Tradiens Research · Data essay · Macro
The Year Stocks and Bonds Both Fell Heavily
In eleven of the twelve years since 1928 in which US stocks fell 10% or more, bonds lost little or nothing; 2022 was the exception.
Of the twelve years since 1928 in which the S&P 500 fell 10% or more, 2022 is the only one in which 10-year US Treasury bonds also lost heavily.
According to the yearly returns published by Aswath Damodaran of NYU Stern, the S&P 500, dividends included, lost 10% or more in twelve calendar years between 1928 and 2025.[1] For the bond side the same source uses a 10-year US Treasury bond whose return is rebuilt from its yield: the interest it pays, set by the yield at the start of the year, plus the change in its price. In eleven of those twelve years the bond lost little or nothing: it gained in nine, and in the other two, 1931 and 1941, it lost at most 2.6%. In 2022 it lost 17.8%, almost as much as the S&P 500's 18.0%, so 2022 is the only one of the twelve in which the bond also lost heavily.
In the twelve years stocks fell 10% or more, bonds lost heavily only in 2022.
S&P 500 down 10% or more: 12 years all other years: 86
The small gray dot on the −10% line, just above zero, is 1966: the S&P 500 lost 9.97% that year, just short of 10%, so 1966 is not one of the twelve.
| Year | 10-year Treasury return | S&P 500 return |
|---|---|---|
| 2008 | +20.1% | −36.6% |
| 2002 | +15.1% | −22.0% |
| 1957 | +6.8% | −10.5% |
| 2001 | +5.6% | −11.8% |
| 1940 | +5.4% | −10.7% |
| 1930 | +4.5% | −25.1% |
| 1973 | +3.7% | −14.3% |
| 1974 | +2.0% | −25.9% |
| 1937 | +1.4% | −35.3% |
| 1941 | −2.0% | −12.8% |
| 1931 | −2.6% | −43.8% |
| 2022 | −17.8% | −18.0% |
S&P 500, dividends included · 10-year Treasury rebuilt from its yield: interest (coupon) plus price change
Each dot is one year: its position from left to right is the S&P 500's return, and its height is the 10-year Treasury's return, both drawn to the same scale. The twelve large dots, left of the line at −10%, are the years the S&P 500 fell 10% or more, and the table lists them. In eleven of them the bond's return sits near or above zero. Only 2022 sits far below zero, about as far down as it is to the left, because that year the bond lost almost as much as the S&P 500.
The figure shows what a 60/40 portfolio relies on. Such a portfolio keeps 60% of its money in stocks and 40% in bonds; in this essay a 60/40 mix means our own calculation of such a portfolio from Damodaran's yearly returns, rebalanced each January. The portfolio counts on the bonds to soften a fall in stocks, and that expectation can mean two different things, which this essay keeps apart because 2022 broke both. The first is a floor: in a year when stocks crash, the bond loses little or nothing of its own value, which is what the twelve years in the figure measure. The second is a hedge: from month to month the bond tends to move the other way from stocks, so that its gains in a bad month offset part of the stock loss, which only monthly data can show. This essay calls the floor and the hedge together the bond's protection.
So the floor held in each of the eleven years before 2022, which raises the question this essay follows: if bonds are the counterweight to stocks, why did both fall heavily in 2022, and why did even the head of Norway's sovereign wealth fund call it very unusual? An older idea says why a second investment helps at all: what it adds to a portfolio depends on whether it tends to fall at the same time as the first. That idea was written down in 1952, by an economist who then did not apply it to his own money.
Harry Markowitz showed in 1952 that diversification depends on how investments move together, yet he split his own retirement contributions fifty-fifty between bonds and stocks to limit regret.
The economist was Harry Markowitz, and the paper was "Portfolio Selection", published in The Journal of Finance in March 1952.[2] He put the point as a rule: "It is necessary to avoid investing in securities with high covariances among themselves." Covariance is a measure of how two returns move together, so the rule says that two investments protect each other only if they do not tend to fall together.
Not long after his 1952 paper, Markowitz had to decide how to invest his own retirement contributions with TIAA-CREF. In a 1998 article in Money magazine, the journalist Jason Zweig quoted him: "I should have computed the historical co-variances of the asset classes and drawn an efficient frontier. Instead, I visualized my grief if the stock market went way up and I wasn't in it—or if it went way down and I was completely in it. My intention was to minimize my future regret. So I split my contributions fifty-fifty between bonds and equities."[3]
The story matters because a fixed split, fifty-fifty or sixty-forty, takes for granted how stocks and bonds move together instead of measuring it. Markowitz's rule is about the hedge, whether the bond tends to move against stocks, and that can be checked month by month. Whether the bond loses little or nothing in a crash year, the floor, is a separate yearly question. The essay leaves aside which split is right and asks whether the hedge that any fixed split relies on has been there, and when; the monthly check shows that before 2000 bonds often fell when stocks did.
Between 1970 and 1999 bonds also fell in about half of the months in which stocks fell, and between 2000 and 2021 in fewer than a third.
To make that check we built a monthly series of our own, starting in 1962. For stocks it uses the whole US stock market, dividends included, as computed from CRSP data and published in the Kenneth French Data Library; for bonds, our monthly 10-year bond, a 10-year Treasury rebuilt from the month-end yields that the Federal Reserve publishes through FRED.[4][5] The simplest test is to take every month in which stocks fell and count how often the bond fell too. Between 1970 and 1999 stocks fell in 139 months and the bond fell in 71 of them, 51%, so in about half of the bad months for stocks the bond gave no offset at all. Between 2000 and 2021 stocks fell in 93 months and the bond fell in 27 of them, 29%, which means that from 2000 to 2021 the bond moved the other way far more often than before.
Bonds fell in half of stocks' losing months before 2000, under a third from 2000 to 2021.
| Period | Stocks' losing months | Bond also lost | Share |
|---|---|---|---|
| 1962–69 | 35 | 15 | 43% |
| 1970–99 | 139 | 71 | 51% |
| 2000–21 | 93 | 27 | 29% |
| 2022–26 | 23 | 17 | 74% |
1962–1969: 43% (15 of 35 months) · 1970–1999: 51% (71 of 139) · 2000–2021: 29% (27 of 93) · January 2022 – August 2026: 74% (17 of 23)
Counted only among months in which stocks fell. Stocks: US total market (CRSP, via the Kenneth French Data Library). Bonds: 10-year Treasury rebuilt from FRED month-end yields.
Each column is one period, and its width is the number of months in which stocks fell, 290 in all, so a narrow column rests on few months. The lower part of each column is the share of those months in which the bond's monthly return, interest included, was negative too; the line across the columns marks 50%, half of the months. The higher the lower part reaches, the less the bond offset the stock losses.
Counting bad months is easy to read but ignores how large each move was, so from here on the essay also uses a standard measure that takes size into account: the correlation between the monthly returns of stocks and bonds over the 36 months up to and including each month, which this essay calls the three-year correlation. A correlation runs from −1 to +1. When it is positive, the two have tended to move together over those three years, and when it is negative they have tended to move in opposite directions, though not necessarily in every month. Recomputing the measure every month gives a line through time. To give a sense of scale, for the months from 1970 to 1999 our line averaged 0.34, and in those years the bond fell in about half of stocks' losing months; for 2000 to 2021 it averaged −0.33, and the bond fell in under a third of them. Our bond series starts in 1962, which is why the monthly counts start there, but a three-year correlation needs three years of data, so the line starts in January 1965.
In 1973 and 1974 bonds still had a positive return before adjusting for inflation while stocks fell, but from month to month the two tended to move in the same direction.
The yearly and the monthly evidence can disagree, and 1973 and 1974, the only years of the 1970s in which the S&P 500 fell 10% or more, show how. On Damodaran's yearly returns, the S&P 500 lost 36.5% over the two years together, while the 10-year Treasury returned about 5.7% in total over the same two years. The bond stayed positive because of how its return is made up. In our rebuilt 10-year bond the price fell in both years, by less than 5 points each time, because yields rose; but the interest the bond paid, about 6½% a year, was larger than the fall in each year, so the total stayed above zero.
Inflation is a separate matter, and neither the floor nor the hedge, as this essay uses the words, protects against it. Consumer prices rose about 22% over the same two years, so after inflation the bond lost about 13% and our 60/40 mix more than a third.
The monthly series shows how little the bond softened those two years. Our three-year correlation stayed positive from August 1966 to August 1974. In nine of the eighteen months of 1973 and 1974 in which the US stock market fell, our monthly 10-year bond fell too. Researchers at AQR, writing in The Journal of Portfolio Management in 2023, describe a pattern they find in the data, though their study does not take those two years apart: "growth shocks drive stock and bond returns in opposite directions, while inflation shocks drive them in the same direction."[6] In 1973 and 1974, then, years in which prices rose fast, the floor held and the hedge was missing, which fits that pattern.
Stocks and bonds tended to move together for decades, then apart from 2000 to 2022.
| Stretch | From | To | Months |
|---|---|---|---|
| above zero | Jan 1965 | May 1965 | 5 |
| below zero | Jun 1965 | Jan 1966 | 8 |
| above zero | Feb 1966 | Feb 1966 | 1 |
| below zero | Mar 1966 | Jul 1966 | 5 |
| above zero | Aug 1966 | Aug 1974 | 97 |
| below zero | Sep 1974 | Sep 1974 | 1 |
| above zero | Oct 1974 | Mar 1975 | 6 |
| below zero | Apr 1975 | Apr 1975 | 1 |
| above zero | May 1975 | Sep 1989 | 173 |
| below zero | Oct 1989 | Dec 1989 | 3 |
| above zero | Jan 1990 | Oct 2000 | 130 |
| below zero | Nov 2000 | Jul 2022 | 261 |
| above zero | Aug 2022 | Aug 2026 (still running) | 49 |
No monthly data of our own before 1965.
Our calculation: 36-month rolling Pearson correlation of monthly returns. Stocks: US total market (CRSP, via the Kenneth French Data Library). Bonds: constant-maturity 10-year Treasury built from FRED DGS10 month-end yields. Series starts January 1965.
For each month the line gives the correlation over the 36 months ending with that month. Where the line is above the zero line, stocks and bonds tended to move in the same direction; where it is below, they tended to move in opposite directions. The two short marks under the line show 1973–74 and 2008, the years the text discusses. In the strip below the line, each block is one unbroken stretch above or below zero, as long as the months it lasted; the shortest are drawn a little wider than their length so that they can be seen.
Reading the correlation line · 1 of 2
Look at the left part of the line, from the late 1960s to 2000. It stays above the zero line almost the whole time, which means that over each three-year window stocks and bonds tended to move in the same direction. The marked years 1973 and 1974 sit inside that stretch. So in those years the bond did little to offset the months in which stocks fell: the hedge was not there.
Our own line starts in 1965, but longer studies point the same way. Molenaar, Sénéchal, Swinkels and Wang, using US data since 1875, conclude that "positive stock–bond correlation regimes have been more common than negative ones", and they also find negative stretches, with a correlation below −0.2, "in the early 1930s, in the late 1950s, and during most of the 2000s".[7] So a hedge did exist before 2000, but over the longer history it was the less common state.
From November 2000 to July 2022 the three-year correlation between US stock and bond returns stayed negative for 261 months in a row, the longest such stretch in our data, which starts in 1965.
From August 1966 to October 2000 our line was positive almost throughout, below zero only in one month of 1974, one of 1975 and the last three months of 1989. That changed late in 2000, when our correlation turned negative and stayed below zero until July 2022, by far the longest negative stretch since the line begins in 1965. Each point on the line sums up the three years that end in that month, so the stretch is a run of 261 readings in a row, each of them negative, and the first of them, for November 2000, already looked back to late 1997. As the monthly counts showed, the run does not mean that the bond rose in every month in which stocks fell.
The year 2008 shows what that pattern did for a 60/40 mix. On Damodaran's yearly returns the S&P 500 lost 36.6% that year, while the 10-year Treasury gained 20.1%, because its yield fell and its price rose. Our 60/40 mix lost 13.9%: the stock side took 21.9 percentage points off it and the bond side added 8.0, so the bond's gain made up a little more than a third of the stock loss. In 2008 both the floor and the hedge held: the bond gained in a crash year, and the three-year correlation was negative.
Reading the correlation line · 2 of 2
Now follow the line where it crosses below zero in November 2000. It stays below zero until July 2022, more than twenty years without a break, which is by far the longest such stretch in the chart. The mark for 2008 sits deep inside it. The bond's gain in the year stocks crashed fits that pattern: for those two decades the two tended to move in opposite directions, so a bad month for stocks was often a good one for bonds.
If inflation news moves stocks and bonds together and growth news moves them apart, then how inflation moves with the economy should matter for the hedge. One study, a working paper by Campbell, Pflueger and Viceira, finds a break in the second quarter of 2001 in how inflation moved with the economy.[8] Before the break, inflation and the output gap, the distance between what the economy produces and what it could produce, moved in opposite directions; after it, they moved together. In their data, which run from 1979 to 2011, the quarterly correlation between bond and stock returns changed from 0.21 before the break to −0.64 after it. In plain terms, after 2001 inflation tended to rise and fall with the economy rather than against it, so bad news about growth usually came with lower inflation, which is good for bonds.
In 2022 the 10-year Treasury lost 17.8% in the same year that the S&P 500 lost 18.0%, after its yield started the year low and then rose sharply.
In 2022 the hedge ended, and in the same year the floor gave way too. That was the worst year for the 10-year Treasury in Damodaran's series, which starts in 1928. An independent series built from Robert Shiller's data, going back to 1872, shows the same.[9] A bond's yearly return is the interest it pays plus the change in its price. At the start of 2022 the 10-year yield was about 1.5%; by the end of the year it was 3.88%, according to the Federal Reserve's daily series. A rise that large cuts the price of a 10-year bond by far more than a year's interest, which is how the bond came to lose almost as much as the stock index.
A bond's interest is fixed when it is bought, so once new 10-year bonds paid 3.88%, one still paying the earlier, much lower rate for about ten more years was worth less, and its price had to fall until a buyer would earn the new, higher yield on it. In our rebuilt 10-year bond the price fell 19.3% over the year while the interest added 1.5%, so the year's loss was the difference between the two. The rise in yields is what produced the loss, and the low starting yield meant only that there was little interest to set against it.
The hedge failed in the same year. In seven of the eight months of 2022 in which the US stock market (CRSP total market) fell, our monthly 10-year bond fell too. In August 2022 the three-year correlation turned positive, which ended the negative stretch that had begun in 2000.
For a 60/40 mix the two failures added up. On a 60/40 mix rebuilt each January from Damodaran's annual returns, 2022 was the third-worst year since 1928, after 1931 and 1937, with a loss of almost 18%; on other constructions it ranks differently, as the Methodology shows. Because the mix held 60% in stocks and 40% in bonds, the stock side took 10.8 percentage points off it, 60% of the index's loss, and the bond side 7.1, 40% of the bond's, so about two-fifths of the loss came from the part that was meant to soften it. One stock-heavy portfolio that reports its results in public shows what that year meant for a real fund.
Norway's sovereign wealth fund, with about 70% of its money in shares and 27.5% in bonds, saw both fall in 2022, which its chief executive called very unusual.
Norway's sovereign wealth fund, managed by Norges Bank Investment Management, describes itself as one of the world's largest funds.[10] At the end of 2022 it held about 70% of its money in shares and 27.5% in bonds, with the small remainder in unlisted real estate and renewable energy infrastructure, so it is not a 60/40, but it is built on the same idea. Measured in its own currency basket, the fund returned −14.1% in 2022: its shares lost 15.4% and its bonds 12.1%. Its chief executive, Nicolai Tangen, said the war in Europe, high inflation and rising interest rates "negatively impacted both the equity market and bond market at the same time, which is very unusual." Its figures differ from the 60/40 mix above in weights, currency and holdings, so the two cannot be set against each other. Of the causes Tangen named, rising interest rates are the one that public data let anyone take apart.
Over 2022 as a whole the 10-year Treasury yield rose 2.36 percentage points, and all of that rise was in the real yield on inflation-protected Treasuries.
Tangen spoke of rising interest rates in general; this section narrows that to one rate, the 10-year Treasury yield, which can be split into two parts using data the Federal Reserve publishes every trading day.[5] One part is the real yield on 10-year TIPS, Treasury bonds whose payments rise with consumer prices, so their yield is what an investor earns after inflation. The other part is the extra yield an ordinary bond pays over a TIPS bond, called the breakeven. It is a market measure of compensation for inflation, which also carries risk and liquidity premia, though Federal Reserve researchers report evidence for the view that it is "a useful proxy for inflation expectations".[11] Because the breakeven is defined as that gap, the two parts always add up to the ordinary yield, so the split shows which part moved but cannot show why.
In 2022 the 10-year yield rose 2.36 points; the real yield rose 2.62.
| Series | December 31, 2021 | December 30, 2022 |
|---|---|---|
| 10-year Treasury yield | 1.52% | 3.88% |
| Real yield on 10-year TIPS | −1.04% | 1.58% |
| Breakeven (peak 3.02% on April 21) | 2.56% | 2.30% |
Only days with a published yield are shown; weekends and the eleven weekdays without one take up no space.
10-year Treasury yield: 1.52% → 3.88% · Real yield on 10-year TIPS: −1.04% → 1.58% · Breakeven: 2.56% → 2.30%, peak 3.02% on April 21
Board of Governors of the Federal Reserve System (DGS10, DFII10) and Federal Reserve Bank of St. Louis (T10YIE), via FRED. The breakeven is by definition the nominal yield minus the TIPS yield.
The top edge of the band is the 10-year Treasury yield and the bottom edge is the real yield on 10-year TIPS, both in percent. The height of the band is the breakeven, so on every day the real yield plus the breakeven equals the 10-year yield. When the band shifts up without getting taller, the real yield rose; when it gets taller, the breakeven rose. The three short rulers measure the band's height, not the level of either yield, at the start of the year, at its thickest on April 21, and at the end.
Over 2022 as a whole, from the last trading day of 2021 to the last of 2022, the 10-year yield rose 2.36 percentage points. The real yield rose 2.62 points over the same days, from −1.04% to 1.58%, and the breakeven fell 0.26 points. So the real yield rose by more than the whole 10-year yield did.
Reading the yield band · 1 of 2
Compare the two ends of the band: the last trading day of 2021 on the left and the last of 2022 on the right. The top edge, the 10-year yield, rose 2.36 points and the bottom edge, the real yield, rose 2.62, so the real yield climbed further than the yield it is part of. That is possible only because the band ends a little thinner than it started: the two outer rulers read 2.56 and 2.30. Over the full year, then, the whole rise in the 10-year yield was in the real yield.
That full-year statement does not hold for every part of the year, because the full-year figures hide a turn in the spring. From the end of 2021 to April 21, 2022, the 10-year yield rose 1.38 points, and 0.46 of them were in the breakeven, which reached a peak of 3.02%; so in those months about a third of the rise was in the inflation part. After April 21 the 10-year yield rose another 0.98 points, so 1.38 and 0.98 make up the full year's rise; in those later months the breakeven fell by 0.72 points while the real yield climbed 1.70. Over the full year, then, the breakeven ended 0.26 points lower, but it got there by rising until April 21 and falling afterwards. A statement that inflation expectations fell in 2022 therefore describes only the months after April 21, and only as far as the breakeven, with its premia, measures expectations at all.
Reading the yield band · 2 of 2
Now look at April 21, 2022, where the band is at its thickest: the ruler there reads 3.02, the breakeven's peak for the year. Between the start of the year and that day both edges rose and the band grew taller, so part of the rise in yields in those months was in the inflation part. After April 21 the bottom edge, the real yield, rises further than the top edge, reaching its high for the year in early November, so the band grows thinner toward the end of the year. That turn is why a statement about the full year cannot be read as a statement about every month of it.
Two other facts set the scene. US consumer prices in June 2022 were 9.1% higher than a year earlier, the highest rate of the year, and the Federal Reserve raised the upper end of its target range for short-term interest rates in seven steps between March and December. Over the full year, the rise was in the real yield, which is consistent with markets demanding a higher real yield as the Fed raised its target range from a top of 0.25% to 4.50%. The data fit that reading without proving it, because the breakeven is an imperfect measure of what investors expected.
The same rise can be read on the stock side as well, but only as valuation logic: higher real yields also lower what future profits are worth today, which would push stock prices down at the same time as bond prices. This essay does not claim that this explains the fall in the S&P 500 in 2022, because nothing in our data splits that fall into causes. What the yield data do show, set beside the two decades of negative correlation before them, supports one general statement and no more: bonds tend to cushion stocks when the bad news is about growth; when it is about rising real yields, both can fall.
The three-year correlation between US stock and bond returns has been positive since August 2022, and was still positive in August 2026, the latest month in our data.
Whether the hedge has come back since 2022 can be checked up to August 2026. The three-year correlation turned positive in August 2022 and has stayed positive in every month since. A positive correlation means that over each recent three-year window the bond has tended to move with stocks rather than against them, so the hedge has not been there. Its highest reading after 2022 was 0.67, in December 2024, and it has been lower since then, but in August 2026, the latest month in our data, it was still 0.45. Between January 2022 and August 2026 the US stock market fell in 23 months, and our monthly 10-year bond fell too in 17 of them, 74%; these counts include the eight losing months of 2022, in seven of which the bond fell too. That share is higher than between 1970 and 1999, when it was about half. The hedge that ended in 2022 has not returned within the period we can measure, and the series cannot say whether or when it will.
Research suggests that bonds move against stocks largely when growth rather than inflation is the bigger source of surprises, so the protection bonds give a 60/40 mix depends on conditions, and whether it has been working can be measured.
In the AQR researchers' model the hedge depends on two conditions. The first is whether growth or inflation is the bigger source of surprises. The second is whether inflation rises and falls with the economy or against it, and this is where the study by Campbell, Pflueger and Viceira finds a break in 2001. On the first of these, the AQR researchers find that in the two decades before 2022, uncertainty about growth outweighed uncertainty about inflation, the condition under which, in their model, bonds tend to move against stocks. In 2022 the pressure came from inflation and rising interest rates, as the head of Norway's fund put it; in the narrower terms of the yield split, the rise in the 10-year yield over the full year was in the real yield. Set against the two decades before it, Tangen's "very unusual" fits the data: the correlation had been negative throughout that period, so a year in which both fell heavily ran against the pattern of that whole period.
The floor needs its own answer, and the arithmetic is plain. In our rebuilt 10-year bond, in six of the other eleven years (1930, 1940, 1957, 2001, 2002 and 2008) bond prices rose as yields fell. In three (1937, 1973 and 1974), prices fell but the interest the bond paid more than made up for it. In the remaining two, 1931 and 1941, the bond lost 2.6% and 2.0%. In 2022 the interest was 1.5% and the price fell by 19.3%. In arithmetic terms, then, the bond ended the year with a gain whenever its price rose or the interest outweighed the fall; in 1931 and 1941 the interest covered about half of the fall, and in 2022 less than a tenth of it. Our sources do not say why yields moved as they did in those earlier crash years, so the essay gives the arithmetic and stops there.
Both answers have limits. Our correlation line begins in 1965, so the hedge before then is known only from the studies cited, and the floor rests on two yearly series, Damodaran's and one built from Shiller's data, which agree about 2022 but not about every other year. Within those limits the evidence supports the point Markowitz wrote down in 1952: what a second holding is worth to a portfolio depends on how it moves with the first. A fixed 60/40 split assumes that the bond will move against stocks when it matters, and the three-year correlation shows, month by month, whether that assumption has held over the three years just past; in August 2026 it still had not.
Methodology
Yearly returns, the bond and the 60/40 mix. Our calculation from Aswath Damodaran's annual returns (NYU Stern, January 2026 release): 60% S&P 500 with dividends and 40% a constant-maturity 10-year US Treasury, rebalanced each January, before costs and taxes. Not the return of any fund.[1] The bond is a 10-year Treasury bond's return rebuilt from its yield: the coupon plus the price change. The coupon, called interest in the essay, is the start-of-year yield on a notional bond. Damodaran does not name his source for the yields of 1928–1952. For 1953–1998 he uses the December average yield as the year-end yield. Recomputed from the last daily yield of the year on FRED, the bond's 2022 return is −17.7% instead of −17.8%, and no ranking in the essay changes. For 2025 we use his spreadsheet, downloaded on 4 October 2026; it differs slightly from the table on his website, and no statement in the essay uses 2025. Values after inflation are adjusted for US consumer prices, December to December (BLS CPI-U, not seasonally adjusted, via FRED). We quote his figures rounded to 0.1 percentage point and do not offer them for download.
The twelve years, and what each bond return was made of. The 10% threshold is a convention, applied before rounding: 1966, when the S&P 500 lost 9.97%, which rounds to 10.0%, is not among the twelve. At 8%, seven more years qualify, 1966 among them; only one of them, 1969, has a bond loss, of 5.0%, and 2022 still stands alone. A cross-check built from Robert Shiller's data, which uses monthly average prices and so a different yearly window, gives a different set of twelve years, and on it too 2022 is the only one with a large bond loss.[9] The table splits our rebuilt bond's return in each of the twelve years into interest and price change, in percentage points; each figure is rounded on its own, so a row may not add up exactly.
| Year | Interest | Price change | Total |
|---|---|---|---|
| 1930 | 3.4 | +1.2 | +4.5 |
| 1931 | 3.2 | −5.8 | −2.6 |
| 1937 | 2.6 | −1.2 | +1.4 |
| 1940 | 2.4 | +3.1 | +5.4 |
| 1941 | 2.0 | −4.0 | −2.0 |
| 1957 | 3.6 | +3.2 | +6.8 |
| 1973 | 6.4 | −2.7 | +3.7 |
| 1974 | 6.7 | −4.8 | +2.0 |
| 2001 | 5.1 | +0.5 | +5.6 |
| 2002 | 5.1 | +10.1 | +15.1 |
| 2008 | 4.0 | +16.1 | +20.1 |
| 2022 | 1.5 | −19.3 | −17.8 |
Where 2022 ranks. On our 60/40 mix, before inflation, 2022 is the third-worst year since 1928; on the construction built from Shiller's data it ranks fifth, which is why the essay gives the rank only with its construction. After inflation, 1974, 1937 and 2022 are each among the four worst years for the mix since 1928 on every construction we checked; 1974 or 2022 comes first, depending on the data and on how inflation is measured. In the table, rank 1 is the worst year for the mix after inflation, 1928–2025.
| Construction | 1974 | 1937 | 2022 |
|---|---|---|---|
| Damodaran, consumer prices December to December | 1 | 3 | 2 |
| Damodaran, calendar-year average consumer prices | 3 | 2 | 1 |
| Shiller, consumer prices December to December | 1 | 2 | 4 |
On the Shiller construction 1946 ranks third.
The correlation and the monthly counts. Our calculation: 36-month rolling Pearson correlation of monthly returns. Stocks: US total market (CRSP, via the Kenneth French Data Library). Bonds: constant-maturity 10-year Treasury built from FRED DGS10 month-end yields. Series starts January 1965.[4][5] Each month is the last month of its window. We publish only statistics derived from the CRSP returns, never the returns themselves. Another window or Spearman's rank correlation gives other numbers. As a check, averaged over 36-month windows, our series comes within 0.02 of the average correlations that Molenaar and co-authors report for 1970–1999 and for 2000 to June 2023.[7] Their picture since 1875 rests on monthly data sold under license by Global Financial Data, a commercial provider, so we cite their results and do not recompute them.
The 2022 yield split. Board of Governors of the Federal Reserve System (DGS10, DFII10) and Federal Reserve Bank of St. Louis (T10YIE), via FRED. The breakeven is by definition the nominal yield minus the TIPS yield. The split therefore tests nothing.[5] Federal Reserve researchers note that "a time-varying inflation risk premium complicates the interpretation of the TIPS breakeven inflation rate"; their samples end before 2022, so we give no size for 2022.[11] The table dates each 2022 rise in the Fed's target range (DFEDTARU) as FRED records it, a day after the decision.
| Date in 2022, as recorded by FRED | Upper end of the target range |
|---|---|
| Start of the year | 0.25% |
| March 17 | 0.50% |
| May 5 | 1.00% |
| June 16 | 1.75% |
| July 28 | 2.50% |
| September 22 | 3.25% |
| November 3 | 4.00% |
| December 15 | 4.50% |
Labels and omissions. Calculated by us: the 60/40 mix, the inflation rates and all values after inflation, the two-year returns for 1973–74, all ranks, the yield changes, the correlation, the monthly shares and the split of the bond's return into interest and price. Reported by the source: Damodaran's yearly returns, the Federal Reserve and BLS data, and NBIM's figures, which are in the fund's currency basket. Claims about the history of the 60/40 that we could not confirm in a primary source are left out.
Sources
- Aswath Damodaran, "Historical Returns on Stocks, Bonds and Bills", NYU Stern,
histretSP.xls, January 2026 release (downloaded 4 October 2026). - Harry Markowitz, "Portfolio Selection", The Journal of Finance 7(1), March 1952, 77–91.
- Jason Zweig, "Investing Lessons from America's Top Pension Fund", Money, January 1998, as republished by the author in "What Harry Markowitz Meant", jasonzweig.com, 2 October 2017.
- Kenneth R. French Data Library, Fama/French research factors, monthly ("created using the 202608 CRSP database"). Copyright Eugene F. Fama and Kenneth R. French.
- Board of Governors of the Federal Reserve System: DGS10, DFII10, DFEDTARU; Federal Reserve Bank of St. Louis: T10YIE; US Bureau of Labor Statistics, Consumer Price Index for All Urban Consumers, not seasonally adjusted (CPIAUCNS); all via FRED, Federal Reserve Bank of St. Louis.
- Brixton, Brooks, Hecht, Ilmanen, Maloney and McQuinn (AQR), "A Changing Stock–Bond Correlation: Drivers and Implications", The Journal of Portfolio Management 49(4), 2023.
- Molenaar, Sénéchal, Swinkels and Wang, "Empirical Evidence on the Stock–Bond Correlation", Financial Analysts Journal 80(3), 2024, 17–36, doi:10.1080/0015198X.2024.2317333 (CC BY 4.0).
- Campbell, Pflueger and Viceira, "Macroeconomic Drivers of Bond and Equity Risks", NBER Working Paper 20070, April 2014, revised August 2018.
- Robert J. Shiller,
ie_data.xls(online data), used as an independent cross-check only. - Norges Bank Investment Management, "Negative returns in a challenging year", press release, 31 January 2023; Government Pension Fund Global, Annual Report 2022; nbim.no, "About the fund" (Internet Archive capture of 6 February 2023).
- D'Amico, Kim and Wei, "Tips from TIPS: the informational content of Treasury Inflation-Protected Security prices", Federal Reserve Board, Finance and Economics Discussion Series 2008-30 and 2014-24 (abstracts).
How to cite this essay
Tradiens Research. (2026). The year stocks and bonds both fell heavily (Version 2.0). Tradiens Data Essays. https://tradiens.com/essays/death-of-60-40
@misc{tradiens2026deathof6040,
author = {{Tradiens Research}},
title = {The Year Stocks and Bonds Both Fell Heavily},
year = {2026},
version = {2.0},
howpublished = {Tradiens Data Essays},
url = {https://tradiens.com/essays/death-of-60-40}
}