The Buffer You Never Spend

The buffer is a return decision

Emergency cash is normally taught as caution, and caution is not a reason a household can act on. The reason is arithmetic. A drawdown and its recovery are not symmetric: fall 35% and you need 53.8% to be back; fall 50% and you need 100%. So the expensive event is not a bad market — the portfolio survives that — it is being *forced to sell* into a bad market, because the shares you sold are not there for the recovery. A buffer changes which asset funds the shock. With three to six months of essential spending in cash, a job loss is paid for out of a savings account, the portfolio is left alone, and the recovery happens to the whole position. Without it, the same shock is a permanent reduction in the number of shares that will be there when the market comes back. That is why the buffer earns its place in a plan even though it pays less than equities: it protects the return on everything else. It follows that the buffer is not “as much cash as feels safe”. Cash is the lowest-return asset in the plan, and holding too much of it for too long has a real cost. The job is to size it to the specific fragility of one household and then hold exactly that — no more as a comfort blanket, no less as an optimisation. What a forced sale costs — Equity portfolio after a 35% fall: $78,000 of $120,000 · Sale made to fund the shock: $18,000 · Gain the shares sold must earn to recover: +53.8% · Price of arriving $4,800 short of a six-month buffer: $4,800 + the recovery on the shares sold The shock is temporary; the sale is permanent. That is the whole case for holding cash you hope never to spend.

Size it to the fragility, not to your nerves

Three to six months of *essential* spending is the standard range, and the range exists because households differ. The input is not how anxious you feel; it is how long a replacement income would plausibly take and how many obligations arrive while it is missing. A dual-income household in a stable industry with no dependents is at the low end. A single earner supporting a family, on commission, in a cyclical industry, is at the high end — and if the household also carries a mortgage and a car payment, a bad year is not one shock but a sequence of them. Three questions do most of the sizing. How variable is the income — salary, bonus, commission, contract? How many people depend on it? And how long do obligations run regardless of income? Anything with a long fixed tail (a lease, a mortgage, childcare, a chronic medical cost) argues for the top of the range, because those payments do not pause while you look for work. The buffer is not insuring the loss of a job so much as buying the time to choose the next one rather than accept the first. Same income, different fragility — Two salaried earners, no dependents, cheap rent: 3 months · Single earner, two children, mortgage and car payment: 6 months · Commission income in a cyclical industry, one earner: 6–9 months ← Cash in the buffer belongs in instruments that cannot lose principal and can be spent within days — a high-yield savings account, a money market fund, a short T-bill ladder. Anything with duration or equity risk is not a buffer: a long-term bond fund lost around 30% in 2022, exactly the year a household needed certainty.

Too much cash is also a decision

The other way to get the size wrong is to leave too much in cash, and it feels like the safe mistake until you price it. Fifty thousand dollars held in cash instead of equities for thirty years, at an assumed 7% equity return against 4.25% on cash, ends at 50,000 × (7.612 − 3.485), or roughly $206,000 of foregone growth. Nobody sends an invoice for that, which is exactly why it goes unmeasured. Notice what the calculation does not say. It does not say the cash was a mistake — it says the *excess* was. The first $14,400 of a six-month buffer is doing a job with a measurable payoff: it prevents a forced sale into a drawdown, and one avoided 35% sale can be worth more than the foregone yield on the whole buffer. Beyond that size, the same instrument stops doing anything and starts costing. The practical rule that follows is to name the buffer as a number, hold that number, and send everything else to the portfolio. Households rarely get into trouble by holding six months of essentials in cash; they get into trouble by holding three years of spending in a savings account “until the market feels better”, which is a market-timing position wearing a safety label. The buffer is a range, not a maximand — Three to six months of essentials: $7,200–$14,400 on these numbers · Six months is not excessive — it is the top of the standard range: the protection it buys is real · $50,000 more than the buffer, held 30 years in cash: ≈ $206,000 of foregone growth ← The same logic applies inside a retirement portfolio, where one to three years of withdrawals in cash is the standard way to keep a bad first year from being funded by selling shares (PF13).

Safe, and safe to spend

A buffer has to do two jobs that pull in different directions: it must be certain not to lose value in the moment you need it, and it must be easy to reach without a penalty. Those two requirements decide where the money sits, and the difference between the places that look interchangeable is mostly about who guarantees them and how fast they can be turned into spending money. Start with insurance, because this is a place where the words are used loosely. Cash in a bank account is covered by **deposit insurance**, which protects a stated amount per depositor per institution per ownership category — so a single account, a joint account and a retirement account at the same bank are separate categories, and the coverage multiplies accordingly. Cash held at a brokerage is a different regime: it is protected against the failure of the *firm* through the customer-asset segregation rules and the investor-protection scheme, but a money market fund is a security rather than a deposit, and buying one means holding a fund that seeks to maintain a stable value rather than a balance the bank has promised to repay. The distinction almost never matters, which is exactly why it is worth knowing before it does. Then access, which is where the tiers separate. A savings or checking balance is spendable in minutes. A money market fund has to be sold and settled, which under the current cycle means the cash may not be available to withdraw until the next business day. A short-dated Treasury bill matures on a known date, and selling one early means going through a market rather than asking for your money back. A term deposit may pay more and lock the funds for months, which makes it an investment rather than a buffer. When a shock arrives — a furnace, a layoff, a medical bill — the difference between minutes and a business day is usually irrelevant, and the difference between a business day and a maturity date is usually not. That produces the shape rather than a single product: one month of spending in the fastest tier, the rest of the buffer in something that settles next day and holds its value, and everything beyond the buffer in the portfolio where the point is the return. Holding all of it in the fastest tier usually means accepting a yield close to nothing; holding all of it in something with a maturity means holding something that is not a buffer at all. The second half of this is a rule for using it, and a buffer without one quietly becomes spending. The workable version is written in advance and has three parts. First, what qualifies — a genuine unplanned expense that cannot be absorbed by the month’s cash flow, as opposed to a want that has been relabelled. Second, the order of use, so the decision is not made under stress: cash first, then the taxable investments, and retirement accounts last because of the tax and the penalty. Third, the refill plan — how much goes back each month and from what, and how long the refill will take, because a buffer that is used and never rebuilt has removed the protection it was built to provide. None of this changes the sizing argument one line up. It decides where the number lives and what happens the day it is called upon, which is the part that turns a balance into a policy. • Bank cash is deposit-insured, and ownership categories multiply the coverage; brokerage cash is protected differently. • A money market fund is a security, not a deposit, even when its price is designed to stay at one dollar. • Access speed separates the tiers: minutes, a business day, or a maturity date. • Write down what qualifies as the emergency, the order of use, and the refill plan. A credit card is not a buffer, but it can bridge an expense by a few weeks without forcing a sale — worth knowing as a sequence rather than as a source, so long as it is repaid from the buffer rather than carried.

Sources

• Emergency savings: sizing and where to hold itConsumer Financial Protection Bureau — emergency savings guidance
• The asymmetry of drawdowns and recoveryStandard drawdown arithmetic (1 ÷ (1 − d) − 1)
• Cash equivalents and the yield they give upTreasury / money-market fund yield data; Vanguard capital market assumptions
• Sequence risk and the case for a cash sleeve in retirementBengen and subsequent safe-withdrawal-rate research