This week, as stocks fell, long-dated US Treasuries rose and made up a good part of the losses. With inflation in the high 3% range, will bonds keep acting as a shield next week?
2026-W42
Bonds were a shield this week, but lost strength on Friday
For a portfolio split 60% S&P 500 ETF (SPY) and 40% long-dated Treasury ETF (TLT), the return this week was about -0.12%. Holding SPY alone would have lost -1.97%. Bonds made up almost all of the equity loss.
Breaking it down by day changes the picture.
| Date | SPY | TLT | 60/40 combined | Share of loss offset by bonds |
|---|---|---|---|---|
| 10-14 (CPI release) | -1.08% | +0.66% | -0.38% | about 41% |
| 10-15 (PPI release) | +0.28% | +1.37% | +0.72% | both rose |
| 10-16 (WTI breaks $100) | -1.11% | +0.10% | -0.63% | about 6% |
The combined figures and ratios are calculated directly from the weekly closing prices. On Friday, when stocks fell the most, bonds barely moved. The same day WTI rose +3.51%. That one day previews the answer to next week's question.
When inflation tops 3%, stocks and bonds normally move together
External research holds that the inflation level determines the relationship between stocks and bonds.
- In months since 1976 when CPI was 3% or higher, the stock-bond correlation was about 0.37, versus about 0.02 in months below 3% (iSectors).
- Invesco's David Aujla said that in data since 1952, stocks and bonds moved in the same direction 98% of the time when 5-year annualized inflation exceeded 3% (Trustnet).
- Robeco's model expects a negative correlation only when inflation is below 3% and real yields are normal. It estimates that each 1 percentage point rise in inflation and real yields lifts the correlation coefficient by about 0.17 (Robeco, 2024-05).
CPI is now 3.7%. By that research, this week's pattern of bonds supporting stocks is the exception. The year the default played out was 2022. The 3-month rolling stock-bond correlation hit 0.45, the highest in 23 years, against a 10-year average of -0.25 (Credit Suisse, cited by Advisor Perspectives 2022-10). That year the 10-year Treasury returned -17.8%, the worst on record (Motley Fool, 2023-01), and the 60/40 portfolio lost 14-16% (InvestmentNews, MSCI).
Why this week was the exception
Bonds were bought because of credit fear, not inflation. Market commentary also held that safe-haven demand outweighed inflation worries, against a backdrop of reassessment of AI infrastructure debt (Investing.com, Kitco, 2026-10-15). Even within bonds, performance split by quality. Investment-grade corporate bonds LQD +0.97%, high yield HYG -0.48%. That means money moved to avoid risk.
In short, the bond shield switches on or off depending on the type of shock. It switches on for growth and credit shocks and off for inflation and oil shocks. Friday was a preview of the off side.
Next week's paths
| Shock driving next week | Stocks | Long-dated bonds | 60/40 defense |
|---|---|---|---|
| AI credit worries spread. HYG falls further, big tech cuts capex | Down | Up | Works |
| Brent settles at $110, hawkish FOMC signal | Down | Down | Fails. 2022-style |
| Tech rebounds, credit worries calm | Up | Down as yields retrace | Stocks make up the loss |
The test for the first path is big tech earnings. One analysis says Meta's Q2 capex was 98% of operating cash flow (Third Pole Markets, secondary source). If signals emerge that spending is being cut, credit worries grow and bonds become a shield again. The second path depends on whether Brent moves above its weekly high of $110 and stays there. In the third path bonds lose money but stocks rise, so the portfolio as a whole has little problem.
The worst combination is the second path. We have not calculated a probability. It is worth remembering, though, that Friday alone was a miniature of that combination.
What investors can do
- Do not treat long-dated bonds as a universal hedge. Long-dated bonds are insurance for credit shocks. For oil and inflation shocks, shorter-maturity Treasuries are better. The 13-week T-bill yield is 4.07%, so the cost of waiting is small too.
- Split hedges by type of shock. For example, holding defensive assets half in long-dated bonds and half in short-term bonds means one side holds up whichever shock arrives. This ratio is an example.
- Watch two signals together. Whether TLT rises on days stocks fall, and where WTI went that day. If TLT fails to rise for 2 consecutive stock-down days, treat the shield as off and raise the short-term bond share. This criterion is a rule set in this answer and is not a validated figure.
- The 10-16 Q&A split the cause of the rate decline on PPI day by day. On a weekly view there is one conclusion. While inflation is above 3%, the bond shield is conditional.