Arnold Kling  

More Unhappiness

PRINT
Communist Economic Policy: Sta... Builders' Rights?...

I continue to have philosophical disagreements with "happiness research." My latest essay takes on a recent paper by Alan Krueger and Daniel Kahneman. I write,


With research into subjective well-being, economists are making statements about what constitutes the good life. In doing so, we are encroaching on territory once claimed by philosophers and theologians -- and, more recently, by self-help gurus. In the 70's, it was I'm OK, You're OK. Now, we are saying "I have positive net affect, you have positive net affect."

...Still, I have a feeling that if happiness research proceeds far enough, it will serve merely to rediscover some eternal truths. For example, this New York Times story cites work by Claudia Senik, who found that "that when people aspire to a better quality of life within the next 12 months, the attempt to reach that goal alone -- the anticipation independent of the outcome -- seems to bestow happiness in the present." Have the sages not been telling us this for centuries?


Comments and Sharing





COMMENTS (20 to date)
Ronnie Horesh writes:

It's only at low levels of income, wealth, nutrition, health, or literacy, etc, that numerical measures correlate strongly with welfare or happiness. Economists and politicians should do what they can to put in place tightly-woven safety nets, and leave the speculative stuff to academics.

conchis writes:

Arnold seems to argue that economists should not be policy advocates... except when he likes those policies.

OK, so that might be a little bit of a straw man. But it's nothing compared to the scarecrow Arnold spends most of that article attacking.

conchis writes:

P.S. Senik's claim is actually rather more interesting than Arnold makes out: she argues that in the transition economies, and other cases where people are subject to high income volatility, the oft-quoted comparision effect (whereby others' income makes us feel worse about ourselves) is dominated by Hirschman's (1973) tunnel effect: other peoples' income actually has a net positive effect on individuals' well-being, because they take it as a signal of their own future prospects.

That strikes me as far from the 'eternal truth' Arnold suggests it is. But then, maybe he's just wiser than me.

conchis writes:

Arnold,

Does your beef with happiness economics extend to those parts of it that seek to examine the behavioural effects of self-reported well-being? For example, Clark et al (1998) suggest (unsurprisingly) that self-reported job satisfaction is a good predictor of quit rates. To the extent that this is true, surely it's of interest whether our usual assumptions about what makes for satisfying jobs are correct. Moreover, the use of self reports as proxies for utility may enable us to distinguish between alternative hypotheses that revealed preference doesn't. (To use Andrew Clark as an example again, he attempts to explain a persistent the disjunct between rising tenure-wage profiles and productivity on the basis of individual preferences for rising wages, rather than the usual bargaining explanations.)

As an aside, does anyone else find it interesting that so many of the economists interested in happiness research (Clark, Oswald, Layard) are labor economists?

Barkley Rosser writes:

Arnold,

Would you be happier if the happiness economists came to conclusions that overturned eternal wisdom?


conchis,

The founder of happiness economimcs research, Richard Easterlin, was originally a demographic economist.

Barkley Rosser writes:

Arnold,

Would you be happier if the happiness economists came to conclusions that overturned eternal wisdom?


conchis,

The founder of happiness economimcs research, Richard Easterlin, was originally a demographic economist.

JohnDewey writes:

"For example, Clark et al (1998) suggest (unsurprisingly) that self-reported job satisfaction is a good predictor of quit rates. To the extent that this is true, surely it's of interest whether our usual assumptions about what makes for satisfying jobs are correct."

That's interesting, but I'm not ready to draw conclusions from it.

After talent is recruited by a firm, the new employee is often asked if they know anybody else at the old firm who would make a good employee. My guess is that the malcontent rarely gets recommended. Firms are not likely to lose unhappy employees due to networking.

Brian writes:

My question is why is this a bad thing? To find statistical validation for sagely advice seems to be a great benefit from my perspective. It adds strength to what the sage says and makes their advice much more persuasive to many. If this helps to convey the wisdom of the ages, isn’t that a good thing? Might it actually end up helping people to identify how to pursue a happier life?

More of my thoughts on this post on my blog.

conchis writes:

Barkley: yeah, I know. I still find it interesting that a lot of the others seem to be labour economists.

JohnDewey: I think you're possibly assuming a bunch of other things about the malcontents that aren't necessarily justified (i.e. that their malcontent is a personality trait, rather than a result of situational factors). In any event, the point was a more general one: subjective well-being is likely to have behavioral implications; if so, then we can be interested in what drives it for behavioral, not merely normative reasons. These are both things that happiness research can help us with.

JohnDewey writes:

conchis,

I'm not assuming anything. I just said that people who are unhappy, justified or not, are not as likely to get job offers through networking as people who express job satisfaction. For me, that's one possible explanation why "self-reported job satisfaction is a good predictor of quit rates."

conchis writes:

JohnDewey,

Sorry, I clearly misinterpreted your point. On the other hand, now I just don't understand it: less satisfied people tend to quit their jobs more, so how does their being less likely to find other jobs explain that?

JohnDewey writes:

conchis,

I was just offering a possible explanation for the report you referenced:

"Clark et al (1998) suggest (unsurprisingly) that self-reported job satisfaction is a good predictor of quit rates."

Have I misunderstood what you wrote? I thought you meant that satisfied workers had higher quit rates.

conchis writes:

JohnDewey,

It seems I have misled you - for which I apologise. The paper suggests that less satisfied workers have higher quit rates. I meant "satisfaction" in the neutral sense of "a point on the satisfaction-dissatisfaction scale", rather than in the sense of "greater satisfaction". I should have been clearer. (Although I would have thought the "unsurprisingly" comment would have helped here.)

JohnDewey writes:

"Although I would have thought the "unsurprisingly" comment would have helped here"

Well, I'm not an economist, so perhaps I'm just not bright enough to keep up.

What's unsurprising to you may differ from what's unsurprising to me. I have observed that my employers couldn't get rid of many malcontents no matter how miserable they claimed to be. These folks just hung around for years, driving the rest of us nuts with their bitching and moaning.

I would guess that, within a small department, we might see a positive correlation between a worker's degree of job dissatisfaction and the quit rates of his otherwise satisfied coworkers.

conchis writes:

Fair enough. As I said, I should have been clearer. Sorry.

jill writes:

Arnold, excellent post!!!

The major problem I have with Behavioral Decision Theory (BDT) as a Cognitive Scientist is that it does not sit on a solid foundation theory. It is a area build upon many interesting empirical findings that are just statically significant. As a graduate student from a top 5 graduate program, I jointly ran the department’s experimental student research pool as a doctoral student. I saw first hand the number of times the average BDT study was run to achieve stat significance.

I am amazed how much social psychology and BDT has changed in the manner in which results are presented since I graduated. As one who was trained with Cook & Campbell and Keppel I am appalled how quickly BDT (now even the great Kahneman) assert complete understanding.

1) BDT theory is very situational in its nature. This immediately calls into question the external validity of the findings. One should then be very cautious in generalizing findings to other situations.

2) BDT does not have a strong theoretical foundation. The cognitive mechanisms are POORLY understood. We do not know much beyond that people employ “heuristics” when reasoning. This suggests that internal validity is not a point of strength in this area of study. Again urging great caution in making inferences outside of the area of study.

3) A third issue that is not a strength of BDT research is construct validity. Many of these studies are

In summary, while I do not question the empirical findings, I am not sure we still understand why we observe them beyond the simple assertion that people employ heuristic reasoning. Heuristic reasoning is a fine first level explanation but it is not a theory. The tons of research offered by Kahneman and Tversky (as well as their fellow researchers) has created a vast catalog of empirical findings but no real that theory. We are merely labeling approaches that may or may not be cognitive different (anchor and adjust, framing, recency, mere similarity, etc.). I do not know how many papers have been published in the area that tweak similarity models originally offered by either Kahneman and Tversky which explain one phenomena better than another. This should raise a flag that we may not have correctly framed the question.

Barkley Rosser writes:

jill,

BDT and happiness research are not the same thing.

jill writes:

Barkley Rosser-

Yes and no. On the yes side, Happiness research is fixated on a dependent variable called "happiness", while BDT focuses on decisions. On the no side, we see that a great deal of BDT is about the nature of utility (i.e., loss aversion, prospect theory) and how it impacts choice. BDT starts with choice but draws upon utility to explain choice, while Happiness research starts with a notion of utility and moves toward choice. They are two sides of the same coin. The fact that Kahneman is into both should be highly suggestive to even the a casual observer with no theoretical leaning.

Barkley Rosser writes:

jill,

But in economics the standard view that BDT competes with is also based on utility. The rational homo economicus maximizes her or his utility subject to budget constraints, knows fully what gives her or him that utility, and knows how to obtain it, and is not distracted by such effluvia as context or framing or all that BDT stuff, and is also never internally inconsistent in her or his preferences.

conchis writes:

jill,

on the yes side - none of the criticisms you offered of BDT appear to be criticisms of happiness research.

Question:

"a great deal of BDT is about the nature of utility (i.e., loss aversion, prospect theory)"

To what extent has loss-aversion found empirical support as a theory about experienced as opposed to decision utility? Maybe I'm wrong, but I was under the impression that (almost?) all the evidence was in fact for the latter. I think Dan Gilbert has suggested that loss-aversion may in fact be irrational, and due to an asymmetric impact bias, whereby we incorrectly overestimate the hedonic effects of future losses relative to gains.

Comments for this entry have been closed
Return to top