Paper of the Day
Andrew Oswald, “On the Common Claim that Happiness Equations Demonstrate Diminishing Marginal Utility of Income,” http://www2.warwick.ac.uk/fac/soc/economics/staff/faculty/oswald/concavity2005.pdf
From the end of the paper (it’s really short):
Psychological wellbeing is not measured in objective units. It is necessary, instead, to listen to what people say. Under the (reasonably mild) assumption that happier people tend to report themselves as happier, we can learn about the direction of influences upon subjective wellbeing. The literature has done this.
However, it requires more stringent assumptions about the nature of people’s answers before we can draw conclusions about curvature. We know little about the shape of the reporting function that human beings use.
Imagine, for example, that there is constant marginal utility of income, but that people, as they get happier, mark themselves happier on a questionnaire scale but do so in a way in which they are intrinsically reluctant to approach the upper possible level on the questionnaire form (the 5 on a 1-5 scale, say). Then the reporting function itself is curved, and we will have the illusion that true diminishing marginal utility of income has been shown.
In conclusion, despite what many articles and textbooks have begun to say, the literature has not established that happiness is curved in income.
I’m glad to see that the conceptual distinctiveness of reported happiness and real happiness, whatever that may be, has really sunk in for Oswald. This directly pertains to the discussion of framing below. The function that maps reported happiness to real happiness depends on the nature of framing effects for self-reports, among other things. I also like Oswald’s hypothesis that people have an aversion to identifying our sense of well-being as falling on the upper limit. After all, we all believe we could be happier, or that others out there are happier.
[Thanks to Tim Hicks for the link, and apologies for leaving his helpful comment in purgatory for the sin of including two links.]
7 Comments so far
Leave a reply
Blogosfera
* Marginal Revolution esclarece um pouco sobre um senso comum bem ruinzinho: o de que McDonald’s é a causa da obesidade norte-americana. Coisa de gente pouco séria como Michael Moore (eu acho que você pode criticar/elogiar Bush, Zico ou Maradona:…
I think Bernard van Praag and Paul Frijters spend some time attempting to deal with the shape of the reporting function more directly in their chapter of Kahneman, Diener, and Schwartz’s “Well-Being: Foundations of Hedonic Psychology”. From memory, they argue against the Oswald-type hypothesis, mostly on the basis of how people respond to these types of scale questions in other domains. But I’d have to check.
Of course, part of this is probably driven by their assumption that satisfaction is bounded above, which - though not entirely implausible - certainly isn’t obvious a priori.
Of course happiness != reported happiness, but it follows from reasonable assumptions and utility theory that happiness should usually be an increasing function of income with diminishing returns. After all, people buy the things they want most first. Subsequent purchases are of things they want less. This is why insurance is generally considered desirable and gambling undesirable. The interesting question is how quickly marginal utility tails off. A logistical distribution of prices and values for goods suggests that exponentially increased spending buys linear hedonic returns, but the logic of finance suggest exponential money buys less than linear returns. A careful reading of Adam Smith and David Hume suggests to me the more nuanced view that a modern person might describe as exponential spending buying a smoothed step function of hedonic returns. Linear utility functions for money suggests that consumers at all wealth levels would consume identical baskets of goods. When a pair of poor people wanted something expensive but couldn’t afford it they would make even odds gambles and one of them would buy it. Unless simply being alive was assumed to have huge positive utility at *any* level of spending (in which case the utilitarian argument for preventing every avoidable death and for preventing birth control including abstinence becomes terribly strong and we are advised to factory-farm ourselves) linear returns to wealth also suggests that the poor should pair up as life insurance beneficiaries and play Russian Rollette.
I just now got onto this discussion about happiness (sorry so late). This seems like a cool project… I’d like to offer some further comments on the business of framing, as well as the wider enterprise of well-being reports.
First, attempts at offering objective measures of subjective states are nothing new. The framing problems that accompany this are common in fields besides psychiatry. A lot of people, for instance, are skeptical of illnesses like fibromyalgia—a condition where people report pain, but once the doctor puts a name to it, their condition may worsen or improve due simply to the doctors’ having created a pathology simply by naming it (the word made flesh, as it were).
Some say pain is pain whether there or not there is a clear pathology. Others argue that self-fulfilling prophecies should be avoided in the absence of an objective pathology. In the case of “happiness states” or depressive episodes, a subject’s reports are all that we have to go on—as well as their wishes regarding treatment. (Note: the psychiatric and physiological questions about well-being and pain are so often connected. A holistic approach is often required in diagnosis).
Ironically, the Szaszian – for example – finds himself in a bit of a pickle because of this problem, as he is committed to the absolute freedom of a subject to improve his well-being, and yet he is normally hostile towards diagnostic criteria in the absence of identifiable pathologies. Some of this boils down to the Szaszian’s dissatisfaction with a physician’s authority. But the Szaszian’s problem reflects the wider psychiatric issue of trying to measure happiness through subjective reports (which you have discussed on this blog). Even if we could say: “you should be happy because your serotonin levels are above normal,” we would have a hard time disputing the claims of the subject. Such is not to argue that there is no correlation between subjective states and physical brain states (there most certainly are), it’s simply to say that the disconnect between people’s actual well-being and how they report it might be as big as the disconnect between reported mental states and whatever complex physical substrate the mental state arises from. The answer? We have to keep winging it with words (reports). What this does to the “progress paradox” assertions, I can’t say for sure.
But in adding further to these complications (and at the risk of introducing categories to holistic and thoroughly subjective systems), many people may not be able to parse the distinction among many types of “well-being” states. As a person who has suffered from depression, (take my subjective reports with however many grains of salt), I know that I can identify a number of different types of well being. The first one you might call “eudaimoniac” states, or the purposeful living-and-doing kind of happiness you alluded to in one of your post. Then there is the mildly euphoric type of happiness that is the general buzz of being “ok” and in a good mood, although this can be pronounced even if eudaimoniac states are troughing (and the reverse). Third, there are what you might term primal states of well-being that accompany fulfillment on the first few rungs of Maslow (sex, food, sleep satiation). And finally, there is a calculative form of well being that some might be able to parse, although this one is tough to describe. It’s not eudaimoniac exactly as it derives from the interesting sensation of thinking or being self-involved in thought. Perhaps it’s an aspect of eudaimoniac states, but I see them as a more general understanding of self, or self-concept as applied to following one’s vocation or avocation. (Maybe self-esteem should be separate, too). Whether or not we should include spiritual or more euphoric states that accompany religious experience and love of wife and family—I cannot say. But the point is, well-being is a multi-faceted gem and part of our difficulty reporting on our subjective states may result in an unwillingness or inability to parse different types of well-being. Thus, the framing issue also nests a category issue. Maybe a better metaphor for happiness is an engine in which the types of well-being are all cylinders, all of which or none of which may be running at any given time.
Finally, I should also say a little bit about another important aspect of depression and well-being. Other people’s reports. When family members “notice a difference in you” and they like it, that should count for something for individuals in treatment. This is so often the story with family members who say “she is just so much cooler on Paxil. Without it she’s a nutcase. I hope she stays on it.” I admit people have said similar things about me (replace “Paxil” with Zoloft and “she” with he).
One possibility with respect to apparently-curved happiness functions is that the 1-10 linear rating scale actually conceals a logarithmic scale.
That is: do people really believe that there are finite, fixed, meaurable upper and lower limits to happiness? This sounds counterintuitive to me. It’s quite possible that many people are really rating their affect on a scale of negative infinity to positive infinity, and thus their ratings are logarithmically compressed into the 1-10 interval.
What results would one expect from such a logarithmic compression? We would expect to see values that cluster tightly around the center, are difficult to shift in either direction and show diminishing increases as they move outward from the center, with very few extreme values. That is precisely what we do find.
To my knowledge, no study has asked subjects if they believe that there actually are absolute limits to happiness and unhappiness, or if they are judging their happiness in the context of imagined infinite values. The fact that nearly all cultures have some concept of “Heaven” and “Hell”, transcendent states of joy and woe, suggests that the propensity to imagine such infinite possibilities is not rare.
generic levitra.
Awesome, man