Is the Flat Trend in Self-reported Happiness a Problem?

In yesterday’s NYT, Darrin McMahon, an historian at Florida State, has some good advice about the pursuit of happiness: don’t pursue it. That’s a wortwhile lesson I associate mainly with Bishop Butler and Henry Sidgwick. However, despite the good advice to “just do it,” as it were, McMahon also repeats a happiness studies chestnut that deserves to die a brutal death. Or, at least, deserves to be confronted with enough skepticism that people will stop repeating it without further explanation or justification. Or a brutal death.

Sociologists like to point out that the percentage of those describing themselves as “happy” or “very happy” has remained virtually unchanged in Europe and the United States since such surveys were first conducted in the 1950’s. And yet, this January, like last year and next, the self-help industry will pour forth books promising to make us happier than we are today. The very demand for such books is a strong indication that they aren’t working.

Should that be a cause for concern? Some critics say it is. For example, economists like Lord Richard Layard and Daniel Kahneman have argued that the apparent stagnancy of happiness in modern societies should prompt policymakers to shift their priorities from the creation of wealth to the creation of good feelings, from boosting gross national product to increasing gross national happiness.

McMahon doesn’t say whether or not he is concerned, or sign on to the Kahneman/Layard program, but he does repeat the peculiar illogic by which one is tempted to move from the fact that self-report numbers are stable to the idea that we ought to shift priorities from creating to wealth to creating good feelings. I wish I’d stop seeing this sort of thing in the New York Times. So let us review the numerous reasons why the stability of happiness self-report surveys need not imply some kind of policy failure that needs correction.


The main explanations for the lack of increasing happiness are also explanations for the lack of talk about increasing happiness. So the self-report data may be reflecting the psychological processes that determine the avowal or self-ascription of subjective states, rather than reflecting the objective character of the subjective states themselves.

The main explanations for the “very happy” flatline are (a) adaptation and (b) social comparison. Now, if people don’t have a reliable “hedonometer,” which we can consult through introspection to discover the objective quality of our subjective experience, then we should expect that self-reports will be subject to habituation and comparison effects.

ADAPTATION. If you ask someone who has been wearing rose colored lenses for a week what color things appear, they will underestimate the objective rosy quality of their subjective experience, due to habituation. If you switch lenses the next week to an even more intense rose, the judgment of the rosiness of experience will likely stay pretty much the same. There is no reason to expect judgments of the hedonic quality of experience to much differ.

SOCIAL COMPARISON. The lack of a hedonometer requires a social rather than a private, internal standard for self-reports. When asked how happy one is, one will compare one’s narrative about one’s inner life, and one’s behavior, against a widely accepted cultural conception of what it means to feel and behave happily. If the entire population is becoming objectively happier over time, then cultural meaning of “very happy” will shift. Our talk about happiness is the result of comparing our representation of ourselves with our representation of the happiness scale, and our representation of where others lie on that scale. So even if people are getting objectively happier (whatever we take that to mean), we should expect self-reports to remain stable due to the shifting goalposts of the social meaning of happiness. It’s the same process by which people who make 150 large a year come to claim that they are middle class.

LIMIT AVERSION. This is closely related to what I’ll call “limit aversion.” Since there is no hedonometer, people have no way of telling whether there is an upper bound to happiness, or where the upper bound might be. So they don’t know if they are at the upper bound. Yet it is natural to believe that there are others who are happier, or to suppose that one might become happier still. So people who are at or near the upper bound may be hesitant to report that they are in the highest category of happiness. An illustrative anecdote: I’m pretty sure that, overall, I’m rather better off now than I was ten years ago. I know that I probably would have reported myself as “pretty happy” and not “very happy” then, and I would report that I’m “pretty happy” now, since I just don’t think I’m the happiest kind of person. And I suspect that I’ll feel the same way in another ten years, even if my well-being improves to the same degree. I know that I dislike the idea that my happiness level may have maxed out. I don’t suspect that others are that different.


The idea that humans have some kind of happiness bank that could possibly have an ever-increasing balance is just silly.

However, the happiness bank assumption seems to be behind all the “paradox of prosperity” books, in which the authors pretend to find it alarming that the balance in our happiness accounts does not have a linear relationship to the balance in our bank accounts. Yet, as far as I can tell, no one has ever really defended the happiness bank hypothesis.

HOMEOSTATIC HAPPINESS. It turns out that no one really has a good account of what happiness really is. The best explanation of the nature of certain kinds of positive affect, which some somewhat vulgarly identify with happiness, is that it is a homeostatic mechanism designed to readjust after achieving a goal in order to keep us always wanting more. The prospect of happiness-as-pleasure is a Darwinian carrot that keeps us pulling hard in harness, and it just wouldn’t work if we stayed happy as clams when we got what we wanted. The homeostatic conception of happiness explains why hedonic adaptation may be adaptive. And it also suggests, to use the dumb analogy, that there may be something like a balance limit to our happiness accounts. Additionally, each person’s limit is likely a largely a function of their individual psychological disposition.

The fact that the wealthy liberal democracies are all toward the top in cross-country comparisons of average self-reported happiness, and that the “very happy” numbers aren’t rising, might indicate that these societies are close to as good as it gets—with a large proportion of their populations at their limit–not stalled and in need of a policy jump start.

And, back to something McMahon said, the fact that self-help happiness books continue to fly off the shelves does not really imply that that they do not work. It may tell us that self-help books are part of what have helped us achieve so much abiding happiness for lo these many decades. Or that people just like reading them. Given the simply embarrassing lack of longitudinal studies (it is, of course, hard to get tenure while waiting 20 years for the data to come in), for all we know people who read self-help books generally are happier, but they are offset in the aggregate by people miserably addicted to Us Weekly. If the happiness bank hypothesis was true, then a flat happiness trend line plus high self-help sales might speak volumes. But since it isn’t, it doesn’t.

So, it may be that we are in fact getting happier, and that the surveys can’t track the shift, due to adaptation, social comparison, limit aversion, and other phenomena governing the self-reporting of subjective states. Or it may be that some form or other of market-based liberal democracy is pretty the best we can do from a policy perspective, happiness-wise. Many of the policies that Layard pefers, for example, are already in place in this or that Scandinavian social democracy. And the difference in happiness between these countries and other wealthy basically liberal market societies is trivial or non-existent.

I’ve obviously gone way off the McMahon hook. But let me take the occasion to share my own conclusions about happiness and policy so far:

  • Many people in rich market liberal societies are getting happier and the surveys miss it.
  • Many people in these societies are at or near their hedonic limits. (Note: I think it may be possible for some people to push the limits through the right combination of diet, exercise, meditation, counseling, adventure, simplification, etc. But there are many others for whom this kind of stuff is pretty much impossible without fundamentally altering the kind of person they are. And I think most people, once they hit a certain threshold, are happy enough that they could just care less about further maximizing the positive qualitative character of their background affect. Dharma just wasn’t that much better off than Greg, once he loosened up a little.)
  • Even if everyone was maxed out, there would still be a distribution of people through all the survey categories due to individual psychological differences. A society in which everyone is as happy as they are likely to get is not a society in which everyone reports that they are “very happy.”
  • There is very little policy-wise that will have a large impact happiness-wise in societies that are already have advanced market institutions and liberal-democratic political institutions.
  • The people who are least likely to be maxed out, and most likely to benefit from upward hedonic mobility in any society, are the poorest people.
  • The policy lever most likely to help poor people out is whatever lever maximizes GDP growth. So prioritizing the creation of good feelings requires prioritzing the creation of wealth.
  • If a society has a class of people who appear economically stuck, such that they tend, generation after generation, to see little benefit from growth, even if benefits are otherwise widely distributed, try to unstick them by removing things like bad welfare policy, or labor policies that price low skilled workers out of the market. Often it is crucial to break down informal cultural norms that discourage the accumulation of human capital, but there’s not much liberal policy can do here.
  • Since most people in rich societies are already pretty happy, people who care about happiness ought to worry less about marginal policy changes in the US and Europe and worry more about people who do not already live in rich societies. The best thing we can do for them is free trade, more hospitable immigration policies, and fiscal policies that maximizes world GDP growth.
  • Races for positional goods shouldn’t concern us.
  • Happiness as we tend to think of it is not a natural kind, but a culturally loaded syndrome of feeling and behavior. It is far from the only moral and political value, and shouldn’t be our sole standard for evaluating policy.

In conclusion . . . You don’t need to be a philosopher of science to know that theory is undetermined by data. But it appears that lots of happiness studies folk either don’t know, or need to be reminded. The most plausible interpretation can’t simply be read off the data, and, to my mind, the most plausible interpretation probably isn’t one that supports actively trying to design policy that prioritizes creating good feelings over creating wealth. An obvious surface reading of the data that requires no fancy framing effects rigmarole says that we are, as far as we can tell, at least as happy as we have ever been, which is very happy. Why prefer “we are getting no happier” over “we have been, and remain, extremely successful at creating happiness?” The main reason why, I take it, is that it’s impossible to use the happiness data to drum up demand for one’s favorite unpopular policies without framing it in a way that makes it look like there’s some kind of problem that needs to be fixed. If you say that data show that we’re just as happy as our grandparents in America’s nuclear family, bowling together, Leave it to Beaver golden age, we’ll never socialize medicine! Anyway, the point is: at the very least, you need to at least tryto eliminate the most plausible competing interpretations of the data before you move on to try to use happiness data to mount your favorite policy hobby horse. No. At the very least, you need to acknowledge that there are alternative interpretations. Until they do that, people trying to sell policy on the basis of happiness research don’t deserve to be taken very seriously.

[Cross-posted from Happiness and Public Policy.]