Institutions
what are they, what are the issues, what insight can we get from Complex
System Dynamics?
GERARD WEISBUCH
Laboratoire de Physique Statistique de l'Ecole Normale Supérieure,
24 rue Lhomond, F 75231 Paris Cedex 5, France.
telephone 331 44 32 34 75, fax 331 44 32 34 33
email:weisbuch@physique.ens.fr
abstract
Institutions structure our life and our societies. They can be seen
as given by God, as a result of some social contract or as attractors of
complex system dynamics...
In the neo-classical vision of economics, they are
no institutions:
-
There are no structures in markets
-
agents are fully rational (full knowledge about the world and other agents)
-
and totally egoist.
This is obviously not the case in real life!
-
Most our behaviours are routinized. We do not take the time at each moment
in our life, to evaluate alternatives using perfect capacities of reasoning
based on complete knowledge. Most of the time, we use standardize routines
that have proven to be succesfull on average (the average might imply average
on time and on members of our social group).
-
Even when we take decisions, they are based on beliefs, not on knowledge.
-
The decision might also be based on pseudo rational procedures such as
imitation.
At this point, I have not yet defined institutions. I will not pretend
to do this later, but rather give you examples of institutions. The above
two examples, routines and beliefs, are institutions which deal with issues
of uncertainty and bounded rationality. Other such institutions
are for instance: norms, signs, languages, and some aspects of cultures.
Another issue facing human societies is cooperation. Game theory
is a formalisation of the issue, but let me recall for you the idea if
you are not familiar with it. In the case of a tax raised to build a road,
the best situation for me is to avoid paying the tax and use the road that
you have paid with your taxes. But if all take the same decision, no road
will ever be built and that situation is worse than the situation of a
road built with everyone's contribution. But how does one establishes cooperation
among a society of egoists? The answer is once more institutions, such
as state, police, hospitals, schools, banks, firms and so on.
On the other hand, institutions is a major theme in social sciences,
where individuals are in situations, i.e. their behaviour is suppposedly
strongly correlated with their social group. In some sense, institutions
are a common issue to all social sciences including economics. Those economists
who study real specific systems do recognize the importance of institutions,
which are most often taken as fixed and given. Others pay attention to
some specific institutions such as property rights and taxes for instance.
Douglass North e.g. has written a fundamental book on the Theory of Institutions
(1990), in which the "role" of institutions is to cope with uncertainty.
What about a Complex System approach?
Let us first note that institutions in social sciences are the equivalent
of patterns in physical systems and functional organisation in biology.
Whatever the language we use, emergence, self-organisation, spatio-temporal
instabilities... cooperative dynamics is basically a description of how
the microscopic properties of simple elements give rise to macroscopic
properties of ensembles. We can then expect that ideas from statistical
physics will be useful in that context. In fact, the methodology is also
based on non-linear dynamics and the combination of these two approaches
is often called Complex Systems Dynamics (Weisbuch 1990).
Let us see the basic scheme for the institutions that may arise on a
market. The loop on figure 1 is typical of a situation of bounded rationality.
An
economic agent will sample data about the market situation and build a
world representation by coding these data into information. The coding
process is equivalent to learning or to interpolation of the data. Decisions
will be taken by the agents according to their representation of the market
and the actions taken by the agents contribute to shaping of the market.
Market dynamics then implies the coupling of cognitive and economic dynamics.
Our description of a market fits into the complex system framework:
the structure of the market is a network of entangled loops, each loop
corresponds to one agent, and agents differ becuase their models of the
world are based on different experience. Complex systems evolve in
time towards attractors. We here interprete attractor configurations
as institutions.
-
For instance, in the case of public information accesssible to all agents
and decisions based on imitation, the dynamics evolves towards strong preferences
for one of the brands or one of the sellers on the market: the preferences
of the agents "emerge" as a result of a dynamics rather than being given
a priori as in standard economics (Follmer 1974, Arthur and Lane 1993,
Kirman 1993, Orlean 1995).
-
Another case is private
information and agents who make their choice according to the experience
they built on the past transactions that they have previously made. A network
of preferential trade appears. Different sellers (or brands) are systematically
preferred by some group of buyers (Weisbuch et al. 1999).
Apart from these two simple paradigmatic examples, a number of other institutions
relate to information processing in markets: brokers or terms of contracts
for instance. Viewing institutions as attractors explains a number of interesting
general features.
-
The stability in time of institutions is a direct consequence of the fact
that they are attractors of the dynamics. Furthermore, their robustness
to change also relates to the same reason. Weak perturbations give little
institutional change, but strong perturbations lead to change of attractors
leading to abrupt changes of behavior, a very non-linear response.
-
For the same set of parameters, different initial conditions might result
in different choices among attractors of the same type: e.g. in a market
selecting one or the other brand.
Figure
2 Time evolution of market share. In this case of an imitation
dynamics, initial market shares above ( resp. below) 50 % evolve towards
an attractor where the market share is 100% (resp. 0%).
-
By contrast, different type of attractors can be observed according to
parameter values, e.g. in a market, fidelity to one brand versus random
selection of brands each time a choice is made.
Figure
3 A regime diagram in parameter space. In this case of private
information and individual buyers choosing from their past experience,
individuals are characterised by a parameter beta describing how much they
prefer to exploit the information they already have against their tendency
to explore to get new information. The T parameter is the time over which
they build information. A sharp transition ( the continuous line) separates
the fidelity to one seller from search among all sellers behaviour. A crossover
(the dotted line) separates eternal fidelity from shorter term relationship.
Parameter space can then be divided in domains which yield the same institutions,
separated by phase transitions. In other words, which type of institution
is observed is a robust discretisation of parameter space. This is an explanation
for the fact that the space of institutions is discrete and not continuous.
To use a biological metaphor, institutions are similar to the well caracterised
species observed in biology. A striking exemple of the stability and discreteness
of institutions is the tribe: all tribes from Siberia to Equator share
a number of common social attributes; and there are no intermediate forms
of social organisation between tribes and chiefdoms or bands.
-
Finally, attractors are generic properties of the dynamics. Their nature
does not depend upon the details of the models, an important invariance
property when one deals with models which are most often strong simplification
of the real world.
References
Arthur, Brian W. and Lane, David .A, (1993), "Information contagion",
Structural Changes and Economic Dynamics, 4, 81-104.
Follmer H. (1974) "Random economies with many interacting agents", Journal
of Mathematical Economics, vol. 1, 1, March, pp. 51-62.
Kirman Alan P., (1993), " Ants , rationality and recruitment ", Quarterly
Journal of Economics. vol. 108, pp. 137-156
North, Douglass C. (1990)" Institutions, Institutional Change and Economic
Performance (Political Economy of Institutions and Decision)"
Orlean A. (1995), ``Bayesian interactions and collective dynamics of
opinions: herd behavior and mimetic contagion'', JEBO, 28, pp. 257-274.
Weisbuch G. (1990), "Complex Systems Dynamics", Addison Wesley, Redwood
City, California.
Weisbuch G., Kirman A. and Herreiner D. (1999), " Market Organisation
and Trading Relationships",
Journal of Theoretical Economics, to appear.
Note: The purpose of the present essay is not the
presentation of neo-classical economics approach to institutions:
these few preliminary and somewhat caricatural remarks are intended to
stress a big difference in perspective.