COORDINATION FAILURES

by

Peter Howitt

Brown University

Revised December 4, 2001

Draft article for An Encyclopaedia of Macroeconomics, edited by Howard Vane and Brian

Snowdon, to be published by Edward Elgar.

During a depression economic activities are badly coordinated. Firms allow plant and equipment

to fall idle despite increasing numbers of able-bodied people willing to operate it in exchange for

less than the value of their marginal product. Savers continue as before to make provision for

extra future consumption while production of the new capital needed to produce more consumer

goods is reduced. Stocks of consumer goods pile up unsold even though the desire to consume

them is, if anything, intensified by rising poverty. Farmers are forced off their land while others

go hungry.

Many economists therefore think of depression as being a state of coordination failure; a

state in which market forces have failed to coordinate the millions of transactors that interact

daily through a web of interconnected markets. What Smith called the ‘invisible hand,’ or

Mummery and Hobson (disparagingly) called the ‘automatic machinery of commerce,’ has not

guided them to a state in which markets clear. Instead, people are somehow led to act at cross

purposes, failing collectively to take full advantage of potential gains from trade. As Keynes put

it, the system is not ‘self-adjusting.’

The first step in understanding how a mechanism can fail is to understand how it works.

Although contemporary economic theory is rather vague on how market forces work, the

beginning student is left in little doubt that they operate mainly through the adjustment of prices.

According to all undergraduate textbooks, a free market will quickly reach a coordinated

(market-clearing) state, because prices rise when there is an excess demand and fall when there is

an excess supply. Analytical accounts of coordination failure therefore focus on why something

might go wrong with the process of price adjustment.

1
 The classical tradition from Thornton through Marshall was to blame prolonged

unemployment on impediments to price-adjustment, particularly impediments to adjusting the

price of labor. This was also the approach of mainstream Keynesianism, which from Modigliani

through Fischer was based on the assumption of sticky wages. But Keynes himself believed that

coordination failure had a deeper reason, namely that wage and price adjustment, which classical

theory pictured as corrective forces, are actually destabilizing. If given full rein they would lead

an economy even further into depression, because a general decrease in wages and prices would

produce ‘debt deflation’ (to use Fisher’s term, which Keynes did not), destabilizing expectations

of further price decreases, and adverse distributional effects.

Patinkin (1948) elaborated on Keynes’s account of coordination failure by portraying the

process of wage and price adjustment as a dynamical system that fails to converge to its (full-

employment) equilibrium. Clower (1965) pointed out another possible reason for non-

convergence, namely that transactors will respond not just to the price-signals of classical theory

but also to quantity signals they receive when their attempts to trade are frustrated by existing

imbalances between supply and demand. Thus excess supply in one market can lead frustrated

sellers to curtail their demands in other markets, causing the excess supply to spread. As

Leijonhufvud (1968) later elaborated, the cumulative decline in effective demand resulting from

this process will tend to amplify deviations from full employment equilibrium rather than

dampening them.

The approach taken by these writers, of analyzing coordination failure in terms of

disequilibrium price adjustment, gained support from the demonstration by Scarf (1960) that

price-adjustment in a Walrasian general-equilibrium setting does not always converge to a

general equilibrium; in effect, adjustments in one market may be continually thwarted by

2
independent adjustments in other related markets. However, work on disequilibrium dynamics

fell out of fashion in the 1970s, largely because its proponents offered no conceptually coherent

account of the many logistical problems that arise when expectations are mutually inconsistent

and markets are not clearing, or of the institutions (firms, shops, money, markets, etc.) that deal

with these logistical problems in real life. The final blow was dealt by Lucas (1972) who showed

that one can provide a conceptually coherent account of at least transitory coordination problems

within a framework of rational expectations with clearly specified informational imperfections, a

framework in which none the awkward problems of disequilibrium theory are visible.

After a decade of relative neglect, the theory of coordination failure re-emerged in the

1980s, when various authors found a way to model it using the rational-expectations-equilibrium

approach which by that time had become de rigueur in macroeconomic theory. Since then, the

term “coordination failure” has taken on a different meaning, with no reference to disequilibrium

dynamics. Specifically, as elaborated by Cooper and John (1988) and later by Cooper (1999), it

now means the existence of multiple equilibria, often Pareto-ranked, of the kind that exist in

games with strategic complementarity.

Suppose for example that there is a strategic complementarity that works through “thin-

market externalities” in the process of search and matching. (See Diamond, 1982 and Howitt,

1985.) That is, when people on one side of a market put more effort into the matching process,

this makes it more worthwhile for those on the other side to do the same thing, because it makes

transacting less costly for them. Then the general expectation on the part of firms that it will be

difficult to find customers can be self-fulfilling. It leads firms to cut back their hiring effort,

which leads to a fall in job vacancies, which makes it harder for unemployed workers to find

jobs. As a result unemployment rises, and the consequent fall in incomes makes people generally

3
less willing to buy goods. This completes the vicious circle by confirming the original

expectation that it will be harder for firms to find customers.

On the other hand, the same chain of reasoning can often be applied to show that the

expectation that customers will be easy to find would also be self-fulfilling. Thus there are

multiple equilibria, some with optimistic expectations, high income and low unemployment, and

others with pessimistic expectations, low income and high unemployment. The latter might be

interpreted as depressions. They persist because they are non-Walrasian equilibria in which

people are interacting not just through prices but also through such non-price variables as the

difficulty of finding customers, or the difficulty of finding a vacancy in the labor market. In a

low-level equilibrium it would be pointless for firms to try lowering their prices since their

problem is not that they have overpriced their goods but that the cost of marketing products is

too high; similarly it would be pointless for workers to offer to work for lower wages since their

problem is not that they are asking too much but that they can’t find a potential employer with an

opening.

Such low-level equilibria imply a coordination failure, in the sense that if only everyone

would get together and raise their expectations in coordinated fashion, they could potentially

reach a high-level equilibrium where everyone is better off. They remain in a depression

because no mechanism exists for bringing about such a coordinated change in beliefs. Thus,

according to this approach, the process of price adjustment fails to coordinate activities because

it fails to deal with the root problem, namely that of pessimistic expectations with respect to non-

price variables.

The contemporary notion of coordination failure as multiple non-Walrasian equilibrium

thus shows the need to go beyond wage and price adjustment if we are to achieve a deeper

4
understanding of depressions. There are however two important, related problems with this

notion. The first is that although, as explained above, multiple-equilibrium models do illustrate a

kind of coordination failure they evade the task of analyzing the coordination process. A

rational-expectations equilibrium is by definition a highly coordinated state of affairs, in which

each transactor has managed somehow to anticipate, as fully as possible given informational

constraints, the actions of others. By focusing exclusively on such equilibria, the modern

coordination-failure literature thus presumes that coordination is managed costlessly by some

unspecified mechanism. This begs the question of how people can achieve such precise

coordination and yet fail in the seemingly simpler task of agreeing that the equilibrium they

coordinate on should be a good one.

The second problem is that any model of multiple equilibrium without some mechanism

for describing which if any equilibrium the economy will be led to lacks empirical content.

Indeed the problem is greater than it might seem at first glance, because the model will have not

just a high-level equilibrium and a low-level equilibrium but also a large number of other

equilibria, in which people randomize between the high-level and low-level equilibrium in

correlated fashion, according to the realization of some extraneous random variable. Because of

this second problem, standard comparative-statics analysis applied to the model cannot predict,

even qualitatively, how the economy will respond to variations in exogenous variables or policy

instruments that impinges on the economy, because the system might respond by changing from

one equilibrium to another.

In short, the rational-expectations-equilibrium theory of coordination failures is

incomplete without an account of the disequilibrium dynamics that the older literature sought to

provide. For it is only by studying what happens out of equilibrium that one can understand

5
which equilibrium will be arrived at, if any, by what route, and with what time delays. Howitt

and McAfee (1992) show how one might add such an account in a highly stylized example, in

which Bayesian learning can lead people to oscillate between a high-level and low-level

equilibrium on the basis of an extraneous random variable that they interpret as “animal spirits.”

Finally, none of the above-mentioned contributions attempts to identify and analyze the

agents that perform the role of coordinating markets in actual economies. A tradition going back

at least to J.B. Say identifies them as commercial enterprises – retailers, wholesalers, brokers,

jobbers, etc. These “shops” are the visible counterparts of Smith’s invisible hand. Howitt and

Clower (2000) show how a coherent network of shops can emerge from competitive evolution.

In their analysis no one has any understanding of the whole economy, yet the adaptive

adjustments made by shops seeking to profit by serving their individual markets often combine

to guide the whole system to a fully coordinated state. Ongoing theoretical research into the

dynamics of such a self-organizing network may provide further clues as to how coordination

normally works in a decentralized free-market economy, and why it occasionally fails.

Peter Howitt

Brown University

Providence, Rhode Island

6
 BIBLIOGRAPHY

Clower, R. (1965), ‘The Keynesian Counter-revolution: A Theoretical Appraisal’ in F. Hahn and

F. Brechling (eds.) The Theory of Interest Rates, London: Macmillan.

Cooper, R. (1999), Coordination Games: Complementarities and Macroeconomics, New York:

Cambridge University Press.

Cooper, R., and A. John (1988), ‘Coordinating Coordination Failures in Keynesian Models’,

Quarterly Journal of Economics, 103, August, pp. 441-63.

Diamond, P. (1982), ‘Aggregate Demand Management in Search Equilibrium’, Journal of Political

Economy, 90, October, pp. 881-94.

Howitt, P. (1985), ‘Transaction Costs in the Theory of Unemployment’, American Economic

Review, 75, March, pp. 88-100.

Howitt, P. and P. McAfee (1992), ‘Animal Spirits’, American Economic Review, 82, June, pp.

493-507.

Howitt, P. and R. Clower (2000), ‘The Emergence of Economic Organization’, Journal of

Economic Behavior & Organization, 41, January, pp. 55-84.

Leijonhufvud, A. (1968), On Keynesian Economics and the Economics of Keynes: A Study in

Monetary Theory, New York: Oxford University Press.

Lucas, R. (1972), ‘Expectations and the Neutrality of Money’, Journal of Economic Theory, 4,

April, pp. 103-24.

Patinkin, D. (1948), ‘Price Flexibility and Full Employment’, American Economic Review, 38,

September, pp. 543-64.

Scarf, H. (1960), ‘Some Examples of Global Instability of the Competitive Equilibrium’,

International Economic Review, 1, September, pp. 157-72.