Impactism
日本語 Join Impactism
Manifesto The Impact Standard Foundations Objections Read For organizations 日本語
Join Impactism
A crowded street under the Ameyoko market sign in Ueno, Tokyo
Photo by Fumiaki Hayashi on Unsplash

Essay

Does Money Buy Happiness? What 2023 Settled

October 12, 20268 min read

Yes. For most people, more money does buy more happiness, and it keeps doing so past every threshold anyone has proposed. The gain from each extra dollar shrinks, because well-being tracks the logarithm of income rather than income itself, so you have to double your money to get the same lift again. There is one group for whom the effect stops. Since 2023 we know roughly who they are, and the answer is not the one the popular version predicted.

We wrote the history of this argument in The Log Curve, including where the $75,000 meme came from and why it is dead. This piece does two things that one does not. It reads the 2023 result closely, because the detail everybody skipped is the interesting part. Then it goes to Japan, which is both the Easterlin paradox’s best evidence and the place where that evidence fell apart on inspection.

What the 2023 collaboration actually settled

Matthew Killingsworth and Daniel Kahneman had published opposite answers. Rather than argue in print, they ran an adversarial collaboration with Barbara Mellers as arbiter, agreeing in advance what would count as evidence, and reanalysed Killingsworth’s data together: 1,725,994 momentary reports from 33,391 employed US adults, collected by pinging their phones at random moments (Killingsworth, Kahneman & Mellers, 2023).

The finding people quote is that happiness flattens above $100,000 for the least happy. Two details in that sentence get lost.

The first is that the split is by happiness, not by income. It is the least happy 15% of the sample, at every income level, whose curve goes flat. Not the poorest 15%. The paper’s reading is that this group carries something money does not touch, and once their material problems are handled, more of it stops helping.

The second detail is the one that should unsettle anyone who likes the “enough” story, and it is in the paper’s own words: happiness “increases steadily with log(income) among happier people, and even accelerates in the happiest group.” For the happiest 30% at each income level, the slope beyond $100,000 gets steeper, not shallower.

So the honest summary is uncomfortable in both directions. Money cannot buy its way past whatever is wrong with the least happy sixth of people. And for the people who are already doing well, the data show no ceiling at all in the range measured. Nobody gets to cite this paper as proof that money stops mattering.

Then the paradox, which Japan was supposed to prove

The other half of the argument is not about people but about countries. Richard Easterlin’s claim was that within a country at one moment richer people are happier, and yet decades of national growth do not raise the national average.

Japan was the flagship case, and the numbers were spectacular. Easterlin, writing in 1995: “Between 1958 and 1987 real per capita income in Japan multiplied a staggering five-fold, propelling Japan to a living level equal to about two-thirds that of the United States… . Despite this unprecedented three decade advance in level of living, there was no improvement in mean subjective well-being.” Layard, Frank and Kahneman all quoted it approvingly.

Betsey Stevenson and Justin Wolfers went back to the source, the Cabinet Office’s 国民生活に関する世論調査, the “Life in Nation” surveys running since 1958. The series was not one series. The questions from 1958 to 1969 asked how people felt about life at home; from 1970 the question asked about life overall. The top response option was rewritten from “Although I can’t say that I am satisfied, if life continues in this way, it will be okay” to “Although I can’t say I am completely satisfied, I am satisfied.” The wording and the allowable answers changed again in 1992. What looked like a flat line for fifty years was four different questions laid end to end (Stevenson & Wolfers, 2008, Brookings Papers on Economic Activity, pp. 46–48).

Read as four comparable stretches, well-being in Japan rises through 1958–63, rises again through 1964–69, rises more slowly from 1970 to 1991, and falls from 1992 to 2007. The fall lands exactly on the end of the growth miracle. Their conclusion is that Japanese well-being largely tracked GDP per capita, and rose fastest when growth was fastest.

Easterlin replied in the same volume and did not concede. We are not going to pretend the dispute is over, because it is not. What we will say is that a piece of evidence Layard, Frank and Kahneman all quoted turned out to depend on a survey instrument that changed under the researchers’ feet. That is worth knowing before anyone builds a policy on it.

What Japan measures now

Japan did something after that argument which few governments have done. In February 2019 the Cabinet Office started a dedicated survey, 満足度・生活の質に関する調査, asking about ten thousand people each year to rate their life from 0 to 10, alongside satisfaction in thirteen separate domains. The stated purpose is to make well-being visible in a way GDP alone cannot.

The first report, published on 24 May 2019, found a national mean of 5.89. Its finding on income is the one to sit with. Satisfaction against household income is 山型, a mountain shape. It climbs with income up to the ¥20–30 million band, tops out there, and then declines gently for households earning more than that. Financial assets show the same shape, rising to the ¥100–300 million band and falling above ¥300 million. At the bottom end the gradient is steep: crossing ¥3 million in household income is worth about half a point on the ten-point scale (Cabinet Office, 第1次報告書, 24 May 2019).

The same report found something else, and it is the part we would put on the wall. Among respondents with five or more people they could rely on, the usual satisfaction penalties did not bite the way they otherwise do. Household income under ¥1 million and poor health, the two heaviest drags in the whole dataset, pulled the score down far less for people who had that many others to lean on.

Seven waves later the aggregate has barely moved. The eighth survey, run 27 February to 14 March 2026 with 10,632 respondents, put the mean at 5.86, and the Cabinet Office describes the series as 一進一退, going back and forth (Cabinet Office, 満足度・生活の質に関する調査報告書2026, published 31 July 2026). Two caveats before anyone reads a paradox into that: the sample is an internet panel, and seven years is a short window for a national average. It is not a controlled comparison. It is what the government’s own instrument says.

Where the evidence runs out

Put the two literatures side by side and notice what neither of them contains.

Neither tells you what “enough” is. The log curve gives you a price list, not a stopping point: the next doubling costs twice as much as the last one and delivers about the same lift. That is a fact about exchange rates, and no fact about exchange rates has ever told anyone when to stop trading. The 2023 paper makes this sharper, since for the happiest people the returns accelerate. If you want a number, the research declines to give you one.

Neither tells you what to do with the money either. Both literatures measure income against a person’s own report of their own life. Nothing in the design asks what the money did to anyone else.

That is where this stops being an empirical question and becomes a cultural one. A society decides what it treats as the top of the mountain, and the decision is not read off a survey. Japan’s own data hint at what the alternative feels like from inside: five people you can rely on did more to protect a life from a low income than the low income did to damage it. Nobody is ranked on that. There is no league table of people you can call at 2 a.m.

We think there will be a yardstick whatever we do, which is the argument in Status Is the Scarce Thing. Since one is coming, the question is what it counts. Money is a fine instrument for moving resources and a poor one for ranking people, and the research above is precisely why: it measures well, it predicts something real, and it runs out exactly where the interesting part of a life begins.

What we are not claiming

Money matters enormously below “enough”, and the Japanese ¥3 million gradient is one more measurement of how much. Our position has never been that money is bad or that a person chasing a higher salary is making a mistake. We also do not claim the Easterlin dispute is resolved; Stevenson and Wolfers challenged the Japanese evidence, Easterlin answered them, and the argument continues in the journals. And we are not proposing a replacement number. A single score of a person’s contribution would inherit the pathologies of the one it replaced and add its own.

The claim is smaller. The empirical question about money and happiness has an answer, and it is not the answer that settles how to live. That part was always ours.

Sources

← All essays

Impactism is free — no paywall, no membership fee, and it will stay that way. If this argument was worth your time, donations are what pay for it. One-time or monthly, any amount.

Support the argument