The current Treasury yield curve is deeply inverted, offering 5.4% for one-year T-bills and just 4.3% for 10-year Treasury bonds.
Investors are caught between the temptation to earn a higher rate for a year and then hope for the best next year, or lock in 10-year rates that are the highest they've been since 2007. That same 2007 investor would have done well focusing on long-term bonds, but the yield curve has been inverted for over a year now, rewarding investors who invested in short-term Treasurys.
An investor has two choices. Buy a 1-year Treasury Bill and earn 5.4% for the next 12 months and then reinvest, or buy long-term Treasury bonds today. Where will interest rates on long-term bonds be a year from now? Are interest rates predictable when the yield curve inverts?
Since inverted yield curves are relatively rare, we can look at the historical evidence from multiple countries to get a sample size large enough to form a conclusion. We find that pivoting to bills when yield curves are inverted has been the smart decision historically, but only if investors keep an eye on their bond portfolio and move back into bonds when the yield curve normalizes.
Future changes in bond yields seem to be random when the curve is inverted, meaning that the 10-year rates a year from today are no more likely to be higher than lower than they are today. However, there is definitely a downward longer-term trend in bills that will likely result in investors in bills experiencing lower returns than bonds.
Investing for the Long Term
Investors generally expect to be rewarded for holding bonds that mature in the more distant future instead of short-term cash. It may seem that the opportunity cost of cash goes away when yields on money markets are higher than 10-year Treasury bonds.
However, rates on 10-year Treasurys never cracked 2% in 2020 and 2021, and an investor can lock in a decade-long yield above 4% today. The so-called expectations hypothesis would suggest that the market believes rates will fall within the next few years, which could punish investors who were tempted into move bonds into cash today.
In other words, investors can get 5.4% if they buy a one-year T-bill. If they invest in a two-year Treasury with a 4.7% yield, the second-year yield expectation is just 4.3% (5.4% the first year and 4.3% the second year).
Between year 2 and year 10, the market expects an annual yield of just 3.8%. So an investor who buys a 10-year bond instead of a short-term Treasury is betting that rates will drop by over 1.5% after the next year or two. Is this a good bet?
The bet is really about whether long-term investors should grab attractive Treasury bill yields today and then reinvest in one year. If the inverted yield curve contains accurate information about future yields, then there will be no advantage to investing in cash.
We can answer this question by looking at historical bond performance data in different yield curve environments using returns from 16 developed countries between 1870 and 2020 using the Jordà-Schularick-Taylor (JST) Macrohistory Database.
The JST dataset includes data on 48 real and nominal returns for 18 countries from 1870 to 2020. Economic data for Ireland and Canada are not available, which is why only 16 countries are included in the analysis.
Bill yields exceed bond yields in only 24% of the historical periods. An inverted yield curve is more common in some countries, such as the United States, than others, such as France. We bin countries individually versus across years so that each country has the same weight.
What is the difference in the future one-year return on bonds and bills in different yield environments?
The exhibit below includes the average annual values in Panel A and the probability of bills outperforming bonds in Panel B.
Difference in Future Returns by Yield Environment
Panel A: Bill Yield Minus Future 1 Year Bond Return Panel B: Probability of Higher Bill Return
Bond Yield minus Bill Yield (%) | Bond Yield minus Bill Yield (%) | |||||||||||
<0 | 0-1 | 1-2 | >=2 | Avg | <0 | 0-1 | 1-2 | >=2 | Avg | |||
AUS | -1.30 | -0.13 | -1.59 | -1.51 | -1.13 | AUS | 43 | 57 | 37 | 46 | 46 | |
BEL | 0.46 | -1.58 | -2.69 | -3.93 | -1.94 | BEL | 53 | 45 | 32 | 25 | 39 | |
CHE | 0.79 | -1.32 | -2.13 | -3.01 | -1.42 | CHE | 60 | 29 | 38 | 26 | 38 | |
DEU | 1.19 | -0.81 | -1.83 | -4.72 | -1.55 | DEU | 62 | 37 | 31 | 22 | 38 | |
DNK | 0.07 | 1.28 | -6.46 | -1.35 | -1.61 | DNK | 61 | 52 | 44 | 43 | 50 | |
ESP | 0.61 | -1.26 | -2.60 | -2.79 | -1.51 | ESP | 50 | 33 | 16 | 36 | 34 | |
FIN | 1.72 | -1.74 | -5.95 | -5.35 | -2.83 | FIN | 64 | 33 | 14 | 29 | 35 | |
FRA | 1.72 | -0.85 | -3.41 | -2.59 | -1.28 | FRA | 58 | 49 | 26 | 41 | 44 | |
GBR | 2.04 | -1.59 | -1.85 | -3.61 | -1.25 | GBR | 71 | 40 | 33 | 42 | 47 | |
ITA | 2.41 | -2.03 | -2.31 | -3.56 | -1.37 | ITA | 68 | 44 | 34 | 31 | 44 | |
JPN | -0.03 | -2.91 | -1.94 | -2.75 | -1.91 | JPN | 59 | 31 | 17 | 21 | 32 | |
NLD | -0.20 | -0.08 | -2.23 | -3.88 | -1.60 | NLD | 55 | 49 | 22 | 35 | 40 | |
NOR | -0.02 | -1.57 | -1.24 | -2.31 | -1.28 | NOR | 58 | 33 | 30 | 21 | 35 | |
PRT | -0.66 | -0.50 | -2.38 | -5.70 | -2.31 | PRT | 34 | 39 | 44 | 27 | 36 | |
SWE | 0.97 | -0.84 | -3.71 | -5.35 | -2.23 | SWE | 68 | 44 | 23 | 35 | 43 | |
USA | 0.94 | -1.73 | -0.32 | -4.68 | -1.45 | USA | 57 | 54 | 48 | 29 | 47 | |
Avg | 0.67 | -1.10 | -2.67 | -3.57 | Avg | 58 | 42 | 31 | 32 |
Source: Authors' Calculations, JST Macrohistory Database, Data as of Dec. 31, 2020