Yesterday a paper I co-authored was published in the British Medical Journal. The paper, available free of charge at the BMJ website, analyzes mortality among Japanese working age males between 1980 and 2005 and estimates the changes that occurred after the collapse of the bubble economy. Our main findings were that a previously existing inequality in health between professional/managerial workers and the remainder of the population was reversed in the 10 years after the economic collapse. This reversal happened not because the health of non-professionals improved, but because professional and managerial workers saw a rapid increase in mortality.
Before 1990 there was a fairly clear pattern amongst the main causes of mortality in Japanese men: the managerial and professional occupations had lower mortality rates. Mortality rates for all groups were largely declining over time, and at roughly the same rate, but managerial and professional occupations on the whole had lower mortality rates. However, after the collapse mortality rates in these two groups suddenly began to increase, while those amongst the non-professional categories largely maintained their previous trajectory. These trajectories and the changes can be seen easily in Figure 1 of the paper, and the changes that occurred at the time of the collapse are summarized in Table 4. For example, before 1995 the relative risk of all cause mortality in managers/professionals was 0.70 (i.e. 70% of that in the other occupations). After 1995 it was 1.18, about 20% higher (and this difference was statistically significant). Table 4 shows that while before 1995 managers/professionals had lower mortality across almost all the major causes of mortality, after 1995 this relationship disappeared or was reversed.
As an aside, the paper also shows that massive increases in suicide rates in all professions coincided with the economic collapse of the late 80s/early 90s.
There is a possibility that so-called “numerator-denominator bias” might have affected the results: if people registered their employment status differently on their death certificate (the numerator) to the population census (the denominator), we might over-estimate the effect of the stagnation in those occupation groups (like managers) that shrank fastest at that time. This effect might be possible if, for example, after the economic collapse managers and professionals moved into other professions or became unempoyed, but after they died their family recorded their profession on their birth certificate as that which occupied the majority of their career. However we checked carefully for this and confirmed that even the most extreme possible effect of numerator-denominator bias doesn’t change the essence of the results, only the magnitude.
It’s dangerous to ascribe reasons and causal relationships to these kinds of phenomena, but the strong implication is that there is a relationship between the economic aftermath of the collapse and this reversal in health inequality in Japan. We postulate that this might be due to the rapid shrinking of the size of the managerial/professional workforce and changes in its working conditions that did not affect the labour/service industries as much. Other possibilities include changes in insurance status and access to healthcare, or perhaps some kind of health-system effect on cancer survival. It’s probably not due to unemployment: unemployment is categorized separately in the labour market statistics and death certificates, so theoretically a person who is sacked in 1985 and dies in 1990 should be counted as an unemployed person, and since we checked for numerator-denominator bias we think we ruled this out.
Japan has very different patterns of mortality to other developed nations, but this paper gives us an indication of the possible large effects that an economic downturn and subsequent stagnation can have on population health. It also shows that an economic downturn doesn’t necessarily affect everyone equally, and doesn’t necessarily affect the poor, or non-professional occupations, more than it does the rich. I guess the results of this paper and its lessons about the role downturn and stagnation can play in health may be applicable to countries like the UK and USA, which are just beginning to experience what Japan did in the 1990s. This paper suggests that we should expect significant effects of the downturn on health, but that we shouldn’t assume it will hit the poorest hardest, and should be aware that every nation’s post-depression experience may follow a unique trajectory. It also tells us that significant health gains made over a long period of time, such as are seen in this data, can be reversed rapidly after a major economic downturn, and economic collapse can undo 20 or 30 years of health gains. In health terms, major economic events are certainly not to be sniffed at!
Compromise and Conceit 
March 8, 2012 at 2:50 pm
Interesting. Based on the post (without reading the article) it highlights that decreases in health inequality can be a bad thing if the decrease is purely a negative for a previously advantaged group.
An interesting follow up would be a cost benefit that compared government assistance to the previously well off to maintain their pre-bust health levels versus government assistance to try to improve the health of people who were worse off during the boom or the general populace.
Depending on the causes of the decline, middle class welfare may have a larger positive impact on the overall populace health than an attempted general improvement. The reason would be that the professionals/managers experiencing the decline were more conditioned to achieve better outcomes – in other terms that it’s a maintenance versus an improvement issue. Maintenance should show a better cost effectiveness than improvement (all else being equal).
It’s also interesting to see what this suggests for the removal of the 30% rebate on private health insurance in Australia. If a move in the southern states to a post-boom footing is decreasing health inequality by successfully killing off the professionals/managers than the decrease in rebate is likely to heighten that effect – assuming that a similar policy change in Japan was not the root cause[1].
Would you like to write to the Health Minister and let him know the middle/upper class will disproportionately suffer? 🙂 I suspect that the Treasurers response would be “Good”.
[1] If it was the root cause, then it’d just cause it in Oz, not heighten it.
March 8, 2012 at 4:43 pm
Interesting. Maybe something to do with loss of status among former professionals? That would accord with evidence that higher status people have better health than lower status ones (see eg Wilkinson, The Spirit Level and Mind the Gap).
March 8, 2012 at 5:29 pm
Congrats on the BMJ paper!
March 8, 2012 at 8:42 pm
Thanks Nick, it’s my second publication there but my first about Japan so I’m quite happy with it. Still not first author though!
Peter T, I don’t think it relates to status – this research shows the mortality rates increased amongst those who remained within their professional career path, where presumably their status was unchanged.
Paul, I think it’s axiomatic that achieving inequality through damaging one group’s health is bad. No one is trumpeting this as a victory in the war against inequality, obviously, even though superficially Japanese society now has less health inequality than it did before. What is interesting is that this can happen at all as a consequence of a generalized economic event. The usual view of health inequalities is that the wealthier groups are protected from the effects of economic crisis on health outcomes (except in the most extreme circumstances, obviously). Not so in Japan. This could reflect a great many things about how Japan’s society is organized, or could be due to its industrial makeup – maybe the blue collar jobs weren’t affected because in a manufacturing economy working at high efficiency it’s much easier to squeeze middle managers than workers? Who knows. Our data was too general to draw conclusions about the mechanism of the inequality reversal.
I can’t imagine means-testing the 30% rebate on private health insurance will be quite the same severity as an economic collapse. But a message that all governments should consider is that the wealthier parts of their society aren’t necessarily immune to the effects of economic crisis, even when they remain wealthier (as likely happened with the managerial/professional class in Japan).
March 9, 2012 at 11:50 am
The reason I highlighted decrease in health inequality as a bad thing in this instance is that the basic description I’ve heard of books such as the Spirit Level is “More equitable societies have more equitable health outcomes” [1]. Now your Japanese health data potentially displays data that support this idea [2] but the take home message with regard to that is “Equally bad is not a good thing. If we have alternatives of the rich being better off and everyone being worse off them equality is a Bad Thing (TM)”.
I didn’t think you were suggesting anything to the contrary of this [3], but it was one point that jumped out at me that wasn’t already clearly spelt out in your summary. I’d prefer to dwell on alternatives and discussion rather than just agree – those are boring posts.
The reduction of the 30% rebate wouldn’t qualify as economic collapse on it’s own, but could heighten the effect of the current economic situation already in place in NSW and Vic. Australia overall is OK, but those two states aren’t in great shape – Vic especially. Your point that ” a message that all governments should consider is that the wealthier parts of their society aren’t necessarily immune to the effects of economic crisis, even when they remain wealthier” is a fair summary of that.
[1] Disclaimer: I’ve gathered this through newspaper articles, so there’s a decent chance this is actually considered in the book but totally escaped the journalist. I would note that it’d probably also escape any politician trying to implement it.
[2] Though I don’t think you’ve got quite the right data for me to say that as it seems you’re talking status rather than material wealth.
[3] Unless there was a hidden bit of the article saying “Class war rocks! Go Wayne Swan!”
March 9, 2012 at 12:12 pm
The focus on inequality usually assumes that rich people are at the best health they can be, and poor people lagging behind, but it doesn’t have to be that way. My other publication in the BMJ found that older people got worse referrals than younger people, even though in general older people are wealthier. Inequality can have many dimensions. I think the idea of the Spirit Level and other books isn’t contradicted by our research: the results in Japan in the 90s could have been worse if it were a less equitable society (e.g. everyone’s mortality increased, but poor people’s increased more; or everyone’s mortality increased but because poor people already had higher mortality, they remained worse-off but at a higher level). I guess that most people would guess the latter event (everyone’s mortality increases but the poor remain worst off) would be the most natural outcome of economic collapse, but that hasn’t happened here, which is interesting.
I don’t think the 30% rebate will have any effect at all, though I probably shouldn’t place any bets. Private health insurance in Australia is a discretionary good and it doesn’t affect some key basics: emergency surgery, general practice and drug subsidies are all independent of private insurance. Also, there isn’t much evidence that the rebate made much difference to private insurance take-up, and so means testing it probably won’t affect consumption decisions unduly.
It is definitely teh case though that governments and health planners shouldn’t assume the health of the rich is 100% safe during economic downturns; and it suggests that there is a complex interaction between labour markets and health which goes beyond one-dimensional ideas that only the people in the worst parts of the labour market suffer changes in their health from changes in the labour market.
March 9, 2012 at 3:41 pm
“I guess that most people would guess the latter event (everyone’s mortality increases but the poor remain worst off) would be the most natural outcome of economic collapse, but that hasn’t happened here, which is interesting.”
Agreed. That’s why I wanted to highlight it by contrasting it to the more normal point made that the poorest in society should be supported during times of economic upheaval (or even normal operations). The results are counter-intuitive when examined at the highest level [1]. And those are the most interesting sort of results.
It does contradict the (simplistic understand I have of the) Spirit Level to the effect that it shows that health inequality is not necessarily something to be strived for. Health inequality is acceptable when the alternative is worse. That premise contradicts the proposal that health outcomes are best when inequality is lowest (as it’s asserting an argument, however marginal) that inequality is (sometimes) good.
Generally speaking, it’s not enough to disprove the Spirit Level, but I never said it was. I said it was interesting.
On another level, I’m tempted to argue that all results are marginal/edge cases (purely because there tend to be so many of them) and that grand visions can get stuffed. But that may be my general cynicism towards grand ideas speaking. Or the beer.
“Also, there isn’t much evidence that the rebate made much difference to private insurance take-up, and so means testing it probably won’t affect consumption decisions unduly.”
Partially true. We can assume it’s a discretionary good that people will prioritise above other discretionary goods, which means the rebate removal won’t impact take up rates. But the impact needs to either be felt on consumption from those previously receiving the rebate or on their savings.
That means that either a) sales (probably of luxury goods given the income bracket we’re talking about) or b) savings/superannuation of the well off are affected. In the event it’s a) then you’re squeezing your already pressured retail sector. If it’s b) then ether moving rich people onto pensions (unlikely at the income being affected) or you’re deferring the impact on the retail sector. I’d bet on a combination of the two where the retail sector is affected in both the short and long term.
Before you start mindlessly defending the Labour government again [2] this isn’t a criticism of the choice or affect I’m describing. It’s simply describing that when the government makes a decision to shift money around money is shifted around and they need to deal with the consequences. It’s like Exalted or Compromise and Conceit at their best in that fashion.
” suggests that there is a complex interaction between labour markets and health which goes beyond one-dimensional ideas that only the people in the worst parts of the labour market suffer changes in their health from changes in the labour market.”
This is both insightful and interesting for its even-handedness. I’m sure that even conservatives can get behind the proposal that this is sometimes acceptable (probably even libertarians, though Objectivists may object) when the net health benefit is significant enough.
[1] Like physics results, that isn’t necessarily true when examined at the micro level, but that’s not how most people think about thinks. Who the hell considers every individual outcome then sums the result for consideration? [3]
[2] Because that is incredibly annoying.
[3] Freaks. That’s who.