ABP, the pension fund for Dutch government and education workers, had €543bn of invested assets at 30 September 2026. On 1 January 2027, it intends to move its 3.2 million members into the new Dutch pension system (ABP Q3 2026 report; summary annual report 2025, p. 3).
The qualification changes the verdict. ABP is not trying to maximise the growth of its assets; it manages assets against pension liabilities, and about half of its interest-rate risk is hedged. In the third quarter of 2026 it lost 4.3% while its funding ratio rose from 126.9% to 132.2%. Rising rates cut the present value of its promises, and revised life-expectancy assumptions added more than 1.5 points to the ratio (Q3 report, pp. 1–2).
The household question is which payments your investments need to cover. Compare your projected pension income with expected retirement spending, then separate the savings you need soon from the personally owned assets you can leave invested for longer. ABP's funding ratio cannot tell you what your own pension will pay.
What is in the fund?
At 30 September 2026, ABP reported 42.9% fixed income, 31.6% equities, 16.2% alternative investments and 9.5% real estate. After developed-market equities, the largest line is a hedging programme: from the third quarter, ABP groups its government bonds and its interest-rate hedge as liability-driven investments, or LDI, at 24.3% of assets. ABP describes LDI as "investments that we hold to limit the impact of interest rate changes on pension coverage" (Q3 2026 report, p. 2).
Actual weights at 30 September 2026: developed-market equities 27.7%; government bonds and LDI 24.3%; corporate bonds and private debt 12.8%; real estate 9.5%; private equity 8.4%; infrastructure 6.7%. Sources: ABP Q3 2026 report, appendix, p. 3 and ABP Jaarverslag 2025, p. 29 for the strategic weights; allocation data (CSV). We grouped ABP's lines into eleven categories; its published category weights sum to 100.2% before a -0.3% overlay.
The strategic mix for 2024-2026 puts 59% in what ABP calls zakelijke waarden, or real assets (equities, property, infrastructure, private equity and commodities) and 40% in fixed income. Actual private equity, at 8.4%, sits above its 6% strategic weight. ABP builds most of its 4% strategic commodity weight with derivatives, so the reported 0.6% is unlikely to show the full exposure (Jaarverslag 2025, pp. 29 and 120).
The hedge is the part a balance-sheet snapshot hides. At the end of 2025, ABP hedged 49.1% of its interest-rate risk. Its investments excluding swaps had a duration of 3.6 years; with interest-rate swaps the portfolio's duration rose to 8.0 years, against 15.2 years for the liabilities. A one-point fall in rates would have added €44.4bn to the investments and €75.0bn to the liabilities (Jaarverslag 2025, pp. 26 and 118).
ABP also hedged 25% of the currency risk on its real assets in the main developed-market currencies in 2025, and all of it on fixed income (p. 119). MSCI World in euros, the comparison below, carries its full currency exposure. Costs were 0.45% of average invested assets in 2025 before transaction costs of 0.08% (summary annual report, p. 4).
Why does ABP use this structure?
ABP's stated objective is a pension, not a growth rate. Its 2025 summary says that for its members it strives "for a balance between purchasing power and a stable pension, partly depending on financial market results" (summary annual report, p. 7). Until the switch, its yardstick is the funding ratio: assets divided by the present value of the pensions it has promised, discounted with the interest-rate curve published by the Dutch central bank.
That explains the hedge. When long rates fall, the promised pensions become more expensive in today's money; the swaps gain value and cushion part of that rise. ABP's 2025 report says the hedge protects pension assets against a possible fall in rates and also anticipates a controlled transition to the new scheme (Jaarverslag 2025, p. 26).
Its investment beliefs supply the rest: risk must be taken to earn returns, asset allocation matters more than selection within a category, a long horizon opens opportunities shorter-term investors lack, and diversification improves the expected return-risk balance, although "the added value of diversification is not endless" (our translation of Jaarverslag 2025, p. 28). ABP reports an average return of about 6% a year over 30 years and expects around 4.5% in coming years in a scenario where high equity valuations normalise (investment results page).
The transition shaped two recent decisions. In its implementation plan of 2 July 2026, ABP says it had raised its interest-rate hedge to 50% in recent years and decided not to reduce investment risk before the switch: "Equity risk is a rewarded risk, but the return on equities is uncertain in the short term" (our translation). Selling equities and buying them back after the transition would expose members to bad timing, it argues (implementation plan, p. 26).
ABP also rejected extra protection of the funding ratio during the transition, first in early 2024 and again on review in 2025 and 2026. It lists the costs and a smaller chance of growth among the drawbacks, and notes that "full protection of the funding ratio is not possible, only a reduction of the risk of a lower funding ratio" (our translation of p. 28).
The bet is visible in the rules: below a 101.5% funding ratio on 31 December 2026, pensions would be cut; above 109%, ABP can compensate for past missed increases and grant everyone an extra increase (financial dashboard).
After the switch, risk will depend on age. Under the plan, all existing pensions, including those already in payment, would be converted into personal pension pots invested collectively (transition plan, p. 33). The degree of protection allocated through a "protection return" rises in a straight line from 10% at age 45 to 75% at 63. Younger members' exposure to the return portfolio may exceed their pot, up to 120%, because the scheme lifts the usual borrowing restriction (implementation plan, table 5, p. 28; transition plan, p. 7).
From December 2007 to December 2025, €100 compounded at ABP's published returns became €244.93. At MSCI World's net euro total returns, it became €487.43. ABP beat the index in five of the 18 years; it lost less in 2008 and more in 2022. The paths reinvest returns and omit ABP's contributions and pension payments; ABP's returns are after asset-management costs, and the index carries no implementation costs.
December 2007 to December 2025: €100 becomes €244.93 at ABP and €487.43 at MSCI World; 2008 cost ABP 20.2% and the index 37.64%, 2022 cost ABP 17.6% and the index 12.78%. Sources: ABP investment results 2008-2025, cross-checked against the multi-year tables in ABP's annual reports for 2014, 2020 and 2025; MSCI World NETR EUR; annual returns and wealth (CSV).
| Calendar years 2008-2025 | ABP | MSCI World net EUR |
|---|
| Geometric annual return | 5.10% | 9.20% |
| Ending value of €100 | €244.93 | €487.43 |
| Sharpe, annual observations | 0.42 | 0.55 |
| Sortino, annual observations | 0.65 | 0.93 |
| Worst year | -20.2% (2008) | -37.64% (2008) |
| Return in 2022 | -17.6% | -12.78% |
Both ratios use the same 18 calendar-year observations, with 3-month Euribor as the cash benchmark and the downside target. They are annual-observation ratios and should not be compared with ratios computed from monthly data. Unlike CalPERS and Yale, ABP's smoother path did not buy it a better ratio: its lower volatility came with an even lower excess return over cash.
The gap opened after 2016. ABP compounded at 6.41% a year from 2008 to 2016, against 7.08% for the index. From 2017 to 2025 it compounded at 3.82% against 11.36%, and from 2021 to 2025 at 1.37% against 13.07%. This year, ABP returned 2.0% to the end of September, while MSCI World in euros gained 15.55% by our calculation (Q3 2026 report, p. 3).
At ABP's measured risk, a 60% MSCI World / 40% modelled euro-bond mix compounded at 6.85% a year, against ABP's 5.10%. Sources: the same ABP and MSCI inputs, the ECB's 10-year AAA euro-area par yield and 3-month Euribor; world/bond frontier data (CSV). These are historical comparisons with annual rebalancing and no costs, not proposed allocations.
ABP sits below this restricted frontier: the 60/40 mix had 11.44% measured risk against ABP's 11.38%. The comparison is narrower than ABP's job. The listed mix carries no liability hedge, no costs, no currency hedge and no property or private assets. Adding back ABP's 2025 costs of 0.53% of assets would still leave more than one point of the 1.75-point annual gap.
ABP's own benchmark points the same way. It reports an excess return of -2.4% against its composite benchmark in 2025, -0.8% a year over five years and -0.1% over ten, and attributes the five-year shortfall mainly to private investments and equities (summary annual report, p. 4; Jaarverslag 2025, p. 69).
The funding ratio tells a different story from the asset return. In each period below, the column a pension member feels moved against the column an investor would read first.
| Period | ABP return | Funding ratio, start to end |
|---|
| 2014 | +14.5% | 105.9% to 101.1% |
| 2022 | -17.6% | 110.6% to 110.9% |
| 2025 | -1.6% | 111.7% to 123.5% |
| Q3 2026 | -4.3% | 126.9% to 132.2% |
| October 2026 | +€10.6bn | 132.2% to 128.8% |
Sources: Jaarverslag 2014, pp. 4 and 6, summary annual report 2025, p. 3, Q3 2026 report and the October 2026 dashboard, which gives the October return in euros only. In 2025, the change in the interest-rate curve added 21.4 points to the funding ratio while the return subtracted 1.7 (Jaarverslag 2025, p. 26).
Copio's reading
Is it optimal for geometric growth?
The asset record trailed MSCI World, but it cannot establish whether ABP chose the best mix for its pension promises. For assets alone, the objective in copio's lens is expected log wealth over a long horizon after costs. ABP's objective is different: stable, inflation-aware pensions paid from a portfolio measured against liabilities, under regulatory capital rules. The fair test is whether a feasible alternative would have grown the funding ratio faster with similar risk to pension payments.
The record gives three tests. First, concentration and joint losses: equities, alternatives and real estate made up 57.3% of assets in September, and ABP's annual returns correlated 0.89 with MSCI World. With about a third in listed equities, the fund took much of the index's annual pattern with much less of its return.
Second, protection: the hedge split published for 2021-2025 shows ABP's investment portfolio compounding at 3.25% a year, but only 1.37% after the hedge, which cost 5.1 points in 2022 and 4.0 in 2025 (Jaarverslag 2025, p. 69). That drag is the cost of insurance against falling rates in years when rates rose.
The funding ratio still rose from 93.5% at the end of 2020 to 123.5% at the end of 2025, because rising rates cut liabilities by more than the hedge lost; in 2025 alone the curve change added 21.4 points.
Third, costs and cash flows: 0.53% of assets went on costs in 2025, and pension payments of €17.9bn exceeded net contributions of €16.9bn (summary annual report, pp. 3-4). An all-equity index is not a like-for-like test of meeting those payments through a downturn.
Over the full period the funding ratio grew slowly: from 105.5% at the end of 2010 to 123.5% at the end of 2025, about 1.06% a year, while the assets compounded at 5.55% a year. The discount rate in ABP's tables changed definition over the period, from a nominal market rate in the 2014 report to the central bank's term structure later, so this is a rough measure.
The decisive comparison still missing is a hedged listed portfolio run against the same liabilities and payments. ABP's total asset growth trailed the simple listed comparison. That comparison does not isolate what private assets contributed or show which mix could better meet the same commitments.
Is it crash aware and protected?
ABP is aware of the crash that matters to a pension fund, a fall in interest rates, and is only half-protected against it; it is not protected against an asset crash. In 2008, ABP lost 20.2% against 37.64% for MSCI World in euros, and over 2008-2009 together it lost 4.08% against 21.47%. The more equity-heavy index lost more; nothing in the fetched reports shows a dedicated crash payout.
In 2022, rising rates turned the hedge into a loss: ABP lost 17.6% while world equities lost 12.78%, and our modelled 30-year AAA euro bond lost 44.15%. The same thing happened on a smaller scale in the third quarter of 2026, when LDI lost 16.9% and world equities in euros gained 2.55%. In both cases the funding ratio held or rose as liabilities fell further; the Q3 2026 figure also includes the revised life-expectancy assumptions.
The mirror image arrived in October 2026. Rates fell, ABP earned €10.6bn, and its funding ratio still dropped 3.4 points to 128.8%, because liabilities rose by €19bn against assets' €11bn (financial dashboard). That is what a hedge of about half the rate sensitivity means: the other half of a rate fall reaches the funding ratio. A lower annual volatility than equities does not earn ABP a crash-protection label; its protection is against one liability risk, and partial by choice.
Is it diversified without giving up growth?
ABP owns many asset classes, but its annual returns still moved closely with world equities and it compounded more slowly than the listed comparisons. On calendar-year returns from 2008 to 2025, ABP correlated 0.89 with MSCI World and 0.40 with our modelled 10-year AAA euro bond. The two listed assets were close to uncorrelated over the period, at 0.06, but fell together in 2022.
Observed correlation of calendar-year returns, 2008-2025: ABP with MSCI World 0.89; with the modelled 10-year AAA euro bond 0.40; MSCI World with that bond 0.06. Source: ABP's published returns, MSCI World NETR EUR and the ECB's AAA par yield curve; correlation matrix (CSV). These are 18 annual observations, not crash correlations; the euro bonds are modelled, not funds.
The positive link to bonds is consistent with ABP's exposure to rates through both bonds and swaps. The 60/40 comparison grew faster at similar measured annual risk, before implementation costs, but did not hedge the same commitments.
Private assets add a measurement problem. ABP's own 2025 report describes a lag in private-equity valuations relative to public markets, which it expects to normalise (Jaarverslag 2025, p. 68). Appraisal values can make the annual path look smoother than the economic loss. Copio does not grade ABP in the portfolio checker: its swap-based rate hedge and private assets cannot be represented faithfully by a portfolio of UCITS ETFs.
What can a household copy?
ABP's structure starts from a list of payments it must make and their value in today's money. A household can make the same list: rent or mortgage payments for the next few years, a dated purchase, and the income it expects to need in retirement beyond the state pension. The funding-ratio question is then which savings are set aside for each payment, not how volatile the whole portfolio is.
That list changes how to read a bond loss. A bond whose maturity and currency match a known bill can fall in price without changing the amount it is due to pay, provided the issuer pays as promised. A long bond that might have to be sold early still exposes you to interest-rate risk: our modelled 30-year bond lost 44.15% in 2022.
The new ABP scheme also writes its age rule in advance: protection rises from 10% at 45 to 75% at 63, and younger members can have up to 120% exposure to the return portfolio. A household can write its own rule before a crash. For a household, exposure above 100% means borrowing, with interest costs and the risk of forced selling; our Kelly criterion guide explains why leverage raises expected log growth only up to a point and lowers it beyond.
Our studies of CalPERS and Japan's GPIF apply the same questions to other pension funds. For why the stock-bond mix behaved so differently in 2008 and 2022, see why bonds protected in 2008 but failed in 2022.
FAQ
What does ABP invest in?
At 30 September 2026, ABP held 42.9% fixed income, 31.6% listed equities, 16.2% alternative investments and 9.5% real estate, with invested assets of €543bn. Its largest fixed-income line, 24.3%, is the liability-driven investment portfolio, which groups government bonds with the interest-rate hedge. Private equity was 8.4% and infrastructure 6.7%.
How has ABP performed against MSCI World?
From 2008 to 2025, ABP's published returns compounded at 5.10% a year, against 9.20% for MSCI World net total return in euros, by our calculation. €100 became €244.93 against €487.43. ABP's annual-observation Sharpe ratio was 0.42 against 0.55, and its Sortino ratio 0.65 against 0.93. ABP lost less in 2008 but more in 2022.
Why did ABP's funding ratio rise when its return was negative?
Higher interest rates lower the present value of the pensions ABP must pay. In the third quarter of 2026, its assets fell from €568bn to €543bn after a -4.3% return, while its liabilities fell from €447bn to €411bn. The funding ratio rose from 126.9% to 132.2%. ABP also revised its life-expectancy assumptions, which added more than 1.5 percentage points to the ratio.
What changes for ABP on 1 January 2027?
ABP intends to move to the solidarity-based contract of the new Dutch pension law, with a personal pension pot for every member, including those already retired. How much it can distribute depends mainly on the funding ratio on 31 December 2026. Below 101.5%, pensions would be cut; above 109%, ABP can compensate for past missed increases and grant an extra increase.
Is ABP's portfolio protected against crashes?
It is built to protect the funding ratio against falling interest rates, not the asset value against every crash. ABP lost 20.2% in 2008, against 37.64% for MSCI World in euros, but lost 17.6% in 2022, when the index lost 12.78%. In the third quarter of 2026 its liability-driven investments lost 16.9% while world equities rose.