Thread by @LevyAntoine

I got tired of the Moody's K-shaped debate.
Still, it's worth repeating some additional, somewhat technical, but important issues.

0. Moody's approach essentially amounts to saying:
Share of Consumption done by top 10%
= Consumption of top 10% divided by Total Personal Consumption Expenditures
= (Disposable income of top 10% - Savings of top 10%) divided by (Total disposable income - Total savings).

To do this, you need:
- to define the top 10% of *what*
- good estimates of the amount of disposable income received by the top 10%, and the share of savings they do *out of that concept of disposable income*. Unfortunately, they have neither.

https://x.com/LevyAntoine/status/2059711058214830119

1. To find out the disposable income of the top 10%, Moody's starts from the BEA's total *personal disposable income* ($23 trillion: post-tax, inclusive of c. $1T owner-imputed rent and $2T in employer insurance/pension/Social Security contributions, excl. of realized K gains).
They then multiply it by the observed share of the *SCF's concept of total income* (pre-tax; exclusive of owner-imputed rent or employer contributions; incl. of realized K gains) accruing to the top 10% in the latest Survey of Consumer Finances.

This is plain wrong:
a. pre-tax income is more concentrated than post-tax (taxes are progressive!)
b. realized K gains are highly concentrated at the top (they are very lumpy and asset ownership is concentrated!)
c. both owner-imputed rent and employer contributions are an equalizing force for the distribution of BEA-style disposable income.

So this methodology yields highly inflated shares of BEA disposable income for the top 10% (about 50% instead of the BEA's own calculation of 34-35%).

NB: Moody's corrected their methodology since March 2026 to at least use a post-federal tax correction after many people, including me, pointed out this.

2. Moody's approach for allocating savings by disposable income groups amounts to then allocating the very tiny amount of BEA personal savings (e.g. 4% of disposable personal income in Q4-2025) across these groups. Note that even if they assumed that *all* personal savings were done by the top 10%, they would find a lower bound of 50-4=46% of consumption done by the top 10%, still above even the share of disposable income computed by the BEA for this top decile.

However, this concept of savings is completely inconsistent with the use of the SCF concept of income. Even abstracting from the tax part, the appropriate to use savings rates as a share of "SCF-style" income would be substantially higher at the top than personal savings rates as a share of "BEA-style" disposable cash income, because:
a. savings rates out of capital gains are much higher than out of disposable cash income
b. savings rates out of owner-imputed rent, employer-paid benefits, and imputed financial services are all zero in the BEA (pushing "BEA savings rates" down), but these are not part of the cash income SCF measure.

3. In practice, Moody's has no way of properly allocating the BEA-style savings. So they allocate BEA personal savings roughly in proportion to the top 10%'s SCF wealth share.
For Q42025, that would be about 68% of the 4% of personal savings. That's how they get 48% of consumption done by the top (0.5-0.68*0.04)/(1-0.32*0.04), a tiny adjustment to what is basically just the income share.

To summarize, you cannot have your cake and eat it.

It's fine to define income *SCF-style*, including highly concentrated, mostly saved items (like realized capital gains) and excluding more evenly distributed, entirely consumed items (like owner-occupied rent or employer health contributions). Then you can get highly concentrated incomes, but you need to allow for much higher savings rates at the top (in addition to much higher taxes).

Or you can also define income *BEA-style*, and you can get the really low overall personal savings rate out of personal disposable income (c.4%) that leads consumption concentration to essentially be income concentration. But then you need to acknowledge the much less concentrated distribution of BEA-style personal disposable income and the substantially lower top 10% share.