Kenneth Singleton: New Research Could Give Policymakers More Tools to Foresee Economic Calamity

Image

The U.S. Federal Reserve building | Reuters/Jonathan Ernst

If Ken Singleton’s current research had been complete in 2007, policymakers would have had a sharper signal that risk premiums in bond markets were anomalously low leading up to the economic crisis — and, perhaps, another clue that the calamity was coming and how severe it would be. In this research, the Stanford Graduate School of Business finance professor, working with Scott Joslin of the USC Marshall School of Business and Marcel Priebsch, a researcher at the Federal Reserve Board, developed a way to more reliably measure and interpret the effects of inflation and economic growth on yields and risk premiums in the U.S. Treasury market.

Singleton is one of the leaders in the field of macro-financial modeling, which studies the links over time between the prices of assets and the macroeconomy. His joint research with the University of Chicago’s Lars Peter Hansen, recipient of the 2013 Nobel Memorial Prize in Economic Sciences, was a pioneering contribution in this field. And his latest research falls within the broad category as well. “Macro-financial modeling is important because it helps open the black box of bond pricing and risk systems by shedding light on the macroeconomic sources of fluctuation in bond yields,” Singleton says.

Singleton, Joslin and Priebsch have been discussing the current research with organizations including the Federal Reserve Board, the International Monetary Fund, and Norway’s sovereign wealth fund, which invests the country’s oil wealth. They are interested because the work could help them better assess worldwide economic risk based on structure of yields on Treasuries that mature at different dates (the Treasury yield curve).

“Policymakers in the world’s economic powers monitor prices in the Treasury market as an indicator of how their decisions affect economic activity. Global development institutions monitor developments in Treasury markets for signals of changing economic risks, or even developing crises,” says Singleton.

Policymakers use data to get a sense of what is going on in investors’ heads — figuring that the collective current unease or confidence of investors has implications for future aggregate demand. One of the key constructs policymakers look at is the risk premium in the Treasury market, which is the expected excess return that investors demand for holding a long-term Treasury bond — say, one that matures 10 years from now — and a short-term bond with maturity equal to their holding period — say, a Treasury that matures in one year.

The risk premium changes over time as Treasuries are bought and sold in response to new information about the state of the global economy. Companies and other governments also look at risk premiums in the Treasury market to help understand their own costs of funds, and traders use market risk premiums to determine their desired portfolios of Treasury bonds and other assets.

Up until about 2005 or so, many of the models of Treasury bond prices used by economic policymakers and academics presumed that all the information about the macroeconomy needed to reliably measure market risk premiums was contained in the yield curve. Their models left out information about the macroeconomy entirely, other than what was implicitly captured in the yield curve.

A second generation of models was introduced in the 2000s that sought to better capture the connection between inflation, cycles in economic growth, and risk premiums in Treasury markets. However, what Singleton and his co-authors show mathematically is that this second-generation approach amounted to making the symmetric and even stronger assumption that the current Treasury yield curve was the only information policymakers needed to forecast future economic activity.

This is clearly counterfactual. A quick look at history shows that the yield on, say, the 10-year Treasury bond is correlated with real economic growth, but economic growth (be it positive or negative) certainly does not always track Treasury yields. In fact, arguably, policymakers’ overreliance on information in the current Treasury yield contributed to policymakers’ underappreciation of the growing fragility of markets leading up to the financial crisis.

Singleton and his co-authors developed a new framework that addresses these empirical limitations of prior models. Specifically, they showed that the part of economic growth that moves separately from current yields is highly informative about where future yields are headed. Exploiting this fact in a rich dynamic model, the team of researchers developed more reliable measures of how the macroeconomy affects risk premiums in Treasury markets.

There is a simple intuition for why the new model outperforms prior approaches to modeling the links between economic activity and risk in bond markets. Imagine an investor who is deciding whether to purchase additional Treasury bonds for her portfolio. Naturally this investor will want to assess the risk-return trade-off, not only for the Treasury market but also to compare this trade-off to those for other investments (equities, real estate, etc). The earlier generations of risk premium models essentially assumed that the bond investor only needed to look at bond yields to decide whether to invest more in Treasury bonds.

Singleton’s model formally recognizes that bond investors will look at the macroeconomy at large to assess the relative risks of investing in bonds versus other asset classes. This broader perspective leads to more accurate measures of the risk compensation demanded by those who hold Treasury bonds.

Specifically, the model includes output growth (precisely, the weighted average of economic activity devised by the Chicago Federal Reserve Board) and inflation as relevant information for evaluating risk, over and above the information in the yield curve. When the researchers applied their model to the 23-year period from 1985 to 2007, they found that output growth has a significant effect on risks related to the slope of the yield curve (the spread between yields on long- and short-term bonds), while inflation was material for understanding the risks associated with the average level of Treasury yields.

Now policymakers are examining this model. Reimagining the nature of risk premiums — a manifestation of investors’ assessments of risk — may lead to more accurate forecasts of how monetary and fiscal policies affect bond yields and, ultimately, the aggregate economy.

Singleton, in collaboration with two PhD students at the GSB, is extending this research. “We are modeling how an investor who is learning about the impact of changing economic policies and major financial shocks on the economy would have priced bonds. Strikingly, our extended model with real-time learning gives substantially more accurate forecasts of future yields than the professional’s consensus forecasts, especially following economic recessions,” Singleton says. At the same time, the research shows that dispersion in predictions of actual professional forecasters tells us about the future path of Treasury yields.

There is the real possibility that variation in bond prices induced by investors’ confusion about how economic surprises will affect bond prices is being misattributed to changing risk appetites of investors.

Kenneth Singleton is the Adams Distinguished Professor of Management at Stanford GSB. Scott Joslin is an assistant professor of finance and business economics at the USC Marshall School of Business. Marcel Priebsch is a researcher at the Federal Reserve Board. Both Joslin and Priebsch earned their doctorates in finance from Stanford GSB. The paper, “Risk Premiums in Dynamic Term Structure Models with Unspanned Macro Risks,” was published in the Journal of Finance in May.