In this article
← All articles
Investment allocation

Howard Marks on defensive asset allocation


Howard Marks frames defensive asset allocation around the balance between owning and lending, and since December 2022 he has argued that higher rates justify far more credit, whose prospective returns he put at roughly 7% for public and 10% for private credit in October 2024. The co-founder of Oaktree Capital Management sets out this view in his public memos.

Key takeaways

  1. Marks reduces asset allocation to ownership versus debt. He calls the choice of a target risk posture the most important portfolio decision.
  2. He changes that posture rarely. He counts five market calls in 50 years and describes Oaktree's normal stance as biased toward defensiveness.
  3. Sea Change argued that higher rates made credit competitive with equities. By May 2023 he wrote that credit should probably be a substantial portion of portfolios, perhaps the majority.
  4. He has warned about valuations without calling for an exit. He cited a forward P/E of 22 in January 2025 and advised a moderate position on AI in December 2025.
  5. Contractual income is not a contractual crash payout. Copio assesses credit after defaults, fees and liquidity costs, against the objective of maximum expected whole-portfolio geometric growth.

What is Howard Marks's approach to asset allocation?

Marks argues that there are only two asset classes, ownership and debt, and that the mix between them should put a portfolio at the level of risk its owner wants. He set this out in Ruminating on Asset Allocation on 22 October 2024.

Owners, he writes, put money at risk with no promise of a return and share whatever profit is left after everyone else is paid. Lenders receive a contractual stream of interest and principal, which he suggests thinking of as "fixed outcome" investing. He calls the gap between them "a difference in kind".

Questions such as domestic versus foreign or public versus private are "just details" to Marks. What matters most is "the selection of a targeted 'risk posture'", the desired balance between aggressiveness and defensiveness, chosen consciously rather than left as a by-product.

Ownership has a higher expected return and a wider range of outcomes, while debt offers a tighter range with little upside: "no one should buy an 8% bond expecting to make more than 8% per year over the long term." Offence is usually played through ownership and defence through debt, and the blend depends on the investor's horizon, finances, needs and "intestinal fortitude".

What is a risk posture, and how often does Marks change it?

A risk posture is an investor's normal balance between aggressiveness and defensiveness, and Marks says it should stay fixed most of the time. In Taking the Temperature (10 July 2023) he writes that departures make sense only when circumstances are compelling and the odds of being right are high, "But such occasions are rare."

He lists five such calls: January 2000, late 2004 to mid-2007, late 2008, March 2012 and March 2020. Each came when markets were "either crazily elevated or massively depressed". After Lehman Brothers failed, Bruce Karsh's team invested about $400m a week from 18 September 2008 to year-end, $6bn in total, and Oaktree as a whole $7.5bn.

Oaktree's normal balance, he writes, "implies a bias toward defensiveness". Its investment philosophy puts "the primacy of risk control" first among six tenets and states that market concerns may tilt portfolios toward defensive investments, "but we never move to raise cash". Compare that with Seth Klarman's approach to cash at Baupost.

Adjustments are gradual. In Calibrating (6 April 2020), Marks noted that high-yield yields excluding energy had risen from about 3.5% to almost 9% and wrote that he no longer felt defence should be favoured. He described the process as "gradual readjustment, not a matter of all-or-nothing".

What did Howard Marks's Sea Change memo say?

Sea Change, published on 13 December 2022, argued that the end of ultra-low interest rates could be the third sea change in Marks's 53-year career. The first two were the rise of high-yield bonds from the late 1970s, with the risk-for-return mindset that came with it, and the four-decade fall in rates that followed Paul Volcker's fight against inflation.

He called 2009-2021 "an asset owner's market and a borrower's market", in which Oaktree spent 2009 through 2019 "in the wilderness". By late 2022 he judged that investors could "get solid returns from credit instruments" without leaning as heavily on riskier assets.

The yields and prospective returns Marks cited across his memos trace the case:

MemoDateHigh-yield yield or credit return cited
Sea ChangeDecember 20224-5% a year earlier; roughly 8% at the time
Further Thoughts on Sea ChangeMay 2023over 8.5% (ICE BofA US High Yield Constrained Index)
Ruminating on Asset AllocationOctober 2024prospective returns of roughly 7% on public and 10% on private non-investment grade credit
Gimme CreditMarch 2025just above 7%, down from roughly 9.5% in 2022

Further Thoughts on Sea Change, sent to clients on 30 May 2023, went further. Marks wrote that credit instruments "should probably represent a substantial portion of portfolios . . . perhaps the majority", and that "significant reallocation of capital toward credit is warranted."

That memo also contains his best-known provocation: telling a non-profit's investment committee in December 2022 to sell all its stocks, private equity and real estate and buy high-yield bonds at 9%. It is sometimes repeated as a recommendation, but Marks wrote in the same passage that "mine wasn't a serious suggestion". He added: "This isn't a call for dramatically increased defensiveness."

He described it instead as a reallocation of capital away from ownership and leverage and toward lending. For how rate rises hit bond prices first, see bonds when interest rates rise.

Does high-yield credit pay enough for its risk?

Marks's answer in Gimme Credit (6 March 2025) was that the historical spread over Treasurys had more than covered credit losses. Citing Barclays data, he wrote that high-yield bonds returned 7.83% a year from 1986 through 2024, against 5.14% for 10-year Treasuries. Defaults averaged 3.5% a year over that period and cost about two-thirds of the money at stake, or roughly 2.3 percentage points a year.

At the time of the memo, the spread was around 290 basis points, one of the narrowest on record. To test whether a tight spread doomed buyers, he and Oaktree's Nicole Adrien looked at returns after the historical low cited in the memo, 241 basis points in June 2007.

Returns after the tightest high-yield spread. Annualised, bought June 2007 | Gimme Credit, March 2025. 1 year: high yield -1.13%, Treasurys 10.19%; 3 years: high yield 5.29%, Treasurys 7.27%; 5 years: high yield 7.26%, Treasurys 7.20%; 10 years: high yield 7.35%, Treasurys 4.15%; 15 years: high yield 6.01%, Treasurys 3.03%.

Annualised returns from the June 2007 spread low, ICE BofA US High Yield Index versus ICE BofA US Treasury Index: 1 year -1.13% vs 10.19%; 3 years 5.29% vs 7.27%; 5 years 7.26% vs 7.20%; 10 years 7.35% vs 4.15%; 15 years 6.01% vs 3.03%.

Source: Howard Marks, Gimme Credit, 6 March 2025, table "Annualized Returns Following All-Time Tight U.S. High Yield Bond Spread" (data: ICE, Bloomberg); chart data (CSV). It covers one starting date, the spread low cited in the March 2025 memo, before the 2008 crisis. It does not show other entry points, fees or any manager's portfolio.

Marks also pointed to quality: the share of the high-yield universe rated BB rose from 32.7% at the end of 1999 to 52.6% at the end of 2024, on ICE data. He concluded that concern about narrow spreads was "very much overblown".

What did Howard Marks say about stock valuations and AI?

Marks has warned that US stocks look expensive and that AI shows bubble traits, but he has argued for moderation rather than exit. In On Bubble Watch, dated 2 January 2025, he cited a J.P. Morgan Asset Management chart of monthly observations from 1988 to late 2014.

It showed "a strong relationship between starting valuations and subsequent annualized ten-year returns", and buyers at a forward P/E in line with the then current 22 "always earned ten-year returns between plus 2% and minus 2%." He still described the P/E as "high but not insane".

The same memo retracts a quote Marks had been attributing to Warren Buffett, that investors get into trouble when they forget profits grow about 7% a year. Marks writes that he asked Buffett for a source "and he told me he never said it".

S&P 500 valuation against its history. Times earnings | 30 September 2026. Forward P/E: 30 Sep 2026 19.0 x, 30-year average 17.2 x; Shiller CAPE: 30 Sep 2026 40.9 x, 30-year average 28.9 x.

At 30 September 2026: S&P 500 forward P/E 19.0x against a 30-year average of 17.2x; Shiller CAPE 40.9x against 28.9x. Source: J.P. Morgan Asset Management, Guide to the Markets US, data as of 30 September 2026, page 5; chart data (CSV). The current guide plots forward P/E against subsequent one-year and five-year returns, not the ten-year chart Marks cited. Neither chart forecasts the return from any given starting point.

On AI, Is It a Bubble? (9 December 2025) concluded that no one should go all-in or stay all-out: "A moderate position, applied with selectivity and prudence, seems like the best approach." In AI Hurtles Ahead (26 February 2026) he judged the hyperscalers unlikely to prove "ruinously excessive" at current prices and called richly valued startups "lottery tickets".

In Shall We Repeal the Laws of Economics - Part III (22 September 2026), he tied the rise in long-term Treasury yields, with the 30-year closing above 5.3% on 17 August, to US fiscal deficits and heavy demand for capital. Asked by a friend whether to sell his stocks, he replied: "That's not the answer".

What risks does Marks see in private credit?

Marks argues that direct lending, the fastest-growing part of private credit, followed a familiar boom pattern and is now exposed to software loans. In What's Going on in Private Credit? (9 April 2026), he estimates that something like $2 trillion of direct loans were made in 15 years, against a private credit sector of about $150bn two decades ago.

He puts software debt at roughly 4-5% of US high-yield bonds, 10-15% of broadly syndicated loans and 20-30% of direct lending, often financing buyouts at around 20 times EBITDA. When AI fears hit software in early 2026, investors in semi-liquid vehicles requested redemptions, and some funds limited them. He says direct lending is about 20% of Oaktree's performing credit and less than 15% of its assets under management.

Who owns Oaktree, and how large is it?

Brookfield owns all of Oaktree. According to Brookfield Asset Management's 10-Q for the quarter to 30 June 2026, Brookfield acquired the remaining 26% on 31 July 2026 for about $3.0bn, the next step in a partnership that began in 2019. Marks is co-chair of Oaktree alongside Bruce Karsh, who remains chief investment officer (Brookfield release, 3 August 2026).

Oaktree had $224bn of assets under management at 30 June 2026, across credit, private equity, real assets and listed equities, according to an Oaktree-advised fund's semi-annual report. Marks wrote in April 2026 that Oaktree's assets had "only" doubled over the last decade.

Is high-yield credit a defensive asset allocation?

High-yield credit promises interest and principal, but defaults and falling prices can coincide with equity losses. Marks's October 2024 figures of roughly 7% for public credit and 10% for private credit therefore need to be assessed after credit losses, fees and liquidity constraints. Replacing equities with lending changes the return distribution; it does not buy a contractual payoff against an equity crash.

Copio's research objective is to maximise expected long-term geometric growth within the investor's constraints. Contractual income earns a place when it improves that whole-portfolio outcome, rather than solely because its quoted volatility is lower. We also assess whether a growth portfolio with cost-controlled convex protection can preserve more productive exposure while generating gains during a covered crash. Our defensive portfolio allocation comparison distinguishes income, reduced exposure and explicit protection.

What can an individual investor learn from Howard Marks?

Marks's memos ask investors to choose their normal level of risk before reacting to market prices.

  • What share of it is ownership, and what share is lending, once funds are looked through?
  • Was its level of risk chosen deliberately, or did it follow from the funds that were bought?
  • What is its normal posture, and what conditions would justify moving away from it?
  • How much of its expected return is contractual, and how much depends on prices rising?
  • If it holds credit, what default and recovery rates would wipe out the extra yield?

For other ways professionals frame defence, see Mark Spitznagel's tail hedging and how asset classes behaved in past crashes.


Sources and method

Research date: 27 October 2026. Primary sources were Howard Marks's memos on oaktreecapital.com: Sea Change (13 December 2022), Further Thoughts on Sea Change (sent to clients 30 May 2023, published 11 October 2023), Taking the Temperature (10 July 2023), Ruminating on Asset Allocation (22 October 2024), On Bubble Watch (2 January 2025), Gimme Credit (6 March 2025) and Is It a Bubble? (9 December 2025).

The other memos used were AI Hurtles Ahead (26 February 2026), What's Going on in Private Credit? (9 April 2026), Shall We Repeal the Laws of Economics - Part III (22 September 2026) and Calibrating (6 April 2020, PDF pages 5-7). Oaktree's investment philosophy page supplied the six tenets.

Ownership facts come from Brookfield Asset Management's 10-Q for the period ended 30 June 2026 (Oaktree Acquisition note) and Brookfield's 3 August 2026 release. The $224bn figure comes from Oaktree Asset-Backed Income Fund's N-CSRS for the period ended 30 June 2026, adviser description.

Valuation data come from J.P. Morgan Asset Management's Guide to the Markets US, data as of 30 September 2026, page 5. The yields and returns quoted are Marks's own figures as of each memo date; we did not recompute them, and the J.P. Morgan ten-year chart he cited was not available for us to check.

The memos express Marks's views at their dates and are not Oaktree's portfolio allocations. Our discussion of ownership, lending and risk posture interprets these disclosures rather than reproducing an Oaktree model. This article is educational, not personalised investment advice.

FAQ

What is Howard Marks's asset allocation approach?

In his memo Ruminating on Asset Allocation (22 October 2024), Marks argues there are only two asset classes, ownership and debt. He calls the choice of a target risk posture, the balance between aggressiveness and defensiveness, the most important decision in portfolio management. Everything else, from geography to public versus private, he treats as implementation.

What was Howard Marks's Sea Change memo about?

Sea Change (13 December 2022) argued that the end of ultra-low interest rates may be the third sea change of his career. He noted that high-yield bonds had yielded 4-5% a year earlier and yielded roughly 8% at the time, which he said could deliver equity-like returns from contractual cash flows. In a follow-up sent to clients on 30 May 2023 he wrote that credit instruments should probably be a substantial portion of portfolios, perhaps the majority.

What did Howard Marks say about stock market valuations in 2025?

In On Bubble Watch (January 2025), Marks cited a J.P. Morgan Asset Management chart of monthly data from 1988 to late 2014. He wrote that buyers of the S&P 500 at a forward P/E near the then current 22 had always earned ten-year returns between plus 2% and minus 2% a year. He did not call it a bubble and listed counterarguments.

Who owns Oaktree Capital?

Brookfield owns 100% of Oaktree. Brookfield Asset Management's 10-Q states that on 31 July 2026 Brookfield acquired the roughly 26% it did not already own, for about $3.0bn in cash and non-cash consideration. Oaktree reported $224bn of assets under management at 30 June 2026.

Is high-yield credit a defensive asset allocation?

High-yield credit has contractual payments but remains exposed to default, interest-rate and liquidity risks. Replacing equities with credit does not buy a contractual equity-crash payoff. Copio judges the allocation by expected whole-portfolio geometric growth after costs and credit losses, including comparison with a growth portfolio supported by explicit protection.