Middle-aged citizens in developed countries are close to their peak earnings, have typically experienced little or no illness, & are in safe countries, but suffer of more sleeping problems, alcohol dependence, concentration difficulties, memory problems, intense job strain, disabling headaches, suicidal feelings, and extreme depression

The Midlife Crisis. Osea Giuntella, Sally McManus, Redzo Mujcic, Andrew J. Oswald, Nattavudh Powdthavee & Ahmed Tohamy. NBER Working Paper 30442, September 2022. DOI 10.3386/w30442

Abstract: This paper documents a longitudinal crisis of midlife among the inhabitants of rich nations. Yet middle-aged citizens in our data sets are close to their peak earnings, have typically experienced little or no illness, reside in some of the safest countries in the world, and live in the most prosperous era in human history. This is paradoxical and troubling. The finding is consistent, however, with the prediction – one little-known to economists – of Elliott Jaques (1965). Our analysis does not rest on elementary cross-sectional analysis. Instead the paper uses panel and through-time data on, in total, approximately 500,000 individuals. It checks that the key results are not due to cohort effects. Nor do we rely on simple life-satisfaction measures. The paper shows that there are approximately quadratic hill-shaped patterns in data on midlife suicide, sleeping problems, alcohol dependence, concentration difficulties, memory problems, intense job strain, disabling headaches, suicidal feelings, and extreme depression. We believe the seriousness of this societal problem has not been grasped by the affluent world’s policy-makers. 

 

CONCLUSIONS

The human midlife crisis seems to be an important and under-recognized phenomenon.
We document longitudinal evidence of extreme distress among middle-aged adults in affluent
countries. These individuals are close to their peak lifetime earnings and in general have
experienced no serious illness. Our findings therefore appear to point to a disturbing paradox
within modern society.

Using eight different measures, an approximate hump-shape in severe distress over the
life cycle emerges in data from industrialized nations such as the UK, Australia, and the USA.
This paper's methods go beyond cross-sectional analyses based on simple measures of
subjective wellbeing (for example, Graham and Pozuelo 2017). As far as we know, our
recurring longitudinal patterns -- they are to be thought of as a collection of complementary
types of evidence -- are not widely known by policymakers.

The late Elliott Jaques (1965) is believed to have coined the term 'midlife crisis' in the
year 1965. He offered anecdotal evidence, and psychoanalytic arguments, for it. Using modern data sets and conventional statistical methods, this paper explores, and provides empirical support consistent with, the hypothesis advanced by Jaques. The paper's analysis finds hill-shaped patterns in data on:

x suicide,
x sleeping problems,
x extreme depression,
x intense job strain,
x disabling headaches,
x suicidal feelings,
x concentration and memory problems,
x alcohol dependence.

In some cases a particular mental-distress marker is available in many nations; in other cases
it is available only for a few nations.

The explanation for the midlife shape currently remains open. Could the paper¶V
empirical result be the product of the stresses of having dependent children, or a country-
specific or new phenomenon, or something to do with selection effects, or an illusion caused
by cohort effects? These are natural and important possibilities. Nevertheless, the balance of
our evidence appears to suggest not. It also does not seem that envy of others causes the midlife shape (Mujcic and Oswald 2018 test for that possibility, although not with extreme distress measures as the dependent variable). The notion of unmet aspirations as part of the explanation does, however, have intuitive appeal, in our judgment (see particularly Schwandt 2016).

Perhaps so also, more speculatively, does some role for rising 'wisdom' seem possible (Jeste and Oswald 2014) in the observed reduction in distress levels later in life.

There is some published evidence for a midlife psychological low in data on
chimpanzees and orangutans (Weiss et al. 2012). So sheer ageing biology in primates may
play some kind of role. That would take the ultimate explanation out of the social sciences and
into the natural sciences. Much is still to be understood.

Scientific caution remains appropriate. The evidence described here is based on a
particular, if large, set of indicators. It is possible to think of objections to those indicators. A
caveat on that, however, should arguably also be entered. It would be incumbent upon a critic
of our chosen extreme-distress measures to suggest what would count instead as a set of better markers of human crisis. Most especially, it would not seem scientifically acceptable to
suggest something like μindicator X is less than perfect so I reject the repeated pattern of these multiple indicators¶.

Finally, we believe it is not currently clear whether:

(i) there is a timeless and innate form of human middle-aged crisis, or
(ii) the midlife pattern documented here is some kind of perplexing, and perhaps temporary, byproduct of today's affluent world.

Whichever of these turns out to be true, the hill-shaped pattern of extreme distress over the
human life-course in rich countries appears to constitute a foundational puzzle for economists,
behavioral scientists, and perhaps other kinds of scientific research.