The Leaky Bucket: Why Retirement Savings Drain Before Retirement

The American retirement system is designed around a simple premise: money goes in over a working life and comes out at the end of it. What the data shows is a bucket with a hole in a specific place. Research covering 162,360 terminating employees across 28 plans found that 41.4% cashed out at job separation — and roughly 85% of those who cash out take the entire balance. Cashouts account for close to 90% of all leakage, dwarfing hardship withdrawals and loan defaults combined. The striking feature is that this is not a habit but a moment: a decision made once, at a job change, frequently on a small balance, with consequences measured across decades. This report examines why the moment produces the decision, and what changes it.

The core thesis

Retirement leakage is usually framed as a discipline problem — people raiding their future to fund their present. Our thesis is that the data doesn't support that framing, and that the dominant form of leakage is better understood as a design failure at a specific transition point.

Three findings drive this. First, cashouts at job separation account for nearly 90% of leakage, not the hardship withdrawals and loans that receive most of the attention. Second, only about a third of cashouts stem from genuine financial emergencies according to industry analysis — meaning the majority are avoidable in the sense that the participant was not forced. And third, two-thirds of cashout leakage involves accounts under $7,000, which are precisely the balances where the friction of moving the money exceeds its felt value to the participant and where the compounding loss is largest in proportional terms.

Put together: the system loses most of its money not from people in crisis but from people changing jobs with a small balance and taking the path of least resistance. That's a plumbing problem, and plumbing problems have plumbing solutions.

The second half of the thesis connects to everything else this desk covers. The population most exposed is the one with the least buffer — and the evidence on that link is unusually clean, with participants holding a modest emergency cushion showing dramatically lower leakage across all three channels. Retirement leakage is, in substantial part, the emergency savings problem arriving years later and denominated in decades of compounding.

Two-thirds of cashout leakage involves accounts under $7,000 — balances small enough that moving them feels like more trouble than they're worth, and early enough that they'd have compounded the longest.

Three channels, one dominant

Channel	What happens	Share of leakage
Cashouts at job separation	Participant takes a lump-sum distribution instead of leaving the balance or rolling it over	Close to 90% of total leakage
Hardship and in-service withdrawals	Withdrawal while still employed, for defined hardship reasons or after age 59½	Second largest, historically a small share of participants annually
Loan defaults	An outstanding plan loan not repaid, treated as a distribution	Smallest — historically around 3% of borrowers default

The proportions matter because attention is allocated inversely to them. Public discussion of retirement leakage focuses heavily on hardship withdrawals — the emotionally legible case of someone in crisis — and on plan loans, which feel like the obviously risky feature. Both are real and neither is where the money goes.

Estimates suggest as much as 2% of assets may leak out of the system annually, against a contribution rate of roughly 10% of assets. That ratio is the whole argument: a fifth of what goes in is coming back out before it's supposed to, and closing the largest channel would be worth more than almost any increase in contribution rates.

Why job separation is the leak

The moment has a specific structure that makes cashing out the default rather than a choice.

The participant is prompted. Leaving a job triggers paperwork asking what to do with the balance, presenting distribution as one option among several. A participant who does nothing at other times is here required to do something.

The alternatives require effort. Rolling over into a new employer's plan or an individual account means opening something, initiating a transfer, and following it through — administrative work at a moment already full of administrative work. Taking the check requires signing one form.

Income has frequently stopped. Job separation and cash need are correlated, which means the prompt arrives precisely when the money is most tempting and least replaceable.

The balance feels small. A few thousand dollars reads as a windfall rather than as retirement savings, particularly early in a career when retirement is abstract.

And a finding worth sitting with: research examining the composition of balances found that leakage increases with the proportion of employer contributions in the account — that a higher match rate is associated with a higher probability of cashing out. The pattern aligns better with behavioral explanations than economic ones, and the implication is uncomfortable: money the participant didn't personally contribute feels less like theirs to protect. A benefit designed to build savings can, at the exit point, make draining them easier.

The small-balance problem

The concentration in small accounts is the most actionable fact in this entire area. Two-thirds of cashout leakage involves accounts under $7,000.

Why small balances leak disproportionately:

• Friction dominates value. The effort of a rollover is roughly constant regardless of balance, so it looms largest on the smallest accounts.
• Plans may force them out. Balances below certain thresholds can be distributed or rolled into a default individual account without the participant's active election, and a participant who receives an unexpected check or loses track of a small account is a participant who has effectively leaked.
• They belong to younger workers, who change jobs most often — the average worker changes jobs many times, with a large share of those changes occurring before age 25 — and who have the longest compounding horizon ahead of them.
• They don't feel like retirement money. The framing effect is real and it scales inversely with balance.

The compounding arithmetic is what makes this consequential rather than trivial. A small balance cashed out in a worker's twenties represents a materially larger sum at retirement — which is why industry analysis estimates that addressing the small-balance problem specifically could retain an estimated $1.6 trillion in the retirement system over 40 years. The leak is small at every individual point and enormous in aggregate, which is the signature of a structural rather than behavioral problem.

The infrastructural response — automatic portability, where small balances follow a worker to their new employer's plan without requiring action — addresses exactly this. It's the same insight as automatic enrollment: the default determines the outcome, so put the default where you want people to end up.

What a cashout actually costs

Three visible costs and one invisible one.

• Ordinary income tax on the distribution, at the participant's marginal rate.
• A 10% early withdrawal penalty generally applying before age 59½, with defined exceptions.
• Mandatory withholding — plans are generally required to withhold 20% of a distribution paid directly to the participant, which means the check is smaller than the balance and the participant may still owe more at filing.
• The compounding forgone, which is invisible, unmeasured at the moment of decision, and by far the largest of the four.

The behavioral failure is precisely that the first three are salient and the fourth isn't. A participant sees a reduced check and reasons about the haircut; almost nobody computes what the balance would have become. The cost that should dominate the decision is the one the decision environment doesn't display — which is a design observation, not a criticism of the participant.

Worth noting alongside: industry analysis finds that only about a third of cashouts stem from genuine financial emergencies, and that participant regret commonly follows the decision. That combination — mostly avoidable, frequently regretted — is what distinguishes this from a rational liquidity choice.

Loans and the job-change trap

Plan loans occupy a more defensible position than their reputation suggests, with one sharp exception.

The case for them: a repaid loan produces limited permanent loss, since the participant is largely paying themselves back, and the rate is frequently lower than the alternatives available to someone who would otherwise use the revolving credit our revolver analysis describes. Historically, roughly 27% of participants have carried a loan at some point and only about 3% of borrowers defaulted — which is a low failure rate for any credit product.

The trap: job separation. An outstanding loan balance that isn't repaid within the applicable window is generally treated as a distribution — triggering income tax and the early withdrawal penalty at the moment the participant has just lost their income. A loan that was performing fine becomes permanent leakage because of an employment event unrelated to the borrower's ability to repay it.

Two practical implications. The loan decision should account for job change probability, which for younger and hourly workers is high. And the real cost of a plan loan is contingent rather than stated: cheap if you stay, expensive if you leave, and you frequently don't control which.

The secondary cost is that borrowed funds aren't invested while outstanding, and some participants reduce contributions while repaying — a compounding loss that doesn't appear as leakage in any statistic but has the same effect.

Hardship withdrawals

Hardship withdrawals receive the most attention and represent a relatively small share of participants annually, though the rate rises noticeably during periods of broad economic stress — which makes it a useful indicator of household distress even where its dollar volume is modest.

The features that matter:

• They're permanent. Unlike a loan, there's no repayment — the money is gone from the account.
• Taxes and penalties apply, so a participant taking a hardship withdrawal to cover a specific expense must withdraw substantially more than the expense.
• Eligibility is defined by plan rules and regulation, covering categories including certain medical expenses, purchase of a principal residence, tuition, and prevention of eviction or foreclosure.
• They're a distress signal — a participant reaching this channel has generally exhausted cheaper options, which connects to the triage ordering in our shortfall guidance.

The honest framing for an individual facing one: a hardship withdrawal to avoid foreclosure or eviction may be entirely rational, because the alternative outcome is worse and less reversible. A hardship withdrawal to pay unsecured debt generally isn't, because that debt is negotiable, dischargeable, and time-limited in ways retirement savings can't be rebuilt — the same asymmetry our equity extraction report identifies when unsecured obligations are converted into secured ones.

The emergency savings finding

The most useful result in the recent literature, and the one that reframes leakage as a liquidity problem rather than a savings-discipline problem.

Analysis of participants in employer-administered plans compared those with at least a modest emergency cushion against those without. Controlling for income, age, tenure, and plan, participants with emergency savings were:

• 19 percentage points less likely to take a plan loan.
• 17 percentage points less likely to take a hardship withdrawal.
• 43 percentage points less likely to cash out.

The cashout figure is the one that matters most, because cashouts are the dominant channel. A relatively small amount of accessible liquid savings is associated with a dramatically lower probability of destroying a much larger amount of long-term savings.

Two contextual notes strengthen the finding's importance. A majority of participants in these plans are hourly workers, who face roughly triple the income volatility of salaried peers — meaning the population most exposed to leakage is the one whose income arrives least predictably, the pattern our income volatility analysis documents. And the causal direction is plausibly bidirectional — people who can maintain emergency savings may differ in other ways — but the policy implication holds either way: building accessible savings alongside retirement savings protects both.

This is why workplace emergency savings features have moved from novelty to mainstream benefit design. The mechanism is straightforward: give the shock somewhere else to go.

What actually closes the leak

Structural, in rough order of impact:

1. Automatic portability for small balances, so accounts follow workers to new plans without requiring action. This directly addresses the two-thirds of leakage sitting under $7,000 and is the single highest-leverage intervention available.
2. Workplace emergency savings, given the magnitude of the association above.
3. Better decision architecture at separation — making rollover the path of least resistance rather than distribution, and displaying the long-run cost at the moment of the decision rather than only the tax haircut.
4. Loan design that survives job change, including extended repayment windows after separation.
5. Consolidation of orphaned accounts, since a worker with balances at four former employers is a worker likely to lose track of several.

Individual, for someone at a job change right now:

• Roll it over rather than cashing out — into the new employer's plan or an individual retirement account. A direct trustee-to-trustee transfer avoids the mandatory withholding entirely.
• Leaving it where it is is generally better than cashing out, if the plan permits and the balance exceeds the forced-distribution threshold.
• If you have an outstanding loan, address it before or immediately after separation, since the window is short and the consequence is a taxable distribution.
• Track old accounts. Orphaned balances at former employers are a common and quiet loss.
• Build the accessible buffer first if you're choosing between goals, because it protects the retirement account from the next shock — and it's the finding above stated as personal strategy.
• Handle a shortfall in the right order. Where a shock has already arrived, the consequence-ranked sequence in our shortfall triage guidance generally puts retirement liquidation well below negotiating with unsecured creditors — because those obligations are negotiable and time-limited in ways a drained account isn't.

Scenarios and what we're watching

Scenario	Shape of the world	Signposts
Base case — slow plumbing improvement	Auto-portability and workplace emergency savings spread gradually; cashout rates decline slowly; leakage remains a material drag	Portability network coverage; emergency savings feature adoption; cashout rates by balance band
Infrastructure case	Portability becomes near-universal for small balances, capturing the two-thirds of leakage concentrated there	Consolidation volumes; small-balance retention rates; forced-distribution practice
Stress case	Household distress raises hardship withdrawals and loans while job mobility raises cashouts; leakage climbs across all three channels at once	Hardship withdrawal rates; loan origination; separation-related distribution volumes

What we're watching: cashout rates by balance band, since the small-balance share is where the fix has to work; auto-portability adoption, which is the infrastructure that would move the number most; workplace emergency savings uptake, given the 43-percentage-point association with cashout behavior; and hardship withdrawal rates, which function as a broad household distress indicator independent of their dollar volume.

The retirement system's largest loss doesn't come from people who couldn't save. It comes from people who did save, changed jobs, and found that taking the money was the easiest form to sign.

Frequently asked questions

---

This report is for general information only and does not constitute financial or tax advice. Figures are drawn from publicly reported academic research, industry analysis, and plan recordkeeper data; distribution rules, penalty exceptions, and plan provisions vary and change. Consult a qualified advisor about a specific decision.