Shocking Look at Average vs Median Household Savings

I have been on a rampage about average vs median income. Here's a report about average vs median savings.

Magnify Money asks How Much Does the Average American Have in Savings?

The question is irrelevant. The story is how unprepared the median person is prepared for retirement. On that score, the article does explain.


  1. The average American household has $175,510 worth of savings in bank accounts and retirement savings accounts as of June 2018.
  2. The median American household currently holds about $11,700 across these same types of accounts.
  3. The top 1% of households (as measured by income) have an average of $2,495,930 in these various saving accounts. The bottom 20% have an average of $8,720.
  4. Roughly 83% of savings are in located in retirement accounts like IRAs and workplace-sponsored retirement savings plans like 401(k)s.
  5. Millennials, who have just started their savings journey, have currently socked away an average of $24,820. Gen Xers have $125,560 in retirement savings. Baby boomers and those born before 1946 have an average of $274,910.
  6. 29% of households have less than $1,000 in savings.

Point number 2 is the most relevant point. 50% of household have less than $11,700 in savings.

Averages Lie

What's wrong with averages? The Skew!

Average and Median Savings by Income Level

The top 1% of income earners have an average savings of $2.53 million and a median savings of $1.16 million.

That average affects people with no savings.

The median savings for 40% of households is zero. The "average" varies by income group but it is much higher.

The "middle" (40-60% of wage earners) median savings is $34,020 but the average is $65,830.

Age Level

This is where the stats get truly depressing.

The average "boomer" headed into or in retirement has $274,910 in savings.

What's wrong with that?

Well, 50% of boomers have less than $24,280 saved up.

Averages lie.

Mike "Mish" Shedlock

Comments (28)
No. 1-14

Yes. It is important to know the difference between average and median, especially in sample sets where extreme outliers skew the whole sample. You know the saying: There are three kinds of lies: lies, damned lies, and statistics.

Speaking of misused statistics, I have lost count of the news reports that roughly state, Trumps tariffs will cause higher prices (inflation) and that may force the Fed to tighten more than expected. What sort of craziness is that? Tariffs are a type of tax that extract money from the economy and economic activity would likely be depressed by tariffs, not stimulated. Higher taxes are in no way the similar to higher prices caused by monetary injections. How can anyone say with a straight face that Fed will tighten more in response to tariffs? Surely the Feds economic models are not that broken? I would like to know your view.


I see those numbers are from the FDIC and the Federal Reserve. Do they consider ownership of businesses which have productive assets to be 'savings'? Or is their definition of 'savings' merely cash deposits in so-called "savings accounts"?


Mish, this is such a great point you make here.

We had a thread on another posting about graphs and how they can mislead. This is an example of how a good graph can be far, far better than a simple number.

The curves that describe income, wealth, and savings distributions (and other things of this sort) are exponential-ish curves. Think, 80-20 rule. A very few entities with very high numbers, lots of low numbers. Smooth transition from beginning to end. Average higher than the median.

With curves like these, even *thinking" about the average will send your mind off the rails. No graph? Use median. Cut people who use averages out of your reading list. You really want the graph. Watch out for various bar-charts and quin-tiles and such. They can be kludges grinding axes. Watch out for fancy analysis that, when you spend a few hours with it, you find someone secretly switched from an exponential/log view to a linear view and, in doing so, did the mental slight-of-hand equivalent of dividing by zero. Just get the raw data in a simple curve.

If you don't have the graph, dollars to doughnuts you'll end up in some kind of fantasy land.


Why should we assume, the statistics is to portray the true picture. In all probability not. Essentially use the statistics to paint the picture you want. When fraud is the business model to peddle something ( in this case to paint a picture of an economy with great savings) then simply twist the facts to tell your tale.


Your story shows why distribution curves are better than averages or medians.