Evolutionary economics

By viewing economics as a cousin of biology, it is easier to see how small causes can have big effects and to grasp the limits of human knowledge
June 18, 2005
Why Most Things Fail by Paul Ormerod
(Faber, £12.99)

In a series of books written over the past decade, Paul Ormerod has criticised orthodox economics for being too mechanistic and divorced from reality and has argued the case for a new approach. As one might expect from a successful business consultant, these books have eye-catching titles—The Death of Economics, Butterfly Economics and now Why Most Things Fail. Their unifying theme is that economics is in thrall to a 19th-century model based on physics when it should really be looking to biology for inspiration. This would make economics focus on what really matters: complexity, ignorance, rivalry, failure and evolution.

In Why Most Things Fail, Ormerod concentrates on failure and extinction in biology and economics. In the biological sphere, mutations lead to species that out-compete other species which eventually become extinct or retreat to some marginal niche. Extraneous events such as climate change may lead to the same outcome because some species are better able to survive in the new environment than others. Ormerod argues that failure and extinction are also pervasive in the economic sphere. Mutations and external events play a role in business life just as they do in biology. The counterparts to biological mutations are new technologies, new forms of organisation and new types of product.

Drawing on the work of Leslie Hannah, Ormerod shows how extinctions are far more common in the economic sphere than is generally realised. Even giant "blue chip" companies are not secure. Of the top 100 industrial companies in the world in 1912, only 19 were still in the top 100 by 1995. Some 29 had gone bankrupt and another 19 had been taken over. Ormerod also reveals some interesting parallels between the extinction of species and of firms. Both follow a "power law" which gives rise to a particular historical pattern of extinction rates. Long periods of comparative calm, during which few extinctions occur, are interspersed with short bursts of frenetic activity in which vast numbers of species or firms are wiped out.

In the biological sphere, most mutations are harmful and reduce the ability of the organism to survive. Some mutations linger on, but many are driven out entirely by competition from incumbent genes. Many potentially beneficial mutations are also driven out by competition before they have time to establish themselves, or before a complementary mutation occurs which allows them to achieve their full potential. The same is true in the economic sphere. Many innovations are harmful, and even potentially beneficial innovations may fail to establish themselves in the face of competition from powerful incumbents. This used to be accepted in orthodox development economics as a justification for protecting "infant industries" in developing countries, so as to give them a breathing space until they were strong enough to compete with foreign producers. The biological counterpart to infant industry protection is allopatric speciation, whereby geographical separation allows a new species to develop by insulating it from competition from an initially superior species.

Ormerod's alternative vision of economics lays great stress on social interaction. Individuals and organisations have only a very limited ability to obtain and process information, so they rely heavily on the information they acquire from encounters with those around them. This leads to herd behaviour and unpredictable consequences. It can make it very difficult for firms to plan their investment and innovation strategies because success or failure may depend on huge swings of fortune that are impossible to anticipate. Ormerod illustrates this with the example of Microsoft Windows. In the late 1980s, Windows was regarded by experts as a failure and Microsoft was on the verge of abandoning it. It was widely expected that development work on it would cease once the new version, Windows 3.0, had been released. On 22nd May 1990, Windows 3.0 was released and more than 2m copies were sold in the first six months.

Ormerod gives many examples of social interaction leading to outcomes which are impossible to predict. The most striking example is Schelling's model of residential segregation. In the US, there are few racially mixed communities and most blacks and whites live in neighbourhoods which are populated almost entirely by their own kind. This might suggest that there is a strong antipathy between the two groups. Yet a large amount of evidence suggests that this is not the case. Most blacks and whites would like to live in neighbourhoods where their racial group is in a majority, but they are perfectly happy to have a large minority of people from the other group as neighbours.

To explore the implication of such preferences, Schelling ran a number of simulations in which individuals were allowed to move house if they found themselves surrounded by too many of the other racial group. These simulations demonstrated two things. In the course of time, the typical result was that blacks and whites spontaneously relocated themselves into highly segregated neighbourhoods. It was impossible to predict where the boundaries of these neighbourhoods would lie or where any particular individual would end up. But it was a safe bet that the bulk of people would end up surrounded largely by people of their own race. This outcome showed clearly that social interaction may magnify small variations into very large differences. It also showed the limitations of the conventional approach to social phenomena, which assumes that large differences must have large causes.

Ormerod made a similar point in his earlier book Butterfly Economics, where he explored the implications of "non-linearity." Most empirical work in the social sciences is based on the assumption that relationships are linear, so that small changes produce small effects and large changes large effects. However, if relationships are non-linear, the link between cause and effect is more complex. Over a certain range small changes may produce small effects, but at a "tipping point" a small change may produce a very large effect. Moreover, this very large effect may be extremely hard to reverse.

This is the vision that underlies the conservative argument on crime. The extent of criminality in a society, it is argued, is partly a matter of material incentives in the form of rewards and punishments, and partly a matter of socialisation. Consider a society in which the crime rate is initially very low and young people rarely meet criminals who lead them into crime. Suppose that punishments are gradually reduced with the result that crime slowly increases. In itself, this may not be a serious problem. However, at a certain point the crime rate may suddenly shoot upwards, perhaps stabilising at a new and very high plateau. Policymakers are likely to respond to this development by reverting to the harsher penalties which they had previously abandoned. Unfortunately, such penalties may have only a limited impact on the crime rate because decades of liberal policy have given rise to a criminal underclass which reproduces itself by transmitting its values to young people.

Conservatives make similar arguments in many other areas, such as divorce law and welfare for lone parents. In each case, they believe that liberal policies set in train social processes that eventually end in disaster and create situations that are very hard to reverse. The liberal response is to dismiss such fears as paranoid and unsupported by the evidence. This is not the place to adjudicate on the issue. The point is that liberals have a rather linear view of social policy in which small changes normally produce small and reversible effects, whereas conservatives have a non-linear view, believing that small changes often give rise to large, unpredictable and irreversible effects. On environmental issues, such as global warming and biodiversity, the positions of these two groups are reversed. Liberals tend to believe that the world is on the brink of disaster and if we do not mend our ways there will be huge and irreversible changes, whereas conservatives take a more relaxed view.

Like Ormerod's previous books, Why Most Things Fail is both intellectually stimulating and entertaining. And despite his initial dismissal of orthodox economics, Ormerod's attitude is actually rather ambivalent. For example, he praises economics for its emphasis on the role of incentives in social life, and he points out that the new economic ideas which inspire his work were all developed by leading and highly regarded members of the economics profession.

Ormerod, despite being a man of the left, is sceptical of human ability to predict and plan. If this is true, what is the role for government? Should it be merely a nightwatchman, defending the polity against internal and external threats, enforcing property rights and preventing crime, or should its role be much wider? Can the state intervene effectively to achieve aims that commend widespread support? Ormerod does not discuss this issue explicitly, although his stress on failure would suggest that most state intervention is pointless. For example, he argues that government attempts to alter the distribution of income have had little long-term impact. Yet surely this is an exaggeration. Human nature and market forces limit what governments can do about inequality, but that does not mean they are powerless. The existence of a strong welfare state in Sweden, for example, clearly helps explain why it is a far more equal society than Brazil. More generally, there are many examples of grand projects which governments have undertaken with conspicuous success, such as the TGV network in France or the D-day landings in the second world war. Ormerod's book is concerned with failure. But what is surprising is not that governments fail, but how often they succeed. The same goes for large corporations, which also perform amazing feats of planning and co-ordination. Successful planning by large public and private organisations, in tandem with markets, have created an environment that is more stable and predictable for many people in advanced economies than at any time in history. Perhaps a better title for Ormerod's book would have been Why Most Things Don't Fail More Often.