Butterfly Economics

by Paul Ormerod

Paul Ormerod, Butterfly Economics: A New General Theory of Social and Economic Behaviour (New York: Pantheon Books, 1998).

On his web site, www.paulormerod.com, he summarizes the book as follows: "I argue that the key weakness of standard economics is the assumption that the tastes and preferences of agents – people, firms – are fixed. Instead, agents adapt their tastes in the light of what other people do. This gives a more realistic and quite different picture of the world."

From the first chapter, that's not the way I would describe his proposition.  Rather, he provides a specific description of individual behaviour that gives "unpredictability in the short run merging imperceptibly over time into a form of regularity":

In most circumstances, a person can either stay with the pattern of behaviour she or she previously followed (an ant visiting its previous site), can decide to switch of his or her own volition, or can be influenced into switching by the observed behavior of others.

Page 10.

My attack on the usefulness of this insight lies in the distinction between the second and third choices.  I don't see how any measurable or perceivable difference can be made between being “persuaded” to do a thing or choosing of one’s own volition (page 5).  How can anyone, human or ant, make up his or her mind to do something without knowing about it, and how can he or she know about it without having seen others do it or been told by other people or ants?  In the context of the ant experiment, how can the ant know about the other site except by observing or being told by other ants?

My quibble aside, in an article called "Revisiting The Death of Economics" (his 1994 book)( published in WORLD ECONOMICS (Vol. 2, No. 2; April-June 2001) he seems to dispense with this tripart division of decisionmaking and states his broad position aptly: "Individuals and firms in standard economic theory can process huge amounts of information in exceptionally complicated ways, but the one thing they are not allowed to do is to alter their behaviour in the light of what others are doing. They respond to the decisions of others only in so far as these affect the prices of the goods and services that the individual buys and/or sells. They do not want a Teletubbie, say, or a hula hoop or, much more seriously, a 30-year US government bond rather than a French one, simply because other people do. But in the real world this sort of behaviour is pervasive. From fashion markets to financial markets to the degree of optimism or pessimism which firms feel about the future, the opinions and behaviour of others affects directly how individuals behave."

In the 2001 article "Revisiting The Death of Economics he also discusses a computer simulation of a market that doesn't yet exist, pollution-permit trading firms: "In other words, in this strictly orthodox world, the outcome nevertheless depends to a large extent upon the particular institutional setting in which the economy is assumed to operate. With one set of institutions, a naïve process of learning enables the equilibrium price to be discovered with a very small number of firms. With another set, it is much harder for them to discover this price."

I take this as going to the heart of a pet project of mine, powertoexchange.com, which attempts to make at least the promised efficiency or 'Pareto optimality' of markets a reality with an institution that creates the market, that provides the "power to exchange." Of course, at this time I'm aproaching it on a very shallow level, economically-intellectually, and I mistrust auctions greatly. I just want an institution that gets the information out there, where all information and prices can be compared. I suppose I shall have to look into what kind of auctions are best, but in general I hope (and expect) that price-setting based on what other people are pricing, how the buying is going, and trial-and-error can beat out auctions in anything that isn't wholly unique.

In the book (pages 15-16) he points out that the great volatility of share prices in a company are an ‘empirical falsification’ of orthodox theory, citing Kenneth Arrow.  The changes cannot be due simply to new information becoming known.  This relates to solutions he presents in his article when talking about permit trading.

Our focus of interest was in a macro phenomenon of this system—the evolution of the price of permits.  This emerged from the decisions and interactions of the individual, micro-level agents in the model.

Alan Kirman has an excellent model of the exchange rate, which emerges from the decisions of individual traders. At any point in time, a trader operates as a fundamentalist—looking at currency fundamentals—or as a chartist, extrapolating recent trends. Individual agents switch between these modes over time, and a key reason for switching is the number of other traders who are operating in any given mode. If everyone else is a chartist, it’s pretty hard to keep your nerve and stay fundamentalist.

These are examples of what I believe will be the future of theoretical analysis in economics. Other examples which I give in Butterfly Economics include the choice of restaurants, the success and failure of Hollywood films, the evolution of crime, how family structures change, and the American business cycle. We are now in the position of being able to create microfoundations of macro phenomena, computer-based models in which individual agents following rules of behaviour interact with each other. And the macro properties of the system emerge from these interactions.

Orthodox economics claims to be able to do this already. Indeed, it can, but only under very special circumstances. All theories are only approximations to reality, and so all theories require simplifying assumptions. A very restrictive assumption in conventional economics is that the tastes and preferences of individual agents are fixed and independent.

[...]

Relaxing the assumption of fixed preferences opens up the possibility of a much better understanding of how the world operates. Conventional theory can be thought of merely as a special case of this far more general approach; it has its greatest validity in circumstances in which fixed preferences offer a reasonable approximation to reality, like the shopper in the supermarket.

[Changes allowable in preferences: ] First, individual agents may decide of their own accord to switch between states of the world, on a probabalistic basis. Second, individual agents may be influenced by other agents to switch to their state of the world, again on a probabalistic basis.

[...]

In this approach, individual agents are given behavioural rules. The properties of the system as a whole emerge from the interaction of these rules. The macro-properties which emerge from micro-behavioural rules can then be compared with the actual macro-properties.

So I find this very exciting. Here is a development that enables orthodox economics to be undermined on its own terms. Yes, conventional economics does have micro foundations of macro behaviour, but these only applicable in special circumstances, and the claim of orthodoxy to generality, to general rules, collapses.

The approach also calls for very careful consideration of the framework in which the model is set up—the institutional setting. In trying to understand the possibilities for the exchange rate, for example, a world of free capital movements implies a different set of behavioural rules for the micro level agents than a world in which capital is controlled.

This means that we must abandon the wholly unrealistic claim of mainstream economics to have discovered a general rule of agent behaviour—namely maximising behaviour—which applies in all circumstances. I do not have space enough here to get into the enormous amount of evidence from the cognitive sciences about the limited ability of agents to process information, which undermines the concept of maximisation except in very special circumstances. The point I am making is that the appropriate rules of agent behaviour will be shaped by the institutional setting in which agents operate.

Personally, I have always interpreted the basic premise of economics to be that people make the best choices they can, so none of this is shattering my world.

The essence of Kirman’s model is that in any given period an agent can continue to behave in the same way as before; he or she can change behavior independently in reaction to news; or the agent can be persuaded to switch by the behaviour of others – by the trails which they leave when their buy or sell decisions appear on the dealing screens.  In other words, we have the identical analytical framework to that of the ants model.

Page 19.

there are many examples of products which are technologically inferior not just surviving, but driving out of existence competitors with distinctly superior qualities.  The free market chooses not the best, but the worst.

Page 20.

He cites his own experience with Betamax VCRs, which he considers to have been far superior in ease of use as well as functions and quality, but which he had to give up for VHS because very soon videos didn't come in Betamax.

Because I focus on what I most dislike, and perhaps due to excess cynicism, I assume that a large part of the reason inferior technologies can beat out superior ones is due to corruption: 'abuse' of market power, using influence over government actions to beat rivals, or anything else that isn't free competition.

What I take from examples like Betamax is that important decisions are made not by all the people operating through the market or a democratic decision-making structure, but instead by a very, very small number of people (executives in U.S. movie companies and electronics distributing companies, perhaps, and maybe public officials).  I don’t know the actual history even to the extent that I imagine here, let alone if there was corruption (which is likely when decisions affecting many people are made by a few people).  I do feel certain that, in the field of video players, if a free and competitive market existed in all the relevant areas, competition could easily have gone on long enough for the superior format to win.  Movie companies facing competition, that is with a different intellectual proprety rights regime, might have to release movies in both formats because someone else will be (releasing the same movies to the audience they’re ignoring).

The extent to which Ormerod dwells on the worst beating out the best (VHS, QWERTY keyboard layout, etc.) surprises me— especially the way he acts as if existing economics has no explanation.  I learned about this phenomenon from Barry Field under the heading of network effects: goods or services that become more valuable to the user the more users there are.  I don't quite get the point of his explanation (page 23), which does not mention network effects or DSIR, demand-side increasing returns.  When network effects are very strong, they imply a 'natural' monopoly.  Network effects are a good thing (see the Internet: the more people on it, the more value it has); the problem is when an entity monopolizes the benefits (Microsoft) or when a bad standard gets locked in (VHS, QWERTY).  To me they show the importance of standards more than anything else, and the necessity for a flexible standard-making system ultimately overseen by the public (but in which no-one, including a democratic institution or a private firm, can be heavy-handed in preventing experimentation.  Remember that Microsoft’s standards are no less standards just because they can change them – and impose them on others – on its own).

Oh yes!  Ormerod’s third example is the continued dominance of chemical pesticides as the dominant method of pest-control strategies, “even though scientists have warned of its dangers for over thirty years.  The alternative which has been developed is integrated pest management, or IPM, which relies upon understanding and enhancing natural controls on species which cause damage.  There is a high fixed cost in setting this technology up, but thereafter it is extremely cheap, for it relies on knowledge.  And once produced, knowledge is easy to replicate.  There have been many studies of the economics of IPM and, at worst, farmers using this strategy rather than chemical control obtain similar profits and rates of return on their investment.”  (Page 20.) You know I’m going to jump all over this one.  He lumps this together with VHS and QWERTY, but unlike those it has little if anything to do with network effects (except in the dissemination of knowledge).

The first point is that the people who decide to use chemical pesticides, the farm owners, do not bear the most significant costs.  The high cancer rate is born by society as a whole.  The people who can be made sick directly by application of pesticides are migrant farm-workers: mostly very poor immigrants, often here illegally or on temporary work permits, who possibly have less influence on anything in this country than anybody.  Possibly, one should look into this, agri-business and the chemical companies may be intertwined in lobbying or ownership or something.  I don’t know the history, but I’m sure chemical pesticides received a significant push by the government after World War II; IPM would need a similar push to be used by many farms.  Again I don’t know the situation, but given the influence Du Pont and Monsanto and others have on the federal government, I’d hardly be surprised if there were some active government barriers— if only in not talking about or talking down IPM.

Just look at the forces pushing genetic engineering to “solve” many of the same problems, such as overuse of pesticides, that Integrated Pest Management already really does solve.  All in all I bet the dominance of chemical pesticides are an excellent example not of the market choosing the worst technology but of corruption and oligopoly and collusion preventing market forces from operating.  (Plus, of course, the problem of costs – cancer, poisoning the environment – not being included in the market price.)

I’ve been a tad too dismissive of Ormerod’s points, largely because I think he picked bad examples.  The argument I think he tries to make is that people do not make their decisions based on maximizing their utility (reflecting what is best for them) but are heavily influenced by what other people do.  (The neoclassical economists would undoubtedly answer that conformity or something is part of what makes up utility— which is why neoclassical economics can be so useless, everything can be incorporated into “utility.”  I respect the ants model and its variations, but I don’t yet see that it applies in these examples, nor how it would helpfully inform policy-making (pretending policy were made with regard to facts) or even prediction in cases where it does apply.

The United States developed with the opposite of free trade

Guidelines for intervention

• “support for companies to get new technologies or new products to market seems to be a sensible strategy in those industries where early success might secure a decisive advatage.”
• encourage strategic alliances “to increase the effective degree of monopoly to allow the profit share in national income to rise.  The current obsession with consumerism – and preventing companies making high profits through such behaviour – is highly damaging to European industries.  Of course, such alliances do need monitoring to make sure that the profits are being used positively and not simply handed out in dividents.  But they are essential in a number of industries to gain the full advantages of increasing returns to scale.”
• “it is not sensible to support industries where other countries already have a lead.  They will already be benefiting from increasing returns in such industries.”

(Page 26.)

That second point is a bit scary, even without the kinship of the business-government alliance to the technical definition of fascism.  Rather than pumping up profit rates and “monitoring” to see they are used to expand the industry, it would seem to make more sense for governments to raise the money necessary and invest directly in the industry, if expansion is that important to competing.

I think he suffers from a worship of big-ness.  I suspect that in many cases it is not a true increased return to scale, but an advantage in bargaining power or influence over government and many other things that I wouldn’t classify as “efficiency.”  This is a tad hard to differentiate, as the ability to swing huge deals that Airbus is now beating Boeing at may have little to do with product quality, but a smaller company doesn’t stand a chance.

Implications of the ants model

The ability to develop models in which the behaviour of indivudals is directly affected by the behaviour of others, exemplified by our ants, represents a very important intellectual advance for the social sciences, and offers a more powerful explanation of a wide range of phenomena than does conventional thingking.  The message of this aproach is by no means easy for policy-makers to digest.  Short-term prediction and control, on which so much public policy is based, is inherently extremely difficult and sometimes literally impossible.  A more studied, less frenzied approach is needed, looking at the longer-term properties of the system as a whole, if any real and consistent success is to be achieved.

Page 27.

Analytically, once the principle that the behaviour of individuals can be affected directly by the behaviour of others is accepted, the properties and features of the collective whole can no longer be deduced simply from the conduct of a typical individual.

Page 28.

Application to Crime Rates

This is quite interesting.  The assumption is that in addition to general economic conditions and the proportion and harshness of after-the-fact punishment, the influence of others also matters in deciding to become a criminal.  The critical factor is the proportion of people the potential criminal comes in contact with who are crime-committing or law-abiding.  Those first two conditions, for instance worsening economic conditions, can tip the balance toward a plurality of criminals, so to speak.  There are thus two paths that can be followed within the same economic and law-enforcement conditions: one where there is a flywheel (my term) of non-criminal citizens and one where there is a flywheel of criminals.

Chapter Three, “To Catch a Thief,” especially pages 44 to 46.

(I’m not sure how much this model has to contribute to understanding crime, but it sure doesn’t speak well of prisons now, does it?)

“Family Values”

Ormerand describes the standard economic theory for the existence of families, which assumes that substantial efficiency results from one person specializing in raising children and the other in bringing in money.  (See Nancy Folbre, The Invisible Heart, for a thorough critique of this efficiency explanation of division of labor between men and women.  My interest is in spinning off the addendums.)  To explain why this ends up as families rather than external firms providing the household work, the standard theory suggests that there are economies of scale in such things as preparing meals.  Another rationalization is that “services produced within the family unit can be more readily tailored to the specific tastes of the consumers – members of the family – than can services bought outside” {Page 49}.

This suggests the importance of the system, or institutions, or whatever you want to call it, in determining whether we have co-operatives and collectives and communities or just firms and families.  For there is no reason for this explanation of non-market organization to stop at the family.  All the same arguments – specialization, customization, mass efficiencies – can be made with equal or greater strength for various communal relationships in addition to (immediate) family units.  (The reason that it might be efficient for one person to concentrate on earning money also depends on the way institutions, corporations, are set up.)

(The whole discussion, of course, also shows why purely economic theories cannot explain everything and shouldn't, perhaps, always try.)

Where Ormerand goes with this is reasonable enough, on the narrow premises of economic theory: increases in womens’ incomes results in decreases in marriage because independent living becomes more feasible and the cost of having children in lost income becomes greater.  {Page 49.}

In different countries and over time (long-term, anyway) marriage and birth rates do not follow the path economic theory predicts.  {Pages 50-51.}  I must point out that perhaps the problem here is taking per-capita GDP to be the figure that matters; there are many other equally economic indicators that might track marriage and birth rates much better.