It is not uncommon for investors to make predictions about the direction of interest rates and then make investment decisions based on those predictions. These predictions may be based on specific forecasting models or simply on gut feelings about the direction of interest rates.
The evidence suggests, however, that predictions about the direction of interest rates are highly unreliable, whether based on quantitative models or feelings about what the future holds.
Researchers have noted this problem and documented the unreliability of interest rate predictions for many years. Roland Robinson summarized the problem in his 1954 article, Forecasting Interest Rates1:
“The obstacles to systematic forecasting of interest rates are such that no well-developed and recognized methodology prevails. The interest-rate theory underlying most forecasting is intuitive, the methods sketchy, and the product often subjective. Furthermore, the published … forecasts of interest-rate changes reveal a sobering proportion of errors and misjudgments.”
The problem is illustrated in the following graphic created by BNY Mellon.
Survey of Professional Forecasters vs. Actual 10-Year Treasury Yields
2000-2023
For illustrative purposes only and does not represent the results of any investment. Source: as of 3/31/23, St. Louis FRED, Federal Reserve Bank of Philadelphia Survey of Professional Forecasters.
The orange dotted lines represent the predictions of professional forecasters for the yield on 10-year Treasury bonds at different moments between 2000 and the first quarter of 2023. The blue line shows the actual yield on 10-year Treasury bonds during that period.
Orange dots that do not fall on the blue line represent inaccurate predictions. As you can see, expert predictions about yields were frequently in error. In fact, BNY Mellon identified 60 times over the last 20 years when the “pros” expected rates to rise and they actually fell.
Michael Bauer, in his 2017 article, Bridging the Gap: Forecasting Interest Rates with Macro Trends,2 noted that, “Neither the short-run movements nor the long-run trends are easily predicted using existing models.”
Bauer offers two reasons for this problem. First, interest rates can vary widely from day to day. This makes it difficult to find correlations between interest rates and economic fundamentals like monetary or fiscal policy.
Second, interest rates have not fluctuated around a constant average, making predictions based on a “mean reversion” approach difficult. Instead, rates have fluctuated based on slow-moving long-term trends, as can be seen from the graphic below.
10 Year Treasury Rate - 54 Year Historical Chart
1962-2023
Source: MacroTrends. https://www.macrotrends.net/2016/10-year-treasury-bond-rate-yield-chart
Bauer further cites research published in The Handbook of Economic Forecasting for the proposition that no approach to forecasting interest rates has been able to consistently improve upon the simple random walk model, which assumes that increases and decreases are equally likely and therefore always forecast no change.
Be careful about building bond portfolios around the “experts” view of interest rates. They’re probably wrong.
- https://www.jstor.org/stable/2351185
- https://www.frbsf.org/research-and-insights/publications/economic-letter/2017/07/bridging-gap-forecasting-interest-rates-with-macroeconomic-trends/
The information contained herein does not constitute investment advice or a solicitation. This article is for dissemination of general information only. Information presented is believed to be factual and up-to-date, but we do not guarantee its accuracy and it should not be regarded as a complete analysis of the subjects discussed. Investments are not guaranteed and are subject to investment risk, including possible loss of the principal amount invested. Past performance is no guarantee of future results.
