The mental health puzzle, part II: happiness?

Last week, I posed the puzzle of the decline in mental health from around 1950 till now in most Western countries (with some countries showing a plateau since the 90s). I was talking in particular about the increase in depression, anxiety, and obesity.

One of the reactions (by Andrew Norton in particular) was on the important clue that we cannot see the mental health decline in happiness data. Indeed, we can hardly see any trend in longitudinal happiness within rich countries: some small ups and downs around recession times, but a basic flat line since the 1950s.

It gets worse: we dont really see any change in the distribution of happiness either. This is not what one would expect from the data on mental health decline: with perhaps 30% of the population in some form of serious mental health problem in any given year, one would expect 30% of the population to be unhappy, perhaps counter-balanced by a deliriously happy 70%. This is not true either: happiness in rich countries looks pretty much like a Bell-Curve whenever it was measured. No elongating ‘left-tail’ as far as I know.

Yet, at the individual level, those with mental problems are a lot less happy than others. Indeed, health in general is the most important of all demographic variables one normally sees in a happiness regression.

Let us first discuss the possible ‘data’ explanations which make sense of this all, many of which are yet to be empirically tested:

  1. There is something wrong with the happiness data because of selection. We can for instance suspect that the mentally unhealthy are under-represented in happiness data and that, over time, the number of people not in the data has increasingly been made up of people too depressed to answer surveys. The problem with the suspicion is that it is probably the other way round: as far as we know, misery loves company in that people who stay in a panel for a long time tend to answer more miserably. Indeed, the prevalence of mental health problems in general surveys is large.
  2. There is something wrong with the happiness data because people lie about their happiness. For this, we can appeal to the same ‘time in panel’ finding as above: people who answer years in a row tend to answer more miserly as time goes by, perchance because they get comfortable with the surveyor and are thus becoming more honest about their unhappiness. This possibility would of course be quite devastating for the happiness literature as it would cast doubt on the many cross-sectional surveys. One would then still need to find a reason though for why this ‘keeping up happiness appearances’ has gone up over time rather than stay constant.
  3. There is something wrong with the health data because of exaggeration. The clear front-runner along these lines is the hypothesis that everyone with a bit of a problem nowadays gets given a label. That would rationalise why the mentally unhealthy are indeed less happy and why we get a measured increase in mental health problems. The problem with this one is both the clinical literature which purports a true increase in the number of people with serious anxiety and depression (of which there are many variants, of course), as well as the undoubted increase in obesity rates which indicate an increased inability to withstand temptations and keep up healthy food and exercise habits (aka willpower), which in turn might be caused by all kinds of factors (including increased temptation and social norms that have lessened the taboo on particular behaviour).
  4. There is something wrong the health data because of changed expression. Here, one can think of a change in the manifestation of unhappiness and mental health problems: it would not be the case that in previous eras people were mentally healthier, but rather they were more actively hiding their mental health problems and these got channelled in different ways, perhaps more destructive ways. The person who would have been a paranoid bully in previous decades would now simply be at home on prozac, equally mentally unhealthy and unhappy but displaying that unhealthhiness differently. This one is very hard to refute or verify because it relies on the possibility of any particular mental health problem being a mere ‘expression’ of underlying factors. It is not clear that is mechanically possible or plausible.
  5. There is nothing wrong with the health and happiness data, but something else is causing both the increased mental health problems as well as some compensating factor that keeps the level of happiness constant. Stories in this direction include the notion that Western societies have seen an explosion in ‘entitlements’ and ‘optimism’. At one level both entitlements and optimism bring happiness because they reduce uncertainty and give people a warm glow in terms of happy beliefs about themselves. On the other hand, there is a sleight of hand involved in both of them in that cashing in on entitlement comes with a knock to self-esteem (sometimes it even comes with the duty to be unhealthy), and optimism comes with the mental cost of having to self-delude constantly. Both can be psychologically draining and thus lead to the bottom of the distribution succumbing to serious mental health problems more often than before. Variations on this theme include the literature on reduce ‘resilience’, the literature on increased ‘temptation’ (thank you Andreas Ortmann), the literature on reduced ‘connectedness’ (which is, after all, for many people a choice), etc. From an economic point of view, one could term all of these in the form of the availability of more psychological choices early in life, with expected value close to zero but a higher probability of derailing. With more psychological risk-taking comes a larger number of people in problems. The mental health industry then keeps the bottom of the distribution up by means of medication and therapy.

My own inclination is to think in terms of option 5: the cross-country data does appear convincing to me, particular the fact that Japan and Korea have fairly low levels of happiness but also low levels of mental health problems compared to the West. Having visited Japan, I basically believe the data on this: the Japanese are neither unhappy nor mentally unhappy. Their happiness levels are just medium but levels of real problems are low. In the West the distribution in that sense is more extreme, but the bottom is kept up by means of the health industry. This of course does still beg the question where changes in these cultural traits of ‘optimism’, ‘temptation’, ‘resilience’, ‘connectedness’, etc. come from and why you get the distribution over countries in these traits.

12 thoughts on “The mental health puzzle, part II: happiness?

  1. I’m not inclined to believe a whole lot from the cross-country data — not because it’s necessarily uninformative, but because there are so many cross-cultural expectations about how one should be and respond to both these surveys and mental health issues that I find it hard to imagine how one would ever decontaminate this sort of data. It therefore seems great data for self confirmation bias versus using for strong arguments. For example, the fact that you happen to think mental health problems are low in Japan might be because they are, or it might be because no-one is willing to admit to having them. I bet the second.

    A good way to pick this apart would be to get disorders that we think have a pretty decent biological basis and then compare them with the reported incidence of these (made all the more difficult due to the behavior interacting with culture in some instances). Off hand, I can’t think of any definitive paper on this, but I could well be wrong.

    • Hi conrad,

      yes, I agree cross-country stuff is difficult to unpick and that its almost impossible to do so convincingly. There is quite a lot of evidence already that there are national differences in how people interpret health survey questions, for instance.

      Dont count on the ‘biological basis’ either though. Lots of mental health problems are of course claimed to be ‘purely genetic’ or have a ‘direct biological cause’ (the studies on obesity and genetics come to mind), but the certainty they seem to bring also evaporates if you closely: the range of mental health problems people have is incredible and highly idiosyncratic; the expression of those problems is highly cultural-specific (anxiety attacks, actual delusions, and psychoses usually fit the fears and delusions deemed somewhat acceptable in the society of its sufferers); the interaction between environment and genes is of an incredible and ill-understood complexity, etc. Good luck unpicking that to a degree that you end up with convincing cross-country data.

      This brings up a general point though: while I agree that cross-country (or cross-history) stuff is fraught I also find it very appealing data because the variation is so high. So though you are right that self-confirmation bias is very high with that kind of data, the potential for learning is also highest. The prettiest flowers bloom right on the cliff!

  2. I don’t trust old data on mental health. Mental health was much less well understood in past years. Our current understanding of pyschiatric disorders is still very poor.
    I had the same thought as Conrad. If we take bipolar disorder or schizophrenia, which we know have a high genetic component, it is hard to believe the percentage of people suffering these disorders changes greatly over time. Whereas, mental health problems with a much higher environmental component could change much more. I haven’t had a look at the data out there. I suspect there is little that could be relied upon.
    Anecdotally, i’ve heard many times of people with a strong family history of these disorders having many people in the family line who were considered crazy or difficult, but would not have been formally diagnosed and included in the statistics.
    The other issue is pretty low boundaries for diagnosis with a mental disorder these days in western countries. People who are not far from normal can be easily diagnosed with depression or anxiety, which blows out the number of people classified as mentally ill.
    While i have a lot of problems with cross country comparisions, i find it interesting that the rate of bipolar disorder apparentley does not vary much between countries, while i think the rate for depression and anxiety vary much more.

    • This is why obesity is so informative; it is the one thing that is measured reasonably well over a long period of time and cross-country comparisons are easily and reasonably accurately made (with some adjustments for build and musculature). It is clearly mental health related. The difficulty with that one, from a research point of view, is more the strong incentives on the part of both patients and medical practitioners to come up with self-esteem protecting stories about it. The constant disinformation around it makes it harder to figure out the more likely underlying factors.

  3. Hopefully this doesn’t come out twice, but I don’t think obesity is a great measure. The reason for this is that some people would argue that it’s a natural tendency of many animals to eat too much. It’s also something that usually happens in small amounts across a lifespan, so it’s a small effect that leads to slow change. This is clearly different from, say, serious depression, which has no obvious ecological benefit.

    Speaking of depression and data that goes against your hypothesis, if you look at suicide rates, you’ll find they have not increased in anyway that would suggest a depression epidemic. In the US, they have gone down: http://www.infoplease.com/ipa/A0779940.html

    In Australia, the figures go up slightly, but nothing huge (most of which I assume is reporting bias given it is now a more open issue).

    • Hi conrad,

      yes, part I talked about the plateau in suicide rates. The ‘usual’ explanation for that one in the US literature is that we have become better at spotting depressed youngsters and are medicating them such that they no longer commit suicide, ie there is a real effect of medical expenditure to lift the bottom.

  4. Obesity , type 2 diabetes and depression/anxiety are in some way linked by lifestyle. However it is not at all clear how they are linked-what is cause and what is effect.

    For example it is possible that the changes to diet common to most English speaking people may, in itself, be causing physiological changes that are not simply down to the calorific value of the diet. It is possible that changes to diet may be changing gut flora. And that the wrong gut flora can do more than give you an ulcer …. it is possible that they can seriously affect the brain and metabolism generally.
    The worrying thing is that Australia keeps cutting funding for research into this sort of stuff.

    Only an analogy- Toxoplasmosis completes its life cycle in the belly of cats, in autumn mice that are infected by Toxoplasmosis suddenly become attracted to open spaces and light – become easy prey for a cat, this change in mouse brain behavior is caused by Toxoplasmosis .

  5. the cross country data is clearly useless rubbish if people try to use it to say one country is “happier” than another. The happiness data don’t indicate anything of the sort. Japan and Korea don’t have a low level of happiness – they have a lower score. The score for any particular country remains pretty constant over time.

    The lack of an effect from mental health on this relatively stable level of declared happiness is not surprising if you look at the literature – I refer you to Robert Cummins work here, the homestasis of happiness indicators is a recurring theme. If indicators of happiness always tend towards the one national homeostatic mean – which all the data suggest they do – then what matters most for policy is the differential levels of happiness within different segments of the population.

    In national terms, if happiness were to rise by even a tiny amount against the homeostatic level, it would represent a huge welfare gain. Happiness data that indicates that a level of happiness rises or falls as a result of any particular intervention or change in circumstances is however very hard to find. If you can get hold of the papers from a NATSEM conference late last year on the topic, they are worth a look.

    Paul: to correction a misapprehension you promulgated in your summary of the previous postings, can I declare I’m not a bureaucrat or involved in government in any way, have not been for many years. The Taskforce on this topic that I chaired last year, and the forthcoming national society that will continue the work, was a private sector initiative sponsored by Global Access Partners.

    • I don’t agree with Bob Cummins homeostasis argument and much of the literature does not agree with him either so don’t get too invested in that one!
      The literature on comparability accross countries is ancient and vast. Whilst comparability in reported health accross country is clearly dicey, there is not in fact clear evidence that we should dismiss comparability in happiness. Indeed, the consensus for a while was that comparability was approximately ok.

      To dismiss cross-country evidence would be fine if we had truly great other sources of variation. Unfortunately, we dont: comparing the US in one period with the US in another period is much like comparing countries. Even following people over time is fraught with difficulties. So in this field you unfortunately dont have undoubted information to fall back on. We must spin our theories with the data we can lay our hands on or generate. And the nice thing about cross-country data is that these are like whole experiments. Yes, one needs to keep track of a lot of things, but that is really no different at the other levels of data.

  6. Paul another way of looking at this might be per/capita alcohol consumption?
    The funny thing is that South Korea is up there with the former Soviet block, what do you make of that? ….Even odder is that Portugal apparently drinks a lot more than Spain?

    I know a few people who are very overweight, and have associated health and mental problems…. it is not a recipe for happiness. But which comes first is a hard question.

    • yes, the alcohol question is difficult to interpret. We know all kinds of conflicting things about it. People who drink a bit are richer than those who dont drink at all, but those that drink an awful lot are at the bottom. We think a bit of alcohol can be healthy (which is not true for a bit of depression!) and happiness and alcohol use at the national level hardly co-move (probably a positive correlation because the rich can afford more and you measure their consumption better).
      It is not even clear what to think of binge drinking, which in several countries is just the socially accepted way for young people to let go of their inhibitions and copulate. So my take on alcohol is that there is not much information on mental health in the usage data.

      Yes, causality is tricky when it comes to mental health.

      • Purely ‘gut’ feeling but we tend to separate mental and physical health too much… provably a continuing hangover from Descartes type mind body opposition.
        In general the significance of the huge amounts of bacteria that, mostly peacefully, co-habit with us is only just starting to be recognised. Which brings me to a question is there a demographic correlation between over use of anti-biotic drugs and ‘poor gut feelings’? :-)

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Notify me of followup comments via e-mail. You can also subscribe without commenting.