4% of Nothing

Someone at Vanguard last year pulled $1,900 from their retirement account. Not to retire. to survive the month. Medical bills, mostly. Sometimes rent. The median hardship withdrawal in 2025 was $1,900, which is roughly one month's rent in a mid-tier American city, or one emergency room visit without complications. The person who takes that withdrawal pays income tax on it, plus a 10% penalty if they're under 59½, and the money that would have compounded over thirty years into something meaningful vanishes to keep the lights on this month.

This is the act of un-saving. Not failing to save. actively consuming what was already saved, because the present demanded it. And it is happening at a scale the American retirement system has never recorded before.

The personal savings rate. the number the Bureau of Economic Analysis publishes every month, the one we track as The Buffer. says Americans are saving. The accounts say otherwise.

What the Rate Actually Measures

The BEA personal savings rate is not what most people think it is. It does not measure how much money Americans put into bank accounts. It is a residual: disposable personal income minus personal outlays. When the BEA reports that the savings rate was 4.0% in February 2026, it means that after tallying all income (including employer pension contributions, imputed interest, and statistical adjustments that have nothing to do with a family's checking account) and subtracting all spending, there was 4.0% left over. The number captures a national accounting identity. It does not capture whether anyone could cover a car repair.

What stopped me here was how far this residual has fallen from where it was before the pandemic. The 2019 average was 7.31%. The 2015–2019 average was 6.14%. At 4.0%, the current rate sits 3.31 percentage points below the 2019 baseline. a gap that represents hundreds of billions of dollars in annual household buffer that simply isn't being generated anymore. The rate bottomed at 2.2% in Q2 2022, and its partial recovery since then has been read as good news. It is good news in the way that a patient's fever dropping from 104 to 102 is good news.

But even that qualified improvement exists only in the accounting residual. The actual reserve. the money in the account when the bill arrives. tells a different story.

The Reserve That Didn't Recover

Every year, Bankrate asks Americans whether they could cover a $1,000 emergency expense from savings. In 2025, only 41% said yes. down from 44% the prior year, interrupting three years of post-pandemic progress. Nearly 1 in 4 Americans reported having no emergency savings at all. Not insufficient savings. Zero.

The Federal Reserve's own Survey of Household Economics and Decisionmaking. the SHED, which has asked the famous $400 question since 2013. found that 63% of adults could cover a $400 emergency exclusively with cash or its equivalent in 2024. That sounds reasonable until you notice the number hasn't moved. It was 63% in 2022. It was 63% in 2023. It peaked at 68% in 2021 and has been stuck five points below its high for three consecutive years, while unemployment fell, GDP grew, and the savings rate itself climbed off its floor.

This is the gap the headline number hides. The rate recovered. The reserve didn't. One is an accounting residual that rises when income outpaces spending in the aggregate. The other is the actual balance in the actual account of a person facing an actual expense. They are not measuring the same thing, and for three years now they have been moving in different directions. or, more precisely, one has been moving while the other sits perfectly still.

The Autopilot and the Override

If the savings rate is the decoy and the emergency reserve is the reality, the 401(k) hardship withdrawal is the confession. It says: the buffer is gone, the credit cards are full, and the only account with money in it is the one I'm not supposed to touch until I'm sixty.

In 2025, 6.0% of eligible Vanguard participants took at least one hardship withdrawal. a record, and triple the pre-pandemic rate of 2.0% in 2019. The rate has risen for six consecutive years. Medical expenses were the leading reason. I've now tracked five separate indicators across three statistics pages this quarter (including credit card default statistics), and the hardship withdrawal series is the only one that has moved in the same direction every single year since 2019. no dip, no plateau, no ambiguity. Straight up.

In aviation, auto-enrollment is the autopilot: it flies the plane without the pilot's hands on the controls. A hardship withdrawal is the emergency override. the pilot grabs the yoke because something demands immediate attention. The problem is that pulling the override disconnects the autopilot, and the plane doesn't automatically return to its original course. A hardship withdrawal doesn't get repaid. The compounding is permanently lost. Jeff Clark, Vanguard's head of defined contribution research, framed this with uncomfortable clarity: "It's inadvertently providing a financial safety net because if they hadn't been auto-enrolled, they might not have had those assets to tap into for an emergency." Read that sentence twice. The system's greatest innovation. defaulting people into saving. created the resource its most stressed users now consume.

And the split screen is staggering. Average 401(k) balances at Vanguard hit $168,000 at year-end 2025, up 13%. Hardship withdrawals also hit a record. Both numbers are true simultaneously. But the averages obscure the K: median balances were $44,115. roughly a quarter of the average. which means a small number of large accounts are pulling the mean skyward while the person withdrawing $1,900 for a medical bill sits far below that midpoint. Balances up. Withdrawals up. The boulder rolls uphill and downhill at the same time.

The Sisyphean Ledger

Albert Camus asked whether Sisyphus was happy. whether the man condemned to push a boulder uphill for eternity could find meaning in the repetition. The American savings system now poses a version of that question. Participation rates are at record highs. Plan design has never been more sophisticated. Forty-five percent of Vanguard participants increased their deferral rates in 2025, according to Vanguard's own workplace data. The system is building retirement wealth more efficiently than at any point in its history. And simultaneously, it is consuming that wealth at record rates to fund the present.

The pre-pandemic baseline makes this visible. In 2019, the savings rate averaged 7.31%, the SHED $400 number was 63% (where it still is), and hardship withdrawals ran at 2.0%. The debt service ratio was 11.63%. Today the debt service ratio is 11.32%. actually slightly better than 2019. which means the debt burden alone doesn't explain why people are raiding retirement accounts at three times the pre-pandemic rate. Something else is doing the work. As I wrote in the Q1 2026 buffer assessment, I think it's the residual effect of the 2021–2023 price level reset: wages have caught up with inflation in the aggregate, but the savings buffer that was drained during the catch-up period was never rebuilt. The rate looks fine. The reserve was consumed and not replaced.

Mark Hamrick, Bankrate's senior economic analyst, put it plainly: "We are essentially a paycheck-to-paycheck nation. Fewer Americans have the equivalent of a financial safety net to cover inevitable unexpected expenses, despite low unemployment and steady growth. This is one of the consequences of elevated prices stemming from inflation, the impacts of which are still being felt." The key phrase is despite. Despite low unemployment. Despite steady growth. Despite a savings rate that technically recovered to 4.0%. The conditions that are supposed to produce savings are present. The savings are not.

The Honest Uncertainty

Here is what I don't know: whether the hardship withdrawal trend is a permanent feature of the post-SECURE Act landscape or a cyclical response to the price-level shock that will fade as real wages compound. Regulatory changes in 2018\u20132019 made hardship withdrawals easier to access. The Bipartisan Budget Act of 2018 eliminated loan-first requirements and six-month contribution suspensions. Vanguard itself noted that withdrawals have risen six years in a row since the rule change. It is genuinely possible that some of the increase reflects access rather than distress. people who would have borrowed from family or carried the credit card balance now have a cleaner path to their own retirement money. The data can't tell us yet how much of the 6.0% is new need and how much is new access to an old need.

Bankrate's Greg McBride offers a frame that captures the bind without resolving it: "With more than one-third of Americans prioritizing both emergency savings and credit card debt, it underscores how many households are in the position of having both high-cost credit card debt and being under-saved for emergencies." These households are not failing at one financial goal. They are failing at two simultaneously, and the goals compete. every dollar toward the credit card is a dollar not in the emergency fund, and every dollar in the emergency fund is a dollar not reducing the average credit card interest rate above 21%. The counterforce to the hardship withdrawal trend would be real wage growth sustained long enough to rebuild the buffer. That hasn't happened yet. But it could.

The Eccles Building, completed in 1937 during the Depression that gave birth to the savings rate as a national statistic, still houses the institution that publishes the SHED survey, still tracks the debt service ratio, still reports the macro numbers that say the American household is managing. and the same institution's own survey finds that the share of adults who can handle a $400 emergency hasn't moved in three years. The rate says 4.0%, the SHED says 37% can't cover $400, and the Vanguard data says 6% of workers are eating their own futures at a median bite of $1,900; these are the indicators worth watching through 2026, alongside the personal savings rate itself and the Bankrate emergency savings share, because if the American Distress Index. currently at 64.0, in the Elevated zone, with buffer depletion accounting for over a fifth of its weight. is measuring the distance between what the system reports and what households experience, then the falsifiable prediction is this: the savings rate could climb back to 5% by year-end 2026 without the SHED $400 number moving at all, because the rate has never been the reserve, and until the reserve itself recovers, the rate is just a number about a country that exists only in the accounting.