December 15, 2021 — As I have mentioned before, Capstone is in the prediction business. We predict what the future policy world will look like and then create a strategy for our clients that matches that future. Understanding how the future may unfold is key to that process.
Prediction, whether in the world of science, politics, or finance, has historically been dominated by a reductionist philosophy. This philosophy tells us if we can predict the smallest element of a system, then that will tell us all we will need to know about the system as a whole. Historically, scientists have studied cells to tell us about organisms, studied atoms to tell us about the nature of matter, and so on. This approach has been fundamental to much of the scientific progress of the past 100 years. Of course, in example after example, the limitations of this approach are also made clear. Studying the internal structure of the neuron has told us little about how consciousness works. Recently, political polls and focus groups failed to predict the rise of Trumpism. And for decades economists have failed to predict asset bubbles by relying on the study of “homo economicus” a fictitious perfectly rational actor.
In 1984 a group of scientists from many disciplines came together in Santa Fe, New Mexico to establish a different approach to explaining our world and making predictions. They accepted that science had made huge leaps from studying phenomena at the smallest scale, however they believed that this approach missed something fundamental about how phenomena behaved at larger, more complex scales. They believed that something new “emerged” when you observe the world as a system. Emergent properties became a phrase used by these scientists to describe how these larger systems behaved in new and surprising ways that could not be predicted by studying their component parts. The small group that met in 1984 became the Santa Fe Institute (SFI), a multi-disciplinary research organization dedicated to studying complex systems science.
Returning to the topic of bubbles, this has been one of the most confounding subjects in modern finance. Economists have struggled for decades to define, identify, or predict asset bubbles. A St. Louis Federal Reserve Bank paper (here) describes the difficulty. However, a close reading of the problem reveals that the difficulty stems from looking at the problem in an overly reductionist way, focusing on the component parts and ignoring how the larger system (e.g., the market) is exhibiting new emergent behaviors. Classical economics has proven ill-equipped to study asset bubbles simply because of its rigid and reductionist approach. As an example, one of the tenets of classical economics is the theory of Rational Expectations which says we all behave in the same hyper-efficient and rational way. Anyone involved in financial markets (or anyone in a relationship) can tell you that is rubbish. However, classical economists “rely on many unrealistic assumptions, such as market clearing, market convergence to equilibrium prices, perfect information, and rationality.”
Complexity science, however, has a solution (here). A group of economists (L. Wang, K. Ahn, C. Kim, and C. Ha) from Korean, American, and Chinese institutions provide a summary of many studies which employed sophisticated computer programs to build “agent-based models” of financial markets and assets bubbles. This technique is a common way to study complex adaptive systems and their emergent properties across many scientific disciplines. The paper is a fascinating survey of the progress complexity science has made in understanding asset bubbles. Although the jargon is technical, the phenomena that are being uncovered are likely more recognizable to a veteran investor than the unrealistic assumptions of classical economics. The paper discusses how sentiment and herding behavior impacts financial markets—intuitive concepts well known to traders.
Predictions of politics and policy have also been a focus of complexity science. A recent paper (here) reveals that zooming out can enhance our ability to predict political outcomes. The paper shows significant increases in predictive power when opinion polls ask a person to estimate how a related group will vote rather than themselves. “We realized that if we ask a national sample of people about who their friends are going to vote for, we get more accurate predictions than if we ask them who they’re going to vote for,” says Mirta Galesic, who is the corresponding author. “We found that people are actually pretty good at estimating the beliefs of people around them.” Researchers discovered that we process data about our social groups very efficiently, and multiple biases prevent us from discussing our own political views with precision. Another example of complexity science’s impact on political science can be seen in another article by researchers from SFI (here). The scientists built a computer model which may explain the recent trend towards political polarization. Lead author Vicky Chuqiao Yang explains “By being ‘inbetweeners,’ independents are viewed as unfavorably as the other party by both sides, and left out. So, Independents get the worst of both worlds, and there are downstream consequences.”
Capstone is the only firm in the world with a unique focus on both policy and business strategy. As such, we are constantly working at improving our understanding of and ability to predict policy and market trends. We have recently written about new techniques we have adopted to improve the accuracy and precision of our policy predictions, and recently our efforts were featured in an article by Entrepreneur magazine. We don’t intend to stop trying to find ways to improve, and we believe that complexity science and the work of organizations like the Santa Fe Institute hold the secret to significant leaps forward in our understanding of markets and politics.