How Far Do Income and Happiness Rise Together — The Full Story of the 75,000 Dollar Debate

Opening — The Number That Became Common Sense: 75,000 Dollars

"Money buys happiness only up to a certain line, and that line is an annual income of 75,000 dollars (about 100 million Korean won)." This sentence may be the most widely cited psychology research result of the past decade or so. Its source is the 2010 paper by Nobel laureate in economics Daniel Kahneman and Angus Deaton.

Then, in 2021, one researcher produced the exact opposite result with better data: happiness keeps rising steadily well past 75,000. Up to this point, it is an ordinary collision in science. The reason I chose this story as the sixth installment of Psychology Through Papers is what happened next — the ending in which the two camps, instead of fighting, became one team and proved that both were right.

Round 1 — 2010, the Plateau Hypothesis

Kahneman and Deaton analyzed telephone survey data that Gallup had collected from 450,000 Americans. The core contribution of this paper is that it split happiness into two things.

• Experienced happiness (emotional well-being): did you feel enjoyment, laughter, stress, worry yesterday? The emotion of the moment.
• Life evaluation: on a ladder from 0~10, which rung is your life standing on? A grade given to your life as a whole.

The results diverged. Life evaluation rose without end with the log of income. But experienced happiness stopped rising near an annual income of about 75,000 dollars. Conclusion: "money keeps buying satisfaction with life, but beyond a certain line it cannot buy everyday laughter." Matching intuition and perfect as a story, this result became common sense in no time.

Round 2 — 2021, Better Measurement Arrives

Matthew Killingsworth is the person you saw in the previous installment. His weapon was the same app. Instead of a phone survey recalling yesterday, a smartphone notification asks about your mood at this very moment on a 0~100 slider. The result of the 2021 paper, which analyzed 1.72 million real-time reports collected from 33,391 American adults, was clear.

Experienced happiness did not stop at 75,000 dollars. It rose steadily with the log of income, all the way to the end of the measured range (annual incomes in the hundreds of thousands of dollars). Not just life evaluation — the mood of the moment as well. So where did the 2010 plateau go? Killingsworth pointed to measurement resolution as the cause: the emotion items in the 2010 data were yes/no binary, which saturates early at the top (nearly everyone ends up answering yes to "did you smile yesterday").

Round 3 — 2023, an Ending Called Adversarial Collaboration

Here the protagonist of this story stops being the results and becomes the procedure. Rather than defending his conclusion, Kahneman proposed an adversarial collaboration to Killingsworth: two people with opposing conclusions reanalyze the data together with a neutral arbiter (Barbara Mellers).

The finding of the 2023 joint paper is elegant. The key was not the average but the distribution.

In other words, the 2010 plateau was real — but only for the unhappiest minority. The unhappiness of this group often comes from pain with a clear limit to what money can relieve — heartbreak, bereavement, depression, illness — so beyond a certain income there is nothing more money can do. For the majority, by contrast, income and momentary happiness traveled together across the whole measured range. The 2010 analysis, because of the saturation of its binary items, had misread this minority's plateau as everyone's plateau.

A textbook resolution of a scientific dispute. No one was refuted, both datasets were respected, and the conclusion became more refined than either side alone.

Understanding It in Code — What the Log Scale Is Saying

Throughout this whole debate, the phrase "the log of income" keeps appearing. Miss what this scale means and you misread the entire result, so let us check it in code.

import math

# a stylized log-linear happiness model (not the real coefficients):
# each DOUBLING of income buys the same fixed happiness increment
BASE_INCOME = 30_000
INCREMENT_PER_DOUBLING = 1.0   # arbitrary happiness units

def happiness_gain(income):
    doublings = math.log2(income / BASE_INCOME)
    return doublings * INCREMENT_PER_DOUBLING

for income in [30_000, 60_000, 120_000, 240_000, 480_000]:
    print(f"income {income:>7,} -> gain {happiness_gain(income):+.2f}")

# output:
#   income  30,000 -> gain +0.00
#   income  60,000 -> gain +1.00
#   income 120,000 -> gain +2.00
#   income 240,000 -> gain +3.00
#   income 480,000 -> gain +4.00
# the same +1.0 lift costs 30k at the bottom but 240k at the top

The 2021 conclusion that "happiness keeps rising" is a story told on this scale. Which also means that the money required for the same increment of happiness doubles as income climbs. Going from an annual salary of 30 million won to 60 million won, and going from 600 million to 1.2 billion, buy roughly the same increment of happiness. "Money buys happiness" and "the happiness return on money deteriorates rapidly" are two faces of the same data.

What We Take Away

1. No need to belittle income growth, and no need to mythologize it. Especially in the low-income range, the happiness effect of money is large and real. "Money cannot buy happiness" is not true in that range, and it is often a luxurious sentence spoken by the comfortable. At the same time, by the law of the log, each further rise in the high-income range gets progressively more expensive. The return on the strategy of betting your whole life on salary growth differs by bracket.

2. Know the list of unhappiness that money cannot fix. What the unhappy minority of the 2023 paper tells us: in the face of lost relationships, depression, and health problems, the power of income bottoms out early. In those areas the investment is not money but relationships, the body, and, when needed, professional help.

3. Learn how to argue. If ego depletion in this series is the case study of a failed dispute, this story is the case study of a successful one. Inviting your opponent not as a target of refutation but as a partner in joint verification — a protocol you can use verbatim in code reviews and technical debates. The moment "who was right" turns into "under which conditions was each of us right", a fight becomes a discovery.

Guide to Reading the Originals

• Round 1: Kahneman, D., & Deaton, A. (2010). High income improves evaluation of life but not emotional well-being. PNAS, 107(38), 16489-16493.
• Round 2: Killingsworth, M. A. (2021). Experienced well-being rises with income, even above 75,000 dollars per year. PNAS, 118(4), e2016976118.
• Round 3: Killingsworth, M. A., Kahneman, D., & Mellers, B. (2023). Income and emotional well-being: A conflict resolved. PNAS, 120(10), e2208661120.

Reading tips: all three are in PNAS, so the abstracts are friendly. In the 2023 paper, Figure 1 (percentile curves of the happiness distribution by income bracket) is the summary of the entire paper. All you need to look at is how the slope of the curve differs across percentiles. The next installment is practical for everyone who studies: The Testing Effect and Spaced Repetition.