Risky Business: Understanding the Economy
Morten Nielsen smiles patiently as I ask him if his work will help avert the next financial crisis. “It might ... but that is a little like asking a theoretical physicist if his work will make stronger bridges. My work in econometrics is theoretical and involves developing mathematical and statistical tools to analyze economic data in various ways.”
Dr. Nielsen, a professor of economics at Queen’s, holds both the David Chadwick Smith Chair in Economics and the Canada Research Chair in Time Series Econometrics. His research creates tools to uncover the intelligibilities in economic data. “And yes,” he adds, “hopefully those tools can be used by others to avoid the next financial crisis.” Nielsen wants to make it clear that his work does not involve recommending which actions governments should take, or what policies they should follow. Rather, his research takes its starting point from revealed shortfalls of current theory and develops tools that empirical economists need to make better predictions about the economy, and that policy makers want to model the effects of new policies.
For instance, even though many economists commented how obvious the warning signs were after the fact, almost no one predicted the economic crisis of 2008, let alone its timing. Assessing and predicting volatility and risk in markets is one of the greatest challenges in modern economics, a fact laid bare by the last crisis. It seemed that certain basic assumptions about the economy didn’t seem quite up to the task.
Nielsen tackles one of these assumptions in his research into the effect of past data on future events in a market. In econometrics, this idea is called “long memory.” “What that means,” Nielsen explains, “is that the past data of a particular variable in the economy is highly correlated to its future behaviour.” In analysis of, say, stock market returns, the assumption is typically the opposite. And for good reason. To suggest that there is memory in returns data is a little like saying that if you flip a coin, whether you flipped heads or tails on the previous turn has some influence on what you’ll flip next. Which is just not how the world works. Even though, as you keep flipping, the ratio of heads to tails approaches 50/50, there is certainly no memory in coin flipping – each coin-flip is independent of the one that came before. More or less, the same is assumed for stock market returns over time – that is, one day’s returns offer no correlation to tomorrow’s. As Nielsen explains, “the assumption is logical in the sense that if future returns were in any way predictable from past events, people would do it and make bazillions. Since no one is doing so, that seems to indicate there is no memory in stock market returns.”
However, for the past 30 years or so, empirical researchers have been finding more and more evidence of long memory in many other economic and financial data. The implication is that the data of economies are not like the data of coin flips – at least not universally. Rather, there is memory in the data that can be used to predict the future. However, the great challenge remains to tease out the variables in the data that are correlated over time. And that requires very sophisticated tools which Nielsen and his colleagues are still developing.
One recent success by Nielsen and his colleagues was to collect data from the pricing of futures and use those data to predict future volatility in the underlying markets. Futures traders try to predict the future volatility of the assets in the underlying market (e.g. stocks) in order to price futures contracts 30 days or so into the future. So by looking at the data from the entire futures market, Nielsen was able to tease out a level of implied volatility in the underlying market and thereby better predict the volatility, or risk, of the assets in the underlying market.
These are exactly the tools needed by economists working in industry and government, whose job it is to be able to see the signs of risk in the market before crises happen. And this progress will only be made by researchers such as Nielsen who are attending carefully to the data to gain mathematical insights that will in time reveal more and more the hidden patterns of our economy.
(e)Affect Issue 5 Spring 2014