social insurance programs might be
reformulated along lines suggested by John Roemer (1993) to protect individuals
from risks over which they have no control, while not indemnifying people against
the results of their own choices, other than providing a minimal floor to living standards. In this manner, for example, families could be protected against regional
fluctuations in home values—the main form of wealth for most people—as Robert
Shiller (1993) has shown. Other forms of insurance could partially protect workers
from shifts in demand for their services induced by global economic changes.
9
9 The Philosophy of State Intervention
There are three philosophies of state intervention in America. Liberals generally
believe that markets wreak havoc with people’s lives and generate extreme income
inequality, state intervention being the appropriate remedy. Conservatives believe
that the market almost always produces efficient and just outcomes, so the state
should be limited to protecting the rules of the game. Economists generally believe that both market and state failure are serious problems, and any proposed
intervention must be thoroughly analyzed in terms of costs, benefits, and feasibility. Beinhocker argues that Complexity Economics offers a philosophy of state
intervention that is distinct from all of the above.
“A complexity perspective,” he asserts, “would distinguish between two types
of government action. Policies that get the government involved in differentiating,
selecting, and amplifying Business Plans would be seen as interfering in economic
evolution…In contrast, policies that shape the fitness environment while leaving
Business Plan selection and amplification to market mechanisms” are useful and
even necessary (p. 426). This position is probably closest to that of the Neoclassical
economics view, except that Beinhocker’s approach is more ecological in that many
of the parameters involved in assessing the effect of an intervention on the fitness
environment of the economy are interdependent and difficult to measure. He conclusion that the “economic role of the state is to create an institutional framework
that supports the evolutionary workings of markets” (p. 427) suffers from a certain
vagueness, but may be workable in practice.
10 Economics for our Grandchildren
Since Beinhocker’s wide-ranging intellect appears to touch on all subjects, it is
surprising that he does not reconsider the model of personal well-being implicit
in Neoclassical economic theory. This model, based on the neoclassical utility
function, proposes that happiness is getting more of what you want, and expending
the least possible effort to do so. Both sides of this assertion are doubtful. A
long line of economists, including Lane (1993), Easterlin (1995), Oswald (1997),
Kahneman (1999) and Kahneman, Diener and Schwartz (1999) have documented
the negligible effect of material wealth on personal well-being for individuals above
the poverty line. Similarly, there is a strong political economy tradition asserting
the centrality of work in personal well-being and self-esteem (Frey 1997).
There is a model of human well-being compatible with the notion that we humans are complex adaptive systems, endowed by our genetic constitution with
certain capacities—cognitive, affective, psychomotor, aesthetic, and spiritual—and
an individual well-being depends on the extent that we have developed these capac10
ities and have the means of exercising them. Happiness, this view, is not what you
have, but what you are. Societies are judged, then, not on what material comforts
they generate, but on the extent to which they foster the development of human
beings fully capable of exercising their personal capacities.
11 Agent-based Modeling of Market Exchange
My own foray into modeling general equilibrium as a complex adaptive system
suggests that the Walrasian model will emerge delimited but enriched rather than
replaced as a result of such research (Gintis 2006a, Gintis 2006b). General equilibrium theory captures important long-term historical aspects of a market economy,
and many of the basic insights of the Walrasian model will be retained, albeit modified. Even in the long run, there will be a strictly positive rate of unemployment,
supply will exceed demand, efficiency will be considerably less than 100%, and
there will be other deviations from equilibrium due to incomplete information and
other “frictions” amplified by local nonlinear resonances. Moreover, the Walrasian
assumption that agents are price-takers, that complete contracts can be written for
all important exchanges and can be costly enforced by a third party, are all unrealistic. Hence, the Walrasian system is a very poor guide to micro-modeling real
economic transactions. In particular, the Walrasian assumptions concerning labor
markets, capital markets, and consumer goods markets are misleading (Bowles and
Gintis 1993, Gintis 2002).
My own recent work has taken the form of developing an agent-based model of
a general equilibrium system with many sectors, firms and consumers, capital and
labor, in which there is no public information (Gintis 2006a). The results are well
described by Beinhocker. Prices in this system are private information, in the sense
that each firm generates its own prices and must engage in costly search to discover
the pricing strategies of its competitors. Moreover, each worker has his own private
discount rate, disutility of labor, and reservation wage. and consumers discover
favorable goods prices through using a search strategy. Major forms of change in
this economy are that agents experiment with their own parameters and assess the
results, and agents copy the behavior of others that have been more successful than
themselves.
Perhaps surprisingly, my dynamical model of general market interdependence,
which is clearly a complex adaptive system in the sense of Beinhocker, stabilizes
rather quickly to a stationary distribution in which prices are approximately at their
market-clearing values, profits are approximately zero, and the unemployment rate
averages about 4%. But the devil is in the details. My economy exhibits “fat tails”
with many large excursion from equilibrium, even in the absence of any systemic
11
Figure 1: Demand and Supply in Sector 1 of a Multi-Sector Economy. Note that
there is considerable period-to-period volatility. Excess supply averages about 8%
of average supply.
shock (see Figure 2). Moreover, as Beinhocker describes in his discussion of the
“Law of Supply and Demand,” (p. 60), the general case is one in which firms
experience excess supply for their product and have generally positive excess demand for labor, even while labor is in aggregate excess supply. This is illustrated
in Figure 1 for one of the sectors in my agent-based model of the economy. In
effect, firms evolve by developing reaction functions to changing economic conditions that maximize profits, but not by solving complicated maximization problems
with extensive knowledge of demand and supply conditions in all input and output
markets, but simply by “muddling through” with judicious jiggling of their reactive
behavior and occasional switching to a strategy of a better-performing competitor.
Like complex adaptive systems in general, my agent-based model does not
achieve optimality. Average consumption reaches about 75% of Pareto-optimal
consumption after about two hundred periods, and oscillates thereafter at between
72% and 80% efficiency (Figure 3).
Finally, my model clearly exhibits the failure of the Law of One Price, as
shown for a ten-sector economy in Figure 2. Note that after about 100 periods the
standard deviation of prices settles to an average level which persists throughout the
3000 period run, punctuated by frequent price excursion of as much as 50% of the
equilibrium price (which is unity for all goods, by construction). These excursions
12
Figure 2: Sectoral Prices. Since conditions of production are the same in all sectors,
relative prices are unity in equilibrium. Taken from Gintis (forthcoming)
are do to any systematic disturbance applied to the underlying model. Rather, they
are local resonances that account for the some of the “fat tails” of stochastic variables
in a complex economy.
12 The Economy is a Biological System
Beinhocker’s critique of Neoclassical theory includes the assertion that the economy
is a biological system subject to the same sort of evolutionary dynamic as other
sociobiological systems, but neoclassical theory models the economy as the solution
to a set of equilibrium conditions. No one has been successful in modeling a bee hive
as the solution to a set of equilibrium conditions. Modeling the market economy,
orders of magnitude more complex than a bee hive, as the solution to a set of
equilibrium conditions is that much less likely to success. Neoclassical theory lacks
explanatory power because it uses the theoretical tools of mathematical physics
while ignoring those of mathematical biology, seeking the sorts of explanations
13
Figure 3: The Efficiency of the Simulated Economy. Note that after only 100
periods, the system has attained nearly 79% efficiency. After 100 periods, average
utility appears to be a random walk between 72% and 80% (the apparent declining
trend in the figure is an artifact of truncation to 3000 periods).
relevant to electro-mechanical systems rather then evolved biological systems.
The stunning success of modern physics and chemistry cannot be replicated
in economics for one simple reason. Physical theory’s success lies in modeling
non-complex, non-adaptive systems. The experimental method in physics is to
create highly simplified laboratory conditions, under which modeling becomes analytically tractable. Physics is no more effective than economics or biology in
analyzing complex real-world phenomena. The various branches of engineering
(electrical, chemical, mechanical) are effective because they recreate in everyday
life artificially controlled, non-complex, non-adaptive, environments in which the
discoveries of physics and chemistry can be directly applied. This option is generally not open to economists, who rarely have the opportunity of “engineering”
economic institutions and cultures.
The recent success of behavioral game theory (Smith 1962, Plott 1979, Camerer
2003) suggests the great value of carrying out controlled laboratory experiments in
economics, but in the absence of “economic engineers” who can reproduce laboratory conditions in the field, economic policy will surely never rival the scientific
accuracy and objectivity of physics and chemistry.
14
13 Why is Neoclassical Theory Sui Generis Theorizing?
The unwillingness of Neoclassical theory to deal with the complex adaptive nature
of economic activity has led a generation of talented theorists to betray a pervasive
indifference to empirical relevance. Beinhocker does not attempt to explain this
development, but it is important to do so if we are to move forward in economic
theory. Therefore, I shall put forward my own understanding of why this occured.
In the natural sciences, theorists appreciate experimentalists because the latter
provide the data that allow them to distinguish among competing models. Economic theorists, by contrast, are indifferent, or even actively hostile, to those who
do controlled experiments. A typical example is the following opinion of Nobel
prize economist Robert Aumann (Hart 2005): “I have grave doubts about what’s
called “behavioral economics”…most of behavioral economics deals with artificial
laboratory setups.…true behavioral economics does in fact exist; it is called empirical economics.…In empirical economics, you go and see how people behave in
real life, in situations to which they are used.”
Neoclassical theorists are not threatened by “empirical economics” because the
lack of controlled laboratory conditions makes it impossible ever to contradict the
theory. Empirical economics is theoretically irrelevant, and hence perfectly safe. I
call this retreat from explaining the world sui generis theorizing, by which I mean
that theorists devise their own standards of adequacy that have nothing to do with
an encounter with real-world phenomena.
Sui generis theorizing flirts with being unscientific. Some theorists defend
their research agenda by claiming to be, or to be like, pure mathematicians. I am
sympathetic to this position, but then what they do should not be taught as basic
economic theory in Ph.D. programs, and a whole new branch of economic theory
that does grapple with the real world should be fostered.
This sorry situation of sui generis theorizing became clear to me in the summer
of 2001, when I happened to be reading a popular graduate text in quantum physics,
as well as a leading graduate text in microeconomics. The physics text began with
the anomaly of black body radiation which was inexplicable using the standard
tools of electromagnetic theory. In 1900, Max Planck derived a formula that fit the
data perfectly, assuming that radiation was discrete rather than continuous. In 1905,
Albert Einstein explained another anomaly of classical electromagnetic theory, the
photoelectric effect, using Planck’s trick. The text continued, page after page, with
new anomalies (Compton scattering, the spectral lines of elements of low atomic
number, etc.) and new, partially successful models explaining the anomalies. This
culminated in about 1925 with Heisenberg’s wave mechanics and Schrödinger’s
equation, which fully unified the field.
By contrast, the graduate microeconomics text, despite its brilliance, did not
15
contain a single fact in the whole thousand page volume (actually, there were two
references to facts, both in footnotes). Rather, the authors build economic theory
in axiomatic fashion, making assumptions on the basis of intuitive plausibility or
consonance with the principles of “rational action.” This is sui generis theorizing
at its most pure.
When the world does not conform to the models, microeconomic theory blames
the world—information is incomplete, competition is imperfect or markets are missing, there are market externalities, or individuals are irrational. Implicit in such
terminology is that there is another world (akin to the physicists frictionless world)
in which information is complete, markets are pervasive, and the costs of information processing are zero. A bounty of excellent economic theory was developed in
the twentieth century based on this assumption. But, this is all in the past. There is
no such idealized world, nor is there any sense in which the economy “works better” as conditions approximate this ideal. We cannot understand a complex adaptive
system by analyzing the system’s behavior in the limit as the degree of complexity
goes to zero.
Why did economic theory take this sui generis turn in the past few decades? The
reason, I believe, is because of the phenomenal success of sui generis theorizing
in the in the early post-WWII period. Let me first note that sui generis theorizing
can be extremely productive even in the natural sciences. Some of Albert Einstein’s
greatest discoveries were sui generis, including special relativity, which follows
from the constancy of the speed of light in inertial frames, and general relativity,
which follows from the equivalence of inertial and gravitational mass and a thought
experiment concerning the path of a light ray as viewed from an accelerating frame.
Indeed, economists themselves developed two stunningly successful sui generis
theories in the first half of the Twentieth century: decision theory and general
equilibrium theory.
Decision theory is an archetypal case of a theory of great explanatory value
developed on the basis of abstract theorizing alone. The creators of modern decision
theory did not pore over reams of data before discovering their underlying unity.
Rather, they drew out the implications of a few intuitively plausible axioms of
rational choice (the most important being transitivity over the appropriate phase
space). Yet decision theory is highly predictive, and when individuals deviate from
its prescriptions, individuals are usually wrong (non-optimizing) and the theory
correct. Nobel prize winning psychologist Daniel Kahneman and his coworkers,
among others, have shown that there are ample real-world deviations from the
predictions of decision theory (Kahneman and Tversky 2000), but these findings
have enriched rather than undermined decision theory which stands as one of the
most successful models in the behavioral sciences (Gintis 2006c).
The second major sui generis accomplishment was general equilibrium theory
16
itself. Walras did not derive the building-blocks of his model from close attention
to empirical findings. Rather, he noted that there were consumers, producers, products, raw materials, labor and capital, and prices. He put them all together in an
appealing package leaving enough puzzles to solve for the next three generations
of economists, culminating in the work of Debreu, Arrow, and their colleagues in
the early post-WWII period.
It is reasonable to say, then, that an economic theorist coming of age in, say
1970, might have thought that sui generis theorizing is the best way to go. In fact,
many took exactly this path, including those working on general equilibrium theory
and game theory. But it has been virtually a dead end. The logical next step in
decision theory was to extend the axioms to strategic interaction, which was the
task of classical game theory. What theorists found was that there is no way to
accomplish this extension that has any logical or empirical plausibility. This fact
remains virtually unknown except among game theorists working in the interactive
epistemology area, because the major results are of post 1990-vintage, whereas virtually every game theory text is based on earlier theory. Most important, economists
expected a rational actor model justification of Nash equilibrium, despite the fact
that Bernheim (1984) and Pearce (1984) had shown that rationality justifies at best
the iteration of strictly dominated strategies. Aumann and Brandenburger (1995)
provided sufficient conditions for Nash equilibrium, but these can be expected to
obtain in only the simplest of situations.
14 The Failure of Sui Generis Game Theory
One of the chief embarrassments of classical game theory was its inability to explain
why an individual would ever play a mixed strategy in a one-shot game. Since
randomizing must be costly (for elementary information-theoretic reasons), and
since all pure strategies in the support of a mixed strategy have equal payoff, a
rational agent should never player a mixed strategy. Nor should he expect his
opponent to play a mixed strategy. Harsanyi (1973) developed the ingenious notion
of “purification” to justify the mixed strategy assumption, but this only applies to a
highly specialized set of games, and virtually never to repeated games where agents
in a Nash equilibrium must play mixed strategies of the stage game. The current state
of affairs is that evolutionary game theory provides a justification both of the Nash
equilibrium concept and of the mixed strategy concept, but only in situations where
a system has achieved equilibrium through an evolutionary dynamic (Gintis 2000a).
Finally, classical game theorists expected a justification of backward induction
on rational actor grounds, and one was provided by Aumann (1995). However,
few theorists accept this justification because it depends on implausible off-the-
17
equilibrium-path reasoning. The fact is that human subjects rarely used more that
two or three stages of backward induction, and this in no way violates the assumptions of rational action.
The substantive achievements of behavioral game theory flow from the fact that
it provides facts that can be used to build a theory of strategic interaction, whereas
the sui generis methodology of classical game theory is powerless to do so.
15 The Failure of Sui Generis General Equilibrium Theory
Just as a 1970 theorist might have expected decision theory to generalize smoothly
to game theory, a general equilibrium theorist might have expected that it would
be feasible to add institutional realism to the general equilibrium model, and to
flush out the dynamic properties of the model in a theoretically satisfying manner.
Both of these expectations failed to materialize. It turns out to be a major mathematical tour de force to add even a small degree of institutional complexity to the
general equilibrium model, and there has been absolutely no progress in providing
a dynamic to a generalized model of production and exchange. Most embarrassingly, attempts to formalize the Walrasian tâtonnement process have led precisely to
complex nonlinear dynamics in which prices are, unlike in observable economies,
generically chaotic.
Of course, if the Walrasian tâtonnement process led to stability, this would give
little comfort to general equilibrium theory. The idea of calling out prices and asking
agents to reveal their demand and supply amounts is not only bizarre but completely
lacking in incentive compatibility. Agents have no incentive to lie, of course, since
in competitive equilibrium no individual makes a difference. However, they have
no incentive to tell the truth either, and they have a strong disincentive to engage in
the costly information-gathering required for an informed response.
Franklin Fisher (1983), in his masterful summary of the dynamics of general
equilibrium, complained that there had been little progress in many years. It is now
a quarter century later and there is still little progress (Saari 1985, 1995)1
16 Conclusion
Economists are envious of the explanatory power of modern natural science, and
rightly so. What we have tended not to notice is that these natural science disciplines
base their success on creating conditions in the laboratory that eliminate complexity,
thus permitting the development of analytical models with high levels of accuracy.
1 There has been progress in obtaining Walrasian equilibrium as the limit of a barter process, but
there are no dynamical implications of this research. See Gale (2000).
18
The engineering disciplines then recreate these non-complex conditions in everyday
life, which accounts for the success of modern technology.
We can partially follow the lead of natural science by studying social life in the
laboratory under conditions favorable to analytical modeling. This is the path taken
by behavioral game theory and agent-based modeling. We can even transport noncomplexity to the economy in special cases, such as devising sophisticated natural
resource and bandwidth auctions. However, in general we must study the economy
as a complex adaptive system directly—a formidable task indeed.
The virtue of this book is that it makes clear the nature of the beast, but it does
not provide us with the tools to undertake its effective analysis. Developing these
tools will be a formidable but breathtakingly exciting task.
References
Andreoni, James and John H. Miller, “Giving According to GARP: An Experimental Test of the Consistency of Preferences for Altruism,” Econometrica 70,2
(2002):737–753.
Arthur, Brian, Increasing Returns and Path Dependence in the Economy (AnnArbor:
University of Michigan Press, 1994).
Aumann, Robert and Adam Brandenburger, “Epistemic Conditions for Nash Equilibrium,” Econometrica 65,5 (September 1995):1161–80.
Aumann, Robert J., “Backward Induction and Common Knowledge of Rationality,”
Games and Economic Behavior 8 (1995):6–19.
Axtell, Robert, “Zipf Distribution of U.S. Firm Sizes,” Science 293 (2001):1818–
1820.
Barone, Enrico, “The Ministry of Production in the Collectivist State,” in F. A.
von Hayek (ed.) Collectivist Economic Planning (London: Routledge, 1935)
pp. 245–290.
Bernheim, B. Douglas, “Rationalizable Strategic Behavior,” Econometrica 52,4
(July 1984):1007–1028.
Bikhchandani, Sushil, David Hirshleifer, and Ivo Welsh, “A Theory of Fads, Fashion, Custom, and Cultural Change as Informational Cascades,” Journal of Political Economy 100 (October 1992):992–1026.
Bowles, Samuel, Microeconomics: Behavior, Institutions, and Evolution (Princeton: Princeton University Press, 2004).
and Herbert Gintis, “The Revenge of Homo Economicus: Contested Exchange
and the Revival of Political Economy,” Journal of Economic Perspectives 7,1
(Winter 1993):83–102.
19
and , “Schooling, Skills and Earnings: A Principal-Agent Approach,” in
Kenneth Arrow, Samuel Bowles, and Steven Durlauf (eds.) Meritocracy and
Economic Inequality (Princeton, NJ: Princeton University Press, 1998).
and , Recasting Egalitarianism: New Rules for Communities, States, and
Markets (London: Verso, 1999). Erik Olin Wright (ed.).
, Jung-kyoo Choi, andAstrid Hopfensitz, “The Co-evolution of Individual Behaviors and Social Institutions,” Journal of Theoretical Biology 223 (2003):135–147.
Brock, William and Steven N. Durlauf, “Discrete Choice With Social Interactions,”
Review of Economic Studies 68,235 (April 2001):235–260.
Camerer, Colin, Behavioral Game Theory: Experiments in Strategic Interaction
(Princeton, NJ: Princeton University Press, 2003).
Cavalli-Sforza, L. and M. W. Feldman, “Models for Cultural Inheritance: Within
Group Variation,” Theoretical Population Biology 42,4 (1973):42–55.
Conlisk, John, “Optimization Cost,” Journal of Economic Behavior and Organization 9 (1988):213–228.
Easterlin, Richard A., “Will Raising the Incomes of All Increase the Happiness of
All?,” Journal of Economic Behavior and Organization 27,1 (June 1995):35–47.
Farmer, J. Doyne and F. Lillo, “On the Origin of Power Law Tails in Price Fuctuations,” Quantitative Finance 4,1 (2004):7–11.
Fehr, Ernst and Simon Gächter, “How Effective are Trust- and Reciprocity-based
Incentives?,” in Louis Putterman and Avner Ben-Ner (eds.) Economics, Values
and Organizations (New York: Cambridge University Press, 1998) pp. 337–363.
Fisher, Franklin M., Disequilibrium Foundations of Equilibrium Economics (Cambridge, UK: Cambridge University Press, 1983).
Fong, Christina M., Samuel Bowles, and Herbert Gintis, “Reciprocity and the Welfare State,” in Herbert Gintis, Samuel Bowles, Robert Boyd, and Ernst Fehr (eds.)
Moral Sentiments and Material Interests: On the Foundations of Cooperation in
Economic Life (Cambridge: The MIT Press, 2005).
Frey, Bruno, Not Just for the Money: An Economic Theory of Personal Motivation
(Cheltenham, UK: Edward Elgar, 1997).
Fudenberg, Drew and Eric Maskin, “The Folk Theorem in Repeated Games
with Discounting or with Incomplete Information,” Econometrica 54,3 (May
1986):533–554.
Gale, Douglas, Strategic Foundations of General Equilibrium: Dynamic Matching
an Bargaining Games (Cambridge: Cambridge University Press, 2000).
Gintis, Herbert, Game Theory Evolving (Princeton, NJ: Princeton University Press,
2000).
20
, “Strong Reciprocity and Human Sociality,” Journal of Theoretical Biology 206
(2000):169–179.
, “Some Implications of Endogenous Contract Enforcement for General Equilibrium Theory,” in Fabio Petri and Frank Hahn (eds.) General Equilibrium:
Problems and Prospects (London: Routledge, 2002) pp. 176–205.
, “The Dynamics of General Equilibrium,” The Economic Journal (In Press)
(2006).
, “The Emergence of a Price System from Decentralized Bilateral Exchange,”
The B.E. Journals in Theoretical Economics (In P ress) (2006).
, “A Framework for the Unification of the Behavioral Sciences,” Behavioral and
Brain Sciences (In Press) (2006).
, Samuel Bowles, Robert Boyd, and Ernst Fehr, Moral Sentiments and Material
Interests: On the Foundations of Cooperation in Economic Life (Cambridge:
The MIT Press, 2005).
Glaeser, Edward L., B. Sacerdote, and J. Scheinkman, “Crime and Social Interactions,” Quarterly Journal of Economics 111 (May 1996).
Gneezy, Uri, “Deception: The Role of Consequences,” American Economic Review
95,1 (March 2005):384–394.
Harsanyi, John C., “Games with Randomly Disturbed Payoffs: A New Rationale
for Mixed-Strategy Equilibrium Points,” International Journal of Game Theory
2 (1973):1–23.
Hart, Sergiu, “An Interview with Robert Aumann,” Macroeconomic Dynamics 9,5
(2005):683–740.
Hayek, F. A., “The Use of Knowledge in Society,” American Economic Review
35,4 (September 1945):519–530.
Henrich, Joseph, Robert Boyd, Samuel Bowles, Colin Camerer, Ernst Fehr, and
Herbert Gintis, Foundations of Human Sociality: Economic Experiments and
Ethnographic Evidence from Fifteen Small-scale Societies (Oxford: Oxford University Press, 2004).
, , , , , and , “Economic Man’ in Cross-Cultural Perspective: Behavioral Experiments in 15 small-scale societies,” Behavioral and Brain Sciences
(2005).
Hodgson, Geoffrey M., “The Approach of Institutional Economics,” Journal of
Economic Literature 36,1 (1998):166–192.
Holmström, Bengt, “Moral Hazard in Teams,” Bell Journal of Economics 7
(1982):324–340.
Kahneman, Daniel, “Objective Happiness,” in Daniel Kahneman, Edward Diener,
and Barry Schwartz (eds.) Well-Being: Foundations of Hedonic Psychology
21
(New York: Russell Sage Foundation Press, 1999).
and Amos Tversky, Choices, Values, and Frames (Cambridge: Cambridge University Press, 2000).
, Edward Diener, and Barry Schwartz, Well-Being: Foundations of Hedonic
Psychology (New York: Russell Sage Foundation Press, 1999).
Kirman, Alan, H. Foellmer, and U. Horst, “Equilibrium in Financial Markets with
Heterogeneous Agents: A New Perspective,” Journal of Mathematical Economics 41,1–2 (February 2005):123–155.
Lane, Robert E., “Does Money Buy Happiness?,” The Public Interest 113 (Fall
1993):56–65.
Lange, Oskar and F. M. Taylor, On the Economic Theory of Socialism (Minneapolis:
University of Minnesota Press, 1938).
Nelson, Richard and Sidney Winter, An Evolutionary Theory of Economic Change
(Cambridge: Harvard University Press, 1982).
Oswald, Andrew J., “Happiness and Economic Performance,” Economic Journal
107,445 (November 1997):1815–1831.
Pearce, David, “Rationalizable Strategic Behavior and the Problem of Perfection,”
Econometrica 52 (1984):1029–1050.
Phelps, Edmund S., Rewarding Work: How to Restore Participation and Selfsupport to Free Enterprise (Cambridge, MA: Harvard University Press, 1997).
Plott, Charles R., “The Application of Laboratory Experimental Methods to Public
Choice,” in Clifford S. Russell (ed.) Collective Decision Making: Applications
from Public Choice Theory (Baltimore, MD: Johns Hopkins University Press,
1979) pp. 137–160.
Richerson, Peter J. and Robert Boyd, Not By Genes Alone (Chicago: University of
Chicago Press, 2004).
Roberts, Kevin, “Voting over Income Tax Schedules,” Journal of Public Economics
8 (December 1977):329–340.
Roemer, John, “A Pragmatic Theory of Responsibility for the Egalitarian Planner,”
Philosophy and Public Affairs 22 (1993):146–166.
Saari, Donald, “Mathematical Complexity of Simple Economics,” Notices of the
American Mathematical Society 42,2 (February 1995):222–230.
Saari, Donald G., “Iterative Price Mechanisms,” Econometrica 53 (1985):1117–
1131.
Sampson, Robert J., Stephen W. Raudenbush, and Felton Earls, “Neighborhoods
and Violent Crime: A Multilevel Study of Collective Efficacy,” Science 277
(August 15, 1997):918–924.
22
Schumpeter, Joseph, The Theory of Economic Development: An Inquiry into Profits, Capital, Credit, Interest and the Business Cycle (Oxford: Oxford University
Press, 1934).
, Capitalism, Socialism, and Democracy (New York: Harper & Row, 1942).
Shiller, Robert J., Macro Markets: Creating Institutions for Managing Society’s
Largest Economic Risks (Oxford: Clarendon Press, 1993).
Smith, Vernon, “An Experimental Study of Competitive Market Behavior,” Journal
of Political Economy 70 (1962):111–137.
Tesfatsion, Leigh and Kenneth L. Judd, Handbook of Computational Economics II:
Agent-Based Computational Economics (North-Holland, 2006).
Tomasello, Michael, Malinda Carpenter, Josep Call, Tanya Behne, and Henrike
Moll, “Understanding and Sharing Intentions: The Origins of Cultural Cognition,” Behavioral and Brain Sciences 28,5 (2005):675–691.
Young, Peyton and Mary Burke, “Competition and Custom in Economic Contracts: A Case Study of Illinois Agriculture,” American Economic Review 91,3
(2001):559–573.
c\Papers\Reviews\Beinhocker.tex October 31, 2006
23