I've been reading the original Easterlin paper * because it looked cool. I have a few thoughts. I haven't looked at the literature, but I'm confident that someone else pointed this out in the 50 years since the paper was published.
The paradox is basically that while the rich are happier than the poor, people don't get happier as they get richer.
Now when you look at Eaterlin data, it sure looks like there is a paradox.
This is the table for the interpersonal comparisons, as you can see, the poorest report a happiness rating of 1.8/4 with 16% saying they're very happy, while the richest report a happiness rating of 2.8/4, and 44% saying they're very happy.
Now let's look at the intrapersonal data.
As you can see, as incomes rise, reported happiness remains flat. I have some thoughts on explanations, but we can talk about those later. What I'm more interested in, is the damn graphs.
Let's look again at the relationship between happiness and income. People's happiness generally increased by 0.2 points for every $10,000. I'm assuming that the "less than 10,000" group is 0 to 10,000, and I would expect people with incomes of $0 a year to be very different from those who make $5,000. I expect way much more heterogeneity from this group than the others. The life cycle data starts from 10,000 anyway, so we can safely ignore it. I'm also ignoring incomes higher than $50,000. Why? Because they go up to 75,000 which is a $25,000 increase. And the $75,000 goes up to infinity??? Is this where the meme that happiness peaks at $75,000 comes from?
So we're stuck with the 10,000 to 50,000 groups. We have a 0.2 increase from 10-20k to 20-30k. The same increase for the 20-30k group and the 30-40k group, and a 0.1 decrease from going up to the 40-50k income group from 30-40k. This gives us a mean increase of 0.1 increase per $10,000. The longitudinal data shows that the sample started from an income of about $12,000 in their 20s, and ended up with maybe $22,000, I'm eyeballing it so a bit unsure.
This gives us an increase of $10,000 maybe $15,000 from ages 22 to 55. We would expect a 0.1, maybe 0.15 increase in happiness, so for example the happiness line would go from 2.3 or to 2.4 or something. This doesn't show in the data. But the y-axis is way too big to see it. It starts from 0 and goes up to 4, but in our rich-poor data, it starts from 1.8 and goes up to 2.8. And when I exclude the less than $10,000 and the more than $50,000 populations. We would have a happiness line that goes from 2.1 to 2.4, a measly 0.3 increase. It's not too surprising when your ginormous y-axis doesn't show that.
I'm not saying that by zooming in we would find the hidden solution to the paradox, but the paradox itself is, from looking at this dataset, exaggerated. And the fact that the 40-50,000 group is 0.1 points less happy than those 10,000 poorer, is evidence for the "there is nothing going on, bad nongranular dataset produces funny results sometimes" hypothesis. Although I would love to download the dataset and tweak it in R to see what's going on, I don't know how to use R.
*This isn't actually the original, which was published in 1974, this is another one that was published in 2001, which includes his hypotheses for the paradox.
