An emergency fund is insurance you underwrite yourself

Not a goal, a shock absorber

Most savings have a destination. A deposit, a car, a wedding, a course — money is set aside because something specific is going to be bought with it at a roughly known time. An emergency fund is the exception, and confusing it with the others is the reason so many are spent before they are needed.

Its function is structural. Every household faces occasional events that cost money and do not announce themselves, and each of those events has to be paid for from something: savings, borrowing, or going without. A buffer exists so that the answer is the first of those, at a moment when the second is usually expensive and the third is often not available.

Seen that way it behaves like an insurance policy with the household as the insurer. The premium is the return the money could have earned somewhere else, and the payout is the ability to absorb a bad month without any of it becoming permanent.

Two different shocks, and only one of them is cheap

It helps to separate the kinds of event a buffer is protecting against, because they behave very differently. An expense shock is a one-off cost: a boiler, a car repair, a vet, a broken laptop that is genuinely needed for work. It is unwelcome, but it has a size and once it is paid, it is over.

An income shock is a different creature. Losing work, or losing the ability to work for a period, does not present a single bill. It removes the flow that everything else was being paid from, and it lasts for an unknown number of months. That open-ended quality is what makes it the expensive case, and it is the case that sets the size of the fund.

A buffer built only against expense shocks will be a few thousand and will hold up perfectly well until the day it meets the other kind. That is why the conventional way of sizing one is in months rather than in a fixed amount.

Where the months-of-expenses rule comes from

The common rule of thumb is some number of months of expenses, and the reasoning behind it is simple enough to reconstruct. If the risk being covered is an interruption to income, then the fund needs to replace income for as long as the interruption plausibly lasts, and the natural unit is a month because that is how outgoings arrive.

The important refinement is that the figure should be built from essential outgoings, not total spending. Housing, food, transport, utilities, minimum debt payments, childcare, medication — the things that continue regardless. Discretionary spending genuinely does fall during a difficult period, and including it inflates the target for no benefit.

The arithmetic then does itself. Suppose essential outgoings come to 2,200 a month. Three months is 6,600 and six months is 13,200, and the difference between those two figures is a judgement about how long a gap the household needs to be able to survive without changing anything fundamental.

What moves the number, in both directions

The right multiple is not universal, and the factors that push it around are reasonably predictable. Income that is variable, seasonal, self-employed or dependent on a single client argues for more months. Two incomes in unrelated fields argue for fewer, because the chance of both stopping at once is lower than the chance of either stopping.

Dependants push it up. So does a long expected search for replacement work, which varies enormously by profession and by where you live. Any sick pay, notice period or statutory protection that applies in your jurisdiction pushes it down, since those effectively cover the first stretch of an income shock already.

Housing matters more than most people account for. A large fixed housing cost raises the monthly figure being multiplied, so it increases the target twice over — once through the multiplier and once through the amount being multiplied.

The price of holding it is the point, not a problem

Money held for immediate access will generally earn less than money committed for longer, and over several years that difference is real. It is tempting to treat this as inefficiency and to put the buffer somewhere it will work harder. Doing so changes what the money is, because a fund that might be worth less on the day you need it is no longer performing the function it was created for.

There is genuine disagreement about the details. Reasonable people argue about how many months are sensible, about whether some of the buffer can sit in a less accessible form once it is large, and about whether an available line of credit can stand in for part of it. Those arguments turn on circumstances rather than on arithmetic.

What is not really disputed is the principle: some amount of money whose value on a given day is certain, reachable within days rather than weeks, is what makes an unpredictable event ordinary. How much a particular household needs depends on things a general article cannot see, and is worth setting out with a regulated adviser if the sums are significant.