Tuesday, May 15, 2018

The Overhyping of _The Happiness Curve: Why Life Gets Better After 50_

Jonathan Rauch of the Brookings Institution has just published The Happiness Curve: Why Life Gets Better After 50, which seems to be getting a major media push from a bunch of completely uncritical reviewers and commenters, some of whom really should know better.  It is not that this book is totally wrong or bad, but that it way overstates its case, cherry picking data and the views of people he has interviewed, with only the slightest of caveats.  Among those falling all over themselves to unequivocally praise the book are NYU Journalism Professor, Pamela Newkirk, in a review in the Outlook section of the Washington Post this past Sunday (May 13), and Tyler Cowen, who plugged it on Marginal Revolution and also provided a blurb on its back cover (the only professional economist doing so).  Here is back cover blurb, which shows the nature of this hype, as well as indirectly revealing one of the major weaknesses of this book, its personally anecdotal quality.

""Do you wish to understand the arc of your life?  And why you are likely to end up happier than you are right now?  If so, The Happiness Curve is the place to start.  And I write this as someone who can vouch that the upper part of the happiness life curve is very glorious indeed." (Tyler Cowen blurb)

OK, I am glad that Tyler is happier now than he was when he was in his forties, and I am also glad that a bunch of people Rauch interviews also report this as well, with Rauch himself making it clear that he himself experienced this.  His affective happiness, his momentary mood, was generally pretty good, but his evaluative happiness, his broader life satisfaction, was not good.  But now after turning 50, his life has turned all roses pretty much. This is all very nice, and the personal stories recounted in the book, as well as a lot of general philosophizing in its later sections ("old people are wise!"), will probably lead to high sales, especially given the hype it is getting from various people publicly, with almost nobody noting that it has some problems and exaggerates the scholarly consensus on this.

Oh, I guess I should state the main premise of the book.  It is that in general there is a U-curve of happiness related to age.  On average people's life satisfaction declines from age 20 to somewhere in middle age (depending on which country they are in), and then rises after that. Rauch very slightly notes some caveats.  This is a "tendency," not a universal law (although at times he seems to suggest that is almost a universal law).  Individuals might be happy at 45 but miserable at 60, but they are the exceptions with their weird personal circumstances (maybe lost a job, a spouse, or got cancer). 

So let us start with the biggest problem with this book: its main premise as stated is false, or at least  not at all clearly true.  The problem is that this result derives from multiple regression studies that have one of the several competing measures of happiness or life satisfaction as the dependent variable, with some long list of variables on the right hand side that are widely thought to be connected to happiness, such as income, employment, marital status, health, and others.  In such regressions for many nations, especially high income ones, the identified coefficients for age support the finding of the U-curve as described above, people tending to be less happy in middle age than when younger or older.  We then have all these personal anecdotes to support this, a bunch of people recounting how miserable they were in their forties, only to be so much happier later.

The problem is that in the book itself, the broadest measure of the relation between age and happiness at the global average level does not support this at all.  I am talking about the raw relation, not that imputed after taking account of all these other variables. So, there it is on p. 68, a figure entitled "Average life satisfaction by age (unadjusted world estimate, 2010-2012)" with Gallup World Poll and the Brookings Institution the source.  It shows a mildly oscillating line that slowly trends upwards from age 20 to a peak at about 64, after which it drops a bit to hold steady to about age 80, after which it goes back up a but with no reporting beyond age 85.  That's right, just plain old unadjusted happiness does not decline as one moves into middle age: it gradually rises according to this huge global data set.  Of course on the next page, we have the adjusted figure, which shows a smooth U-curve bottoming out around age 50.

But what then is the point or meaning of all these personal tales of woeful misery in the forties but joy and happiness afterwards?  Nothing.  They are irrelevant and go against the evidence reported in this figure on p. 68.  not a single one of these is a report of "taking into account your income, employment, marital status, health, and other variables such as your social relations, how happy were you at different ages?"  No, these are accounts of the unadjusted states of happiness of these individuals.  And indeed, without dragging through particular persons' accounts (some of whom I know personally and have even heard these accounts), I note that quite a few of them involve people having something bad going on with one or another of these other variables, in several cases marital breakups or personal unhappiness related to personal relationships, with this being overcome one way or another after the person hits 50.  But that is not what this is supposed to be about.

Indeed, the not so clearly mentioned point is that in middle age most people are doing well on these other variables.  They are near their maximum income for life.  They are employed.  They are married. And while their health may not be what it was when they were 25, it is not too bad and better than it will be ever again.  In short, on lots of things, most people are doing pretty well on most of the things that are associated with happiness and life satisfaction,so, big surprise that on average around the world we see them being at least slightly happier than they were when younger overall on average.  The people in the book recounting their woeful forties and their rapturous fifties are the exceptions, the people who had one or another of those other variables go bad on them in their forties, but then get better in their fifties.

There is also serious distortion and misreporting of the data for different countries and regions around the world.  The key figure is on p. 79, where we see purported happiness curves (adjusted, of course) for US, UK, Latin America and Caribbean, China, Germany, and Russia, all of this supposedly again from the Gallup World Poll..  Aside from identifying a low point for each, this also shows life expectancy.  A relation is claimed to be found that in the generally happier countries the low turning point is at an earlier age,  Only in Russia does the low point hit just after life expectancy in the late 60s.  Rauch quotes a wisecrack, "Don't be Russian."  Latin America, whose curve looks quite flat (as does Germany's), has a supposed low point just under 50, with a life expectancy just under 75.  So, it is admitted that the happiness curve is not quite universal, almost, but "don't be Russian."

Oh, except there have been serious competing studies that dispute some of this.  A 2014 paper in The Lancet by Andrew Steptoe and Angus Deaton, "Subjective wellbeing, health and ageing," shows strictly declining age lines not just for Russia but all of the former Soviet Union and Eastern Europe. Ah ha, we can dismiss them as just whiny old people who liked communism and got unhappy when their health care and old age pension systems collapsed!  But in fact Steptoe and Deaton (yes, Nobel Prize winner, Deaton) find a clearly declining line all the way for Latin America and the Caribbean, in contrast to what Rauch reports with no hint that his report is not universally accepted.  Maybe these nations are suffering from lack of old age health care and pension systems like Russia, but in contrast to Russia, this region tends to do better compared to what its income is than does Russia and neighbors, who tend to do worse. Also, sub-Saharan Africa simply has a flat line, no relation with age at all. I may have missed it, but Rauch simply never mentions this region of the world, which tends to have low levels of reported life satisfaction anyway.*

Having dumped pretty hard on this book let me say it makes a good read.  The anecdotes are generally interesting. Much of the philosophizing has a feel good character to it.  I do not think anybody is actually misquoted, although I know that some people he quotes about other matters disagree with his main point, and he carefully does not quote them on it.  But vague philosophizing with anecdotes does not prove much. My late ex-mother-in-law claimed she was happiest in her early thirties.  She had youthful vigor and good health (middle age officially starts at 35), but was mature enough to be taken seriously by others and to understand life and be somewhat established.  It all sounded pretty reasonable to me, and still does.

Along those lines it is curious that the first figure Rauch shows (p. 65) agrees with my late ex-mother-in-law  It is a figure from 2014 of apparently unadjusted happiness numbers by age from the UK.  This one shows not a U-curve or a gradually rising while fluctuating one, but a distinct M-curve.  Happiness rises from 20 to peak at the 30-34 age range.  It then falls to 50-54, then rises to another peak at 65-69, after which it declines.  None of this long-increasing misery in the early 30s that Rauch at times talks about, although he mostly focuses on people unhappy in their forties.  In any case, this figure fits pretty well with one in a paper I recently published based on longitudinal studies across several western European nations, "Experience life cycle satisfaction in Europe," by Robson Morgan and Kelsy J.O'Connor, Review of Behavior Economics, 2017, 4(4), 371-396.  They find the earlier peak to be more in the late 20s, and the later peak more about 70, however with lots of variation across individual nations (they do find a low point in middle age for all of them).

Anyway, maybe this book will help depressed Gen-Xers feel better.  Or maybe it will depress older people who are mysteriously unhappy.  I do not know. I know that when I turned 50 two decades ago I saw an article in a newspaper or magazine (not a scholarly one, but then Rauch claims his book is "reportorial" not "scientific") that claimed that 50 was the age of maximum happiness. I felt pretty happy at the time, so I liked it, although vaguely worried about my impending decline of happiness.  Now I can look at this book and other studies and figure out whether I am now at the peak of happiness with a downturn coming, or maybe it is up, up, and away all the way (at least to 85) as Rauch suggests.  I guess I shall find out, :-).

Addendum:  Another oddity about the personal anecdotes in this book: several of those telling them are active happiness economics researchers now in their fifties or early 60s.  When they were in their forties, such research was not getting much attention, and they were professionally frustrated.  Now it is hot stuff, as seen by the publicity about this book (and the attention they get in it personally).  Their careers have done much better after 50 than happens with most people, who are supposed to deal with the failure of their professional dreams as they see themselves reaching their highest level job and are comparing themselves with others of their age who are doing better, but, after 50 they wisely learn to accept the end of their dreams.  OTOH, this group has done much better than their dreams; they are exceptional weirdos on the upside, thus not models for the average wannabe reader.  I note this holds for Tyler Cowen as well, who "personally vouches" for how great it is to be over 50, but like these now renowned happiness researchers in the book, he has done exceptionally well with his career since 50.  His personal vouchsafing is of limited use to the more average possible reader of the book.

*Second Addendum: At one point Rauch discusses a carefully done and econometrically sophisticated longitudinal study of South Africa by Powdthavee.  This one does find a low point in middle age for the adjusted estimate.  Of course, South Africa is not a typical sub-Saharan African nation.

Barkley Rosser

No comments: