Cutting cycles and learning to love bonds again

There’s no perfect historical analogy for the current environment. Fiscal and monetary institutions have changed, as have technology, geopolitics, and financial markets behavior. Plus, the economy is still normalizing from the effects of unprecedented stimulus measures following the pandemic.

Still, we found commonalities across cutting cycles that could inform current asset allocation decisions. We studied 12 Fed cutting cycles spanning over 70 years, which are identified in red in the chart below.

The most obvious takeaway is that bonds outperformed cash in all 12 cutting cycles, as shown below, by an average of 8.1% annualized.

And they outperformed when the yield curve was inverted at the start of five of the 12 cycles. Periods when the yield curve was inverted at the onset of Fed cuts are highlighted in grey. In those cases, bonds outperformed cash by an average of 7.8%.

So, my thinking is, let’s learn to love bonds again. This is not about stocks versus bonds. Rather, it’s about cash versus bonds. While past performance doesn’t tell us what will happen in the future, bonds could resume their role as diversifiers to stocks. I expect this as markets start responding more to growth scares than inflation risks.

I recognize this trade is a tough sell in the near term. At below 4%, the 10-year U.S. Treasury yield is near its lowest level in a year. If it continues to trade within a range, as it has for the last two years, its next trend could be higher, not lower.

However, long-term investors who are less sensitive to the entry point could phase in the trade. I expect Fed cuts to lower the range over time.

On the other hand, tactical investors may want to wait for a move up in yields as an entry point—a potential last hurrah in the 10-year yield toward 4.5%–5.0%, which I believe could happen around the election or following a geopolitically induced inflationary oil price shock.

We’re in that camp. Our Asset Allocation Committee remains underweight duration and positioned for upside risk in inflation. There is no consensus within the committee on duration. For now, the 10-year yield is too low to justify moving money from cash to bonds, but based on our analysis of cutting cycles, we’re preparing for that scenario.

Other economic and investment takeaways aren’t as conclusive. On average, however, the Fed cut during economic slowdowns. Hence, during cutting cycles:

• Unemployment rose. The average increase in unemployment was 1.6%. It rose during 10 out of 12 cutting cycles.
• Inflation fell. The average decline in inflation was 1.7%. It declined during 10 out of 12 cycles.
• The equity risk premium compressed. The average outperformance of stocks over bonds across all 12 cutting cycles was 2.2%, annualized, with stocks outperforming bonds 58% of the time. Over the full sample (January 1940–July 2024), however, the average outperformance of stocks over bonds was 7.3%, annualized, with stocks outperforming bonds 70% of the time. (Stocks are represented by S&P 500 Index. Bonds are represented by the Bloomberg U.S. Aggregate Bond Index.)
• High yield (HY) bonds underperformed investment grade (IG). The average underperformance of HY was 7.2%. HY underperformed IG in four out of six cutting cycles (HY data start in 1983). In the full sample (August 1983–July 2024), HY outperformed IG by 2.4%, on average, 62% of the time. (High yield bonds are represented by the Credit Suisse High Yield Index. Investment-grade bonds are represented by the Bloomberg U.S. Aggregate Bond Index.)

See Figure 2 for sources; analysis by T. Rowe Price. Past performance is not a reliable indicator of future performance.

Looking forward

Which tactical asset allocation positions, besides the potential of bonds to perform better than cash, could outperform this time? To help us answer this, we need to forecast:

1. the direction of rates and
2. the sensitivity of asset classes to rates (i.e., their duration).

To generate alpha, tactical asset allocators must get both right.

The duration of stocks, for example, is a matter of debate. Declining rates should mean a lower discount rate on future cash flows (earnings) and thus rising valuations—resulting in a positive duration, correct? But what if rates decline because economic growth slows, which hits the expected value of future earnings and their growth rate? In this case, the empirical duration could be negative.

Or take small-cap stocks. Historically, they’ve had lower duration than the tech‑heavy, large-cap stocks. Tech companies’ cash flows typically extend further into the future, which means a higher sensitivity to the discount rate. But recently, due to their shorter‑term debt and higher leverage, small-caps have been more sensitive than large‑caps to changes in interest rates.

And here’s another duration distortion: Over the last year, growth stocks have become less sensitive to rate movements than value stocks, as growth stocks’ returns have been driven by artificial intelligence (AI) innovation. Empirically, their sensitivity to rates has flipped.

So many factors drive stock returns, not just interest rates. For many asset classes, isolating the effect of interest rates (i.e., estimating duration) is extremely difficult and requires judgment calls.

The chart in the appendix may help. If you believe that rates are going down and want to add duration to your tactical asset allocation, you could add to the pairs with positive duration (long Treasuries versus the overall investment-grade market, for example) or reverse the pairs with negative bars (e.g. underweighting floating rate versus investment grade). Of course, as with any of these one-factor analyses, you should be aware of the influence of other factors that have been omitted, e.g., valuation, macro, and sentiment.

Takeaways

• During Fed cutting cycles, as unemployment typically rose and inflation declined, high yield historically performed poorly, and stocks sometimes underperformed bonds.
• Nonetheless, given strong earnings, reasonable valuations in certain market sectors, and continued AI spending, we remain comfortable at neutral (slightly overweight, according to our target allocations) stocks versus bonds.
• An important takeaway is that, historically speaking, bonds consistently outperformed cash during Fed cutting cycles, averaging 8.1% annualized returns over cash. Even when the yield curve was inverted, bonds still outperformed cash by an average of 7.8%.
• If yields rise again to 4.5%–5%, we may consider reallocating cash to a diversified bond portfolio, especially as market focus shifts from inflation to growth concerns.

Appendix

The green bars show our estimate of recent duration. Here’s the methodology:

• Analysis based on daily data.
• Uses rolling five-day returns to reduce the impact of asynchronous markets.
• Regression model controls for stock return sensitivity (beta).
• Exponentially weighted with a six-month half-life.

The interest rate used when calculating empirical duration is the change in the 10‑year yield (sourced from the Federal Reserve Board) and controlling for equity returns as represented by MSCI ACWI Local returns.

Relative valuations are defined in terms of the earnings yield spread (for equities) and yield spread (for fixed income). The blue lines show the historical range of these rolling durations, from the 10th to the 90th percentiles. Notice how U.S. growth versus value is outside its historical range, reflecting the impact of AI. Also, in general, ranges for fixed income pairs are tighter than for stocks, because fixed income returns are more easily explained with the interest rate factor.

If you believe that rates are going down and want to add duration to your tactical asset allocation, you may consider adding to the pairs with positive green bars or reverse the pairs with negative bars (e.g. underweighting floating rate versus investment grade).

Thank you to James Whitehead, Cesare Buiatti, Rob Panariello, and Charles Shriver for their help with this analysis.

References

Empirical duration is the sensitivity of the asset’s return to changes in the 10-year bond yield.

Regression models describe the relationship between measurable variables. At a basic level, a regression model allows for estimating how one variable is expected to change as another independent variable changes. The models can also help identify if it’s a strong or weak relationship between the two.

Relative valuation is the concept of comparing the price of an asset with the market value of similar assets.

Indices

In the figure, the names refer to the indices as follows:

Treasury Long—Bloomberg U.S. Treasury Long Index

Bonds—Bloomberg U.S. Aggregate Bond Index

Int’l. Growth—MSCI EAFE Growth Index

Int’l. Value—MSCI EAFE Value Index

U.S. Growth—Russell 1000 Growth Index

U.S. Value—Russell 1000 Value Index

RAF—Real Assets Combined Index Portfolio

EQ (Equity), Stocks—a mix of 70% Russell 3000 Index, 30% MSCI ACWI ex-US

EMD (Emerging Market Debt)—J.P. Morgan Emerging Markets Bond Index Global Index

EMB (Emerging Market Bonds)—J.P. Morgan Emerging Markets Bond Index Global Index

HY (High Yield)—Credit Suisse High Yield Index

U.S. Floating Rate—Morningstar LSTA Performing Loan Index

EM (Emerging Markets) Equity—MSCI Emerging Markets Index

U.S. SC (Small-Cap)—Russell 2000 Index

U.S. LC (Large-Cap)—S&P 500 Index

Int’l. Bond Hedged—Bloomberg Global Aggregate ex-USD Index (USD Hedged)