The Quiet Storm Makes Landfall
Duration Risk Realized and the Fiduciary Playbook for a Five Percent World
“He who observes the wind will not sow, and he who regards the clouds will not reap.”
— King Solomon, Kohelet (Ecclesiastes) 11:4
Executive Summary
Thirteen months ago, in The Quiet Storm in Fixed Income, we argued that the multi-year rise in interest rates had left a large, mostly unrealized loss sitting quietly inside long-dated institutional bond portfolios, that a Federal Reserve easing cycle would relieve the short end of the curve but not the long end, and that the fiduciary response was to shorten duration, upgrade credit quality, and build the governance for liability-driven hedging before the loss became realized. This report returns to that thesis with the discipline we ask of any manager we diligence: we score every call against what happened, we say plainly where we were wrong, and we restate the house view for the year ahead.
The storm is no longer quiet. On September 16, 2026, the Federal Open Market Committee raised the target range for the federal funds rate by 25 basis points to 3.75 to 4.00 percent, by a 12 to 0 vote, the first increase since July 2023 and the end of an easing cycle that delivered only 75 basis points. The 10-year Treasury yield, 4.04 percent on the day our original report was published, closed at 5.27 percent on October 6, 2026, having touched 5.31 percent the day before, its highest since 2002. The 30-year rose from 4.65 percent to 5.64 percent and has closed above 5 percent every day since July 7. The Bloomberg US Aggregate Index has returned -2.91 percent in 2026 through September 30; Treasuries of 20 years and longer -7.91 percent; the Bloomberg Municipal Index posted its worst quarter since 1981. Investors who held the long end as ballast through 2026 learned what Silicon Valley Bank learned in 2023, that a high-quality bond is a safe asset only to the holder who never has to sell it.
The proximate cause was a supply shock the original report did not forecast: the war with Iran that began on February 28, 2026, the closure of the Strait of Hormuz, and an energy spike that took headline CPI to 4.2 percent in May and Dated Brent to $131 on September 15. But the structural causes were the ones we named. The New York Fed’s estimate of the ten-year term premium rose from 0.53 percent at the end of September 2025 to 0.82 percent a year later, its highest month-end reading since June 2014, and to 0.96 percent on October 5. The 10-year real yield rose 124 basis points to 2.91 percent while the breakeven inflation rate was unchanged at 2.36 percent. The long end repriced compensation for risk, not expected inflation, and the risks being compensated are fiscal and structural: a federal deficit near $2 trillion with net interest outlays above $1 trillion, a Treasury that dealers expect to raise coupon sizes in 2027, a central bank with a new chairman and a new communications doctrine, and a corporate sector that is now, in the words of the market commentary recorded in the FOMC’s September minutes, competing for capital through “heavy private debt issuance to finance the development of artificial intelligence (AI) infrastructure.”
Our scorecard is mixed in the way honest scorecards are. The direction of the long end, the role of the term premium, the breakdown of the stock-bond relationship, the rising pressure on endowment spending, and the fiduciary failure of under-hedged pensions were called correctly and in several cases understated. The shape of the curve was called wrong: we expected steepening, and beyond the bill sector the curve bear-flattened as the two-year yield rose 128 basis points on a hiking cycle nobody in September 2025 expected. The front-end gains we promised arrived in the fourth quarter of 2025; in 2026 the two-year yield rose 128 basis points and the short sleeve merely held its value. The model allocation we published protected capital relative to a long-duration portfolio by six to seven percentage points, but its intermediate Treasury and investment-grade sleeves carried more duration than the thesis warranted, and its municipal sleeve was a year early.
The house view. We hold with conviction that the long end of the Treasury curve has entered a regime of structurally higher compensation that will not retrace to the post-2008 norm. Goldman Sachs’ rates strategists wrote in August that the drivers of the higher term premium “will prove persistent, with steeper curves a durable feature,” and that is also our reading. A 10-year yield near 5 percent is, as Goldman’s asset-allocation team observed, “higher but not high” against a 250-year average of 4.7 percent; the anomaly was the decade that preceded this one. What has changed is that duration is now compensated. A real yield near 3 percent is the highest since 2008, and the question for a fiduciary is no longer whether to avoid duration but how to add it without buying the fiscal tail. Our answer, held provisionally and with explicit triggers, is a staged extension from bills into the three-to-seven-year sector and inflation-protected Treasuries, with the 20-to-30-year sector reserved for liability matching and expressed through swaps rather than physical bonds. In credit we hold the conviction that quality must dominate beta: investment-grade spreads of 84 basis points near their historical tights sit against 1,211 basis points for CCC-rated debt and first-lien recoveries of 43 percent, the lowest in Fitch’s decade of data; this is a market paying for quality and punishing the reach for yield. We narrow the private credit sleeve we recommended a year ago, and we say why. And we repeat, with more urgency than before, that the central fiduciary failure of this cycle is governance, not forecasting: only 28 percent of the public pension plans in the CFTC’s 153-plan study use interest rate swaps at all, and 2026 was the year that gap was paid for.
This report proceeds through the six pillars of our standing framework. It scores the original thesis, sets the macroeconomic and policy backdrop, examines the Treasury curve and the fiscal arithmetic beneath it, quantifies what 2026 did to bond portfolios, assesses public and private credit, and reads equities and alternative investments through the lens of a 5 percent discount rate. It then turns to the institution’s own balance sheet, which has changed as much as the market has, before stating the opposing case at full strength and closing with positioning and a prioritized list of actions.
One Year On: Scoring the Quiet Storm
The published archive is one developing house view, and a view that is never scored is not a view. Table 1 sets each claim of the September 2025 report against dated evidence.
Table 1: The Quiet Storm scorecard
| September 2025 claim | What happened | Verdict |
|---|---|---|
| The Fed pivots to easing; some forecasts called for as much as 100 bp over four meetings | Cuts of 25 bp on Sept 17, Oct 29 and Dec 10, 2025 to 3.50-3.75%; hold from January through July 2026; hike to 3.75-4.00% on Sept 16, 2026 | Validated for one quarter; the cycle ended at 75 bp and reversed |
| The short end appreciates | 2-year yield 3.51% (Sept 16, 2025) to 3.47% (Dec 31, 2025); 1-3 year Treasuries returned +5.14% in 2025. Then 4.79% (Oct 6, 2026), +128 bp; 1-3 year Treasuries +0.48% in 2026 through Sept 30 | Validated in 2025; reversed in 2026 |
| The long end stays “higher for longer” despite cuts | 10-year 4.04% to 5.27%; 30-year 4.65% to 5.64%; 30-year above 5% every close since July 7, 2026; target range 50 bp lower over the same window | Validated and understated |
| The curve steepens | 3-month/10-year slope 0 to +106 bp; 2s10s 53 bp to 48 bp; 5s30s 106 bp to 61 bp. The 5-year point rose most (+144 bp) | Level right, shape wrong beyond bills |
| Term premium rises on deficits and fragmentation | ACM 10-year term premium 0.53% (Sept 30, 2025) to 0.82% (Sept 30, 2026), highest month-end since June 2014; 0.96% on Oct 5, 2026. 10-year TIPS yield 1.67% to 2.91%; breakeven 2.37% to 2.36% | Validated; the move is real-rate and term-premium driven |
| Stock-bond correlation breaks down | Five-year S&P 500/Treasury correlation 0.53 (AQR, to Feb 28, 2026); March 2026 joint sell-off (equities -4.5% to -6%, 10-year +38 bp); Q3 2026 bond rout alongside an equity rally. Aggregate -2.9% vs S&P 500 +12.7% YTD to Sept 30 | Validated: duration failed as ballast in both directions |
| Endowment spending pressure rises | Cambridge Associates Endowment Radar 2025: median dependence on distributions 17.0% (from 16.3%), effective spending 5.4%; NACUBO FY2025: withdrawals $33.4 billion, +11%; new 8% excise tier effective for tax years after Dec 31, 2025 | Validated and strengthened by the tax |
| Public pensions gain discount-rate relief | Public plans discount at expected return (NASRA median still 7.00%); their record funded ratios (Milliman PPFI 88.2% at July 31, 2026) came from assets. Corporate plans received the direct relief: Milliman 100 at 114.5%, discount rate 6.47%, highest since May 2009 | Right mechanism, wrong sector |
| Derivatives adoption for duration hedging is “dramatically low” | CFTC study: 43 of 153 public plans (28%) hold interest rate swaps; mean position 0.09% of liabilities vs ~252% for a full hedge. No 2025-26 adoption found | Confirmed and unchanged |
| A 250 bp collateral buffer for leveraged LDI | UK schemes carried ~300 bp buffers into the March 2026 gilt shock and the October 1 move to a 6.07% 30-year gilt; limited cash calls, no forced selling | Validated by a live test |
| Model allocation: 10-20% short, 20-30% intermediate, 20-30% IG corporate, 10-15% municipal, 10-20% alternative credit | 2026 YTD total returns to Sept 30: bills +2.72%, 1-3 year +0.48%, 7-10 year -4.51%, IG corporate -4.36%, Bloomberg Municipal Index -4.18%, 20+ year -7.91%; Cliffwater Direct Lending Index +3.0% in H1 | Direction right; too much duration in the middle; munis a year early; private credit needs tighter definition |
The table reads as a validated thesis with two plain failures: the shape of the curve and the timing of the front end. Both failures share a cause, a supply shock that converted an easing cycle into a hiking cycle within nine months. The lesson is not that the framework was wrong but that a framework built on the policy rate is hostage to the policy rate; the structural claim about the long end held regardless.
Three further points belong on the record. First, the original report’s duration table illustrated a 300 basis point move; the realized 2026 move on the 30-year was 80 basis points from December 31 and 99 basis points from publication, and it was enough to produce a nine-month loss of nearly 8 percent on Treasuries of 20 years and longer and to take the iShares 20+ Year Treasury ETF, on September 28, to the lowest price since its 2002 launch. Second, the report framed the Fed’s September 2025 cut as a response to a cracking labor market; the labor market did not crack. Payrolls averaged roughly 45,000 a month over the past twelve months, the unemployment rate drifted between 4.1 and 4.3 percent, and the Fed’s own September 2026 statement described job gains that “have kept pace with the workforce.” Third, the report treated energy as a disinflationary tailwind and gave no weight to an oil shock. A fixed-income framework that omits the single variable most capable of converting cuts into hikes is incomplete, and we have corrected that below.
Macroeconomic Environment: An Oil Shock, a New Chairman, and the First Hike Since 2023
The macroeconomic story of the past year divides cleanly at the last day of February. Through the fourth quarter of 2025 the economy behaved roughly as the street expected: three cuts, the end of quantitative tightening on December 1, a 43-day government shutdown that erased the October 2025 CPI from the statistical record, and a consensus that Goldman Sachs summarized in January as “Solid Growth, Low Inflation, Shaky Labor Market,” with two further cuts and a 10-year yield pinned at 4.20 percent. Then the United States and Israel struck Iran on February 28, Iran closed the Strait of Hormuz within days, and the oil market experienced what the International Energy Agency called the largest supply disruption in the history of the global oil market. Brent rose from roughly $72 to about $120 in March; a ceasefire on April 8 collapsed into a renewed blockade threat that took Brent futures to $126 intraday on April 30; Gulf shut-ins peaked at 10.9 million barrels a day in May; and after a June memorandum of understanding gave way to renewed strikes in July and Houthi attacks on Saudi Arabia’s East-West pipeline in September, Dated Brent printed $131 on September 15. Brent averaged $114 in September and stood near $101 on October 6; the Energy Information Administration’s October outlook has it at $105 in the fourth quarter before easing to $74 by the end of 2027. Gasoline averaged $4.33 in September, a record for the month.
Inflation followed. Headline CPI rose from 2.4 percent in February to roughly 4.2 percent in May and stood at 3.4 percent in August, with energy up 16.3 percent year over year and gasoline up 27.4 percent. Core CPI, held down by shelter at 3.0 percent and by a shutdown-distorted base, stayed between 2.4 and 2.8 percent all year. Core PCE, the Fed’s preferred gauge, rose from 2.7 percent in October 2025 to a peak of 3.2 percent in May and printed 3.0 percent in August under the Bureau of Economic Analysis’s revised methodology, which lowered the year-over-year rate by 0.36 percentage point, mostly by changing how portfolio-management fees are measured. Goldman Sachs estimates that “tariffs and the Iran war account for 0.9pp of core PCE inflation’s 1pp overshoot of 2%,” and that core inflation has run at a 2.0 to 2.5 percent annualized pace over the summer. That is the dovish reading. The hawkish reading, which carried the FOMC, is that inflation has now exceeded 2 percent for more than five years, that the share of the PCE basket rising faster than 3 percent remains well above its pre-pandemic norm, and that households have noticed: the University of Michigan’s long-run inflation expectation reached 3.9 percent in May, its one-year expectation 4.6 percent in September, and consumer sentiment fell to 44.8 in May, the lowest reading in the survey’s history, and remained at 48.1 in September.
The Federal Reserve that confronted this shock was a different institution from the one that cut rates in September 2025. Kevin Warsh was confirmed as chairman on May 13, 2026 by a 54 to 45 vote and sworn in on May 22; Jerome Powell remains on the Board as a governor, and the Supreme Court’s 5 to 4 ruling in Trump v. Cook on June 29 left Governor Cook in her seat. Warsh has dispensed with the chairman’s dot, shortened the policy statement, and told the press on September 16 that he is “not in the forward guidance business” and that “data point dependence is a dangerous preoccupation.” Goldman Sachs’ rates strategists argue that this “less transparent communication approach” raised policy uncertainty and “contributed to the recent rise in term premium that took long-end yields to multi-cycle highs.” We agree, and we note that the chairman’s own account of why long yields have risen, growth, hyperscaler competition for capital and geopolitics, concedes the structural half of our thesis. His Jackson Hole debut on August 28, in which he said “the Fed’s predominant focus right now should be on prices,” moved the two-year yield 7 basis points and the probability of a September hike from roughly 30 percent to above 50; the committee delivered on September 16 without dissent. The projections showed a median funds rate of 4.1 percent at the end of both 2026 and 2027, a longer-run estimate raised from 3.06 to 3.25 percent, and 16 of 18 participants projecting at least one more increase this year. The minutes released on October 7 record that “most participants assessed that another increase in the target range for the federal funds rate would likely be appropriate by year end” and that “several participants” view the current rate as “not restrictive or only mildly restrictive.” Futures on October 7 priced roughly a 22 percent probability of a hike on October 28 and 75 to 80 percent odds of one by December.
Growth has been the puzzle. Real GDP rose at a 2.2 percent annualized rate in the second quarter, revised up from 1.5 percent; the Atlanta Fed’s GDPNow estimate for the third quarter stood at 3.7 percent on October 6 and Goldman’s at 3.1 percent; business fixed investment is rising at a 7 percent pace, with equipment spending up 11 percent, led by the data-center build-out. Goldman’s September 28 note asked why the economy has been so resilient to higher rates and higher oil, and answered that the drag from higher interest rates “has been nearly offset by the effects of higher equity prices, narrower credit spreads, and a weaker dollar.” Its October 5 analysis is less sanguine: the offset has faded, higher rates should subtract about 0.2 percentage point from 2027 growth under Goldman’s baseline of a December hike followed by cuts in late 2027, and “slightly over 0.5pp” if current rates persist. The labor market is at breakeven. September payrolls rose 29,000 against expectations near 85,000, with the prior two months revised down by 60,000; the unemployment rate is 4.2 percent; average hourly earnings are growing 3.0 percent, the slowest since 2021. The economy is growing above potential on investment and upper-income consumption while housing contracts, lower-income households absorb a gasoline tax, and the labor market neither adds nor sheds. That is not a recession. It is a configuration in which the Fed can keep hiking, and that is what bond markets have priced.
The fiscal backdrop completes the picture, and it is the part that does not change with the oil price. The Congressional Budget Office estimates the fiscal 2026 deficit at roughly $2.1 trillion, above 6 percent of GDP, after the Supreme Court’s February 20 decision in Learning Resources v. Trump struck down the IEEPA tariffs and forced refunds of roughly $110 billion. Net interest outlays were $1.05 trillion through August, up 12 percent, and will exceed both defense and Medicare for the year. Total federal debt crossed $40 trillion on August 18; debt held by the public was $32.43 trillion on September 30. The average interest rate on marketable Treasury debt rose from 3.41 to 3.52 percent over the fiscal year, and it rises with every refinancing at current coupons. The CBO’s long-term baseline takes debt held by the public from 101 percent of GDP today to 120 percent in 2036 and 175 percent in 2056; a one-percentage-point increase in interest rates would take the 2056 figure to 222 percent. Goldman Sachs’ September 10 analysis reaches 123 percent by 2035 on its own rate forecast, with the average interest rate on the debt overtaking nominal GDP growth “by the early 2030s” if current rates hold; its October 5 follow-up puts the 2035 ratio at 132 percent in that case. The September analysis also names the risk an endowment CIO should carry in mind: that if the required consolidation becomes large enough, “there could be pressure on the Fed to keep rates too low, resulting in high inflation.”
The Long End Reawakens: Curve, Term Premium, and the Arithmetic of Supply
The original report’s central claim was that a fall in short-term rates “will not necessarily translate to a significant capital appreciation in long-duration portfolios if the long end remains elevated.” It did not. Table 2 sets the curve on the day of publication against the curve today, with the policy rate alongside.
Table 2: The Treasury curve, then and now (par yields, percent)
| Tenor | Sept 16, 2025 | Dec 31, 2025 | Sept 30, 2026 | Oct 6, 2026 | Change since publication |
|---|---|---|---|---|---|
| Fed funds (effective) | 4.33 | 3.64 | 3.88 | 3.88 | -45 bp |
| 3-month | 4.04 | 3.67 | 4.20 | 4.21 | +17 bp |
| 2-year | 3.51 | 3.47 | 4.88 | 4.79 | +128 bp |
| 5-year | 3.59 | 3.73 | 5.09 | 5.03 | +144 bp |
| 10-year | 4.04 | 4.18 | 5.29 | 5.27 | +123 bp |
| 30-year | 4.65 | 4.84 | 5.64 | 5.64 | +99 bp |
| 10-year TIPS (real) | 1.67 | n/a | n/a | 2.91 | +124 bp |
| 10-year breakeven | 2.37 | n/a | n/a | 2.36 | -1 bp |
| ACM 10-year term premium | 0.53 (Sept 30) | 0.67 | 0.82 | 0.96 (Oct 5) | +43 bp |
Sources: U.S. Treasury daily par and real yield curves; Federal Reserve Bank of New York (EFFR, ACM term premium). The table reads as an upward shift of roughly 100 to 145 basis points from the two-year point outward, led by the belly, with the policy rate 45 basis points lower than when we published. Every basis point of the 10-year move is accounted for by the real yield; none by inflation expectations.
Three features of this move matter for positioning. The first is that the front end did not stay anchored. The two-year yield rose more than the ten-year, because the oil shock converted roughly 60 basis points of priced easing into 15 basis points of priced tightening by mid-May and then, after Jackson Hole and the September hike, into a market that by late September was pricing “100bp+ hiking cycles in the US, UK and Europe,” in the words of Goldman’s Dominic Wilson and Kamakshya Trivedi. The yield curve from two to thirty years therefore flattened even as the curve from bills to bonds steepened. Our steepening call was a call on the policy rate, and the policy rate went the other way.
The second feature is that the long-end move was real. The 10-year TIPS yield rose from 1.67 to 2.91 percent; the 30-year real yield from 2.40 to 3.35 percent. Breakevens did not move, and the FOMC’s staff noted in September that “changes in real rates contributed to most of the net increase in longer-maturity Treasury yields.” This is the signature of a term-premium and supply repricing rather than an inflation scare, and it matches the mechanism we described in 2025. The New York Fed’s ACM estimate of the 10-year term premium, 0.53 percent at the end of September 2025 and 0.96 percent on October 5, remains well below its long-run average of 1.35 percent. We take that gap seriously in both directions. It says the market has not yet priced a full normalization of risk compensation, which is a reason for caution about the long end; it also says that a further 40 basis points of term premium would leave the 30-year near 6 percent, a level at which the opposing case, set out below, becomes compelling.
The third feature is that the move was global, and that matters because the marginal buyer of a 30-year Treasury now has a choice. Thirty-year gilts traded at 6.07 percent on October 1, the highest since 1998; the 30-year Japanese government bond reached a record 4.235 percent on October 5, after the Bank of Japan raised its policy rate on September 18 to 1.25 percent, the highest since 1995; the 30-year Bund reached 3.79 percent in August, its highest since 2011; and the French 10-year touched 4.99 percent on October 2 with the OAT-Bund spread at 152 basis points, a level Goldman’s rates desk compared to “the most acute phase of the European sovereign crisis in late 2011.” Peter Oppenheimer’s September 17 strategy paper made the structural point: “As recently as 2022... 30-year bond yields in Germany and Japan were close to zero.” The era in which a Japanese life insurer or a European pension fund had no domestic alternative to a Treasury has ended, and the Treasury International Capital data show it: foreign holdings of Treasuries fell to $9.25 trillion in July, with China’s holdings at $618 billion, the lowest since 2008-09, and foreign official holdings down from $4.01 trillion in February to $3.77 trillion.
Against that demand picture stands supply. The Treasury’s August 5 refunding held coupon auction sizes unchanged “for at least the next several quarters,” but the Treasury Borrowing Advisory Committee recorded a $1.45 trillion financing shortfall for fiscal 2027-28 at current sizes, and dealers expect coupon increases “sometime in 2027.” The Treasury’s response to long-end stress was to at least double its liquidity-support buybacks in the 10-to-30-year sector; Goldman’s William Marshall estimates the program offsets about 30 percent of gross 20- and 30-year supply and is worth about one basis point on the 10-year yield, “more micro than macro,” because buybacks “do not address the fundamental drivers and macro risks that have been the primary factors behind higher long-end yields globally.” The auction tape agrees: the 30-year auction on August 13 cleared at 5.216 percent, the highest since 2001, and the 20-year on September 15 cleared at 5.42 percent with indirect bidders taking a record-low 52.5 percent. Marshall’s September 11 analysis puts combined net issuance of Treasuries, agency mortgages and investment-grade corporates to private investors at $2.6 trillion in 2026 and $2.85 trillion in 2027, record dollar amounts even if not records as a share of GDP, and the main driver of the increase is not the Treasury but the corporate sector, specifically the “long tenor of AI-related issuance.”
This is the new variable in the long-end equation. Hyperscaler capital expenditure will reach roughly $800 billion in 2026 and, on Goldman’s estimates, $1.2 trillion in 2027; it “now exceeds cash flow from operations,” which means the marginal dollar is borrowed. AI-related issuers accounted for about a quarter of gross investment-grade supply this year, pushing Goldman’s full-year IG issuance forecast to $2.3 trillion; its credit strategists expect roughly $1.3 trillion of hyperscaler debt over 2027 to 2029, and JPMorgan puts 2027 hyperscaler issuance near $315 billion. The FOMC minutes name this competition for capital as a contributor to higher term premiums, and Wilson and Trivedi draw the conclusion that governs our view of the long end: “Until that boom fades, upward pressure on real rates is likely to persist and overly dovish central bank policy may simply transfer pressure from the front to the back of the yield curve.”
Where does the street stand now, having been wrong a year ago? Table 3 records the December 2025 forecasts against the outturn and the revised views.
Table 3: The street’s December 2025 forecasts for 2026 against the outturn
| Forecaster (date) | Fed funds end-2026 | 10-year end-2026 | Inflation end-2026 | Other | Outturn / current view (Oct 7, 2026) |
|---|---|---|---|---|---|
| Goldman Sachs (Dec 18, 2025) | 3.00-3.25% (cuts in March and June) | 4.20% | 2.2% | S&P 500 7,600; IG 90 bp, HY 315 bp; Brent $56 | Funds 3.75-4.00% after a hike; 10-year 5.27%; core PCE 3.0% (Aug); S&P 500 7,819; IG 84 bp, HY 312 bp; Brent ~$101. GS now: one more hike in December, cuts from late 2027; 10-year 4.75% end-2026, 4.40% end-2027; 30-year 5.15% / 4.90% |
| J.P. Morgan Asset Management (Nov 13, 2025) | 2-3 cuts | 4.00-4.50%; “modest curve steepening” | CPI 2% by end-2026 | “80% of investors remain underweight duration” | CPI 3.4% (Aug); hike not cuts; 10-year 5.27% |
| Goldman Sachs Asset Management (Nov 2025) | -75 bp to 3.00-3.25% (base) | neutral; steepening bias | n/a | long end “more susceptible to fiscal concerns” | Steepening bias right vs bills, wrong vs 2s; direction of policy wrong |
| Silvercrest (Dec 4, 2025) | no December cut; ~3.0% over three years | 3.8% end-2025; 3.0-4.5% three-year range; 10-year “will generally follow the path of Fed policy” | 2.2% (three-year view) | IG over HY | 10-year 5.27%; the long end did not follow the Fed |
| FOMC median (Sept 2025 SEP) | 3.4% | n/a | 2.6% | long run 3.0% | 3.875% midpoint; Sept 2026 SEP median 4.1%, long run 3.25% |
| Wind River (Dec 22, 2025, The Architecture of Fragility) | accepted GS’s 50 bp of cuts | “higher for longer”; no level | n/a | spreads to “widen significantly”; gold $4,900; silver $60-70 | IG/HY spreads near tights (84/312 bp) but CCC +400 bp to 1,211 bp; gold peaked at $5,595 in January and trades ~$4,190; silver ~$62 after a $122 peak |
Sources: Goldman Sachs Global Investment Research, “Macro Outlook 2026” (Dec 18, 2025) and portal forecast pages (Oct 7, 2026); J.P. Morgan Asset Management, “2026 Year-Ahead Investment Outlook”; Goldman Sachs Asset Management, “Investment Outlook 2026”; Silvercrest, “Economic & Market Outlook 2026 & Beyond”; Federal Reserve SEP; Wind River Capital Strategies. The table reads as a uniform miss on the direction of policy and the level of the long end, including our own December report’s acceptance of the easing consensus. The forecaster who tied the long end most explicitly to the policy rate missed the mechanism as well as the level, which is the point of this report.
We are not scoring the street to embarrass it. We are scoring it because the same houses now forecast a 10-year yield of 4.75 percent at the end of 2026 and 4.4 percent at the end of 2027, with a terminal funds rate near 3.4 percent, and a fiduciary deciding whether to extend duration should know how the forecasters did the last time they called for lower long yields. Goldman’s own rates desk is candid about the asymmetry it sees: valuation and positioning show “a building asymmetry towards lower yields,” five-year-forward five-year rates sit about one standard deviation above the desk’s fair-value model, and “the missing ingredient for duration relief is a catalyst.” We find that argument persuasive at the belly of the curve and insufficient at the long end, for reasons the next three sections set out.
Duration Risk Realized: What 2026 Did to Bond Portfolios
The original report’s first table illustrated how a 300 basis point move would affect Treasuries of different maturities. The move that arrived was smaller and still sufficient. Table 4 pairs the realized 2026 returns of each duration bucket with the sensitivity of the same buckets at today’s yields, so that a committee can see both what happened and what the next move would do.
Table 4: Duration realized and duration ahead
| Sector | 2026 total return to Sept 30 | 2025 total return | Yield, Oct 6, 2026 | Duration, years (approx.) | Price change, +100 bp | Price change, +200 bp | Price change, -100 bp |
|---|---|---|---|---|---|---|---|
| Treasury bills (0-3 month) | +2.72% | n/a | 4.21% | 0.2 | -0.2% | -0.4% | +0.2% |
| 1-3 year Treasuries | +0.48% | +5.14% | 4.79% (2-yr) | 1.9 | -1.9% | -3.8% | +1.9% |
| 5-year Treasury (par) | n/a | n/a | 5.03% | 4.4 | -4.3% | -8.3% | +4.5% |
| 7-10 year Treasuries | -4.51% | +8.20% | 5.27% (10-yr par) | 7.7 | -7.4% | -14.0% | +8.1% |
| Bloomberg US Aggregate | -2.91% | +7.30% | 5.55% (YTW) | 5.7 | -5.7% | -11.4% | +5.7% |
| US agency MBS | -3.31% | +8.58% | 5.98% | 5.9 | -5.9% | -11.8% | +5.9% |
| US IG corporates | -4.36% | +8.03% | 6.03% | ~6.5 | -6.5% | -13.0% | +6.5% |
| US high yield | -0.07% | +8.83% | 8.15% | ~3.5 | -3.5% | -7.0% | +3.5% |
| Municipals (Bloomberg index) | -4.18% | +3.90% | 4.04% (10-yr AAA) | ~6 | -6% | -12% | +6% |
| Treasuries 20+ years | -7.91% | +4.29% | 5.64% (30-yr) | 14.4 | -12.9% | -23.4% | +16.1% |
Sources: index total returns via iShares fund pages and Bloomberg press reporting as of Sept 30, 2026 (Treasury bills, 1-3 year, 7-10 year, 20+ year, Aggregate, MBS, IG and HY via iBoxx, Bloomberg Municipal Index); yields from the U.S. Treasury, ICE BofA (via FRED) and iShares (AGG, MBB) as of Oct 5-6, 2026; price changes computed for par bonds at the stated yields with convexity included for the 5-, 10- and 30-year, duration-only for the others. Corporate and municipal durations are approximate; the 7-10 year row pairs the index return with 10-year par-bond sensitivity. The table reads as a near-perfect ordering: the shorter the bucket, the better the 2026 return, and the asymmetry of the long end remains stark in both directions. A 100 basis point rally in the 30-year is now worth about 16 percent; the same move the other way costs 13.
The arithmetic has two uses. The first is retrospective: a fixed-income portfolio built to the original report’s model allocation, roughly 15 percent bills, 8 percent short Treasuries, 25 percent intermediate Treasuries, 25 percent investment-grade corporates, 12 percent municipals and 15 percent alternative credit, would have returned between -1 and -2 percent in 2026 through September on the index figures above, against -7.9 percent for a long-Treasury portfolio and -2.9 percent for the Aggregate; a bills-only portfolio returned +2.7 percent. That is the six-to-seven-point protection relative to long duration that we claimed in the summary, and it is also the admission that the model’s middle sleeves carried duration the thesis did not require. The second use is prospective. At today’s yields the income cushion is real. A 10-year note yielding 5.27 percent absorbs roughly 70 basis points of yield increase over a year before its total return turns negative; a 30-year bond at 5.64 percent absorbs about 40 basis points; in September 2025 those breakevens were roughly 50 and 30 basis points. Duration is still a wager on the direction of rates, but the wager is now paid for.
The banking system illustrates the unrealized-loss mechanism we used in 2025, and it illustrates something more uncomfortable about timing. The FDIC’s Quarterly Banking Profile for the second quarter of 2026, released August 25, reported unrealized losses on securities of $326.7 billion, down from a peak of $689.9 billion in the third quarter of 2022 and from $413.2 billion in the first quarter of 2025, as the easing cycle lifted bond prices and portfolios matured. That figure is as of June 30, when the 10-year yielded 4.44 percent. The 10-year ended the third quarter at 5.29 percent. On the system’s $5.8 trillion securities book, assuming a duration near five years, an 85 basis point rise implies, as an illustrative calculation rather than a forecast, something on the order of $250 billion of additional mark-to-market loss, which would put the total back above $500 billion, a level last seen in 2023, when the third-quarter report arrives in late November. Barron’s estimated in September that Bank of America’s held-to-maturity portfolio alone carried an unrealized loss of about $82 billion, and that a further 50 basis point rise would push it past $90 billion. Six small banks have failed in 2026, the most since 2017, including Nano Banc of Irvine, California on September 25, though none for the reasons SVB failed; deposits have been stable and the system’s net interest margin stands at 3.32 percent. The point is not that a banking crisis is imminent. It is that the quiet storm’s defining feature, an unrealized loss that reappears whenever rates rise and vanishes from the discussion whenever they fall, is still operating three years after the lesson was supposedly learned, and institutional bond portfolios carry it in exactly the same way.
For a fiduciary, the operative difference between a bank and an endowment is the liability. The bank’s depositors can run; the endowment’s beneficiaries cannot. But the endowment’s spending draw, its capital calls, and now its excise tax can, and the 2026 experience of two kinds of institution shows both faces of the coin. Corporate defined-benefit plans, which discount their liabilities at high-grade corporate yields, saw the Milliman 100 funded ratio reach 114.5 percent at September 30 with a discount rate of 6.47 percent, the highest since May 2009 and 111 basis points above a year earlier; their liabilities fell faster than their assets. An endowment has no such offset. Its liability is a spending rule indexed, in practice, to higher-education inflation of 3.5 percent and to a new federal tax, and a 2026 bond loss simply subtracts from the pool that funds both. The LDI framework we described in 2025 exists precisely to make an endowment’s balance sheet behave more like the corporate plan’s: to recognize that the institution’s obligations have a duration, and that a mismatch between that duration and the portfolio’s is a risk the committee is taking whether or not it has chosen to.
Credit: Quality Holds, the Bottom Cracks
The original report recommended high-quality corporates “for a small increase in credit risk” and warned that bonds’ historically low correlation with equities could break down. Credit has in fact been the steadier leg of fixed income in 2026, and the way it has been steady tells us where the risk now sits. Table 5 assembles the dashboard.
Table 5: Credit dashboard
| Measure | Sept 16, 2025 | Dec 31, 2025 | Oct 5, 2026 | 2026 range |
|---|---|---|---|---|
| ICE BofA US Corporate OAS | 76 bp | 79 bp | 84 bp | 73-94 bp |
| ICE BofA BBB OAS | 96 bp | 101 bp | 104 bp | 92-116 bp |
| ICE BofA US High Yield OAS | 279 bp | 281 bp | 312 bp | 260-346 bp |
| ICE BofA CCC & lower OAS | 807 bp | 885 bp | 1,211 bp | 838-1,215 bp |
| IG effective yield | 4.70% | 4.84% | 6.03% | 4.71-6.03% |
| HY effective yield | 6.41% | 6.53% | 8.15% | 6.43-8.22% |
| S&P US speculative-grade default rate (trailing 12m) | 4.6% (Aug 2025) | n/a | 3.9% (July 2026); 3.75% forecast for June 2027 | |
| Fitch US private credit default rate | n/a | n/a | 6.3% (Aug 2026), a record | |
| First-lien loan recoveries (Fitch, 2025) | 43%, lowest since 2016 | |||
| Distressed exchanges as share of defaults (Moody’s, 2025) | 65% | |||
| JPMorgan distressed loan universe | $139.8 billion (Oct 6, 2026), +90% YoY | |||
| Bank loans to non-depository financial institutions (Fed H.8) | $2.07 trillion (Sept 23, 2026), +20% YoY | |||
| Public BDC price-to-NAV (Raymond James, Q2 2026) | 0.80x; non-accruals 4.2% of cost; PIK 8.0% of income |
Sources: ICE Data Indices via FRED (BAMLC0A0CM, BAMLC0A4CBBB, BAMLH0A0HYM2, BAMLH0A3HYC, BAMLC0A0CMEY, BAMLH0A0HYM2EY); S&P Global Ratings (Aug 20, 2026); Fitch Ratings via LSTA (Aug 2026, Apr 6, 2026); Moody’s; JPMorgan (Oct 6, 2026); Federal Reserve H.8; Raymond James via press. The table reads as a market that has repriced the bottom of the capital structure by 400 basis points while leaving the top almost untouched: all but 8 basis points of the 133 basis point rise in investment-grade yields is Treasury yield.
The investment-grade market has absorbed a historic volume of issuance at spreads near the tightest of the century: $1.22 trillion in the first half, up 30 percent, with a record $145 billion in August. The spread has not widened because the buyers have shown up; foreigners own about 29 percent of U.S. corporate bonds and bought a net $251 billion in the first half, on pace to rival 2025’s record, and Goldman’s Amanda Lynam sees “a floor” under that demand. Hyperscaler spreads remain wide, as the FOMC minutes noted, but the market has distinguished between an AA-rated borrower funding a data center and the rest of the index. For a fiduciary the conclusion is practical: an all-in yield above 6 percent on investment-grade paper is attractive as income and unattractive as a spread trade, and the duration of the index, around 6.5 years, is the risk that matters. Goldman’s mortgage strategists make a point we endorse: with the IG index yielding only 16 basis points more than the agency mortgage index against a ten-year average gap of 50, and with mortgage duration a year and a half shorter, agency MBS is the better-compensated way to own high-quality spread. Oracle’s downgrade to BBB- in July, with its 2055 bonds at 77 cents on the dollar, is the reminder that even the AI complex has a bottom tier.
The high-yield market is where the quiet storm has become audible, and the sound is coming from the lowest tier. The broad high-yield spread of 312 basis points is only 33 basis points wider than when we published, but the CCC spread has gone from 807 to 1,211 basis points, above the April 2025 tariff peak and the widest in three years; Bloomberg’s CCC index crossed 1,000 basis points on September 30 for the first time since March 2023. JPMorgan counts $139.8 billion of distressed loans, up 90 percent in a year and just $4 billion below the May 2020 peak, with technology a 39 percent share, and the leveraged loan market’s distress is concentrated in software, where roughly 45 percent of sub-80 debt now sits. Defaults look benign in the headline, with S&P’s speculative-grade rate at 3.9 percent and forecast to drift to 3.75 percent by mid-2027, but the headline flatters. Distressed exchanges were 65 percent of 2025’s default events; 45 percent of this year’s global defaults were repeat defaulters, the highest share since 2019; first-lien recoveries averaged 43 percent in 2025, the lowest in Fitch’s 2016-2025 data; and the loan market’s dual-track default rate including liability-management exercises runs at 2.9 percent against a payment-default rate below 1 percent. This is what a default cycle looks like when creditors have agreed not to call it one. Our December report, The Architecture of Fragility, said spreads would “widen significantly” in 2026; on the index it did not happen, and we record the miss. In the tier where the small-cap maturity wall and floating-rate exposure actually sit, it did.
Private credit is the sleeve of the original allocation that most needs revision, because the year has delivered precisely the evidence Howard Marks asked for. His November 2025 memo, written after First Brands and Tricolor, said the problem was not “systemic” but “systematic,” and that a market of $2 trillion built in a benign era had never been tested. The test arrived. The Federal Reserve’s May Financial Stability Report counted roughly $1.4 trillion of U.S. private credit loans and $425 billion in semi-liquid vehicles, and reported that in the first quarter of 2026 redemptions from those vehicles exceeded inflows for the first time ever. Blue Owl’s technology income fund received redemption requests for 40.7 percent of its NAV in the second quarter against a 5 percent gate; Blackstone’s BCRED and Cliffwater’s interval fund have run at 10 and 16 to 17 percent; Moody’s moved its outlook on the BDC sector to negative on April 7. Publicly traded BDCs trade at 0.80 times NAV with non-accruals at 4.2 percent of cost, while the Cliffwater index and Goldman’s BDC universe report non-accruals between 1.7 and 1.8 percent, a dispersion that is itself the finding. Fitch’s measure of private credit defaults reached a record 6.3 percent in August; the Cliffwater Direct Lending Index, which returned 9.3 percent in 2025, returned 3.0 percent in the first half. First Brands’ bankruptcy was converted to Chapter 7 liquidation on August 24, with asset sales of about $194 million against more than $9 billion of liabilities; Tricolor’s former chief executive goes to trial on October 19.
Two further facts shape our revision. The first is that direct-lending spreads have widened to roughly 540 basis points on Goldman’s measure, a premium of 245 basis points over syndicated loans and the widest since 2023, so the entry point for a new, senior, drawdown-structured commitment is the best in three years. The second is the July paper by Pranjal Drall and Andrew Granato that Michael Burry has championed, arguing that private-equity-affiliated life insurers hold roughly $700 billion of opaque private credit and that state guaranty-fund mechanics would socialize losses “more sharply than banking’s federal deposit insurance”; the NAIC puts insurers’ private credit exposure at $1.21 trillion. We need not adopt the paper’s conclusions to take its lesson, which is that the private credit sleeve of an endowment portfolio is not an uncorrelated income stream; it is a claim on the same sponsor-backed, software-heavy, floating-rate borrowers that the CCC market is repricing, intermediated through vehicles whose liquidity terms were written for a different decade. We therefore reduce the sleeve, narrow it to senior secured and asset-based strategies in drawdown rather than evergreen form, and treat its liquidity as equity-like in the institution’s cash-flow model. Blackstone’s mid-year view that there is “a gap between headlines and reality” in private credit may well be right about the median loan. The fiduciary’s exposure is to the tail and to the gate, and both have arrived.
Municipals: The Worst Quarter Since 1981 and the Best Entry Since 2011
We recommended a 10 to 15 percent municipal sleeve in September 2025 on the basis of “compelling tax-equivalent yields” and strong state and local credit. The credit part held; the timing did not. The Bloomberg Municipal Index fell 4.36 percent in September 2026, its worst month since 2008, and 6.35 percent in the third quarter, its worst quarter since 1981, leaving it down 4.18 percent for the year. Supply did it: issuance ran at $458.5 billion through September, up 6.3 percent on the pace of 2025’s record $586 billion year, with a record $55.5 billion in September alone, and the 21-week streak of fund inflows ended in the week of September 30. The 30-year AAA benchmark yield crossed 5 percent on September 24 for the first time since January 2011 and stood at 5.13 percent on October 1; the 10-year stood at 4.04 percent; municipal-to-Treasury ratios reached 77 and 91 percent.
For a taxable investor, those levels are extraordinary: a 30-year AAA municipal yields the equivalent of 8.1 to 8.7 percent on a taxable bond at the top federal brackets, and a 10-year the equivalent of 6.4 to 6.8 percent against a Treasury at 5.27. The One Big Beautiful Bill Act left the tax exemption intact. State finances are weaker than a year ago but not weak: NASBO projects fiscal 2026 general-fund revenue growth of 2.2 percent, negative in real terms, with rainy-day balances of $191 billion and sixteen states budgeting withdrawals for fiscal 2027; Moody’s carries a negative outlook on cities and counties and on higher education; Brightline Florida’s $4.4 billion Chapter 11 filing on September 25 is the year’s notable default, with senior holders unimpaired. The sector is cheap for reasons of supply and duration, not solvency.
For a tax-exempt endowment, the calculus is different, and our original table did not make the distinction it should have. A 30-year municipal at 5.13 percent yields less than a 30-year Treasury at 5.64, and a tax-exempt institution captures no exemption. The case for municipals in an endowment or foundation portfolio is therefore a relative-value case, that ratios above 90 percent have historically reverted toward the 70s and that taxable municipals offer spread over Treasuries with a credit profile the sector’s default history supports, and it is a case for a tactical allocation rather than a structural one. We revise the sleeve accordingly below: zero to 10 percent for tax-exempt pools, concentrated in taxable municipals and in crossover purchases of tax-exempt paper when ratios exceed 90 percent, and a structural 10 to 15 percent only for the taxable family-office and operating-company pools that some of our clients also steward. The entry point is the best in fifteen years; the holder for whom it was built is the one who pays tax.
Public Equity Markets Through the Rates Lens
The S&P 500 closed at a record 7,818.93 on October 6, 2026, up roughly 14 percent for the year in price terms and 18 percent since the Fed’s first cut in September 2025, which is the single fact most at odds with a 5.3 percent 10-year yield and a hiking Fed. The reconciliation is earnings. Index earnings grew 51 percent year over year in the second quarter and consensus expects roughly 27 to 30 percent for the third, the highest pre-season estimate since 2021, with AI-infrastructure beneficiaries accounting for more than half of the growth and the top ten contributors for two-thirds. Forward earnings have risen faster than prices, so the forward multiple has fallen from 22 times at the start of the year to 19 times, its ten-year average, while the Shiller CAPE sits at 41.9, shy only of December 1999. Goldman’s Ben Snider describes companies as “over-earning” on gains from private AI stakes and a semiconductor margin windfall, and forecasts earnings growth slowing to 11 percent in 2027; his colleagues note that fully expensing hyperscaler capex would cut consensus 2027 earnings by 22 percent, and that breadth is the narrowest since the dot-com bubble, with the median stock 17 percent below its high while the index sits 2 percent from its own.
The rates lens sharpens this. Goldman’s dividend discount model puts the equity risk premium near 3 percent, where it has sat for two years; the simpler measure, the forward earnings yield of about 5.26 percent against a nominal 10-year yield of 5.27 percent, puts it at zero, a configuration last seen in the late 1990s and one in which, as Oppenheimer wrote on September 17, equity markets are “more vulnerable to further increases in bond yields.” The speed of the move matters as much as the level: the real 10-year yield rose 53 basis points in the month to September 24, crossing what Goldman calls the “two standard deviation speed limit” historically associated with negative equity returns, and the most rate-sensitive sectors, homebuilders, REITs, utilities and unprofitable technology, have already fallen hard while the index has not. Street year-end targets cluster at 8,000 to 8,100, and Goldman’s twelve-month target of 8,700 assumes no multiple expansion at all. Our May report, Overpriced Resilience, Underpriced Fragility, argued that downside risk had become the dominant theme to posture for and recommended an “insensitive portfolio,” index put overlays and elevated cash; the index has since risen about 4 percent while the median stock has fallen, the valuation case has not changed, and the discount rate has risen 80 basis points. We hold that view. For an endowment the relevant point is simpler still: a bill yielding 4.2 percent and a 5-year Treasury yielding 5 percent are real competitors for the marginal equity dollar, and the equity market’s 2026 resilience has been bought with an earnings surge that its own strategists expect to fade.
Alternative Investments
Venture capital. The venture market of 2026 is the most bifurcated in its history, and the bifurcation is a fixed-income story because it is about liquidity. U.S. venture deal value reached $412.7 billion in the first half, more than any prior full year, with artificial intelligence taking 86 percent and megadeals 87.5 percent of the dollars; OpenAI raised $122 billion at $852 billion and is reportedly seeking $30 billion at $1.4 trillion, Anthropic raised $65 billion at $965 billion and has slipped its IPO to November at a $2 trillion target, and SpaceX’s $75 billion listing on June 12 at a $1.7 trillion valuation was the largest in history. Exit value in the first half was $2.19 trillion, almost all of it SpaceX. Yet distributions ran at 7.9 percent of NAV against a long-run 14.5 percent; PitchBook puts the DPI of 2021-vintage venture funds at 0.05 times, the lowest five-year multiple this century; the median venture-backed IPO trails the growth index by 38 percent within 120 days; 86 of 110 IPO-ready companies are still raising privately; and fundraising reached only $72.4 billion across 405 funds in the first half, with Andreessen Horowitz, Thrive and Founders Fund taking 48 percent of it and first-time funds at their weakest pace since 2016. Our October 2025 report, The Power of Selection, argued that in venture “who you invest with is every bit as important as what you invest in”; 2026 has made manager concentration the asset class’s defining fact. The link to this report is the discount rate. At a 5 percent risk-free rate, a 2021-vintage fund that has returned 5 to 14 cents on the dollar in five years is a liability on the institution’s cash-flow statement, not an asset on its allocation chart, and the LP-led secondary market prices venture stakes at 79 percent of NAV.
Private equity. Bain’s mid-year review counted roughly 33,000 unsold portfolio companies and a fourth consecutive year of distributions below 15 percent of NAV; PitchBook finds that about 40 percent of U.S. PE-backed companies, more than $860 billion of NAV, have been held for more than seven years, the highest share since 2016. Exits recovered sharply in the third quarter, to $462.9 billion globally, the second-highest quarter on record, and continuation-vehicle volume fell to $47.4 billion year to date from 2025’s $97.6 billion record as sponsors found buyers again. Fundraising did not follow: U.S. PE raised $222 billion through September, on pace for the first sub-$300 billion year since 2020, and across private markets 78 percent of first-half capital went to funds above $1 billion. Secondary volume reached a record $118 billion in the first half at 91 percent of NAV for buyout stakes, and leveraged-loan dividend recapitalizations, the sponsor’s substitute for an exit, ran at $28.7 billion through September 2, down 39 percent, in what PitchBook calls a “higher-for-longer rate path.” The 2027-28 maturity wall that The Architecture of Fragility described has not produced the default cycle we expected at the index level; it has produced the liability-management cycle described in the credit section, and a Fed hike adds 25 basis points to the floating-rate cost of every loan in the stack.
Real estate. Private real estate marked a modest recovery through mid-year, with the NCREIF ODCE index returning 1.49 percent in the second quarter and Green Street’s commercial property price index up 4.7 percent year over year through September. Green Street’s own October 6 commentary is the one that matters: “If this rise in yields sticks, property prices are likely to go down.” Public REITs fell 5.7 percent in September. CMBS delinquency reached 8.02 percent, the highest since November 2020, with office at 12.16 percent, and the Mortgage Bankers Association counts $875 billion of commercial real estate debt maturing in 2026 and $652 billion in 2027, into a 30-year mortgage rate of 7.28 percent and a 10-year Treasury above 5. Data centers are the exception that proves the rule; they are up 28 percent this year and they are the reason the rest of the capital stack is competing for funding.
Private credit. The credit section covered the asset class; here we add only the allocation data. Private debt was the one strategy whose fundraising rose in the twelve months to June 30; in the first half of 2026 it was up 33.7 percent in dollars but down 34.6 percent in fund count, with funds above $5 billion taking 57 percent of the capital and emerging managers a record-low 5.5 percent. Evergreen direct-lending assets, which tripled from 2022, pulled back in 2026, and PitchBook’s Evergreen Index shows direct lending lagging infrastructure. The asset class is consolidating into the hands of the largest managers at the moment its underwriting is being tested, and the two facts are related.
Hedge funds. Industry assets reached a record $5.6 trillion at June 30 after the largest quarterly increase in industry capital on record, and the HFRI Fund Weighted Composite returned about 8 percent through August after a 7.5 percent first half, the best since 2021. July’s AI and momentum unwind, when Goldman’s hedge fund VIP basket trailed the equal-weight S&P 500 by 11 points in “one of the sharpest hedge fund de-grossing episodes of the past decade,” is the cautionary data point: gross leverage, net leverage and AI exposure remain above long-run averages, crowding hit a record in the second quarter, and funds carry a near-record net short in Nasdaq-100 futures. For an allocator the case for the diversifying strategies, macro, relative value and trend, is stronger in a world where Treasuries have, in SEI’s phrase, “lost their flight-to-quality positions”; AQR’s data show trend-following with a -0.28 correlation to equities over five years, the only major category with a negative sign.
The Institution’s Own Balance Sheet
The original report argued that higher operating costs were forcing endowments to lean harder on distributions, and that the pressure would convert a mark-to-market loss into a realized one. The pressure has grown, and a liability we did not anticipate has been added to it. Table 6 collects the balance-sheet data.
Table 6: The institutional balance sheet, 2026
| Measure | Latest | Prior reference |
|---|---|---|
| NACUBO-Commonfund FY2025 return (657 institutions, $944 billion) | 10.9%; 10-year 7.7%; effective spending 4.9% | FY2024 11.2% |
| Share of operating budgets funded by endowments (NACUBO FY2025) | 15.2%; withdrawals $33.4 billion, +11.0% | FY2024 14.0% |
| Cambridge Associates Endowment Radar 2025 (Mar 27, 2026) | Median dependence 17.0%; effective spending 5.4%; top quartile 5.7-9.0% | 2023 edition: 16.3%, distributions +10.5% |
| Fixed income allocation, NACUBO FY2025 | 10.7% overall; 8.2% for endowments above $5 billion | |
| FY2026 returns reported to date | Stanford 31.7%; Penn 27.4%; Washington University 37.3%; UVIMCO 27.0%; MIT 10.3%; UTIMCO 14.9%; Cambridge Associates preliminary median 15.5% | FY2025: Harvard 11.9%, Yale 11.1%, Princeton 11.0% |
| Section 4968 excise tax (tax years after Dec 31, 2025) | 1.4% / 4% / 8% tiers at $500,000 / $750,000 / $2 million per student; Yale’s FY2027 budget: $52 million to $299 million; MIT booked an $858 million increase in its deferred tax liability | 1.4% flat |
| Higher Education Price Index | FY2025 3.6%; FY2026 forecast 3.5% | CPI 2.6% (FY2025) |
| LP-led secondary pricing, H1 2026 (Jefferies) | 87% of NAV; buyout 91%; venture 79%; endowments and foundations 16% of sellers; volume $118 billion | 2025 full year $240 billion |
| Milliman 100 corporate pension funded ratio | 114.5%; discount rate 6.47%; surplus $160 billion (Sept 30, 2026) | 106.5%; 5.36% (Sept 30, 2025) |
| Milliman Public Pension Funding Index | 88.2% (July 31, 2026); record 89.1% (May 31) | 85.4% (Sept 30, 2025) |
| Public plan assumed return (NASRA, Apr 2026) | Median 7.00%; FY2025 the first year more plans raised than cut | |
| Public plans using interest rate swaps (CFTC, Carrion and Coughlan) | 43 of 153 (28%); mean position 0.09% of liabilities vs ~252% for a full hedge | |
| Pension risk transfer (LIMRA) | 2025 $48.8 billion; H1 2026 ~$6.8 billion, down 26-47% |
Sources: NACUBO-Commonfund Study of Endowments FY2025 (Feb 12, 2026); Cambridge Associates; Commonfund HEPI; university announcements (Sept 17 - Oct 2, 2026); 26 U.S.C. 4968; Yale Daily News (Oct 2, 2026); Jefferies Global Secondary Market Review (July 2026); Milliman (Oct 7, 2026; Aug 27, 2026); NASRA; CFTC Office of the Chief Economist (Aug 2024); LIMRA. The table reads as an institution whose assets had an excellent fiscal 2026, whose liabilities grew faster, and whose peers have begun to sell private assets at a discount to meet them.
Three developments define the year. The first is the tax. The One Big Beautiful Bill Act’s tiered excise tax took effect for tax years beginning after December 31, 2025, and it bites first in fiscal 2027 budgets. Yale’s tax bill rises from $52 million to $299 million; MIT recorded an $858 million increase in its deferred tax liability; Princeton, which falls below the 3,000 tuition-paying-student threshold, nonetheless cut PRINCO’s long-run return assumption from 10.2 to 8 percent, a decision that lowers the payout by roughly $500 million a year. Yale has completed 78 percent of a three-year cost reduction; Stanford cut $140 million and 363 positions; Northwestern and Duke have made comparable cuts. For a portfolio, an 8 percent tax on net investment income is, illustratively, a 36 basis point annual drag on a fund realizing 4.5 percent in taxable income and gains, and a tax that falls on realized gains changes the cost of rebalancing. The second development is federal funding: Harvard’s appeal over the government’s research-funding freeze was argued on October 5 with no ruling; Columbia, Brown, Cornell and Northwestern settled for a combined $406 million; the House’s fiscal 2027 marks cut NSF by 20 percent; and new international enrollment fell 17 percent in the fall of 2025, with NAFSA projecting a further 9.5 percent decline. The third is the response our original report predicted: endowments have become sellers. Endowments and foundations were 16 percent of LP-led secondary sellers in the first half of 2026. Yale sold roughly $1 billion of private fund stakes to HarbourVest on September 25, after $3 to 4 billion in 2025; Harvard sold about $1 billion in 2025 and issued $675 million of tax-exempt bonds in March; CalPERS sold $2.3 to 3 billion in February. These are not distress sales, and the fiscal 2026 returns that Stanford, Penn and Washington University have reported are among the strongest since fiscal 2021. But they are sales of the least liquid assets at 79 to 91 cents on the dollar to fund obligations that a fixed-income portfolio of the right duration was supposed to fund, and that is the quiet storm’s realized loss arriving through a different door.
The pension side of the ledger shows the mechanism we described working as designed where the discount rate is a market rate, and failing to engage where it is not. Corporate plans discount at high-grade corporate yields, and the rise in those yields cut their liabilities faster than their assets fell; the Milliman 100 funded ratio of 114.5 percent is a seventeen-year high, and a surplus of $160 billion has prompted a debate about how to use it that would have been unthinkable in 2020. Public plans discount at their assumed return, which the median plan still sets at 7.00 percent; their record funded ratios, 88.2 percent on Milliman’s index and 85.0 percent on Equable’s, came from the asset side, from a fiscal year in which CalPERS returned 14.8 percent and CalSTRS 13.9 percent. That is welcome and it is fragile, because an asset-driven funded ratio reverses when assets do, and a liability discounted at 7 percent did not get cheaper when Treasuries reached 5. The discount-rate relief we anticipated in 2025 went to the corporate plans; the public plans received an assumption-channel version of it, with more plans raising than cutting their assumed returns in fiscal 2025 for the first time in the data.
What has not changed is the governance gap. The study behind our 2025 claim, by Allen Carrion and John Coughlan of the CFTC’s Office of the Chief Economist, found that 28 percent of the 153 public plans it examined use interest rate swaps at all and that their mean position was 0.09 percent of liabilities against roughly 252 percent for a full hedge, which the authors called “dramatically too low.” We found no public plan that adopted swaps-based liability hedging in the year since, and the trend among the sophisticated is in the other direction: CalPERS’s total portfolio approach went live on July 1 with a 5 percent strategic leverage allowance and a 15 percent active one, Texas TRS runs at 104 percent and UTIMCO raised its leverage ceiling to 115 percent on September 1. Leverage is a return decision; a swap overlay is a risk decision. The UK’s experience in 2026 is the evidence that the second works. British schemes entered the year with collateral buffers near 300 basis points under the Pensions Regulator’s standard of at least 250, the standard we cited, and they absorbed the March gilt shock with, in Reuters’ summary, limited collateral calls; the Bank of England’s July Financial Stability Report judged the system resilient. The buffer held because it was designed to a stress that arrived. The U.S. institutions that most needed that design still do not have it, and with the MOVE index near 114, we now recommend 300 basis points rather than 250.
The last change on the balance sheet is the relationship between the two asset classes the Endowment Model relies on. The correlation between stocks and bonds that rose to 0.6 over 2022-23 has not reverted on any multi-year window. AQR measures the five-year correlation of U.S. equities and Treasuries at 0.53 through February 2026; Morningstar measures the three-year correlation of equities with long Treasuries at 0.65 at the end of 2025; J.P. Morgan’s Guide to the Markets shows the 24-month average at 0.33 for 2021-25 against -0.31 for 2000-20. Morningstar’s shorter window, January 2025 to mid-2026, shows 0.11, a reminder that the relationship is regime-dependent rather than fixed. In March 2026 equities fell 4.5 to 6 percent while the 10-year yield rose 38 basis points; in the third quarter bonds suffered their worst quarter since 1994 while the S&P 500 rose 2 percent. Bonds failed as ballast in an inflationary risk-off and then failed to matter in a rates-driven risk-on. Goldman’s asset-allocation team concluded in September that “higher bond allocations have added volatility rather than buffered portfolios” and that there is “unlikely to be an immediate return to the simple 60/40 portfolio template.” AQR’s conclusion is the more useful one for a fiduciary: a positive correlation is historically normal, bonds still diversify at any correlation below one, and the substitutes that institutions have reached for, private credit chief among them, carry an equity correlation of 0.68 and an equity beta of 0.70. The answer to a bond that no longer hedges equities is not a credit instrument that behaves like equity. It is a shorter bond, a real bond, a trend-following sleeve and a hard-asset sleeve, held in the proportions the institution’s liabilities require.
Strategic Synthesis: The Case Against Us, at Full Strength
A view that has been largely validated is the view most at risk of confirmation bias, so before restating ours we set out the opposing case as its best advocates would make it. The opposing case is that this is the moment to extend duration decisively, into the long end, and that a firm that called the storm is now at risk of standing in the rain after it has passed.
The argument runs as follows. Real yields of 2.9 percent on the 10-year and 3.35 percent on the 30-year are the highest since 2008 and above any plausible estimate of trend real growth; the 10-year yields more than its 250-year average. The Fed is hiking into a supply shock, which is historically the way hiking cycles end, and the growth data are turning: payrolls are at breakeven, residential investment is contracting, Goldman has cut its fourth-quarter growth forecast to 1.4 percent and estimates that persistence of current rates would subtract half a point from 2027 growth and push the federal debt ratio to 132 percent of GDP, an outcome the political system would answer with consolidation or with pressure on the Fed to ease. Inflation’s drivers are one-off: Goldman attributes 0.9 of the 1.0 point core PCE overshoot to tariffs and the war, trimmed-mean PCE runs at 1.8 to 2.2 percent, core CPI has run at 2.0 percent annualized for three months, and tariff pass-through is “mostly done.” Positioning is light; Goldman’s rates desk finds real-money duration underweights at levels historically consistent with a rally and five-year-forward rates a standard deviation above fair value. J.P. Morgan’s long-term assumptions, published a year ago, described “the best outlook for intermediate treasuries since the global financial crisis” and said investors are “better compensated for taking duration risk than they were in recent cycles”; Bridgewater argued that bonds “can once again play their pivotal role” at normal yields. And the arithmetic of Table 4 is symmetric: a 100 basis point rally returns 16 percent on the 30-year. If Goldman’s forecast of a 4.4 percent 10-year and a 4.9 percent 30-year by the end of 2027 is right, a long Treasury portfolio returns roughly 15 to 20 percent over fifteen months while a bill portfolio returns 5. A fiduciary who stays short for another year because the last year was bad is doing what retail investors do, and the Endowment Model exists to do the opposite.
We have examined that case for the signs of motivated reasoning in our own, and we find three places where it lands. First, our 2025 report did not forecast oil and our December report accepted the easing consensus; a firm that misses the proximate cause of a move should hold its structural explanation more loosely than it is tempted to. Second, the term premium at 0.96 percent is already the highest since 2014, and the 2026 move was, by Goldman’s account, partly a volatility and communication premium that can decay. Third, every house in Table 3 that tied the long end to the policy rate was wrong last year, and we should not assume that a view which was right once about structural factors is immune to the same error in reverse.
Having granted all of that, we do not reach the opposing conclusion, for four reasons that are about the long end specifically and not about duration in general. The first is that the compensation is not yet full. The term premium has recovered from zero to below 1 percent against a long-run average of 1.35 percent, and the structural drivers Goldman’s own rates desk calls persistent, a $2 trillion deficit, record coupon issuance from 2027, and a corporate sector borrowing $1.3 trillion to build data centers, have not been resolved by a 100 basis point move; its judgment that the rise in term premium “is unlikely to retrace” and that long forwards will “remain somewhat sticky even if inflation outcomes justify a more dovish-than-priced policy path” is our judgment too. The second is that the marginal buyer has changed. Foreign official holdings have fallen by $240 billion since February and China’s are the lowest since 2008-09; Japanese and European institutions now have domestic 30-year bonds at 4.2 and 3.8 percent; the Treasury’s own buybacks have “done little to encourage lower yields.” A long-end rally needs a buyer, and the buyer who shows up at 5.6 percent is price-sensitive in a way the Fed and reserve managers were not. The third is the oil tail. Goldman’s baseline takes Brent to $85 by December, but its upside scenario is $115 by year-end and above $120 in the second quarter of 2027, and under that scenario the Fed does not stop at one more hike; a fiduciary should not own 14 years of duration against a two-sided energy market in which one side leads to a 6 percent 30-year. The fourth is that a growth break, the scenario in which the long end rallies most, is also the scenario in which the belly rallies most per unit of risk. Goldman wrote in December 2025 that bonds hedge recession “particularly at shorter maturities” and that the long end offers “optionality” against fiscal and independence risk; the first half of that sentence is the reason to extend into the five-to-seven-year sector now, and the second half is the reason to express the long end through options rather than bonds.
The synthesis is therefore not a refusal to extend. It is a sequencing. The belly of the curve, three to seven years, offers real yields of roughly 2.5 to 2.9 percent, captures most of a rally if growth breaks, loses 4 to 7 percent if rates rise another point, and carries the income to absorb a year of that. The long end offers a higher real yield, a larger payoff in the rally and a larger loss in the fiscal tail, and its risk can be bought more cheaply through receiver swaptions, which Goldman’s cross-asset team notes are “more convex than S&P puts,” than through the bonds themselves. We extend into the former, we hedge with the latter, and we hold the physical long bond only to the extent the institution’s liabilities require.
The Conviction View: Positioning for a Five Percent World
Our view, graded for conviction, is as follows. With conviction: the long end of the Treasury curve has repriced to a structurally higher range and will not return to the 2010s; the drivers are fiscal and structural and have outlasted the supply shock that triggered the move. With conviction: in credit, quality dominates beta, the investment-grade and agency mortgage markets offer all-in yields matched only briefly at the 2023 peak, and the lowest tier of leveraged credit is in a default cycle that the headline statistics are not recording. With conviction: the gap between what institutions know about their duration mismatch and what they have done about it is the defining fiduciary failure of this cycle, and the UK’s 2026 experience proves the remedy works. Provisionally: duration is now compensated, and a staged extension into the belly of the curve, with explicit triggers, is the right response for institutions whose liquidity is secure. Provisionally: private credit’s repricing has made new senior commitments attractive while making legacy evergreen exposure a liquidity risk; the sleeve should be narrower, more senior and in drawdown form. As a tail scenario that is no longer out of the realm of possibility: a fiscal dominance regime in which the Fed is pressed to hold rates below where inflation warrants, for which gold, as The Sovereign Reset argued in January, remains the governed hedge.
Table 7 revises the model allocation we published in September 2025 for the fixed-income portion of an endowment or foundation portfolio.
Table 7: Fixed-income strategic allocation, revised October 2026
| Sleeve | Sept 2025 range | Oct 2026 range | Rationale and triggers |
|---|---|---|---|
| Treasury bills and cash (0-1 year) | 10-20% | 15-25% | Yields of 4.2% fund 12-24 months of spending, capital calls and the excise tax; the liquidity floor that prevents forced sales of private assets at 79-91 cents |
| Short Treasuries and TIPS (1-3 years) | within “short-duration” | 15-25% | 4.8% nominal; the holding sleeve from which the staged extension is funded |
| Intermediate Treasuries and TIPS (3-7 years) | 20-30% (intermediate Treasuries) | 15-25%, built in stages | Real yields of 2.5-2.9%; add one-third now, one-third if the 10-year exceeds 5.50%, one-third on evidence of a growth break (three months of negative payrolls or unemployment above 4.6%) |
| Long Treasuries (10-30 years), physical | within intermediate | 0-10% | Only to the extent of a modeled liability match; the balance of long-end exposure through receiver swaptions and swaps |
| Investment-grade corporates and agency MBS (1-7 years) | 20-30% (IG corporates) | 15-25% | All-in yields of 6.0%; favor agency MBS over corporates while the IG-MBS yield gap is 16 bp against a 50 bp norm; no reach into BBB cyclicals with 2027-28 maturities |
| Municipals | 10-15% | 0-10% for tax-exempt pools; 10-15% for taxable pools | For tax-exempt institutions: taxable municipals and crossover purchases when ratios exceed 90%; for taxable clients: 30-year AAA at 5.13% is an 8.1-8.7% taxable equivalent |
| Private and alternative credit | 10-20% | 5-10% | Senior secured and asset-based, drawdown structures only; treat liquidity as equity-like; new commitments at 540 bp spreads are attractive, legacy evergreen exposure is a gate risk |
| Hard-asset hedge (gold), held outside fixed income | n/a | per The Sovereign Reset | The hedge for the fiscal dominance tail; rebalance within bands after the 25% drawdown from January’s $5,595 peak |
The table reads as a portfolio that is shorter at the core, more deliberate in the middle, and more selective at the edges than the one we published a year ago. Its purpose is unchanged: to make the fixed-income allocation the part of the portfolio that funds the institution’s obligations in every scenario rather than the part that surprises the committee in one of them.
What would make us wrong, and what we will watch. If the 10-year rises through 5.75 percent with the term premium above 1.25 percent, the long end will have reached the compensation we think it lacks, and the sequencing above accelerates. If core PCE re-accelerates above 3.5 percent and the Fed moves past 4.5 percent, the staged extension pauses at its first tranche and the view that inflation’s drivers are transitory, which we have borrowed from Goldman, will have failed. If payrolls turn negative for three consecutive months or unemployment exceeds 4.6 percent, the growth break has arrived, the belly rallies, and the risk shifts to being too short; the third tranche deploys. If investment-grade spreads exceed 150 basis points, the quality call becomes a buying opportunity in credit beta rather than a caution. And if the Treasury’s November 4 refunding announces coupon increases earlier than the “next several quarters” language implies, the supply argument for caution at the long end strengthens. We will report against each of these in the next cycle.
Prioritized recommendations for endowment and foundation CIOs
Re-run the asset-liability model with a 5 percent world as the base case, not the stress case. Model liabilities with the institution’s effective spending rate (the Endowment Radar average is 5.4 percent), HEPI inflation of 3.5 percent, the applicable excise tax tier, and the capital-call schedule of the private portfolio. Stress the result against a further 200 basis point rise, a 100 basis point rally, and a 2027 growth break. Present the duration of the liabilities and of the assets side by side; the gap is the institution’s largest unmanaged risk.
Fund the next 12 to 24 months of obligations in bills and 1-to-3-year Treasuries before any other decision. At 4.2 to 4.8 percent this costs nothing in expected return relative to the Aggregate and it is the only reliable protection against becoming the 16 percent of secondary sellers who are endowments and foundations.
Extend duration in three stages, into the belly, with the triggers above written into the investment policy statement in advance. The first third moves now from the 1-to-3-year sleeve into 3-to-7-year Treasuries and TIPS. Pre-authorizing the second and third tranches removes the committee’s timing decision from the moment of maximum stress.
Hold the long end through derivatives, not bonds, except for a modeled liability match. Adopt a swaps and swaptions policy now, with a collateral buffer sized to 300 basis points of adverse movement, pre-agreed collateral sources, and a standing operating procedure. The 2025 report said of this step that “we cannot underscore this enough”; 2026 has shown the cost of not taking it, and the UK has shown that the framework holds under stress.
In credit, buy quality for income and refuse beta. Short and intermediate investment-grade and agency mortgages at 6 percent; no additions to high yield or leveraged loans while CCC spreads exceed 1,000 basis points and first-lien recoveries sit at 43 percent; a full look-through of existing managers for software concentration, payment-in-kind income and liability-management exposure.
Reduce and restructure the private credit sleeve. Lower the range to 5 to 10 percent; direct new commitments to senior secured and asset-based strategies in drawdown vehicles at today’s 540 basis point spreads; submit redemption requests on evergreen vehicles whose gates are already binding rather than waiting for them to lift; and model the sleeve’s liquidity as equity-like in the cash-flow plan.
Treat municipals as a relative-value position for tax-exempt pools and a structural one for taxable pools. Crossover and taxable-municipal purchases when ratios exceed 90 percent; a full 10 to 15 percent allocation only where the tax exemption is captured.
Keep the hard-asset hedge, governed. The gold sleeve described in The Sovereign Reset and refined in our March gold update is the hedge for the one scenario this report cannot hedge with bonds, a Fed pressed into accommodating the fiscal position. Rebalance within bands; do not chase.
Educate the committee before the next shock, not during it. The mechanics of duration, swaps, collateral and liquidity budgeting belong in a standing board curriculum. The committees that acted in 2026 were the ones that had done this work in 2025.
The quiet storm has made landfall, and the institutions that weathered it were not the ones that predicted the oil shock or the hike. They were the ones whose balance sheets had been built so that neither mattered. That remains the whole of the fiduciary playbook, in a 5 percent world as in any other.
We would be honored to serve as a resource for your organization. Contact us today to discuss:
Contact | Connect with Financial Experts — Wind River Capital Strategies