In my last post (here), I looked at the mounting evidence that GDP per head is correlated with happiness when tracked for individual countries through time—a finding that goes against the previous orthodoxy that went under the moniker of the Easterlin Paradox (if we all get richer, none of us get happier).
The U.S. and China are sometimes argued as key countries that show no such improvement in happiness, but anti-Easterliners explain away the U.S. by pointing to stagnant median income growth through time (GDP per head has risen, but it has all gone to an elite, so most people haven’t secured any income-induced extra happiness), and view the China findings as irrelevant due to a lack of sufficient data.
The situation is ironic since it is only recently that advocates of the Easterlin Paradox have made headway in transferring their ideas out of academia and into the public domain, so catching the attention of politicians. Here is the economist Andrew Oswald in an Op-Ed in The Financial Times in 2006 (here):
But today there is much statistical and laboratory evidence in favour of a heresy: once a country has filled its larders there is no point in that nation becoming richer.
The hippies, the Greens, the road protesters, the downshifters, the slow-food movement – all are having their quiet revenge. Routinely derided, the ideas of these down-to-earth philosophers are being confirmed by new statistical work by psychologists and economists.
Justin Wolfers, the Easterlin Paradox’s great nemesis, would beg to differ. Accordingly to him, GDP per capita has captured human welfare as encapsulated in the idea of self-evaluated happiness quite well. Indeed, he views the happiness literature as maturing to a stage where it aligns well with GDP and, indeed, the old stalwarts of economic analysis ‘utility‘ and its first cousin ‘revealed preference‘—as such happiness has become respectably boring and quite neo-classical economics in tone.
“Utility’ and ‘revealed preference’ are the two trump cards of orthodox economists when confronted by arguments from non-economists that money can’t buy you happiness. Such economists will say “don’t listen to what people say, look at what they do”. And what people frequently do is buy, buy, buy—or work like hell so they are able to buy, buy, buy— to the exclusion of all those things that are supposed to bring happiness like hanging out with the kids, communing with nature, going for a jog, catching up with old school friends and taking up charity work.
Nonetheless, while Wolfers appears to relish his bar fight with Richard Easterlin, he has been very reluctant to take on the titan of behavioural economics, Nobel Laureate in Economics Daniel Kahneman. In Wolfers last major paper on happiness written with his wife Betsey Stevenson, the conclusion purposefully avoided any confrontation with Kahneman:
To be clear, our analysis in this paper has been confined to the sorts of evaluative measures of life satisfaction and happiness that have been the focus of proponents of the (modified) Easterlin hypothesis. In an interesting recent contribution, Kahneman and Deaton (2010) have shown that in the United States, people earning above $75,000 do not appear to enjoy either more positive affect nor less negative affect than those earning just below that. We are intrigued by these findings, although we conclude by noting that they are based on very different measures of well-being, and so they are not necessarily in tension with our results.
This is interesting, because Kahneman says some quite specific things about the use of the word ‘utility’ by economists in his magnum opus “Thinking Fast and Slow”.
As economists and decision theorists apply the term (utility), it means “wantability”—and I have called it decision utility. Expected utility theory, for example, is entirely about the rules of rationality that should govern decision utilities; it has nothing to say about hedonic experiences.
Kahneman goes on to make a distinction between the ‘remembering self’ and the ‘experiencing self’. The latter is concerned with the immediate emotions of joy, love, hate, sadness and so on and is completely distinct from the former’s happiness calculus gleaned from a balancing of a perceived life’s worth.
The book highlights an example of this dichotomy: the contemplative question of whether one’s happiness would increase if one moved to sunny California from the weather-challenged Midwest. The example is played out as a husband and wife spat. The wife believes that all will alter in a move to a sunnier clime, the curmudgeon of a husband says nothing will change. And on this occasion, the data suggests that Kahneman is right. Weather (and climate) is the wallpaper of our lives: it is something that we will barely give thought to for more than a few minutes per day—and most often we see it as a given in our lives: neither a subtracter of happiness nor an additor.
Here is Kahneman filling out the different concepts of happiness:
So what happens if we start to measure experiential happiness rather than remembered happiness? The former is sometimes divided into positive affect—joy, love, hope, amusement and so on—and negative affect—pain, sadness, hate, regret and so on. What we find out, accordingly to Kahneman, is that the correlation between the remembering self and the experiential self is only 0.5. Events that will maximise self-evaluation of happiness will not necessarily maximise experiences. That is why people choose to take a job with a long commute or work for a bulge bracket investment bank like Goldman Sachs, even though both choices may be very negative in terms of experiential happiness.
In a classic paper with Angus Deaton, Kahneman actually teased out the impact of a rise in income for the remembering self and experiencing self. He came up with this chart (click for larger image) from this seminal 2010 paper (here):
And for those who like numbers, we have this table below from the same paper. What you see is a reasonably high correlation between income and how we perceive our lives (the Cantril ladder of life satisfaction from one to 10) but a very low correlation with positive affect (joyish kind of stuff) and blue affect (sadness kind of stuff).
So Justin Wolfers may have felt he had won the war, but has he in fact just won an insignificant battle? More to come on this.
Risk and Well-Being 
The debate is conducted in terms of income and consumption, but I think we have also moved from a world where work was risky to physical health to one where many suffer work-related mental-health issues. Plus people find any change traumatic, regardless of whether it is change for the better or for the worse. There are (at least) these three different stands to try and untangle – change in income inequality, change in occupational health, change in change!
Great post and thanks for the link to Kahneman’s TED talk.
It looks like money can’t buy (experiential) happiness after all, but can buy increased satisfaction with relative status. I think that most people associate the word “happiness” mostly with the experiential kind; I know I do.
Also very interesting is the lack of correlation between the experiential and remembered kinds of happiness. It’s strange how many folks will make themselves miserable on a daily basis just to get the money that buys them increased status and remembered happiness. Also interesting how many religions try to get people to eschew pursuing remembered status so as to avoid “suffering” and just “be here now”.
So, after I answered to your comment on my blog, I’m going to make some remarks about your last two posts about sustainability and happiness.
1. What strikes me about Wolfers’ and Stevenson’s analysis is that, in all but one regression presented by you in the charts their line of best fit has a negative intersect (and a rather small positive slope), which means that, up to a point, GDP per capita growth is correlated with declining happiness. This is a finding worth explanation/consideration and particularly important since I see in the 1982-2005 chart mainly rich countries in the lower right quadrant (decline in happiness despite increase in GDP per capita).
2. I would love to see the p-values for the coefficients they estimated there. Indeed, I would bet that if you add further explanatory variables to the regression, the effect of GDP per capita on happiness would diminish if not disappear.
3. As you mentioned the case of the U.S., my first thought was: if we really have to talk about income, let’s take median income or a related measure, not GDP per capita.
4. How do Deaton and Kahneman measure experiential happiness? I cannot easily imagine how it should be done (and then meaningfully correlated with income)… And, a second important question: what is the relevance of this experiential happiness? I mean, Jeremy Bentham would love it, but the limits of hedonic utilitarianism have been demonstrated impressively since he died. I don’t see the point in granting this aspect of wel-being much attention. Actually, remembered happiness seems relatively more convincing to me (though not really convincing, as should be clear from our discussion on my blog).
Bartosz: Wolfers test statistics are all in the original papers, but if you need to delve into them more your really need to contact him directly. I had an e-mail exchange with him a couple of years ago about betting markets and climate change and he was quite accessible then, so hopefully the hubris is not too high now 🙂
As for experiential data methodology, you need to read Kahneman’s “Experienced Well-Being” chapter in “Thinking Fast and Slow” which references all the literature.
I want to stress that I think we are on the same page here in terms of being skeptical about GDP as a welfare indicator. Yet I also feel we need to realise that for many people within a capitalist economy, the dominant urge is to maximise income and consume. To move to a steady-state or no-growth economy requires a set of policies that confront this fact. To get people to not ‘do what they do’ requires persuading them that a) maximising current consumption is not the same as maximising current happiness (however you want to define the term ‘happiness’) or b) maximising consumption now will have a cost in terms of maximising happiness in the future or c) maximising consumption now is causing a loss of happiness of other parties. In many ways, I don’t think economists are the best people to tackle these questions; such questions appear to be more in the realm of psychology or social psychology. That’s why I think Kahneman’s work is so fascinating.
Lazy as I am, I hoped for answers to my questions without having to read the papers;-) But I just recently ordered Kahneman’s book, so at least this question will “answer itself” when I read it. As for Wolfers, maybe I’ll find the time some day to run a regression myself and see whether my hypothesis about the effect of income on happiness is true. As a young researcher, I can only need stuff like this to publish;-)
Regarding the last paragraph of your comment: I agree. I don’t think that the differences in our views are really serious. Indeed, the points you made about what should be done to achieve sustainability are basically right from my perspective. Which does not mean that we cannot discuss the details;-) I also fully agree with what you write about conventional economists’ role in the debate. Our professional training has limited potential here, unless we let more behavioral economics/psychology insights in (Elisabeth Gsottbauer and Jeroen van den Bergh once wrote an interesting – even though rather poorly written – paper on this).