>Forecasting Uncertainty

New research from Maxim Ulrich shows that inflation uncertainty impacts bond prices and yields more dramatically than GDP uncertainty.

Columbia Business School – Ideas at work
June 18, 2010

Treasury bonds are typically viewed as among the safest investments, providing lower returns than the stock market but avoiding high degrees of risk and volatility. But no investor can escape uncertainty altogether, and bond investors use GDP and inflation forecasts as rough barometers for term structure, or bond prices and yields.

Dozens of forecasters provide quarterly estimates of how much inflation and GDP are likely to change in the short term, providing bond investors a means to gauge future yields. But when forecasts vary significantly, investors are left to determine which of these conflicting stories represents the best estimate. All forecasts can’t be correct, and few are ever spot-on, so what’s a savvy investor to do?

“Some argue that investors should simply take the average of forecasts to get at the best projection of GDP and inflation,” Professor Maxim Ulrich says. But some forecasts are very close to each other, suggesting low uncertainty, while in other quarters the forecasts diverge widely, suggesting high uncertainty. An average taken in a quarter with a great deal of variance in forecasts is likely to be less reliable than one taken in a quarter with low variance.

Nor do GDP and inflation estimates fluctuate at the same time or rate, so there is no clear means to determine how each measure exerts influence on term structure. Research has typically focused on the impact of GDP uncertainty, rather than inflation uncertainty. Ulrich attributes this focus to the greater variance in GDP forecasts than inflation forecasts. “That variance suggests that there is greater uncertainty about GDP than about inflation,” he says, “and that it should be easier for investors to predict how inflation will move than how GDP will move.”

Ulrich modeled the term structure for U.S. government bonds, taking into account the many estimates for GDP and inflation that investors are confronted with. Using the Survey of Professional Forecasters, published quarterly by the Federal Reserve Bank of Philadelphia, Ulrich observed the variance among all forecasts as a measure of general uncertainty. By comparing this modeled GDP and inflation, Ulrich was able to distinguish GDP uncertainty from inflation uncertainty and show how each affects real and nominal bond prices and rates. He concluded that while GDP forecasts may vary more from quarter to quarter, uncertainty about inflation has a greater long-term impact on term structure.

“The quarterly forecasts for inflation don’t vary a lot, but over the long term, the misspecifications will have a greater effect on yields,” Ulrich explains. He found that investors demand a bigger premium for long-run inflation uncertainty than for long-run GDP uncertainty.

The model also provides a simple method investors can use when consulting the forecasts. “The Survey of Professional Forecasters provides our best empirical proxy for determining the amount of uncertainty bond investors face,” Ulrich says. “Investors can look at all key forecasts of GDP and inflation each quarter and, using a simple calculation, observe the magnitude of uncertainty.”

Maxim Ulrich is assistant professor of finance and economics at Columbia Business School.