Can I retire at 47 with $2.1 million?

Plan Mirror · Updated 2026-10-27

Steady returns

$1.9M

left at 95, in today's dollars

A century of markets

94.6%

194 of 205 start dates last to 95

1,000 random futures

65%

650 of 1,000 futures last to 95

Ask most retirement calculators whether $2.1 million is enough to stop working at 47 and you get one number back. We ran one household through all five of Plan Mirror's tests, and the tests disagree. On steady returns the money lasts to 95 with room to spare. Replayed through a century of real markets it survives 194 of 205 start dates. Against 1,000 random futures built on today's more cautious return outlook, one in three runs short.

That disagreement isn't noise. Each test asks a different question, and between them they show exactly where an early retirement like this one is strong and where it could break.

The household

A married couple in Texas, which has no state income tax, retiring this year at 47 and planning to 95. Counted out, that is 49 years of retirement, through 2074.

Household

A couple, both 47

Retires

At 47, in 2026

Plans through

Age 95 (2074)

Invested

$2,100,000: $700,000 taxable brokerage, $1,200,000 traditional 401(k)/IRA, $200,000 Roth

Spends

$84,000 a year in today's dollars, taxes and healthcare included

Social Security

$32,000 a year combined, in today's dollars, from 67

401(k)/IRA before 59½

Allowed, with the 10% early-withdrawal penalty

Assumes

6% average return, 2.5% inflation

Taxes

Married filing jointly, Texas (2026 rules)

The spending figure is all-in. It covers the mortgage, groceries, travel and, until Medicare at 65, health insurance bought on the open market. Plan Mirror also pays the household's income tax out of it, so $84,000 is what leaves the portfolio, not what is left after tax.

$84,000 is 4% of $2.1 million. If that number rings a bell, it's the famous 4% rule. But the research behind that rule tested 30-year retirements. This one has to last 49 years, and for the first 20 of them the portfolio pays for everything.

Answer one: steady returns

The first thing Plan Mirror does with a plan is the simplest: grow the portfolio at 6% a year, raise spending with 2.5% inflation, work out each year's taxes and see what is left. On that path the couple finishes at 95 with $6.1 million, which is about $1.9 million in today's dollars.

Underneath that green line, the plan moves through three stages.

The first years: no income tax. Plan Mirror spends the taxable brokerage account first. Most of each sale is the couple's own money coming back, and the gains are small enough that the engine estimates no income tax at all until 2035.

From 56 until Social Security: the long bridge. When the brokerage account runs dry, withdrawals come from the 401(k) and IRA, and every dollar of those is taxed as income. The estimated tax bill jumps to about $18,000 a year. It is also the stretch where access matters. Withdrawals before 59½ normally carry a 10% penalty on top of the income tax, and in these runs we told Plan Mirror this couple would simply pay it. That costs about $11,000 a year for the four years until 59½, $37,392 in all, and it comes out of the same $84,000. There is a penalty-free way across those years, and we test it after the simulations.

From 2046: Social Security takes some of the load. At 67, Social Security starts paying $52,436 in that year's dollars, and the amount pulled from the portfolio drops from $134,287 to $85,208.

Dollar figures in the table are in each year's own dollars, so they rise with inflation. The $6.1 million at the end is the same buying power as $1.9 million today.

Two quick stress tests

A steady return is a useful baseline and a fiction: no portfolio earns 6% every year. So Plan Mirror reruns the plan twice more, and draws both on the same chart.

With spending 10% higher, the money still lasts, and the couple ends with about $795,000 in today's dollars. With returns 2 points lower, 6% becomes 4% and the portfolio runs out in 2065, at 86.

That asymmetry is the first real lesson. This plan can absorb a lot of extra spending, but it is sensitive to the return it earns over a 49-year horizon. Which raises the obvious question: what returns should we expect?

Answer two: a century of real markets

The historical replay stops pretending returns are smooth. It takes actual U.S. stock market returns and actual inflation, quarter by quarter, and runs the plan from every start date since 1926 Q3 that has a full 49-year window behind it. Retire in 1966, and the plan lives through the inflation of the 1970s. Retire in 1929, and it meets the Depression in its first year.

194 of those 205 start dates last to 95. That's 94.6%.

The failures are the interesting part. The worst start was 1929 Q4, right after the crash, and in that history the money is gone by 2040, at 61. That is before Social Security ever arrives.

In this plan, the first ten years decide most of what follows.

Plan Mirror measures this directly. Take the quarter of start dates with the worst first ten years of returns: 21.6% of them ran out of money, against 5.4% of all start dates. Most of those failures don't arrive during the bad decade itself. They come years later, because a portfolio that shrinks while it pays the full $84,000, with Social Security still 20 years away, has too little left to recover. Retirement researchers call this sequence-of-returns risk. An early retirement, with decades of withdrawals ahead and no other income at the start, is especially exposed to it.

Two things to know about this replay. It holds 100% U.S. stocks, which is bolder than most retirees' portfolios: a mix with bonds would have cushioned the worst starts and given up some of the best. And the past century was very good to U.S. stocks. The median history ends with $16.4 million in today's dollars, far more than the couple could ever spend. Nobody should plan on the next fifty years repeating that.

Answer three: 1,000 possible futures

The Monte Carlo test asks the question the replay can't: what if the future is less generous than the past? It generates 1,000 random market paths. The typical one grows at the same 6% a year as the steady projection, about 3.5% after inflation, roughly half the 7.2% a year, after inflation, that U.S. stocks delivered over the last century. Around that typical path, returns swing as much as the stock market's do (15% volatility a year), and inflation holds at 2.5%.

650 of the 1,000 paths last to 95: 65%. The median path ends with $1.5 million in today's dollars, a little below the steady projection's $1.9 million. That gap is the cost of uneven returns: a bad year hurts more when you're withdrawing from the portfolio than a good year helps.

The paths that fail tell the same story as history, only louder. Among the quarter of paths with the weakest first ten years, 82.4% run out of money. Yet most failing paths make it past Social Security: 9.7% of paths have run dry by 67, against 35% by 95. The damage is done in the first decade, and the money takes years more to run out.

So the two tests don't contradict each other. The replay says this plan would have survived almost anything the last century threw at it. The Monte Carlo run says that if the next half century earns half the real return of the last one, a 4% withdrawal starting at 47 works in two futures out of three. In your own plan you can set the return and volatility you believe in and rerun both.

The penalty or a 72(t) schedule

The tax code has one penalty-free way into a 401(k) or IRA at any age. Under section 72(t), you take substantially equal periodic payments: an amount set by an IRS formula from the account balance, paid every year for five years from the first payment or until 59½, whichever is longer. Change the payment before then and the 10% is charged on every payment so far, with interest.

So we gave the same couple a second plan. Instead of spending the brokerage account first, they start a 72(t) schedule the year they retire: $60,000 a year from 2026 through 2039, when they are 60. The IRS works out the payment one account at a time, so they first roll the 401(k) into their IRA and run the schedule from that one account. At the 5% interest rate the IRS always permits, its fixed amortization method gives up to $70,518 a year on the IRA's $1.2 million, so the payment fits. The brokerage account pays the rest of each year's spending. With no penalty to pay, we told Plan Mirror they would take nothing else from the 401(k) or IRA before 59½. Here are both plans through all five tests:

On steady returns, the two plans end in the same place. Both finish with about $1.9 million in today's dollars. The difference is the tax bill on the way there. The first plan pays no tax until the brokerage account runs out, then about $18,000 a year in tax and penalty once that money is gone. The 72(t) plan pays a small bill from the first year, $2,840, because each year's payment is taxed as income a little at a time, mostly in the lowest brackets. In 2036 it pays $1,878. Their spending is all-in, so whatever isn't paid in tax or penalty is theirs to spend.

In the historical replay, the 72(t) plan is at least as safe. 194 of the 205 start dates last to 95, the same count as the first plan. The failures come later. In the worst start, 1929 Q4, the money runs out in 2042, at 63, against 2040 on the first plan. By the time Social Security starts, 1.5% of the 72(t) histories have run short, against 2.4%.

The Monte Carlo futures agree. 652 of 1,000 paths last on the 72(t) plan, against 650. By 67, 8.9% have run short, against 9.7%.

So for this couple, a 72(t) schedule started at retirement saves the $37,392 penalty and costs nothing any of the five tests can see. What it asks for is commitment. The couple takes $60,000 from the IRA every year from 47 to 60, in good markets and bad. They can't take more in a year that needs it, and the only change allowed is a one-time switch to the IRS's required-minimum-distribution method, which recalculates the payment each year.

So can they retire at 47?

On the past century's markets, comfortably. On today's more cautious outlook, it's a plan that works most of the time and fails in a recognisable way. All five tests side by side:

What decides it is the first ten years: the return the portfolio earns while it carries the whole $84,000, with Social Security still 20 years away. That's useful to know, because a weak first decade is visible while it's happening. A couple in this position should decide before they hand in notice what they would do if it comes, whether that's spending less or earning something for a few years, and test that response rather than assume it. They also need to decide how they'll reach the 401(k) before 59½. Paying the penalty is the simplest route. A 72(t) schedule started at retirement does at least as well in every test and saves the $37,392, if they can live with a fixed draw from the IRA until they are 60.

What these runs leave out

• One tax profile. The household files jointly in a state with no income tax. A state income tax would come out of the same $84,000.
• Your real Social Security number. $32,000 a year is an example. The estimate on your Social Security statement assumes you keep earning until you claim, so stopping at 47 lowers it.

What we'll test next. First, the other way to time a 72(t): spend the brokerage account first and start the schedule only when it runs low, which in a bad decade comes sooner. These runs also leave out the two classic defences against a weak first decade. One is a cash buffer: keep a few years of spending in cash and draw on it after a bad quarter, instead of selling stocks while they're down. Whether that actually pays for itself is less obvious than it sounds. The other is a spending rule that trims withdrawals when the portfolio falls behind its path. And there's the opposite kind of test: a large one-off cost, such as a new roof or a health crisis, landing at a random point in retirement. Those are the next questions we'll put to this household.

Every figure on this page comes from 2 recorded cases run together in Plan Mirror under 2026 tax rules. Each was run through the plan with its two stress tests, a historical replay of U.S. stock returns and inflation from 1926 Q3 to 2026 Q2, and 1,000 Monte Carlo paths. It is an illustration, not financial, tax, investment or legal advice. See the methodology.

Related scenarios

• 72(t) vs the rule of 55: one couple at 55, three ways into the 401(k)
• Can I retire at 55 with $2 million?
• Does a cash buffer make retirement at 55 safer?
• Can I retire at 60 with $1 million?