Alcohol costs grossly exaggerated
June 17th, 2009 at 3:32 pm by David FarrarTwo economists – Eric Crampton and Matt Burgess, has scrutinised a report by BERL, which cost the Government $135,000. The BERL report concluded the annual social costs of alcohol was $4.79 billion (and has been quoted as a reason to tax alcohol more etc), while Crampton and Burges says BERL have exaggerated costs by 30 fold. Crampton blogs:
“What we found shocked us. BERL exaggerated costs by 30 times using a bizarre methodology that you won’t find in any economics textbook,” Dr Crampton said.
“BERL has virtually assumed its answer. The majority of the reported social costs rest on two very strange assumptions which BERL has asserted without any reason or evidence,” said Dr Crampton said.
“The report assumes that one in six New Zealand adults drinks because they are irrational; that is, they are incapable of deciding what is good for themselves. BERL further assumes that these individuals receive absolutely no enjoyment, social or economic benefit from any of their drinking,” Dr Crampton said.
“These assumptions allowed BERL to count as a cost to society everything from the cost of alcohol production to the effect of alcohol on unpaid housework. That’s bad economics.”
Among other serious flaws, Dr Crampton said the report’s external peer review was done by the authors of the report’s own methodology, important findings in academic literature that alcohol had health and economic benefits were ignored, BERL did not properly warn readers about the limitations of its methodology, and used language in the report that was frequently misleading.
And that is just from the press release. The actual report is as savage as I have seen in critiquing an economic work:
This paper reviews BERL’s report, finding it contains serious deficiencies. For reasons of time, we focus exclusively on BERL’s tabulation of the costs of alcohol. Methodological errors account for approximately forty percent of BERL’s listed costs: double-counting of the costs of insurance and the costs of insured losses; counting as costs all of the alcohol consumed by harmful drinkers rather than just the portion harmfully consumed by those drinkers; incorrect use of multipliers; not accounting for cohort differences between serious alcoholics and the rest of the population in labour force characteristics; and, assuming an implausibly large reduction in crime in the absence of alcohol.
And further:
First, for alcohol consumers BERL uses an epidemiological basis to define the threshold for economic harm. This definition is crossed after 1.8 pints of beer and is low enough to catch one New Zealand adult in six. …
Second, BERL assumes all harmful alcohol and drug consumption is irrational. Irrational consumers are incapable of detecting private costs in excess of private benefits. To the extent those private costs exceed benefits, they are counted as social costs. Third, BERL assumes irrational consumers enjoy zero gross (not net) benefits, meaning all private costs are counted as social costs. The second and third assumptions are not justified – they are simply asserted by BERL. The effect of these assumptions on BERL’s cost estimate is profound. An analysis that would otherwise be confined to externalities is instead inflated by private costs.
And even more:
The credibility and independence of BERL’s work is also questionable, further limiting its usefulness. The analysis ignores most of the large body of peer-reviewed economic literature in favour of a few (mostly commissioned) reports by a very small subset of health economists whose reports have been subject in that literature to many of the same criticisms leveled here. BERL’s report can be reasonably characterized as a New Zealand implementation of a methodology developed by Professors Collins and Lapsley, cited over 100 times in the BERL report. These same authors provided the external peer review of the report.
And finally the summary:
It is customary in reviews like this to offer at least some praise, but BERL’s report has few redeeming features. Beneath its professional veneer, BERL’s report fails in multiple dimensions. Its conclusion is assumed. Its core assumptions defy both reason and the body of peer-reviewed literature. Its headline figures are overstated by an order of magnitude. The methodology is without foundation in the economics discipline, and the report has been peer-reviewed by the authors of its flawed methodology. Its literature review is highly selective. The report contains elementary errors and misunderstandings of economics, and policymakers are likely to be misled by the report’s loose terminology and spurious comparisons1
2. The BERL Study . In sum, these flaws render the report of negligible use for subsequent policy-making.
Now that is brutal. And the Government paid $135,000 for this report and the Law Commission has been citing it as a rationale for its advocacy.
I suspect the BERL report is just one of money where only costs are looked at, benefits ignored, and costs inflated to the maximum.
June 17th, 2009 at 3:48 pm
Thank you for that second to last line – I wondered if this was the basis for the absurd rantings of that barking mad old fool Geoffrey Palmer. It was.
Vote:June 17th, 2009 at 4:07 pm
Sanity at last. With “social cost” at $160 million (roughly) and alcohol tax-take at an annual $785 million, could someone tell ALAC and Alcohol Healthwatch and all those doctors who have been writing to the papers to quit bleating about raising taxes to offset the cost of alcohol? Pick another point and argue that, but please stop saying there needs to be a tax hike.
Vote:June 17th, 2009 at 4:12 pm
Pretty pathetic. I love research, think it is vital for a healthy, rational, functioning society. Bad research, though, is worse that simply a waste of time; it can be a vehicle for fools who have made up their minds already to try and skew the numbers to superficially back their blinkered outlook, thus feeding the public lies with the comforting veneer of numbers and percentages.
Vote:June 17th, 2009 at 4:16 pm
Clearly alcohol was involved in the writing of the report.
Vote:June 17th, 2009 at 4:16 pm
@bearhunter: Our net measure includes the excise tax take (2005/6 of $516 million), so please don’t weigh it against the excise tax take. We find that there remain net external costs. However, such a finding is wholly inadequate for arguing that taxes should be increased: any tax increase is very likely to impose greater harm on moderate drinkers than the benefits to harmful drinkers. A thorough cost-benefit analysis would be required to justify any tax increase. Such an analysis has been conducted by nobody.
Vote:June 17th, 2009 at 4:21 pm
This is amazing stuff. It seems to me that BERL must now either come out swinging, or else disappear into the woodwork and attempt to retrieve their reputation some other way. Having read none of the reports, I hope for their sake that the methods used by Crampton & Burgess stand up to scrutiny.
Vote:June 17th, 2009 at 4:23 pm
@transmogrifier: I hope that that one wasn’t directed at Matt and I. All of our workings are available online: I’ve released the excel sheet that has all of our calculations.
Vote:June 17th, 2009 at 4:26 pm
@ Crampton: Ah, gotcha. I’ll stop foaming now.
Vote:June 17th, 2009 at 4:28 pm
I’m sure the “Hospitality Industry” will be jolly chuffed about your report Crampton,
Vote:June 17th, 2009 at 4:30 pm
I wonder where is this “University of Victoria at Wellington” with which the second author is associated. Anywhere near Victoria University of Wellington, do you think?
Vote:June 17th, 2009 at 4:30 pm
Christ, don’t tell me you’ve got a hard-on for the hospo industry as well, Sonic. What about all them flat-capped men enjoying a well-earned pint after a hard day’s work down t’mines? Or have Labour completely ditched the working classes?
Vote:June 17th, 2009 at 4:30 pm
@Crampton. No, at the original report.
Vote:June 17th, 2009 at 4:34 pm
Sonic’s got a cat, and a whippet! Lad.
Vote:June 17th, 2009 at 4:34 pm
As one of their best customers I certainly do not have anything against those good people who sell our main dangerous drug bearhunter.
Vote:June 17th, 2009 at 4:37 pm
Alcohol abuse is the leading factor in about 80%+ of all crime. It leads in assaults, it leads in burglaries it leads in traffic accidents and deaths. If it were possible to make it illegal I would. It would probably be classified as a class B drug. There is a significant health factor in alcohol abuse as well including- liver/kidney damage, brain damage and injury. I do not know the costs of all this but in terms of drug abuse causing costs and harm to the community – alcohol abuse would lead.
Vote:June 17th, 2009 at 4:37 pm
“Alcohol abuse is the leading factor in about 80%+ of all crime.”
Quote your sources.
Vote:June 17th, 2009 at 4:39 pm
@sonic: I haven’t a clue. This isn’t a commissioned report. I just don’t like it when policy is founded on reports inadequate to that purpose. My salary as an academic economist is utterly invariant to whatever position I might take on the costs or benefits of alcohol.
It might well be the case that a thorough cost-benefit analysis would support an excise tax increase. I’d be a little surprised given the likely effects on benefits to moderate drinkers (whose consumption is more elastic than harmful drinkers), but it’s possible. The point is that nobody’s produced anything approaching a cost-benefit analysis, and a cost-benefit analysis would be needed for saying that increasing the tax would do more good than harm.
Vote:June 17th, 2009 at 4:44 pm
Thats fair enough Crampton, I’ll go off and read it, I’d say it was virtually impossible to do a rigorous cost/benefit analysis on this subject, I’d certainly not want to try. How would you factor in the good results of alcohol for example?
Vote:June 17th, 2009 at 4:47 pm
@tvb: the BERL report attributes to alcohol any crime in which the offender, on survey, says that alcohol contributed at least ‘somewhat’ to his offending (on a scale that runs “not at all”, “a little”, “somewhat”, “a lot”, “completely”). Is it really realistic that all crimes in those latter three categories would have been eliminated if alcohol ceased to exist?
Vote:June 17th, 2009 at 4:51 pm
@sonic: I agree that nothing in this area can be done precisely. But BERL ignores benefits entirely and that can’t be right. Moreover, they zero out any possible health benefits from alcohol even for moderate drinkers. That last bit we haven’t even attempted to correct for: it would be a big job and we haven’t the time. To that extent, we overestimate costs and underestimate benefits. Running to get Ira from daycare now but I’ll be checking in later this evening after he’s gone to bed.
Vote:June 17th, 2009 at 4:58 pm
There are beneficial aspects to alcohol and crime as well.
What about the criminals who get so comered out on alcohol that they are unable to go out and commit crime?
Vote:June 17th, 2009 at 5:28 pm
@bharmer. The methods Eric and Matt use are standard stuff. Its the methods used by BERL that I find odd. I don’t think you will that you will find their methods in any economics textbook.
Vote:June 17th, 2009 at 5:31 pm
“Only costs are looked at, benefits ignored, and costs inflated to the maximum” sounds like all those reports on the social cost of illegal drugs as well.
Vote:June 17th, 2009 at 5:41 pm
Lyndon: yes. Many of our concerns with BERL’s alcohol study apply to their report to the Police last year on drug harms which used a similar methodology.
Vote:June 17th, 2009 at 5:51 pm
# Paul Walker (19) Vote: Add rating 0 Subtract rating 0 Says:
June 17th, 2009 at 5:28 pm
@bharmer. The methods Eric and Matt use are standard stuff. Its the methods used by BERL that I find odd. I don’t think you will that you will find their methods in any economics textbook.
Happily take your word for it Paul. I guess my point was that a report that attacks another report must surely stand up to the same scrutiny that the first is subjected to.
Vote:June 17th, 2009 at 6:11 pm
@bharmer Happily take your word for it Paul. I guess my point was that a report that attacks another report must surely stand up to the same scrutiny that the first is subjected to.
Of course, but having read Eric and Matt’s report a couple of times now, I don’t have many worries on that front. I will be interested to see how BERL respond.
Vote:June 17th, 2009 at 6:14 pm
@bharmer, you’re exactly right, which is why we’ve released the spreadsheet that has calculations. BERL wouldn’t give us access to their calculations so Eric had to use the text description in the main report only. It’s possible and perhaps likely that some revision will be necessary. Hard to see the headline conclusions changing too much though, unless we and all of our reviewers have missed something extraordinary.
Eric is doing a series of posts on his blog that set out in more detail what we did. It’ll spare you having to read the full article.
Vote:June 17th, 2009 at 6:23 pm
People are not all completely dumb – especially crims. They will always say that alcohol is a factor in their offending. They will say anythign to reduce the blame for their actions from their own behaviour to some external source that “makes them do it”. I know plenty of people who drink way too much who are not criminals. Is their alcohol different to that consumed by people who criminally offend?
Vote:June 17th, 2009 at 6:42 pm
whoops, sorry. I already posted so have edited out duplicate.
Vote:June 17th, 2009 at 7:05 pm
Brian Smaller
People are not all completely dumb – especially crims. They will always say that alcohol is a factor in their offending. They will say anythign to reduce the blame for their actions from their own behaviour to some external source that “makes them do it”.
To be fair to BERL they anticipate your criticism and at fn 51 they write “While the attribution is based on self-report, Hales and Manser (2007) note that there is a high degree of correspondence between self-reported drug use and positive urinalysis”
Our concern is not so much that self reporting is unreliable but that some crimes may still have occurred anyway even without alcohol. BERL rather unconservatively assumes none of the crimes in which alcohol was a factor would have occurred absent alcohol, thus maximising the costs attributable to alcohol.
Vote:June 17th, 2009 at 7:10 pm
I frequently see experts from BERL being interviewed on TV relating to business/economic and I have always wondered how they do their forecasting. They’re usually being asked to give projection about the next few n-th time-steps ahead (ie, next 6 months, next 2 years, etc,…) and they seem to be very confident in their answers. It would be good if they show the methods of their forecast or even include their codes so that they can be scrutinized when they go wrong (big margin). Economists frequently made their codes available (either request directly or downloadable from their site) when they make claims in their papers. The reason they make their papers/codes available to others is that other researchers can scrutinize their claim. I don’t know if BERL frequently make their codes/models available from their site as many businesses/financial market traders just simply follow the economic projections that they make out in public with or without any close scrutiny.
Vote:June 17th, 2009 at 7:37 pm
” BERL rather unconservatively assumes none of the crimes in which alcohol was a factor would have occurred absent alcohol,”
But how would you even begin to quantify that, if you are suggesting replacing that assumption one has to ask with what, a guess?
It’s fine to pick holes in BERL’s methodology, but not so fine to present no alternative.
Vote:June 17th, 2009 at 7:48 pm
Sorry reading more I also have an issue with this
” BERL assumes all harmful alcohol and drug consumption is irrational. Irrational consumers are incapable of detecting private costs in excess of private benefits.”
I’d argue that committing “harm” to yourself is almost by definition irrational.
Also
“counting as costs all of the alcohol consumed by harmful drinkers rather than just the portion harmfully consumed by those drinkers;”
Again I’d stress Harmful, are you aiming to say that the first 10 beers an offender had were fine, we should only count the 11th (just before he burned his house down) as a “cost”?
Vote:June 17th, 2009 at 8:04 pm
@sonic
I think sections 3.3 and 4.2.8 and 4.4 are the sections you want to read.
Vote:June 17th, 2009 at 8:07 pm
Does BERL have the figures for Chardonnay sales, I bet they have taking a dive from November 08. Not much tax out of the bloody socialists now.
Vote:June 17th, 2009 at 8:14 pm
Thanks Paul but I still wrestle with
“we do not attempt to estimate consumer surplus enjoyed by harmful drinkers whose consumption
nevertheless leaves them better off in utility terms.”
What is the utility of harmful drinking? I cannot think of any
Vote:June 17th, 2009 at 8:30 pm
@sonic What is the utility of harmful drinking?
Basically the same as non-harmful. I’m guessing but the fact that I have a glass or two of wine at night with my meal puts me close to being in the harmful group according to BERL. Why do I drink the wine, because I enjoy it. Remember the utility in this sense just means the satisfaction, or “jolly” as a professor of mine put it, that you get from doing what you do. So if you enjoy drinking then you get utility from doing that.
Vote:June 17th, 2009 at 8:47 pm
@sonic: let me help a bit here. They define harmful drinking as any drinking that they define as crossing one epidemiological threshold for harm: 40 grams per day for men, 20 for women. That’s two pints for men, one glass of wine for women. Does it really seem plausible that somebody who has three pints (or for that matter 2 pints and a sip) a day gets zero enjoyment from any of it?
As for presenting an alternative: please read the report and the accompanying spreadsheet, or follow my updates over on Offsetting Behaviour. On crime, we don’t have BERL’s underlying survey data so we don’t know what proportion of criminals or, equally importantly, the breakdown across types of crime for each type of survey response. All caveats about survey data aside (which also are important), it seems impossible that all crimes that were “somewhat” affected by alcohol would disappear absent alcohol. Similarly, at least some crimes in the higher categories would still have occurred even if there were no alcohol, although the reduction there would be much smaller. What did we do? We assumed that a third of crimes would still have occurred even if alcohol had disappeared. I think that’s a small reduction. There are other areas in the crime category that seem similarly odd. Would folks really spend a whole lot less on locks for their houses if there weren’t so many drunks around? Hmmm.
Vote:June 17th, 2009 at 9:04 pm
tvb – WTF ? Who drinks and burgles ???
Vote:June 17th, 2009 at 9:37 pm
@sonic: I can add something here. BERL is not just saying that for harmful drinkers benefits are outweighed by costs. As you say, that much is obvious. BERL goes considerably further and says harmful drinkers get nothing in the way of enjoyment, economic or health benefits. And not just from the 40th gram of alcohol onwards – from the first sip.
In BERL’s world, its just pure cost for harmful drinkers, which one NZ adult in six according to BERL. This is the main reason their cost numbers are so enormous – they’ve assumed away nearly all offsetting benefits.
Its one thing to say that costs exceed any enjoyment you get from an activity. Its quite another to say there is no enjoyment at all.
Vote:June 17th, 2009 at 9:56 pm
People get drunk they lose their judgment they break into a house. It happens a lot. Domestic violence alcohol is the leading contributing factor and violence generally. People simply do not know how the other half live especially those that abuse alcohol. I would just love to ban the stuff.
Vote:June 17th, 2009 at 10:55 pm
@tvb: we do not dispute that that happens. We do dispute that we could eliminate 100% of crimes where the offender says alcohol contributed “some” to his offending by eliminating alcohol.
Vote:June 17th, 2009 at 11:51 pm
“@sonic What is the utility of harmful drinking?
@Crampton
” and says harmful drinkers get nothing in the way of enjoyment, economic or health benefits. And not just from the 40th gram of alcohol onwards – from the first sip.”
Again you fail to demonstrate any utility for drinking by the harmful drinker (by which I would mean those that had a problem with alcohol that had a seriously detrimental effect on their health or wellbeing) “the first 11 pints of beer had no influence, we must only count the 12th?
I’m sure anyone who dealt seriously with problem drinkers could put you straight on that. the first drink leads to the rest.
BERL might not be perfect, but on first reading neither is your report.
Vote:June 18th, 2009 at 5:43 am
Sonic, section 5.2 of our review cites evidence for economic and health benefits that persist into the range that BERL defines as harmful.
Vote:June 18th, 2009 at 7:24 am
Brilliant thought process here. Move over Stephen Hawking. Why don’t we just ban crime. That will work. I know plenty of alcoholics. Have one or two in my own extended whanau, and one or two more to whom alcohol is a wee bit too much of a comforter. Doesn’t mean that I can’t drink responsibly (or not) when I choose to.
Vote:June 18th, 2009 at 11:01 am
@sonic: do you really seriously believe that somebody drinking 2 pints of beer per day doesn’t enjoy any of it?
Vote:June 18th, 2009 at 11:46 am
@sonic: Again you fail to demonstrate any utility for drinking by the harmful drinker
What are you trying to say? That anyone in the range that BERL considers harmful drinkers gets zero utility from all the drinking they do or that utility net of costs are zero for harmful drinkers?
Vote:June 18th, 2009 at 12:04 pm
Sonic, get a grip. What these guys are saying is that BERL have assumed that:
1. Anyone who drinks more than 2 pints (for a man) or 1 glass of wine (for a woman) is a harmful drinker
2. Further, that person, if they drink 3 pints in a night, is getting no pleasure, and it is all cost.
That is clearly an inane assumption. I would frequently drink 3 pints in a night (not frequently as in 3 times a week, but certainly frequently as in a couple of times a month). The missus and I will also frequently drink a bottle of wine over dinner – which is 2.5 glasses for me, and 1.5 glasses for her. Both of us are therefore harmful drinkers, and we get no pleasure out of that. I’m clearly irrational, because I spend $30 on that bottle, and get no pleasure.
It is just flat wrong. If they’d made another assumption we could dispute whether it was accurate – but the assumption they’ve made has no dispute – it’s wrong.
Same on crime. If they’d assumed 1/3 reduction, 2/3 reduction and then defended that, it is at least arguable. All alcohol influenced crime disappearing is clearly flat out wrong.
What Crampton and Burgess are saying is that these flawed assumptions actually drive the entire report. They aren’t a couple of small assumptions buried in the detail – if you make different assumptions the result is totally different. BERL should at a minimum have done a sensitivity assessment on those assumptions.
I can understand it is easier to make those assumptions than work out the real number, particularly when you’re working within a reasonably tight dollar budget. But it is unprofessional to not at least work out whether those poorly based assumptions are driving your entire result.
Vote:June 18th, 2009 at 1:02 pm
PaulL: In fairness to BERL, they do do a lot of sensitivity analyses; just not on the bits that particularly matter. In further fairness, it’s two pints per night on average. You have to be consuming 2 pints every night to hit harmful use. We still think that rather a weak standard.
Vote:June 20th, 2009 at 5:45 pm
I think we should have a whip round and commission BERL for more research:
*Cost of fags
*Cost of Government
Would be a hoot.
Vote: