The plateau is in the tail, not the mean

In 2010, Daniel Kahneman and Angus Deaton published the most quoted number in the economics of happiness. Working from roughly 450,000 responses to Gallup's daily tracking poll — the same survey behind every figure on this page — they reported that day-to-day emotional well-being stopped improving once household income reached about $75,000, even though people's overall ratings of their lives kept climbing. The finding escaped the journals instantly. It became a salary negotiation talking point, a sermon, a personal-finance commandment.

Then it broke. In 2021, Matthew Killingsworth measured experienced well-being with smartphone check-ins — 1,725,994 reports in the moment — and found no plateau at all: well-being kept rising past $75,000, in a straight line against the logarithm of income. Two careful studies, same question, opposite answers. So Kahneman and Killingsworth did something rare: they teamed up, with Barbara Mellers refereeing, and re-analyzed Killingsworth's data together. Their 2023 resolution is the punchline of this article. Both sides were right — about different people. Among the unhappiest fraction of the population, money genuinely stops helping at higher incomes, with the flattening setting in somewhere around $100,000. For everyone else there is no plateau, and for the happiest there may even be an acceleration.

The moral is not about money; it is about averages. A mean curve is one line summarizing millions of different lives, and a plateau confined to a minority simply vanishes inside it. The original $75,000 paper and its refutation were both, in the end, arguments about a mean. So here is the same question, re-staged on the original battlefield — 1,808,814 Gallup interviews from 2009 through 2017 that carry both a life rating and an income bracket — showing the whole distribution of well-being at every income level, not its average.

↓Chunky scale, honest shares

Percentiles of an 11-step scale are awkward creatures — they jump in lumps, and our smooth curves above lean on an interpolation convention (stated in the notes). The sturdier framing is to count people past a fixed rung. Two shares tell the whole story. The share rating their life 4 or below — the unhappy tail — collapses from 24.6% in the poorest bracket to 5.0% by $60–90k, and then nearly stops: only 3.3% in the top bracket. Per doubling of income, that's a fall of 5.9 points across the middle of the range but just 1.7 points over the final doubling. The floor is in sight, and more money barely moves it.

The share rating their life 8 or above does the opposite. It rises from 33.7% to 49.9% by $60–90k — and then keeps rising, 54.6% between $90,000 and $120,000 and 61.5% at the top: +11.6 points over the final doubling, faster than the +7.3 points per doubling it managed across the middle incomes. No satiation, no slowdown — if anything, the opposite. One distribution, two regimes: the unhappy minority's income curve flattens right where Kahneman and Deaton said happiness stops; the thriving majority's curve never does, exactly as Stevenson and Wolfers, and later Killingsworth, insisted.

↓Not a composition story

A skeptic should ask whether the rising top is just composition — richer brackets hold more college graduates and more 55-year-olds, groups that rate life differently. So cut the sample into nine age-by-education cells and re-run the top end of the curve inside each one. The answer survives everywhere: from the $60–90k bracket to the top bracket, the share rating life 8 or above rises in all nine cells, by +3.1 points (high-school educated, 55 and over) up to +15.4 points (college graduates, 35–54). The unhappy share, meanwhile, moves by at most 2.6 points in any cell. The split between a flattening bottom and a climbing top is not an artifact of who sits in which bracket.

↓The claim that started it was about feelings — and it still holds

Kahneman and Deaton's satiation point was never about ladder ratings; it was about emotions yesterday. This file carries those too, and inside a single survey regime — 2009 through 2012, about 1.1 million interviews with an income bracket — the original result re-appears almost exactly as advertised. The share who experienced enjoyment a lot of the day yesterday climbs from 75% in the poorest bracket to 89% by $60–90k, then creeps up barely a point across the entire top half of the income range. Worry falls from 47% to 26.3% by $60–90k and then refuses to fall further — flat to the top bracket. Sadness bottoms out around 11%. And stress is the strange one: it falls only to about 35% in the $36–48k bracket, then curls back up to 37.8% among the highest earners.

Set side by side, the two constructs split cleanly. Daily feelings satiate near the famous threshold, on schedule. Life evaluation — once you look past its mean to its distribution — satiates only for the unhappy tail, while the share of lives rated 8 or above keeps compounding with every doubling of income. The 2010 paper, its 2021 refutation, and the 2023 truce all live comfortably inside one dataset, each describing a different slice of it.