Tuesday, April 27, 2010

Marsden Jacob on alcohol

The Law Commission commissioned Aussie consultants Marsden Jacobs to weigh in on the costs and benefits of alcohol. Marsden Jacobs previously issued this report which cited approvingly the Collins and Lapsley measures of the social harm of alcohol (including the amount that drinkers spend on their own booze as a social harm) and which dismissed reduced alcohol consumption in Australia and other countries post liberalisation by noting that, absent liberalisation, the reduction could have been even greater. But I suppose they're a step up from Brian Easton.

Some initial thoughts on the Marsden Jacob (available here and here) report:
  • The Marsden Jacobs report fundamentally mischaracterises our approach to rationality, following Easton in arguing that we require strong rationality assumptions. We don't. All we need is that the harms of irrational excess consumption on the latter portion of a harmful drinker's consumption, taken on average across all drinkers categorized as harmful, roughly balance the surplus they enjoy from the earlier portions of their consumption. Heck, MJ's graph at p. 18, figure 7, basically replicates the graph we have at the link above, except representing irrationality as shifting the marginal benefits curve inwards rather than the marginal costs curve outwards. Either one works.

    We noted that BERL was too quick to assume irrationality, but nothing we did required strong rationality. If you wanted to restrict the set of harmful drinkers to being only those for whom the total costs of consumption are higher than the total benefits, that would be different. But when the range of harmful drinkers is set so broadly as to include folks who occasionally have three or more pints, we really have to take seriously the consumption benefits that these folks get.
  • MJ then go on to conflate economically irrational consumption with epidemiologically harmful consumption. This is a serious error on their part. It's not crazy to count internalities where there are demonstrated irrationality issues. I don't like it much, and I'd argue against it, but it's not crazy. But taking consumption beyond an epidemiological threshold as prima facie evidence of irrationality is crazy. They move directly from the non-crazy "if people are irrational, they may well consume in excess of where they'd consume if they were rational; if irrational, some apparent consumer surplus needs adjusting downwards" to this:
    The above discussion highlights that, to assess the impact of policy measures, such as an increase in the rate of excise, it is important to know the proportion of total alcohol consumption that can be considered to involve high lifetime or high short-term risk. This is especially important when exploring and quantifying the implications of value judgements that allow for irrationality in consumption decisions on alcohol when intoxicated or over the longer term.
    For MJ, consumption beyond an epidemiological norm is irrational. Interesting value judgement. I wonder whether Palmer had a good idea about their particular values when he hired them.
  • MJ seriously and uncharitably misreads our report at paragraphs 55 & 56. At page 31 of our report, we noted that our measure of external costs also included some external costs that would be taken as pecuniary rather than technological were we to take the Buchanan and Stubblebine approach to delineating the two. All of our numbers provided the figure that included matters Buchanan and Stubblebine would have taken as being pecuniary, like costs to the public health system. We noted that revising things to remove all pecuniary externalities, both positive and negative, would have had little effect on overall results because we'd also then be knocking out tax revenues as being only a transfer. Net costs would be little changed.

    MJ then characterizes us as arguing strongly in favour of the lower "technological externalities only" position and only begrudgingly conceding the possibility that others might consider also the fiscal externalities that Buchanan and Stubblebine labeled pecuniary; they say this is out of line with Treasury and the Business Roundtable's prior work.
    Thus, Crampton and Burgess exclude from consideration lost output, alcohol production costs, costs of crime prevention, health care costs, most road crash costs and excise taxes collected.
    This is a serious mischaracterization of our work. At Table 2, we tallied costs of lost output due to harmful alcohol use: $173 million less consumption resources saved of $299.7 million. We did indeed deem alcohol production costs to be fully internalized: people do buy their own alcohol without subsidy. Costs of crime preventative expenditure we tallied as $24.7 million at Table 3. Section 4.5 lists external health care costs of $254.8 million. We did discount most road crash costs to include only the costs on innocent bystanders: $33.2 million. Finally, we counted collected excise taxes against the total measure of external cost. It's only in the one-paragraph: "Oh, if you want to follow Buchanan and Stubblebine instead, it won't make much difference" that we wipe all of those costs and benefits aside.

    It's not like we made a big secret about this; it's all laid out very plainly in the report, we released a spreadsheet with all supporting calculations, and I've blogged extensively on it. Marsden Jacob and Associates either is incompetent at basic reading comprehension, or deliberately promulgates lies. It would have taken considerably less than a month for us to fisk the BERL report had we been able to dismiss all of those costs out of hand. Rather, we adjusted them so they made some kind of sense, then gave our best guess at the portion that fell externally.
  • MJ's treatment of cost internalisation at paragraph 58 is interesting. They say our approach is flawed for assuming that drinkers internalize costs; they say rather that costs are imposed on family, friends and others.
    They do not bear the full costs because friends, families, partners and governments act to offset these costs. A rational fully informed drinker would recognise and anticipate this support when making his consumption decisions. If the subsidy from the welfare system and/or from the support of other individuals were removed, the individual would make different choices. This is a form of moral hazard since drinkers (even if perfectly informed of the costs and risks) know that they will not bear the full costs. The individual drinker does not count the cost of these subsidies since they are a benefit to him, but they are a cost elsewhere.
    Is there any consumption - cars, movies, anything at all - for which this would not be true? Do we say that the cost of McDonald's food is not internalized because people on welfare would buy less McDonald's food if they didn't get welfare payments? Do we reckon that Corvettes are horrible things because men in their late 40s buy them and impose those costs on their families? This is ludicrous.
  • Next in the list of items headed Marsden Jacob and Associates either is incompetent at basic reading comprehension, or deliberately promulgates lies: paragraph 59, where they claim that our results rely too strongly on assumptions of strong cardioprotective effects of alcohol. Again, this either is a deliberate lie or they're incompetent. Here's what we did say in our report:
    As noted earlier, Corrao et al (2000) provide reasonable evidence of health benefits of alcohol consumption extending well beyond the hazardous threshold used by BERL; no health benefits are counted by BERL. Net health care costs of “harmful” alcohol use will consequently be rather lower than those cited by BERL. However, putting a value on this would require serious research beyond what we here are able to do.
    We specifically said that we think the health costs are an upper bound because of the beneficial effects of alcohol for cardioprotection even on heavy drinkers (note that the benefits for moderate drinkers are left to one side as well): we do not tally them ever in our figures. Nothing in our results rely on strong or even ANY cardioprotective effects at high consumption levels. MJ is starting more and more to look like a deliberate hit piece rather than a serious piece of analysis.

    In Box 3, they go on to argue against the health benefits of moderate drinking, which was entirely beside the point for tabulating costs and benefits of heavy drinking. But, they raise every canard that I blew up here. It's hopeless. It seems as though there's a result that they were paid to find, and they were sure to find it. It does discredit to the Law Commission that they commissioned and endorsed this work.
  • Finally, they chide us for ignoring the work of the Sheffield team in cost simulation. I will happily confess to ignorance about that work. We were fisking a particular report on the costs of alcohol. If we'd been paid anywhere in the range that BERL were paid, or Marsden Jacob (oh, I'll look forward to getting that price by OIA), we might well have canvassed a bit more broadly.
  • Marsden Jacobs argue that tax increases on alcohol can be efficient: they argue that while it's true that moderate drinkers are more price responsive, the Harberger Triangle is small relative to the tax take, and the tax take is better than just a transfer, it's an opportunity to offset less efficient taxes. Seamus previously warned us of the problems in taking an undergraduate understanding of Ramsey taxation to policy: it isn't absolute price elasticity that matters so much as cross-price elasticity with untaxed leisure. In any case, the Law Commission goes on to undermine whatever efficiency case there might be by arguing that at least some of the excise tax take should be earmarked for alcohol harm reduction. Treasury explicitly reminds LC of this in their note preceding the Marsden Jacobs report.
  • MJ seriously downplay the health benefits of moderate drinking and make no attempt to estimate the reduction in those benefits with an increase in excise tax
Skimming ahead (haven't read the last third in any depth as yet), I see calls for scenario modelling of results where half to 80% of alcohol is irrationally consumed; I also see them arguing that the existence of alcohol advertising proves that consumer preferences are too malleable to be the basis for welfare:
the integrity of the concept [consumer surplus] can be questioned for several reasons including that the New Zealand alcohol industry believes it is profitable to spend substantially (more than $30 million a year) on alcohol advertising and promotions to shift consumer preferences;
Wow. Advertising means that consumer surplus is all wrong. Just wow. I need a drink.


  1. Look at it this way EC, the amount of time and effort that the Wellington Wowsers are putting into attacking Burgess and Crampton they are clearly very worried about your results.

  2. Perhaps the Law commision could look at the laws regarding control of public sector procured reports. I think a recommendation of steep excise taxes on the fees associated with their production would naturally fall out.

  3. "the integrity of the concept [consumer surplus] can be questioned for several reasons including that the New Zealand alcohol industry believes it is profitable to spend substantially (more than $30 million a year) on alcohol advertising and promotions to shift consumer preferences"


    The rest was the sort of "to be expected" rubbish. But this last para is insane.

    I wish analysts could get past the idea that "they know what is best". Reading stuff like this indicates to me that the author thinks they are so much smarter than everyone else that they understand other peoples preferences more fully than the individual making the choice.

    Makes me so angry - I think I also need a drink.

  4. What is amusing is that this is no different from what is used much of the time when economists wish to restrict some activity. The mistakes and fallacies exposed here, are very common. The advertisement one is used very commonly by social democrats in the US to attack whatever business that don't happen to like that day; lately its been pharmaceutical research companies.

  5. @Doc If there were a professional licence for economists, pulling this kind of nonsense would have it revoked. Fine to make those critiques with your economist hat off. Or, to make those critiques in econ journals to other economists to try and change what the orthodox technique is. But to use them with economist hat on in policy reports? Revoke the licence.

  6. Agreed with Eric - you don't come up with subjective policy conclusions while still wearing your economics hat.

    It is like keeping your socks on while making love.

  7. @Matt: But what if they're business socks?


As I randomly delete anonymous comments, please use a handle. Comments on older posts go into moderation; sorry if there are delays in hoisting your comment from the moderation pool.