What a 60/40 Portfolio Really Owns

Two investors both say they own a 60/40 portfolio. Both are telling the truth. In 2008 one of them lost 13.9% and the other lost 23.3%.

What separated them was the composition of the 40. Using a single consistent dataset of annual total returns, here is what a 60% S&P 500 portfolio did in the two most instructive recent years, depending on whether the bond sleeve held ten-year Treasuries or ten-year Baa corporate bonds.¹

In 2008 the Treasury sleeve was worth 9.4 percentage points. In 2022 it cost 1.1. A single number cannot describe a bucket that behaves in opposite directions in the two worst years of the last two decades.

Reasonable people put credit risk in that bucket on purpose, and they have evidence on their side. The problem starts when the second number is treated as a complete description of the defensive position, and it gets worse when someone else is managing the money against it.

There is no canonical 60/40

Start with two mainstream implementations that both qualify for the label.

Both are 60/40. One holds nothing outside the United States and the other puts 36% of the whole portfolio overseas. That is a large difference in currency exposure, in economic exposure, and in what a dollar crisis would do to the account, and the label carries none of it. In 2022 the two funds lost 16.97% and 16.00% respectively, a gap of roughly a percentage point between products that answer to the same two digits.²

What “40% bonds” resolves to

The most common answer is a fund tracking the Bloomberg U.S. Aggregate, which is where the ambiguity gets concrete. As of August 20, 2026, the iShares AGG portfolio was 46.4% Treasuries, 23.0% agency mortgage-backed pass-throughs, and 24.8% corporate credit split across industrials, financials and utilities, with the remainder in CMBS, sovereign, supranational, agency and asset-backed paper. Effective duration was 5.70 years.³

Put that inside a 60/40 and the portfolio holds roughly 18.6% Treasuries, 9.2% agency MBS, and 9.9% corporate credit, with about 2.3% spread across everything else. Those weights are set by who happens to be issuing debt, which is a reasonable way to build an index and a strange way to size your own defensive position.

The index rules matter as much as the current weights. The Aggregate requires investment-grade ratings and fixed-rate coupons, and it explicitly excludes inflation-linked bonds, floating-rate issues, and tax-exempt municipals.⁴ A portfolio whose entire defensive sleeve is the Aggregate holds exactly zero explicit inflation protection. That is a design choice, and almost nobody who says “60/40” has made it consciously.

The history has the same problem. The Aggregate begins in 1976, so any multi-decade 60/40 chart splices something onto the front of it.⁴ One Vanguard research series uses the S&P High Grade Corporate Index for 1926 to 1968, the Citigroup High Grade Index for 1969 to 1972, and the Lehman Brothers U.S. Long Credit AA Index for 1973 to 1975. Their retail model-portfolio page splices a different set, running intermediate-term government bonds through 1972 instead.⁵ Both are defensible ways to build a long series. Both also mean a century-long chart labeled 60/40 does not hold the same 40 throughout, and in this case the pre-1976 segment is long-dated corporate credit in one version and government bonds in the other.

Where the risk sits

Sixty percent of the capital is not sixty percent of the risk. Because equities are several times more volatile than high-quality bonds, the equity sleeve dominates the variance of a conventional balanced portfolio, which is the observation the entire risk-parity literature starts from.⁶

How badly the capital weights understate the equity share depends on the correlation regime. Molenaar, Sénéchal, Swinkels and Wang measured US stock-bond correlations across decades and found the sign flips: positive through the 1970s to 1990s, negative through the 2000s and 2010s. Holding return and volatility assumptions constant, their stylized 60/40 carried about 10.5% annual volatility under the positive-correlation regime and about 8.4% under the negative one. Their variance decomposition of the same portfolio makes it sharper: the equity sleeve supplied 71% of variance over 1970 to 1999 and 111% over 2000 to mid-2023, with the stock-bond covariance term swinging from +18% to -25%.⁷ A contribution above 100% is not a typo. Negative covariance subtracts from total variance, so the equity sleeve can account for more than all of it. Identical capital weights, materially different portfolios.

We have written that argument up separately in The Risk Parity Reality Check, which walks the risk-contribution math and the case for and against acting on it. Take that as given. The question here is what sits inside the 40.

Why a corporate bond is not a Treasury with a better coupon

Merton gave the structural answer in 1974. He opens by naming the three things a corporate bond’s value rests on: the required return on riskless debt, the provisions of the indenture, and the probability the firm cannot satisfy them. For a given maturity, he then shows, the risk premium reduces to two variables, the volatility of the firm’s operations and the ratio of the promised payment’s present value to the current value of the firm.⁸ The bondholder is short the firm doing badly, on the same underlying the equity sleeve is long.

Merton also describes the region every real bond lives in, between riskless and ruined:

“Between these two extremes, the debt will behave like a combination of riskless debt and equity, and will change in a continuous fashion.”

Merton (1974), Section V.⁸

Fama and French put a number on where along that continuum the equity behavior starts. They identified term and default factors in bond returns that also capture common variation in stock returns, then tested whether the stock-market factors add anything on top:

“Except for low-grade corporate bonds, the stock-market factors have little role in returns on government and corporate bonds.”

Fama and French (1993), conclusion. Their below-Baa portfolio carried a market beta of 0.30 and the market factor explained 19% of its return variance, against high-grade corporates where the term-structure factors explained nearly everything.⁹

An investment-grade corporate bond and a share of stock in the same company are different instruments. A high-yield bond sits much further along Merton’s continuum toward the share, which is worth knowing before it goes into a bucket labeled “bonds.”

The case for owning credit anyway

Rick Ferri has made this argument for years, and made it well. Writing on Bogleheads in 2015 he put it directly:

“Some investors mistakenly view credit risk as an extension of equity risk. In fact, prices of corporate bonds are not a simple mirror image of equity risk, thus an analysis of corporate bond returns requires more than a simple extension from equity returns.”

Rick Ferri, Bogleheads forum, April 6, 2015.¹⁰

He is right, and the evidence has strengthened since he wrote it. Ghaderi, Plante, Roussanov and Seo reconstructed the US corporate bond market from 1895 to 2022, hand-collecting archival quotes into a panel of 101,395 bonds and more than seven million bond-month observations. Recent samples make corporate bond excess returns look like mostly term premium. Their long sample turns up a sizable and statistically significant credit risk premium, and they attribute its disappearance in shorter studies to small-sample limits and duration-matching problems rather than to an absence of compensation.¹¹ Investors have been paid for taking credit risk.

Vanguard reached a similar verdict from a different angle. Decomposing bond-portfolio volatility into an interest-rate component and a credit component, they found a 100% U.S. aggregate bond sleeve is about 96% interest-rate risk and 4% credit risk, and that credit excess returns have been negatively correlated with Treasury returns, minus 0.30 for intermediate credit over January 1994 to July 2022. Their conclusion is that credit exposure diversifies Treasuries. They pair it with a constraint worth stealing: cap credit at no more than 25% of the bond sleeve’s volatility, so that most of the sleeve’s risk still comes from interest rates.¹²

Our own data agrees, and it agrees more strongly than we expected. Because Damodaran builds the ten-year Treasury and the Baa corporate series the same way, the comparison below holds duration fixed and varies only credit. Over 1928 to 2025, at 60% equities:

The obvious rebuttal to a credit sleeve is that you could take the extra risk in the equity sleeve instead and keep the bonds genuinely defensive. Over this sample, that rebuttal loses. Dialing equities up to 66% with a pure Treasury sleeve reproduced the Baa version’s volatility and delivered 0.39 percentage points a year less. Matching its return with Treasuries alone would have taken roughly a 74/26 split, which came with 14.5% volatility and a 50% drawdown against the credit sleeve’s 13.1% and 44.8%. On this data, credit bought its extra return more efficiently than more stock did.

One caveat has to travel with those numbers. Damodaran’s Baa series is a yield index repriced each year at constant ten-year maturity, so it does not subtract realized default and downgrade losses.¹ The advantage shown is an upper bound. We can put a number on how much of it is available to lose: the volatility-matched comparison above breaks even at about 1.0 percentage point a year of credit losses on the sleeve. Whether the real sleeve cleared that bar is the question Ghaderi and coauthors answer with actual bond-level returns, and their answer is that a credit premium survives.¹¹

The cost shows up in one specific state

Averages and standard deviations are not what a defensive sleeve is for. Split the same 98 years by what equities did.

The credit sleeve beat Treasuries by 3.20 points in the average year stocks rose and trailed by 1.02 points in the average year stocks fell. After inflation, the credit sleeve’s average return in down-equity years is negative. The premium is real, and it is delivered on a schedule that is correlated with when you would rather not be paying for it.

That is what a risk premium looks like from the inside. The extra yield is compensation, and compensation implies a state of the world in which you hand it back.

Three crises, three different lessons

2008: credit

The opening table covers it. Treasuries rallied hard as investors reached for safety, corporates did not, and the two portfolios diverged by 9.4 percentage points while carrying the same label.

March 2020: liquidity

Investment-grade credit does not need large default losses to stop acting like ballast. Haddad, Moreira and Muir document that in the first half of March 2020 the cumulative return on investment-grade corporate debt was about -20%, roughly as much as the stock market over the same stretch, and that over the three weeks from March 1 to March 20 investment grade fell about as far as high yield and as far as equities, with investment grade suffering slightly larger losses.¹³

Their evidence points away from a pure default story. The CDS-bond basis for investment-grade companies widened to around 300 basis points, and bond ETFs traded at discounts to net asset value averaging around 5% across several categories of debt, with the discounts larger for the safer and more liquid instruments. Investors were selling what they could sell, and the safest holdings are the easiest to sell.¹³ The disruptions reversed quickly once the Federal Reserve announced it would buy corporate debt.

Liquidity is a third dimension, separate from credit and from duration, and it goes wrong at the worst possible moment by construction. Anything in the bond sleeve that is priced infrequently, including private credit, inherits this problem while reporting a smoother return series than it has.

2022: rates and inflation

This is the year that stops the guide from being an argument for Treasuries. Nominal government bonds hedge a deflationary recession. An inflation shock raises discount rates on stocks and bonds together, and in 2022 the ten-year Treasury lost 17.8% while the S&P 500 lost 18.0%. The credit sleeve, with less duration risk per unit of yield, lost less. The defensive sleeve and the growth sleeve failed together, and the bucket that held credit was the one that held up.

Safe against what?

By this point “safe bonds” has stopped being a useful category, because each candidate is safe against a different thing.

The Aggregate holds several of these at once, in weights set by issuance patterns rather than by anything about your situation. If you want an inflation-protected sleeve you have to add it, because the index rules exclude inflation-linked bonds.⁴ We cover the mechanics of matching bond duration to the spending it has to cover in Asset-Liability Matching and How to Build a TIPS Ladder.

What the label does for a manager who is paid on returns

Everything so far applies to an individual choosing a bond fund. The problem sharpens when someone else is choosing on your behalf against a category constraint.

Consider the structure. A manager is told to hold 40% fixed income. The manager is evaluated on return. Fixed income is defined by category, not by economic exposure. Every unit of credit risk, duration, leverage, or illiquidity added inside that 40% raises expected return without violating the letter of the constraint. The mandate is satisfied on paper and drained of meaning underneath.

This is not hypothetical, and it has been measured. Becker and Ivashina studied insurance companies, whose regulatory capital requirements depend on credit-rating categories. Conditional on rating, insurer portfolios were systematically biased toward the higher-yielding, higher-CDS-spread bonds available within each bucket. The behavior was stronger for firms whose capital requirements bound more tightly. The bonds they preferred did not perform better afterward. They carried higher volatility and higher exposure to market returns, which is the authors’ evidence that the extra yield was payment for risk.¹⁴ The constraint was a rating category, so the optimization happened inside the rating category.

Choi and Kronlund found the same pattern in corporate bond mutual funds, which tilt toward bonds yielding more than their benchmarks. Funds that reached for yield generated higher returns and attracted more inflows, especially when rates were low, and their risk-adjusted returns nonetheless tended to be negative. They also held less cash and less liquid bonds, which made them more fragile to redemptions.¹⁵ Reaching for yield looked like skill in the flows and like risk in the attribution.

Two things this evidence does not show. It does not show that retail financial advisers routinely hide equity beta from clients to preserve a 60/40 label. Both papers study delegated institutional portfolios, insurers and mutual funds, under explicit category constraints. And it does not show that credit exposure is illegitimate. What it shows is that a coarse category constraint plus a return-based incentive reliably produces more risk inside the category, and a 60/40 mandate is a coarse category constraint.

The fix is to write mandates in terms of the exposures you care about. If the bond sleeve exists to cushion a growth shock, say so, and cap spread duration, below-investment-grade holdings, and illiquid allocations explicitly. If you only write down the percentage, the percentage is the only thing you have agreed on.

Look inside your own 40

The calculator below holds the headline allocation fixed and lets you change what the bond sleeve holds. The metric worth watching is the sleeve’s average return in the years stocks fell, which is the state the sleeve exists for.

What we recommend

Own credit if you want it. Own an aggregate bond fund if you want that. What we would not do is treat the second number as a complete description of your defensive position. A short checklist:

1. Give the sleeve a job. Cushioning a recession, matching a known liability, protecting against inflation, and providing liquidity are four different jobs with four different right answers. You cannot decide whether credit belongs until you know which one you are hiring for.
2. Look through the fund to the exposures. Treasury share, credit quality, effective duration, spread duration, MBS convexity, inflation linkage, currency, leverage, and liquidity. Every fund publishes these. “40% bonds” publishes none of them.
3. Test against scenarios as well as volatility. Run 2008, March 2020, and 2022 separately. They rank bond sleeves in different orders, which is the point. Volatility alone can be gamed, especially by anything with appraisal-based marks.
4. Price credit against the alternative of holding more equity. If the reason for corporates is expected return, compare it explicitly to a Treasury sleeve with a higher stock weight. Over 1928-2025 credit won that comparison on this data, but it is the comparison that forces the trade into the open, and over your horizon it may go the other way.
5. Write mandates around exposures. Constrain the economics you care about, not only the bucket percentages, or you have created an incentive to optimize inside the bucket.
6. Keep asset location separate from asset allocation. Treasury interest is taxable federally but exempt from state and local income tax, so there can be a good tax reason to shelter corporates first and hold Treasuries in taxable accounts. That changes your after-tax return. It does not change either asset’s economic risk, and it is not a reason to hold one instead of the other.

When this argument does not apply

A broad aggregate bond index fund is cheap, diversified, conventional, and entirely reasonable for most households. If your bond sleeve is 20% of a portfolio you will hold for thirty years, the difference between an Aggregate fund and an all-Treasury fund is small next to your savings rate and your equity allocation.

Simplicity that someone will actually stick with also has real value. An investor who understands and holds “60/40” through a bad year is better off than one who abandons a more precisely specified portfolio in month four. Compression is useful. The failure mode is forgetting that it is compression.

The argument bites hardest in three situations: when the bond sleeve is large because you are near or in retirement and it is doing real spending work; when someone else manages the money against a percentage mandate; and when the sleeve holds anything priced infrequently, where the reported volatility is not measuring what you think.

Key takeaways

• A 60/40 label fixes the capital split and leaves the economics open. Currency, duration, credit quality, inflation linkage, and liquidity can all change without the label moving.
• Credit is a compensated risk. The long-run credit premium holds up in bond-level data going back to the nineteenth century, and over 1928-2025 a Baa sleeve delivered more return per unit of volatility than raising the stock weight did.
• The compensation arrives on a correlated schedule. In that same sample the Baa sleeve beat Treasuries by 3.20 points in the average up-equity year and trailed by 1.02 points in the average down-equity year, and its average real return in down-equity years was negative.
• 2008, March 2020, and 2022 rank bond sleeves in different orders. Credit risk, liquidity risk, and inflation risk are three separate exposures, and no single bond percentage tells you which ones you own.
• A percentage mandate plus a return incentive produces risk inside the bucket. Insurers and bond mutual funds both reach for yield within their constraints. Constrain the exposures you care about, not only the weights.

How Summitward helps

Frequently asked questions

Is a 60/40 portfolio with corporate bonds riskier than one with Treasuries?

On the historical record, modestly and in a specific way. Over 1928-2025, substituting a Baa corporate sleeve for a ten-year Treasury sleeve in a 60/40 raised volatility from 12.06% to 13.11% and deepened the worst drawdown from 40.1% to 44.8%. The more useful difference is conditional: the credit sleeve averaged 1.02 percentage points less than Treasuries in the years the S&P 500 fell, and more in the years it rose.

Does a total bond market fund count as Treasuries?

No. A fund tracking the Bloomberg U.S. Aggregate holds Treasuries alongside agency mortgage-backed securities, investment-grade corporate credit, and asset-backed paper, in weights determined by who is issuing. It also excludes inflation-linked bonds entirely, so it provides no explicit inflation protection.

Should I just hold Treasuries and more stocks instead of corporates?

It is the right comparison to run, and on the long US record it did not win. Matching the Baa sleeve’s volatility with an all-Treasury sleeve took about a 66/34 split and gave up 0.39 percentage points of annual return. The case for the swap is not that it earns more; it is that it puts your risk where you can see it and keeps the defensive sleeve genuinely defensive.

Why did bonds and stocks both fall in 2022?

Nominal government bonds hedge a growth shock. In an inflation shock, discount rates rise for both stocks and bonds at once, so the hedge goes the wrong way. Longer duration made that worse, which is why the ten-year Treasury lost more in 2022 than a shorter-duration credit sleeve did. Are Bonds Still Good Diversifiers? works through the correlation regimes in detail.

What does “reaching for yield” mean?

Buying the riskiest securities available inside a category you are required to stay in. An insurer constrained by credit-rating buckets buys the highest-yielding bonds within each rating; a bond fund benchmarked to an index buys bonds yielding more than the index. Both raise expected return without technically breaching the constraint, and both have been documented in the academic literature.

How should an institution write a 60/40 mandate so it cannot be gamed?

By constraining exposures rather than only weights. Set limits on effective duration and spread duration, on the share below a given credit rating, on leverage, and on illiquid or appraisal-priced holdings, and add scenario-based risk limits alongside the volatility target. A volatility target on its own is also gameable, because infrequently priced assets report artificially smooth returns.

Related guides

• Are Bonds Still Good Diversifiers? It Depends What Risk You’re Hedging: the correlation regimes behind why bonds sometimes rally in a crash and sometimes do not.
• The Risk Parity Reality Check: the risk-contribution math showing how much of a 60/40’s variance the equity sleeve supplies.
• 60/40, Target-Date Funds, or 100% Stocks Forever?: how much bond exposure to hold, once you know what kind you want.
• Asset-Liability Matching: choosing bond duration from the spending it has to cover.
• Compensated vs. Uncompensated Risk: where credit sits among the risks that pay you and the ones that do not.
• Every Portfolio Is Long/Short: the same gap between what a portfolio is called and what it is exposed to, applied to equities.

Sources

1. Damodaran, A. “Historical Returns on Stocks, Bonds and Bills: 1928-Current”. NYU Stern. The source for every figure in this guide computed from 1928-2025 annual data. Damodaran computes both the ten-year Treasury and the Baa corporate return as the annual coupon plus the price change on a bond whose maturity is held constant at ten years, so the two series are duration-matched and the Baa series does not subtract realized default or downgrade losses.
2. Vanguard. LifeStrategy 60/40 Fund (VSMGX) fact sheet (June 30, 2026) and Balanced Index Fund (VBIAX) summary prospectus (April 28, 2026). The source for the 36/24/28/12 LifeStrategy target allocation, for the Balanced Index Fund tracking the CRSP US Total Market Index and the Bloomberg U.S. Aggregate Float Adjusted Index with no international equity sleeve, and for the 2022 returns of -16.00% (VSMGX) and -16.97% (VBINX). VSMGX was formerly named LifeStrategy Moderate Growth.
3. iShares. Core U.S. Aggregate Bond ETF (AGG) product page. The source for the sector breakdown and the 5.70-year effective duration, both as of August 20, 2026, and for the benchmark Bloomberg U.S. Aggregate 2022 return of -13.01%. Sector weights move with issuance, so check the current figures before relying on them.
4. Bloomberg. “Bloomberg US Aggregate Index” (June 2024). The source for the index’s eligible sectors, its investment-grade and fixed-rate requirements, its exclusion of inflation-linked bonds, floating-rate issues and tax-exempt municipals, and its history back to January 1, 1976.
5. Vanguard. “Constructing return-target portfolios: A time-varying, valuation-aware approach” (October 2023), and Vanguard’s model portfolio allocation page. The source for the two different pre-1976 bond splices described above.
6. Asness, C., Frazzini, A., & Pedersen, L.H. (2012). “Leverage Aversion and Risk Parity”. Financial Analysts Journal, 68(1), 47-59. The source for the claim that a conventional balanced portfolio is dominated by equity risk despite its capital weights.
7. Molenaar, R., Sénéchal, E., Swinkels, L., & Wang, Z. (2024). “Empirical Evidence on the Stock-Bond Correlation”. Financial Analysts Journal, 80(3), 17-36. The source for the 10.5% versus 8.4% stylized 60/40 volatility figures and for the variance decomposition in their Table 5. Each variance component is computed over 36-month rolling windows and averaged within each period, and the sample ends in June 2023.
8. Merton, R.C. (1974). “On the Pricing of Corporate Debt: The Risk Structure of Interest Rates”. The Journal of Finance, 29(2), 449-470. The source for the three determinants of a corporate bond’s value, for the risk premium reducing to firm volatility and the ratio of promised payment to firm value, and for the quoted sentence from Section V. The familiar “riskless bond minus a put” restatement of this model does not appear in the 1974 paper, which works in terms of a call on the firm.
9. Fama, E.F., & French, K.R. (1993). “Common Risk Factors in the Returns on Stocks and Bonds”. Journal of Financial Economics, 33(1), 3-56. The source for the term and default bond factors, for the quoted conclusion, and for the below-Baa portfolio’s 0.30 market beta and 19% explained variance. Sample is July 1963 to December 1991.
10. Ferri, R. (2015). “Is there Smart Beta in Corporate Bonds?” Bogleheads forum, April 6, 2015. The source for the quoted position that corporate bond prices are not a mirror image of equity risk.
11. Ghaderi, M., Plante, S., Roussanov, N., & Seo, S.B. (2026). “Reconstructing a Century of U.S. Corporate Bonds: Credit Risk in Historical Perspective”. NBER Working Paper 35578, August 2026. The source for the 1895-2022 sample, the 101,395 bonds and 7.3 million bond-month observations, the sizable and statistically significant long-run credit risk premium, and the attribution of its absence in modern samples to small-sample and duration-matching problems. Working paper rather than peer-reviewed canon, so read it as strong recent evidence rather than settled. Elements previously circulated under the title “Pricing of Corporate Bonds: Evidence From a Century-Long Cross-Section.”
12. Maciulis, V., DiCiurcio, K., & Tan, Z. (2023). “Credit risk premia: Considerations for multiasset portfolios”. Vanguard research, April 2023. The source for the 96% versus 4% interest-rate and credit split of aggregate bond volatility, the -0.30 correlation between intermediate credit excess returns and Treasury total returns over January 1994 to July 2022, the conclusion that credit diversifies Treasuries, and the 25% cap on credit’s contribution to bond-sleeve volatility. Linked to an archived copy because Vanguard has taken the original offline.
13. Haddad, V., Moreira, A., & Muir, T. (2021). “When Selling Becomes Viral: Disruptions in Debt Markets in the COVID-19 Crisis and the Fed’s Response”. Review of Financial Studies, 34(11), 5309-5351; NBER Working Paper 27168. The source for the -20% cumulative investment-grade corporate return in the first half of March 2020, the roughly 300 basis point CDS-bond basis, and the roughly 5% average bond-ETF discounts to net asset value.
14. Becker, B., & Ivashina, V. (2015). “Reaching for Yield in the Bond Market”. The Journal of Finance, 70(5), 1863-1902. The source for insurers holding higher-yield, higher-CDS bonds conditional on credit rating, more so where capital requirements bind, and for the finding that the bonds they preferred carried higher volatility and higher market exposure without better returns.
15. Choi, J., & Kronlund, M. (2018). “Reaching for Yield in Corporate Bond Mutual Funds”. The Review of Financial Studies, 31(5), 1930-1965. The source for reaching funds generating higher returns and attracting more inflows, especially in low-rate periods, while their risk-adjusted returns tend to be negative and they hold less cash and less liquid bonds.