{original squeezed contributor: davidw}

Key Points

• Failure is all around us - species, companies, brands and government policies fail. To understand success, “we must first understand the pervasive existance of failure”.

• “Individual behavior is not fixed, like a screw or cog in a machine is, but evolves in response to the behavior of others. Control and prediction of the system as a whole is simply not possible.”

• The world is complex, and in continuous evolution. People are, in turn, part of complex and constantly changing networks, and their ability to understand this environment is inherently limited. This is why things fail.

• Decentralized, “market like” systems may not deliver a perfect outcome, but they are the most likely to deliver a satisfactory outcome, compared with a centrally controlled model.

• “Innovation, evolution, and competition” are the hallmarks of a successful system.

Summary

Nearly 100 years ago, the first true multinational corporations appeared in an age of corporate expansion that could be termed the “Edwardian Explosion”. Nearly all of those companies, many of which were just as large as the giants of today, are now gone. They failed despite having had millions or billions of dollars at their disposal, and management willing to innovate and respond to a changing world and markets. Of the 100 largest industrial companies in 1912, by 1995, 29 had gone bankrupt, 48 disappeared (mergers, acquisitions and so on), and 52 survived, but only 19 remained in the top 100. Once you discount the large number of small companies that fail in their first few years, the average lifespan for small companies is similar to that of large firms - and most of them eventually fail.

Most modern economics deals with “how to succeed”, “the correct strategy”, and so on, and looks very little at the persistance of failure, despite its pervasiveness. Economics prefers to deal with an idealized “state of equilibrium”, and to look at periods of change as a temporary pause between returns to the natural order of things. However, change is all around us, and is much more of a constant than any sort of equilibrium. For instance, economics textbooks often quickly cover the price of a product in relationship to the demand curve, as if it were easy for a company to determine the demand for their product at given prices. It isn’t, as the large market research and advertising sectors attest. Were it easier, firms wouldn’t need to spend such large sums of money attempting to determine what will sell and what won’t (and still end up wrong quite often). It isn’t easy for most firms to gauge production prices for complex products, even though the process is under their control, let alone gather accurate information about something so “basic” as the demand curve.

After looking at companies and their failures, and the difficulties of economics in dealing with the most basic aspects of a firm’s functioning, we look at government policies. Governments in nearly all western nations continue to grow. Government spending as a percentage of the economy was lower during the most socialist government Britain has ever had, in the 1950s, than under Mrs Thatcher in the 1980ies. Despite the public sector growing to be over twice as large in the post war years, the unemployment rate - the primary cause of poverty - was virtually unaffected. Higher spending on education as a means to increase social mobility has had similar results. Spending has increased, however, social mobility has not. By looking at the Gini Coefficient (a measure of the equitable distribution of wealth, where a higher number means more inequality, and 0 means complete equality) over time in the US, we see that, during the 20th century, it has changed - both increasing and decreasing, but in a fluid and dynamic way, with irregular fluctuations. When looking at the world as a whole, the Gini coefficient decreased for the first time in the history of capitalism, with the economic success of Asia lifting millions of people out of poverty. Various theories describing inequality have been proposed, from Marx to the mainstream general equilibrium theory, yet none convincingly explains the ebb and flow of inequality over time - either between citizens of one country, or between countries. The evidence points to this inequality moving in irregular, unpredictable ways. One of the main reasons why it is difficult to create a theory that successfully matches the evidence is that inequality in any society is the result of millions of interactions. There are so many interacting variables that predicting the outcome of the system is nearly impossible.

Segregation - be it racial, or along class lines as in Britain - is another area that social reformers have attempted to tackle over the years, but in some ways, with little success. Despite many efforts over the years, racial segregation remains very prominent in the US. Economists such as Gary Becker have examined the issue through the lense of their field, which has as one of its principal tenets the law that agents respond to incentives. Given economics’ view of agents as rational, segregation can continue only if the people involved desire this outcome. However, the evidence suggests that people generally do not have widespread, strong racial preferences. This apparent discrepancy - weak racial preferences at an individual level coupled with relatively stronger effects at a systemwide level - are explained in theoretical work by Thomas Schelling that shows how small, local preferences can cause larger effects on the overall system. Furthermore, the precise outcome of the type of experiment he performed is impossible to predict from the inputs to the system. Uncertainty rules.

The “laws of supply and demand” are taught as basic economics, but things in reality really aren’t that simple. Once again, there are many variables. For instance, a crop might be expensive one year due to a poor harvest, then cheaper than expected the next year, because farmers plant a lot of the crop commanding a high price following the bad year. The third year, farmers may plant less of the crop due to the low prices in the ‘good harvest’ year, sending prices up again as less of the crop is brought to market. In this way, prices might never reach a classic “supply and demand” equilibrium, but move around it. Game theory is a branch of study that proposes solutions to difficult problems inherent in the interaction between people. However, it too fails to be of much use for a large number of practical situations. Game theory techniques require clear and distinct rules, but in real life, these are often lacking. People, either acting alone or in groups act in their own interests, with the purpose of obtaining specific goals, yet it seems that it is “difficult or impossible to predict the consequences of decisions in any meaningful sense”.

When comparing humans and their actions and interactions with evolution in the biological sense, one of the chief differences is that people act consciously in response to their environment, whereas animals clearly cannot guide their own evolution. Even so, people face massive uncertainty, something that standard economics doesn’t take into account, working instead with rational agents with nearly perfect information about decisions they must make. This is fine in certain circumstances in order to have a clearer view of a specific problem. “Bounded rationality”, where agents do not have access to all information increases the relevance of economic theory. In reality, “maximizing” (obtaining the absolute best result) is difficult or impossible compared to “satisficing” (obtaining a good enough result). In games such as chess, where it is impossible even for computers to calculate every move, grandmasters often make the most reasonable move, avoiding obvious traps, rather than trying (unsuccessfully) to determine the best possible move in light of all possible moves that the opponent might make.

If potential chess moves are difficult to predict, where the rules are relatively simple, and known, think of the difficulties involved in determining an accurate view of the present, let alone charting a course for the future, for the head of a company, or government. The Hotelling location model is described as a long, continuous beach (1000 meters, say), with two ice cream vendors who both want to maximize the customers who will go to their respective establishments. Clearly, placing their shops in the middle is the best choice. Go clear to one end, and you risk losing many of your customers to the other vendor who places their shop in the middle. This model can be an effective way of describing markets, or even the political spectrum. Go too far to one extreme, and you will get no votes from the other end, nor the centrists, who will all vote for the more moderate candidate. It’s pretty easy to see this logic with only two ice cream shops (political parties/companies/agents of some kind), but even increasing the number to 5 yields a vastly more complicated problem because the solution depends entirely where your rivals place themselves. And that depends, in turn, on your own placement. And so it goes. Individual agents aren’t totally powerless - by thinking about the problem, and analyzing it, perhaps we can gain some slight advantage.

Evolution is random. Genes do not mutate of their own accord, and species do not choose how to evolve. The fittest individuals and species only evolve over long periods of time thanks to random mutations and natural selection. On the other hand, in the “perfect world” of classic economics, agents know everything and thus are able to choose the absolute best option. This would make comparing the two difficult, but for the sake of argument, we can imagine two players betting on coin tosses. There is no real way to tell whether either of the players is really a better performer, or just lucky. Individuals and firms must lie somewhere on this continuum, between pure randomness and omniscient decision makers. There is strong evidence that the truth is closer to random actors, that, however, do glean some advantages in terms of information from time to time, and thus are more likely to survive as the ‘fittest’ of their generation.

Like economists, biologists often look at the survival of the ‘fittest’, rather than extinction. An interesting finding regarding success is that a particular gene’s likelyhood of “invading” a population (becoming present in all or most individuals over time) is very low. Extinctions, like we saw earlier with company failures, happen at irregular intervals - many at once, then fewer, then many again, but without an entirely predictable pattern. The frequency of species extinction, does, however, fit a power law curve, meaning that the frequency with which we see a given number of extinctions decreases according to the square of the size. Interestingly, the failure rates of firms fit the same pattern - almost as if they were just as random as extinction amongst species that are, unlike firms, completely unable to plan their own futures.

Conventional economic theory points to external shocks as the ultimate cause for business cycles, because, as previously discussed, the economy “ought to” tend towards an equilibrium. When modeled with simple rules, however, it is possible to demonstrate complex emergent behavior with regards to extinction events, with endogenous causes internal to the system, rather than external causes (asteroids, floods, climate change).

Things fail because of the inherent incertainties involved in any complex system. Despite our best intentions, outcomes often do not match desired effects. It is impossible to get around this simple fact, and no amount of intelligent analysis will change the situation.

We are not completely helpless in the face of the Iron Law of Failure, however. Sometimes failure is even beneficial, for instance when an old, fossilized firm fails and a newer, more dynamic one takes its place, or replaces its entire industry. Indeed, it is the dynamic of competition, innovation and experimentation that must be promoted.

Links

Paul Ormerod’s page: http://www.paulormerod.com/