As we prepare to celebrate the 200th anniversary of Charles Darwin's birth, it is important to reflect on the impact of evolutionary principles on economics, and in particular, finance. The basic ideas of natural selection have played an important role in economics, and their origins go back to Adam Smith. It is also interesting, that both in biology and in economics the basic principles of evolution have often been over simplified and misunderstood. In no area is this more apparent than in finance where many of the basic theoretical foundations are built from evolutionary principles. Currently, as we try to better understand the functioning of financial markets it is important to reassess these basic evolutionary principles.
Proponents of unfettered competition and industry self-regulation have often used arguments based on survival of the fittest in defense of limited regulatory environments. In much of economic organization this principle appears to serve us relatively well. The unregulated world self-orgainzes into surprisingly robust institutions which do everything from getting bread to our table to producing iphones. Emergent economic systems are often able to handle small shocks and disturbances applied to them. As in biological evolution, some mistakes are made, but generally we seem to find amazing ways to ``get the job done''.
The question is how well financial markets can self-organize into efficient allocation mechanisms absent any outside control or direction. I think there are several key differences in financial markets which make them distinct from markets for bread or gasoline.
First, evolutionary dynamics are not as simple as most people interpret them. In most situations, evolution is thought of as a game played against a stable, unchanging environment. However, in many scenarios evolution is about adapting to the behavior of others. This continual pattern of behavior adapting to other behavior is known as coevolution. In finance it can be as simple as trading strategies played against each other, and as complicated as the evolution of new institutions and securities. Few general principles can be stated for coevolutionary dynamics. What little we still know from computer simulations suggests very rich and interesting patterns can be generated. However, little is known about convergence to any sort of efficient outcome. On the contrary, computational models point to instability and crises as the normal state for financial markets, and policies that would eliminate these might be difficult to implement.
The second special characteristic of financial data is noise. Evolution requires fitness levels to be accurately measured for the process of natural selection to operate. The process cannot be asked to do anything optimal if it is difficult to figure out in which directions improvements can be found. Financial data is notoriously noisy. The question of luck versus skill for many professional investors is an old debate which is often a greatly ignored problem in finance. In the many years I have analyzed strategies on both real and simulated returns series, it has been clear that making decisions about optimal investment strategies is very difficult, if not impossible, given sample lengths we have. Modern markets with complex derivative instruments can further complicate this problem. Consider a scenario where an investor has access to a fair coin strategy losing $1 and gaining a $1 with probability one half on each event. Fairly simple derivatives can be used to convert this to losing $100 with probability (1/100), and gaining $1/99 with probability (99/100). The expected gain of this strategy is again zero. However, 99 times out of a 100 this will look like a risk free payoff. In a world with noise, this transforming investor could be seen as a financial genius and highly compensated, while a complaining risk manager is quietly shown the exit. Charles Darwin would definitely not be happy with this situation.
The final special character of financial markets is that evolution selects for survival which is not necessarily aligned with utility maximization. This distinction is complicated, but may be extremely important. In theoretical evolutionary races over agent wealth shares the agents left standing in the end could actually look irrational from the standpoint of having incorrect beliefs. Only in special cases is it likely that fitness based selection will align well with "optimal" behavior from the perspective of utility maximization.
In summary, the message from the young world of agent-based research on financial markets would be to treat these markets with great care from a policy perspective. Their ability to evolve nearly optimal, robust, solutions to problems of allocation and security design should never have been relied on. Unfortunately, no agent-based model is detailed enough to predict the existence of incorrectly rated securitized debt instruments, or many other curious features of the current crises, but they do give us basic insights into the potential instabilities that are generic to all financial systems.
Many writers have been encouraging all of us to put Keynes back on our reading lists. The message from the complexity/agent-based world says this is fine, but you should also add Simon and Schelling to your reading program as well if you want to really understand the dynamics of modern financial markets.