Monday 30 April 2012

Can't kill a bad stat

I thought I'd put a stake through the Ministry of Health's estimate that smoking costs the health system $1.9b per year. But, the Dominion Post cites it again this weekend.
Of course, reducing tobacco sales will reduce the $1 billion a year the Government gets from taxing it, but that needs to be weighed against the nearly $2b a year smoking-related illnesses cost the health sector and the untold cost of lost working hours and productivity.
The Ministry of Health figure seemed to assume that smokers, if they never smoked, would never impose end-of-life costs on the health care system; their figure counted the brought-forward end-of-life costs as a cost of smoking but didn't net from the figure the end-of-life costs that otherwise would have obtained a decade or so later.

MoH has been pretty quiet about that number lately. I can't find reference to it on the site - my previous links to their documents that had it are now deprecated; website restructurings build memory holes. They didn't cite it in their documents listing what would be necessary to achieve the Smokefree 2025 goal; we might have expected to see it in the preamble listing the harms done by smoking.

Where can I still find the figure, or something close to it? The RIS on last year's excise changes:
A 2007 estimate put the cost of smoking to the health system at $300 to $350 million per annum; however current work within the Ministry of Health suggests that figure may be as high as $1 to $1.6 billion per annum [2].
2. Please note that this analysis is work in progress and methodological issues are currently being addressed.
I'll point again to O'Dea's estimate: smoking costs MoH about $350 million. And the Dom is lazy and wrong when they say the costs of lost productivity are "untold". They're very extensively tabulated in the O'Dea study. His "social costs" figure has gotten a reasonable amount of attention in the press. Here's his breakdown, my editorialising:
Here are the components of O'Dea's $1.7 billion figure (Table B.1, p.44); you judge for yourself whether the Herald's right to call these costs on the public:
  • Reduced production from mortality: $570m (I call private)
  • Reduced production from morbidity: $280m (I call private)
  • Resources diverted for tobacco consumption: $650m (I call batsh*t insane to consider this public: it's what smokers spend on their cigarettes net of excise taxes)
  • Resources required to treat induced diseases and other consequences: $350m (public external transfer cost. This is the real cost to the health system)
  • Smoking-induced fires: $15m (largely private, barely worth arguing about as such a small part of the overall figure)
So more than a third of the $1.685 billion is smokers' spending on cigarettes and only $350m are real external costs through the health system. [Note: O'Dea nets from the $1.865 cost $180m in presumed benefits to smokers of smoking.]
So the "untold" cost are maybe around $850m, largely borne by smokers through lower earnings. If that's a social cost, deciding to work part time and enjoy more leisure instead of working very hard, or deciding to take vacations, would also impose social costs.

I expect that MoH is a bit stuck. Associate Health Minister Tariana Turia would be angry if they publicly backed away from the figure. I suspect that at least some at MoH don't trust the figure and don't want to have to defend it. So they're just not saying much about it. And that's fine where the number isn't already floating around being cited and building public pressure for policy measures that would enjoy less support if voters didn't think that smokers were imposing net costs on the tax system.

By MoH's last published note, they've retreated from their prior $1.9b estimate to $1-$1.6b: much lower than the Dom's cited number. I'd be curious whether MoH ever completed the work in progress and addressed the methodological issues.

Meanwhile, the South Australian government wants to bring suit against tobacco makers to compensate them for smokers' health care costs.

That's an interesting one. The 2008 Collins & Lapsley report finds very high "social costs" of smoking, mostly consisting of things like smokers' expenditures on tobacco, the presumed valuation of reduced mortality, and deceased smokers' forgone household production. But actual medical costs aren't all that high.

Here's Table 42 of the Collins & Lapsley report.

Even if you include reduced taxes paid by smokers* due to excess mortality and morbidity, tobacco excise dwarfs everything else. Counting GST here is a bit tough; in the absence of smoking, smokers would instead spend their money on a mix of products some of which attract GST and some of which don't. But even ignoring the GST entirely, it seems hard to find much merit in suing the tobacco companies for health care costs incurred by state governments. You could maybe fault the federal government for not kicking back part of the collected excise revenues to defray state-level health care costs, but that's hardly the fault of the tobacco companies.

* I am not endorsing their numbers here; I've not examined them closely. We found problems in their estimates of reduced productivity due to alcohol use, but haven't looked at their similar tobacco estimates.


  1. It needs to be said, and it needs to be said again.

    Chris Auld's post on this a while ago is wonderful:

    "The idea that a healthy lifestyle substantially decreases demand on the health care system has been repeatedly shot, stabbed, and poked at with sharp sticks, but it won't just die."

    1. It is an excellent post. Note that I'd provided the first comment there!

  2. I just read Chris Auld's piece. It is very good, but it did occur to me while reading that if the concern from the public is that "smokers cost me more tax", then the cost to the health system is not the only consideration.

    Chris points out that smokers cost more per year but for less years, while healthy people cost less per year but the extra years eventually see them costing more. However, it may be the case that, by surviving to pay tax for longer, healthy people are more likely to cover the costs of their own health care.

    On the other hand, in countries will national superannuation, those healthy people will also be claiming more in retirement income.

    At any rate, Chris's point is that the gross costs of smoking to the health system is an incomplete way of viewing the problem. I assert that even net costs to the health system is insufficient to determine whether smoking drains the public purse.

    1. I am far happier simply ignoring fiscal externalities than deeming them an appropriate basis for policy. The corrective measures for mitigating them wind up being very intrusive. Consider the costs to the health system of STDs and what regulatory or tax measures might thereby be justified.

  3. Agreed.

    If I didn't agree for other reasons then I would agree simply because it seems profoundly wrong to impose a socialised health system out of which one cannot opt, then regulate behaviour based on costs to that system.