Ormerod’s Why Most Things Fail


Jason Collins


July 28, 2012

After sitting on my reading list for a few years, I have finally read Paul Ormerod’s Why Most Things Fail. Ormerod’s basic argument is that failure is all around us and given the complexity of the world, there are limits to how much corporations can control their fate or governments can control the success of their policies. Governments, firms and households lack complete information. They do not have the cognitive power to process the available information to determine the optimal choice. As a result, when you look at their success, the outcomes look more like the result of chance than of rational strategic decisions.

Ormerod’s argument is built upon some interesting work done by himself and others in which he examined the extinction rate of United States firms (and ultimately a wider suite of global firms). Firm death tends to follow a power law distribution, and when mapped against the historical extinction of species, which we know is built upon chance events, the pattern looks similar. In models of firm extinction involving networks of interconnected firms, if firms are given much more than 10 per cent of the available information about their relationships with other firms and are able to affect those relationships, the patterns of firm death cease to mirror those which we see. This suggests that firms act with little control over their success or failure.

While this is an interesting and important observation, is the mapping of firms to species the right mapping for the analogy? For example, a species is defined (roughly) as a group of organisms that are capable of interbreeding and producing viable offspring. Thus, the units of selection, the genes, are limited to within that species. In the case of a firm, if we consider the unit of selection to be a strategy, these are able to spread to any firm. All firms are capable of interbreeding and producing fertile offspring. So are firms more akin to members of a species than to each being a species on their own? And if so, what implications does this have for the model? If individual organisms within species have similar patterns of death without reproduction to that observed for the extinction of species (which I expect is roughly the case), then the implications may be small. However, without exploring these types of questions, Ormerod has not convinced that his comparison is the right one.

Ormerod takes some time to build to his exploration of firm extinction and some detours of varying interest along the way. One of his building blocks is an exploration of Schelling’s models of segregation, which Ormerod uses to show that simple rules can result in surprising and complex phenomena. This example forms ones of the pillars of Ormerod’s case about the complexity of the world, but I wondered at times if this was the most convincing example available. Despite the complex behaviour in Schelling’s models and the difficulty of predicting which person will end up living where, the model does allow some prediction at the macro level. It is also the case for other models explored in this area, whether that being the first-order difference equations investigated by Robert May or Brian Arthur’s El Farol bar. Predicting specific results is near impossible, but picking the pattern and the effect of parameter changes on that pattern is possible.

The detours also includes some bashing of the neoclassical economics straw man. Ormerod’s choice of supporting evidence is interesting, but the omissions are often obvious. Take his quoting of Vernon Smith on the flaws with existing models of the operation of markets, but no mention of Smith’s experimental work which suggests how well markets seem to find an equilibrium despite the knowledge shortfalls and bounded rationality. Similarly, when discussing bounded rationality, Ormerod does not explore the success or failure of heuristics (Also strange was the crediting of bounded rationality to Akerlof and Stiglitz with no mention of Herbert Simon). Ormerod could still have made his case with a more in-depth discussion, and then it might have felt more convincing.

Once Ormerod has established that companies have little control over their fate, and that the world is too complicated for governments to make decisions (both arguments I am sympathetic to), he dedicates little space to ask what this means. In the company case, it comes with a call to innovation and flexibility. But given that strategic choice has little to no effect on the probability of firm survival, why will that particular approach work?

When it comes to government, again the questions left unanswered are more interesting than those addressed. If governments are likely to achieve success only by chance and cannot possibly achieve success through detailed planning, what should they do? We have a host of government interventions ranging from legislation to enable joint stock companies to protection of property rights, each arguably important for our wellbeing. How would these be facilitated in a world where we otherwise throw up our hands in despair? Ormerod’s hints at some ideas but instead of exploring them, he sticks to denouncements of governments acting as though Soviet Russia was a success. Fair enough, but I sense the book sells Ormerod’s thoughts on this question short.