You should not have an emergency fund.

It is pretty standard advice. “Have 3 months (or 6 or 9 or 12; no one can seem to agree on exactly how big it should be) in an emergency fund.” Usually you are told to build up an emergency fund before you turn your focus to real investing.

This advice is almost always wrong.

The basic story is

1. Your emergency fund doesn’t grow much. (Since it is invested in a high-yield savings account earning 1%.)
2. Emergencies are rare.
3. So having an emergency fund is a bad idea.

The best analysis of emergency funds that I’ve read comes from Charles B. Hatcher in his 2000 paper, “Should Households Establish Emergency Funds?” (His answer: No.)

you need to have an emergency more frequent than once every four years for the emergency fund to make sense

Hatcher tries to calculate the “liquidity premium” — the cost per year of having a highly-liquid (but low interest) emergency fund versus alternative investments. He compares this against other ways of funding emergencies: getting a loan.

His headline result is: with a 3-month emergency fund, you need to have an emergency more frequent than once every four years for the emergency fund to make sense.

Part of a table from Hatcher’s paper

He has a table showing how often an emergency needs to happen for an emergency fund to make sense. The way to read this table:

• The left column is how much interest/return you are giving up by having an emergency fund.
• The top row is the interest rate on a personal loan. (Which you would have to take out if you didn’t have an emergency fund.)
• Putting it all together: if you give up 2% in order to have an emergency fund and you can borrow money at 10% interest, then an emergency needs to have at least a 39% chance of happening every year for the emergency fund to make sense.
• Obviously: the more interest you give up to have an emergency fund, the more often emergencies need to happen. The cheaper you can borrow money, the more often emergencies need to happen.

Hatcher’s conclusion is clear: unless you alternative investments have very low yields or you expect frequent emergencies, an emergency fund doesn’t appear to make sense.

More recently (2013) Scott et al builds on Hatcher’s work in “Is an All Cash Emergency Fund Strategy Appropriate for all Investors?”

From Scott et al

Their findings agreed with Hatcher. Having a six-month emergency fund almost always has a significant opportunity cost. The average investor who had 60% equities would have total wealth almost 20% smaller due to maintaining a six-month emergency fund.

Scott et al had a more surprising finding: having an emergency fund actually increases the standard deviation of a portfolio. That is: it increases risk. Why is that the case? After an emergency you need to refill the emergency fund, which temporarily diverts normal investments. That results in dollar-weighted returns on your investments being less consistent.

Maintaining the six-month cash reserve strategy translates into a proportionate lower standard of living during retirement

A final finding from Scott is that emergency funds often fail to cover the total cost of the emergency anyway.

One of the problems is that people are rarely very crisp on exactly what emergencies the emergency fund is hedging against. The most common answer is “sudden unemployment”. That…actually seems pretty reasonable on the surface.

But in that case, why are we making up numbers for how big the emergency fund should be? Three months? Six months? Something else? We have actual data about the length of unemployment.

So in a Big Emergency™ (i.e. the 2008–9 Global Financial Crisis), it peaks at 41 weeks for the average unemployed person; that’s a 10-month emergency fund. It currently stands at 28.1 weeks; that’s a 7-month emergency fund.

Of course, you’d actually apply for Unemployment Insurance. In California (to pick a random state) that would last you 6 months and give you $1,800 a month. So your 3-month emergency fund would actually stretch out to 4 or 5 months. A 6-month emergency fund would stretch out to maybe 8 months. (That’s a back-of-the-napkin calculation; in reality you’d need to start paying for private health insurance but your commuting and food costs would likely go down.)

In any case, there doesn’t seem to be much congruence between recommended emergency fund sizes and how long unemployment actually lasts in practice.

So lots of people seem to be saying that the “liquidity premium” of an emergency fund isn’t worth it. What does that look like in practice?

Here’s a simple scenario with two people. Person A put $15,000 into an emergency fund 5-years ago. Person B put the same $15,000 into a normal investment portfolio that is 50% stocks and 50% bonds.

Person A   Person B
$15,000    $15,000
$15,300    $15,639
$15,606    $17,226
$15,918    $19,904
$16,236    $21,714
$16,561    $21,779

Person B is over $5,000 ahead. What started out as a fund big enough to handle 3-month emergency ($5,000 a month) is now able to handle a 4-month emergency.

At this point emergency fund advocates will usually bring out worst case scenarios. What if the stock market tanks 50% and then you lose your job? Remember you’re invested in a 50/50 portfolio. So a 50% stock market crash results in losing 25% of your portfolio. Meaning your drastically reduced “virtual emergency fund” is now only worth $16,334…which is only $227 less than the traditional approach.

Not exactly a resounding victory for the traditional approach.