Forecasting


The current issue of The Economist carries an article titled, “Riders on a swarm.”  The article describes the use of swarm intelligence–the collective behavior that results from the individual actions of many simple “agents”–that is inspired by the behavior of insects like ants and bees or flocks of birds.  Although–unlike a column that appeared in a previous issue –“agent-based simulation” is not mentioned by name, these models have all of the relevant attributes of agent-based simulations, and you can find example models of collective insect and flocking bird behavior in agent-based toolkits such as NetLogo

As noted in the article, these models have found some business applications in logistics and problems like traffic control.  Ant-based foraging models, for example, have been applied to solving routing problems for package delivery services.  Route optimization, given a set of delivery locations, is a fixed problem with a large number of potential solutions that probably can be solved analytically (or by simple brute force) with enough computing power.  Swarm models have the advantage that they can arrive at a good and often optimal solution without needed to specify and solve a linear programming problem.  By programming simple individual agents, such as artificial ants, with a simple set of rules for interacting with their environment and a set of goal-directed behaviors, the system can arrive at an optimal solution, even though no individual agent “solves” the problem. 

Something that was new to me in this article is “particle swarm optimization” (PSO) which is inspired by the behavior or flocking birds and swarming bees.  According to the article, PSO was invented in the 1990’s by James Kennedy and Russell Eberhart.   Unlike the logistics problems, there may be no closed form or analytically tractable solution to problems such as finding the optimal shape for an airplane wing.  In that case, a simulation in which thousands of tiny flowing particles follow a few simple movement rules may be just the ticket.

This stuff is fascinating, but it’s not clear that there are many useful applications for this type of modeling in marketing or marketing research, at least as long as the unit of analysis is the intersection of an individual “consumer” and a specific purchase or consumption occasion.   Of course, if imitation and social contagion are at least as important in our purchase decisions as the intrinsic attributes of t products and services (as research by Duncan Watts and his collaborators has shown in the case of popular music), then agent-based simulations may turn out to be one of the best ways to understand and predict consumer behavior.

Copyright 2010 by David G. Bakken.  All rights reserved.

Advertisements

The “economic focus” column in the July 24th-30th (2010) issue of The Economist is titled “Agents of change.”   As many of us have come to believe over the past couple of years, the “dynamic stochastic general equilibrium” (DSGE) economic models used by central banks and other economists more or less fell apart when it came to predicting or anticipating the credit-fueled meltdown that we are just now beginning to recover from.  The Economist reports on a June workshop sponsored by the National Science Founcation and attended by central bank economists (from the Fed and the Bank of England), policy advisors, and computers scientists who convened to explore the potential of agent-based models of the economy.

Agent-based models have emerged from the intersection of computer science and social science and have been applied to population dynamics, epidemiology, species extinction, wealth creation, the formation of communication networks, and a host of other problems not well served by traditional economic models.  In contrast to the DSGE approach, which represents the economy as a series of equations to be solved using highly aggregated data as inputs, agent-based models of economic systems are “bottom up”–they generate complex behavior by creating  populations of autonomous agents, giving them simple behavioral rules, and then simulating (over thousands of iterations in many cases) the interactions of these agents.  Under some starting conditions, an agent-based simulation may produce results similar to a DSGE model.  For example, Joshua Epstein and Robert Axtell (one of the NSF workshop organizers) found that agents operating under rules that permitted bargaining for the exchange of two commodities arrived at prices that fluctuated around a sort-of equilibrium point.  By the way, their book, Growing Artificial Societies:  Social Science from the Bottom Up (1996), is one of the best introductions to agent-based modeling.

In some ways, agent-based models of the ecomony are generating new interdisciplinary thinking.  In an Op-Ed piece in the New York Times in October of 2008, Mark Buchanan (a theoretical physicist)  titled “The Economy Does Not Compute,” we learn about an agent-based model developed by Yale economist John Geanakoplos and two physicists, Doyne Farmer (another of the NSF workshop organizers) and Stephan Turner designed to explore the influence of the level of credit or leverage in a market on the market’s overall stability.   A typical objective for an agent-based model is to develop an understanding of the sensitivity of a complex system to changes in one or more of the model variables, and these researchers found that greater levels of credit leads to greater interdependence among the actors (or agents) and this pushes the market toward instability.  The DSGE models are not very good at capturing this kind of process.  And this model revealed something even more striking, but perhaps not surprising to those who have used agent-based models to capture the non-linear nature of complext adaptive systems.  What Geanakopolos, Farmer and Turner found is that the leverage-induced market instability does not develop gradually but arrives suddenly–with the economy essentially falling off a cliff.

Buchanan goes on to cite two other applications of agent-based modeling.  One involved testing the impact of small transaction taxes in foreign exchange markets, and the second looked at deregulation in a state’s electricity market.  In both cases, the simulations provided insight that challenged the prevailing ideology-based assumptions, and could lead to better policy outcomes.

I’m not sure how readily mainstream economists or central bankers will embrace the agent-based way of thinking, but it’s pretty clear that at a minimum, agent-based approaches to understanding complex systems like the economy can only add to our ability to make better policy decisions.

Copyright 2010 by David G. Bakken.  All rights reserved.

Nassim Nicholas Taleb introduced a new term into the lexicon of business forecasting, the “black swan event.”  The metaphor comes from the apparent fact that, for some reason, black swans should not exist, but they sometimes do.  In THE BLACK SWAN:  The Impact of the Highly Improbably, Taleb expounds for  366 pages on what is, for the most part, a single idea:  the normal (bell-shaped) distribution is pretty much worthless for predicting the likelihood of any random occurrence.  Taleb augments this idea in various, occasionally entertaining ways, acquaints the reader with power law and “fat tail” distributions, and takes excursions through fractal geometry and chaos theory.

Taleb tells us he aspires to erudition, and he introduces the reader to plenty of “great thinkers” that history has failed to credit.  You can come away from this book feeling that it is mostly about showing us how erudite Taleb is.  For me, one of the key shortcomings is Taleb’s tendency, via style, to claim that we should accept his arguments on faith.  There are plenty of concepts, especially involving numbers, that would benefit from concrete examples.  There’s just a little too much “Take my word for it” in his writing.  Still, if you’ve got time to kill, this is not an unrewarding read.

David Orrell tackles the very same subject–our inability to predict the future–in The Future of Everything:  The Science of Prediction (which has a sub-sub title: “From Wealth and Weather to Chaos and Complexity”).  For a mathematician, Orrel has an entertaining style and writes with clarity.  This book is far more focused than THE BLACK SWAN, which is sort of meandering.  The book is divided into three main parts: past, present and future.  The past provides a history of forecasting, beginning with the Greeks and the Oracle at Delphi.  The present considers the challenges of prediction in three key areas: weather, health (via genetics), and finance.  Orrel did his dissertation research on weather forecasting, and after reading this book, I think you’ll agree that it’s a great case study for revealing everything we think we know about the “science of prediction.”

Orrel’s main point is that a key problem in prediction is model error (the basis of his dissertation), which far outweighs the influence of chaos and other random disturbances.  In a nutshell, the complexity of these systems exceeds our ability to specify and parameterize models (models are subject to specification error, parameter error, and stochastic error).  Weather is a great example.  While there are only a few components to the system (temperature, humidity, air pressure, and such), the interactions between these components are almost impossible to predict.  Another problem is the resolution of the model; conditions are extremely local, but it it very difficult to develop a model that resolves to a volume small enough to predict local conditions.

Orrell educates.  The reader comes away with an understanding of the logic and mechanics of forecasting, as well as the seemingly intractable challenges.  Orrell provides clear explanations of many important forecasting concepts and does a good job of making the math accessible to a general reader.  There are a couple of shortcomings.  Orrell gives only passing notice to agent-based simulation and similar computational approaches to complexity.  And, in the third part of the book (the “future”), after spending the preceding two parts on the near futility of prediction (but for different reasons than Taleb), Orrell offers his “best guesses” for the future in areas such as climate change.

While I embrace the basic premises of these books, some new developments are cause for optimism.  Economists using an agent-based model of credit markets were able to simulate the fall off the cliff that we’ve experienced in the real world, as just one example.  While not truly “predictive,” these models can help us understand the conditions that are likely to produce extreme outcomes.

THE BLACK SWAN has its rewards, but The Future of Everything has far more value for the forecasting professional.  As a chaser, you might try Why Most Things Fail:  Evolution, Extinction and Economics by Paul Ormerod.

Copyright 2010 by David G. Bakken.  All rights reservcd.