Top 1 percent net worth: the published threshold, by household and age

A threshold is the value at which a percentile begins: the figure above which you are in the top one per cent. To publish one you need the distribution at that resolution, and the Bulletin does not report it. It reports groups: the least wealthy quarter, the next two quarters, the 75th to 89.9th percentile, and the top decile.

For the top decile the published figures are a median of $3,794,600 and a mean of $7,810,500. Those two numbers together say something precise about the top-percentile question without answering it: the mean of the top decile is roughly twice its median, which can only happen if wealth inside that decile is itself concentrated at the very top.

So the honest published answer is a bound rather than a threshold. The top 1 percent net worth threshold lies well above the top decile’s median, and the Bulletin does not say where.

The rising ratio across these rows is the concentration itself: the further up the distribution, the more the mean exceeds the median.

Three routes exist and each has a cost. The survey’s public microdata can be used to compute percentile points directly, which is legitimate and produces a figure that is the analyst’s rather than the Federal Reserve’s. The Distributional Financial Accounts publish wealth shares held by the top percentiles, which is a different quantity: a share of total wealth, not a household threshold. Or the figure can be taken from a private estimate, which is not a public document at all.

This site prints none of the three as a published threshold. What it prints is the top-decile household figures above, the note that the mean-to-median ratio implies heavy concentration inside that decile, and the routes a reader can follow if they want a computed number and the label that goes with it.

A top 1 percent net worth by age table runs into sample size immediately. A survey of a few thousand households can support a national top-decile figure and cannot support a top-percentile figure within each age band: the cells become too small to publish reliably, which is exactly why the Bulletin does not publish them.

What can be said is directional, and it comes from the age table itself: concentration rises with age. In the older bands the mean is four to five times the median, against roughly three times in the youngest. Wealth accumulates unevenly, and the unevenness compounds over a working life.

The reason the survey stops where it does is worth stating, because it is a sampling limit rather than an editorial one. A national figure rests on the whole sample; a figure for one age band rests only on the households in that band; a figure for the wealthiest fraction of one age band rests on a handful. Each cut divides the sample again, and the published tables stop at the point where the remaining cell would carry a margin of error wider than the number itself. That is why the Bulletin prints deciles nationally and medians by age, but never the two crossed together.

One further distinction decides which published number answers which question, and the two are routinely swapped. A threshold is the amount a household must hold to enter a group: a boundary, and a single figure. A share is the proportion of all wealth a group holds between them: a ratio, which says nothing about any one household. The Bulletin publishes shares for the top percentiles and medians for the deciles, so a question phrased as "how much do you need" asks for a boundary it does not print, while "how much do they hold" asks for a ratio it does. Reading one as the other is where most of the confident figures circulating on this subject come from.

The by-age page carries the full table those ratios come from.

Because the alternative is to publish a figure and attribute it to the Federal Reserve, which would be false, or to publish it unattributed, which would make it indistinguishable from the estimates this site exists to avoid.

It is the same discipline applied everywhere here. Where a register publishes a band, the band is printed. Where it publishes a group, the group is printed. Where a reader wants a finer cut than exists, the page says which document would have to contain it and does not improvise. The verification standard is what makes that refusal consistent rather than arbitrary.