How Long Will $1 Million Last in Retirement? Illiquid Assets 2026
Some links pay us a referral fee; each one says so. Disclosure
Quick Answer
On October 22, 2026 the U.S. Treasury's 10-year par real yield was 2.64% (5-year 2.33%, 20-year 2.91%, 30-year 3.05%). If $1,000,000 earned exactly 2.64% after inflation and you took a fixed real amount out at the start of each year, it would pay 39 full years at 4% ($40,000 a year), 27 at 5%, 21 at 6%, 14 at 8% and 11 at 10% (our arithmetic). It would not run out only if spending were 2.57% or less of the starting balance. The answer scales with the starting amount: $2,000,000 at 4% is $80,000 and lasts the same 39 years. History is less smooth than a yield: in the 69 rolling 30-year windows starting in 1928 through 1996 (NYU Stern's S&P 500 and Treasury return series), a 60% stock, 40% Treasury-bond mix that spent 4% of its starting balance a year, raised with inflation, lasted all 30 years in 65 windows and ran dry in the four starting in 1965, 1966, 1968 and 1969; at 5% it ran dry in 19 windows and at 6% in 33 (our arithmetic). Then comes money that cannot leave on demand. Apollo Diversified Real Estate Fund, which offers to buy back 5% of its shares each quarter, repurchased about 27% of the shares tendered at its May 5, 2026 deadline. Re-tendering everything at that fill returns half of a position in 3 quarters and 90% in 8, and if 40% of the $1,000,000 sits in a sleeve that pays nothing, the other $600,000 alone covers a 6% ($60,000) rate for 11 years, not 21 (our arithmetic). Its next window closes November 3, 2026. As of October 22, 2026.
Key Takeaways
- The real yield moves, and so does the answer. The 10-year par real yield was 2.88% on October 1, 2026 and 2.64% on October 22 (U.S. Treasury). At a constant return equal to each, $40,000 a year from $1,000,000 lasts 42 years on the first date and 39 on the second (our arithmetic). A constant real return equal to a par real yield is our simplification: a portfolio earns something else, and nothing here includes taxes, fees, Social Security or a pension.
- Years against spending rate at the 10-year real yield of 2.64%: 3.5% lasts 50 years, 4% lasts 39, 5% lasts 27, 6% lasts 21, 8% lasts 14. With no real growth at all, 4% lasts 25 years. The break between lasting and not lasting is r / (1 + r): 2.28% at the 5-year yield, 2.57% at the 10-year, 2.83% at the 20-year and 2.96% at the 30-year (our arithmetic).
- The IRS divisor is a different spending rule. For a traditional IRA, IRS Publication 590-B Table III divides the prior December 31 balance by 24.6 at age 75 (4.07% of the balance), 16.0 at age 85 (6.25%) and 8.9 at age 95 (11.24%), so on $1,000,000 the minimum is $40,650 at 75 (our arithmetic). It is a floor on pre-tax accounts, not a plan.
- Order matters. In our 69 historical windows a 60/40 mix spending 4% a year lasted 30 years in 65 and ended with a median 1.64 times its starting real balance, but the window starting in 1973 ended with 0.23 times and the one starting in 1982 with 5.36 times; the four that failed started in 1965, 1966, 1968 and 1969 (our arithmetic from Damodaran's series). The same three returns in a different order leave $804,100 or $835,600 after taking $50,000 a year, which is a $31,500 gap.
- Fill rates are the gate. Apollo Diversified Real Estate Fund repurchased approximately 30% of shares tendered on November 4, 2025 and 27% on both February 3 and May 5, 2026. If every share still held is re-tendered each quarter at 27%, half of a position is out after 3 quarters, 71.6% after 4 and 90% after 8; a $300,000 position pays $81,000 in the first quarter and $8,948 in the eighth (our arithmetic). BREIT, by contrast, fulfilled all $2.1 billion of repurchases requested in the first half of 2026.
- If the sleeve that cannot be sold produces nothing, the rest has to carry the spending. At 6% spending, a pot with 60% illiquid has $400,000 that can be sold and lasts 7 years, against 21 years for the whole $1,000,000; with 80% illiquid and 8% spending the sellable $200,000 lasts 2 years (our arithmetic at the 2.64% real yield).
CSV · 220 rows
How long $1,000,000 lasts: Treasury real yields, IRS Uniform Lifetime Table, 69 historical 30-year windows and illiquid fund repurchase terms
220 rows: the October 2026 Treasury par real yield curve, the IRS Uniform Lifetime Table and the share of the balance it implies, years of withdrawals by spending rate and real return, results of 69 historical 30-year windows for three portfolios and six spending rates, SEC Rule 23c-3 terms, Apollo Diversified Real Estate Fund fill rates and BREIT repurchase limits with accession numbers, and the re-tender and frozen-sleeve tables (our arithmetic).
The arithmetic: how many years each spending rate buys
The method is simple enough to check by hand. Start with $1,000,000. At the start of each year take out a fixed amount in today's dollars (the dollar figure rises with inflation, so the real amount is constant), then let what is left earn a real return. Count the full withdrawals paid before the balance cannot cover the next one. The returns in the columns are the Treasury's par real yields for October 22, 2026, the yield a Treasury Inflation-Protected Security of that maturity would pay above inflation if bought at par. We use each as a constant return for the whole period, which assumes the yield never changes, so read it as the floor-case arithmetic of a ladder that earns the yield and not as a forecast of any portfolio (our reading). Figures are before income tax and fees.
| Spending rate | Per year on $1,000,000 | 0% real return | 5-year yield 2.33% | 10-year yield 2.64% | 20-year yield 2.91% | 30-year yield 3.05% |
|---|---|---|---|---|---|---|
| 3% | $30,000 | 33 | 61 | 74 | 99 | 100+ |
| 3.5% | $35,000 | 28 | 45 | 50 | 57 | 62 |
| 4% | $40,000 | 25 | 36 | 39 | 42 | 44 |
| 5% | $50,000 | 20 | 26 | 27 | 29 | 29 |
| 6% | $60,000 | 16 | 20 | 21 | 22 | 22 |
| 7% | $70,000 | 14 | 17 | 17 | 18 | 18 |
| 8% | $80,000 | 12 | 14 | 14 | 15 | 15 |
| 10% | $100,000 | 10 | 11 | 11 | 11 | 11 |
Source: U.S. Treasury, daily par real yield curve rates, October 22, 2026; years of full annual withdrawals are our arithmetic. 100+ means money is still left after 100 withdrawals. Because spending is a share of the starting balance, the same years apply to $500,000, $2,000,000 or $3,000,000: only the dollar amount changes.
Two readings of the table. First, between 3% and 4% the answer changes by decades, and between 6% and 10% it changes by 10 years: the gap is widest at the low rates, where the real yield does most of the work. Second, the line between a pot that outlasts the spender and one that does not is r / (1 + r), the most a start-of-year withdrawal can be without eating principal: 2.28% against the 5-year yield, 2.57% against the 10-year, 2.83% against the 20-year and 2.96% against the 30-year (our arithmetic). Anything above that line uses up principal and the only question is how fast. This page gives no spending rate to use; the table shows what each one does.
What the IRS divisor takes out of a traditional IRA
If part of the $1,000,000 is in a traditional IRA, the IRS sets a minimum. IRS Publication 590-B (2025), Table III, Uniform Lifetime, gives the “Applicable Denominator” by age. The prior December 31 balance divided by the denominator is the required minimum distribution (RMD); the full rule, including what happens when an IRA holds a fund that does not sell in time, is in our guide to the required minimum distribution with illiquid IRA assets.
| Age | Denominator | Share of prior December 31 balance | RMD if the balance were $1,000,000 |
|---|---|---|---|
| 73 | 26.5 | 3.77% | $37,736 |
| 75 | 24.6 | 4.07% | $40,650 |
| 80 | 20.2 | 4.95% | $49,505 |
| 85 | 16.0 | 6.25% | $62,500 |
| 90 | 12.2 | 8.20% | $81,967 |
| 95 | 8.9 | 11.24% | $112,360 |
| 100 | 6.4 | 15.63% | $156,250 |
Source: IRS Publication 590-B (2025), Table III (Uniform Lifetime), page 67; shares and dollars are our arithmetic on a balance that stays at $1,000,000.
At 75 the divisor takes 4.07%, more than the 2.57% a 10-year real yield can carry without eating principal, and the share rises every year: 6.25% at 85 and 11.24% at 95. The RMD is a minimum for pre-tax accounts and applies to the IRA balance, not to the whole pot, and it is taxable income (our reading). Social Security or a pension covers part of spending in many cases and lowers the rate that must come from savings; our Social Security claiming age chart shows the benefit side.
Order of returns: the same average, a different balance
Sequence of returns risk is the arithmetic of withdrawing while prices fall: dollars taken out after a loss do not come back when prices recover. A three-year example with invented numbers makes the mechanism visible. Both columns have the same three real returns and the same compound factor (0.80 x 1.10 x 1.10 = 0.968); only the order differs. Each year $50,000 comes out at the start (5% of the original $1,000,000).
| Item | Loss first | Loss last |
|---|---|---|
| Real return in years 1, 2, 3 | -20%, +10%, +10% | +10%, +10%, -20% |
| Balance after year 1 (take $50,000, then apply the return) | $760,000 | $1,045,000 |
| Balance after year 2 | $781,000 | $1,094,500 |
| Balance after year 3 | $804,100 | $835,600 |
| Balance after year 3 with no withdrawals at all | $968,000 | $968,000 |
Illustrative inputs, not historical; the arithmetic is ours.
Now with real history. The NYU Stern dataset (Aswath Damodaran) gives annual returns for the S&P 500 with dividends and for a constant-maturity 10-year Treasury bond from 1928 through 2025, and the annual CPI change from FRED. We deflate each return by that year's CPI change and run every 30-year window that starts in 1928 through 1996, which is 69 windows, spending a fixed share of the starting balance each year, raised with inflation, with the mix rebalanced annually. A window “runs out” when the balance cannot cover the next full withdrawal.
| Spending rate (on $1,000,000) | 60% stocks / 40% bonds | 100% S&P 500 | 100% 10-year Treasury bond |
|---|---|---|---|
| 3% ($30,000) | 0 of 69 | 0 of 69 | 11 of 69 (worst start 1941: 24 years) |
| 4% ($40,000) | 4 of 69 (worst start 1966: 25 years) | 5 of 69 (worst start 1929: 21 years) | 42 of 69 (worst start 1941: 18 years) |
| 5% ($50,000) | 19 of 69 (worst start 1969: 17 years) | 15 of 69 (worst start 1929: 15 years) | 52 of 69 (worst start 1940: 15 years) |
| 6% ($60,000) | 33 of 69 (worst start 1973: 13 years) | 22 of 69 (worst start 1929: 12 years) | 59 of 69 (worst start 1941: 12 years) |
| 7% ($70,000) | 40 of 69 (worst start 1973: 11 years) | 28 of 69 (worst start 1929: 10 years) | 64 of 69 (worst start 1940: 11 years) |
| 8% ($80,000) | 55 of 69 (worst start 1973: 9 years) | 38 of 69 (worst start 1929: 8 years) | 68 of 69 (worst start 1941: 9 years) |
Source: Damodaran, NYU Stern, histretSP.xls (annual returns 1928 to 2025; CPI percent change from FRED); windows and counts are our arithmetic. The 69 windows overlap heavily, so they are not 69 independent histories, and the series is a published academic compilation, not a government source. Taxes and fees are not deducted.
At 4% the 60/40 mix failed in four windows, those starting in 1965, 1966, 1968 and 1969, the first of them lasting 27 years and the worst, the 1966 start, 25 years. The same 4% left a median of 1.64 times the starting real balance in the 65 windows that lasted, with a range from 0.09 to 5.36 times, and the 1973 start survived 30 years with only 0.23 of the original real balance left. At 5%, 19 windows failed and 16 of them started between 1956 and 1973; at 6%, 33 of 69 failed. The columns show that the answer depends on what the money is in: all stocks failed in 5 windows at 4% but in 38 at 8%, and 10-year Treasury bonds alone failed in 42 windows at 4%, the worst being the 1941 start.
When part of the money cannot be sold in a quarter
A table of years assumes every dollar can be spent. Vehicles that hold real estate or loans often cannot return cash on demand, and the rules differ by structure.
Interval funds. SEC Rule 23c-3 governs the periodic repurchase offers of interval funds: the offer amount “shall not be less than five percent nor more than twenty-five percent of the common stock outstanding on a repurchase request deadline” (17 CFR 270.23c-3(a)(3)). If holders tender more, the fund “may repurchase an additional amount of stock not to exceed two percent” and otherwise repurchases “on a pro rata basis” (270.23c-3(b)(5)); proceeds are due seven days after the pricing date (270.23c-3(a)(4)). So a fund that offers the 5% minimum and receives requests for 18.5% of its shares (ignoring the optional extra 2%) fills about 27% of every holder's request (our arithmetic). That is what Apollo Diversified Real Estate Fund's Form N-CSRS for the period to March 31, 2026 reports: it repurchased approximately 30% of the shares tendered on November 4, 2025 and 27% on February 3, 2026, and in a later note, approximately 27% of the shares tendered at the May 5, 2026 deadline, or 7,026,204 shares for $172,360,589. Its Form N-23C3A of September 24, 2026 opens the next offer for up to 5% of outstanding shares, closing November 3, 2026 at 4:00 p.m. Eastern. Our Apollo Diversified Real Estate Fund repurchase schedule follows every window.
Non-traded NAV REITs. BREIT's Form 10-Q for the quarter ended June 30, 2026 limits repurchases to “no more than 2% of our aggregate NAV per month” and “no more than 5% of our aggregate NAV per calendar quarter”, with a 98% price for shares held under a year, and says it fulfilled $2.1 billion of repurchases requested in the first half of 2026, which was all requests received. The same plan has pro-rated before. See our BREIT review and NAV REIT redemption status.
| Vehicle | Cap on what leaves | Latest result we read | Filing |
|---|---|---|---|
| Apollo Diversified Real Estate Fund | Offer of 5% of shares each quarter (Rule 23c-3 allows 5% to 25%) | About 30%, 27% and 27% of tendered shares repurchased at the November 4, 2025, February 3, 2026 and May 5, 2026 deadlines | Form N-CSRS 0001398344-26-010317; Form N-23C3A 0001398344-26-017504 |
| Blackstone Real Estate Income Trust (BREIT) | 2% of aggregate NAV a month and 5% a quarter | All repurchase requests fulfilled in the six months ended June 30, 2026 ($2.1 billion) | Form 10-Q 0001662972-26-000111 |
What does a fill rate do to a holder who wants out? Assume every quarter you tender everything you still hold and the fund fills the same share (our assumption; real fills change each quarter and NAV moves between request and pricing). The share of the position still held is (1 - fill) raised to the number of quarters.
| Fill rate of shares tendered | Still held after 1 quarter | After 2 | After 4 | After 8 | Quarters until at least half is out | Quarters until at least 90% is out |
|---|---|---|---|---|---|---|
| 100% (BREIT, first half of 2026: all requests fulfilled) | 0% | 0% | 0% | 0% | 1 | 1 |
| 30% (Apollo, November 4, 2025) | 70.0% | 49.0% | 24.0% | 5.8% | 2 | 7 |
| 27% (Apollo, February 3 and May 5, 2026) | 73.0% | 53.3% | 28.4% | 8.1% | 3 | 8 |
| 10% (hypothetical) | 90.0% | 81.0% | 65.6% | 43.0% | 7 | 22 |
Fill rates from Apollo Diversified Real Estate Fund Form N-CSRS and BREIT Form 10-Q; the remaining shares and quarter counts are our arithmetic.
In dollars, a $300,000 position tendered in full each quarter at a 27% fill (our arithmetic):
| Quarter | Cash received | Still held | Cumulative cash |
|---|---|---|---|
| 1 | $81,000 | $219,000 | $81,000 |
| 2 | $59,130 | $159,870 | $140,130 |
| 3 | $43,165 | $116,705 | $183,295 |
| 4 | $31,510 | $85,195 | $214,805 |
| 5 | $23,003 | $62,192 | $237,808 |
| 6 | $16,792 | $45,400 | $254,600 |
| 7 | $12,258 | $33,142 | $266,858 |
| 8 | $8,948 | $24,194 | $275,806 |
The next table turns the question around. If the sleeve that cannot be sold pays nothing and spending continues, the sellable part has to carry it. The table gives full annual withdrawals the sellable part alone can pay, at the 10-year real yield of 2.64%, for $1,000,000 split by the share that is illiquid (our arithmetic). The first row is the whole pot, from the first table.
| Illiquid share | Sellable | 4% spending ($40,000) | 6% ($60,000) | 8% ($80,000) | 10% ($100,000) |
|---|---|---|---|---|---|
| 0% ($0) | $1,000,000 | 39 | 21 | 14 | 11 |
| 20% ($200,000) | $800,000 | 27 | 16 | 11 | 8 |
| 40% ($400,000) | $600,000 | 18 | 11 | 8 | 6 |
| 60% ($600,000) | $400,000 | 11 | 7 | 5 | 4 |
| 80% ($800,000) | $200,000 | 5 | 3 | 2 | 2 |
Our arithmetic from the Treasury par real yield; the illiquid part is assumed to pay no cash at all. It is not a forecast that any fund will suspend repurchases.
Compare the two tables. At a 27% fill the exit of 90% takes 8 quarters, two years. Where the sellable part pays only two or three years of spending (3 years for 80% illiquid at 6%, 2 at 8% and 2 at 10%), spending would meet that wait; where it pays 8 years or more, as in the 40% rows at 8% spending and below, the gate is a delay and not a stop (our reading).
Verdict: what the numbers settle and what they leave open
- The table is the arithmetic, not a rate. At 2.64% real, each extra point of spending takes years off: 39 at 4%, 27 at 5%, 21 at 6%. What rate fits depends on how long the money must last, what else pays you, and how much of it you could sell in a bad quarter, none of which this page knows.
- History says a constant yield hides the bad starts. The four 60/40 failures at 4% began in 1965 to 1969; in the data, 1966 had an S&P 500 return of -9.97% with CPI up 3.46%, and 1969 had -8.24% with CPI up 6.20%.
- An IRA has its own clock. The IRS divisor rises with age whatever the pot earns; if the IRA holds an illiquid fund, the December 31 value and the repurchase terms matter. See RMDs with illiquid IRA assets.
- The share that cannot be sold is a spending input. Counting a sleeve as spendable only after its fill rate is known turns a gated fund from a surprise into a number.
What a reader can do with this
- Put your own spending on the table. Divide a year of spending from savings by the starting balance, find the row, and read the years at the 10-year column, then at the 0% column to see the no-growth floor. Re-run it when the Treasury real yield changes; the daily curve is published at home.treasury.gov.
- Split the pot by what can be sold in a quarter. List each holding as sellable now, sellable in a window (a quarterly fund) or not sellable on a schedule (a rental, an LLC interest, a private note). Use the frozen-sleeve table for the last group and the fill-rate table for the second.
- Read the fund's own numbers. The repurchase cap and each window's result are in the Form N-23C3A, the N-CSR or N-CSRS and, for a non-traded REIT, the 10-Q on SEC EDGAR. Our real estate crowdfunding liquidity guide compares how platforms and funds let money out, and what an interval fund is explains the mechanics.
- Check the share of net worth. Our retirement wealth by age page shows how large the real estate share is for households aged 55 to 74.
- See how a single-premium annuity is taxed. Our annuity payout calculator shows how a single-premium annuity is taxed; this page does not weigh it against anything.
FAQ
Update alert · free
An email when the retirement withdrawal math for illiquid assets numbers change
When a rate, rule or filing behind this page changes: what changed, the one number that matters, and the source to check it yourself.
Sources, read and saved on October 23, 2026: U.S. Treasury, daily par real yield curve rates for October 2026 (home.treasury.gov); IRS Publication 590-B (2025), Table III, and 26 CFR 1.401(a)(9)-9(c) (eCFR, current to October 6, 2026); Aswath Damodaran, NYU Stern, Historical Returns on Stocks, Bonds and Bills, 1928-2025 (histretSP.xls), with CPI from FRED; 17 CFR 270.23c-3 (eCFR, current to October 6, 2026); Apollo Diversified Real Estate Fund, Form N-CSRS (0001398344-26-010317) and Form N-23C3A (0001398344-26-017504); and Blackstone Real Estate Income Trust, Form 10-Q for the quarter ended June 30, 2026 (0001662972-26-000111). Years, shares, windows and dollar tables are our arithmetic; the scripts are saved with this page. This is analysis of public documents, not investment, legal or tax advice.
Keep reading.
- 0120 min read
ACAP Strategic Fund (XCAPX): $11.9B, 25% Offers, 1.3% Taken
ACAP Strategic Fund (XCAPX), the $11.9B Alkeon long/short interval fund: 25% repurchase offers, 1.3% taken in March 2026, fees, shorts and swaps in the filings.
- 0222 min read
AFA Asset Based Lending Fund: Mount Logan Vote, Fees, Tender
AFA Asset Based Lending Fund (AMCLX) will vote on a Mount Logan takeover of its adviser. The proxy's fees, the November 6, 2026 tender and what is still blank.
- 0324 min read
Asset Depletion Mortgage 2026: Fannie, Freddie, Lender Formulas
How Fannie Mae, Freddie Mac and four lenders turn IRA and savings balances into mortgage income: divisors of 60 to 360 months, and the tax cost of withdrawing.