From the Hand: BERL on Crampton-Burgess on BERL
Following the incessant blather on the BERL report into the cost of alcohol Ganesh Nana has come out to defend the report (Hat Tip this guy).
Why bother? After all we all know that any report on costs will be filled with wild conjecture and pointlessly misrepresented by the public.
Well it turns out the Dr Nana felt the report has been too heavily misrepresented and wanted to clear some things up, so I guess that’s fine.
Except that there are two significant ways that his reply leaves me wanting:
- His criticism of rational choice is nonsensical,
- He refuses to admit that BERLs own work was the result of wild value judgments in the same way as the Crampton-Burgess reply.
On the first issue he states “We would suggest that it is nonsense to argue that a drunk driver who wraps themselves around a power pole has made a fully informed, rational choice that is consistent with their long-term welfare”.
However, this is ridiculous. Crashing your car is a suckful outcome – agreed. But the drunk driver doesn’t make the choice to crash their car – they make the choice to drive their car drunk given some probability of crashing. Ex-ante the choice to drive (following the choice to drink) may have been optimal, even if the ex-post outcome sucked.
This suggests that saying that it is nonsense illustrates either a lack of understanding regarding the 1st year difference between ex-ante and ex-post or that the author simply tried to frame an issue to make other people agree with them. Given Nana is an economist, and therefore must be extremely intelligent, it must be because of the second reason.
On the second issue he states “It is for readers to make their own judgments about the benefits of harmful alcohol use. It is for policy makers as well as the New Zealand society, not BERL, to judge what set of values they use in this assessment.”.
Now this seems true. Matt and Rauparaha loves to bang on about this rubbish all the time. But this isn’t what the BERL report does.
Doesn’t the BERL report actually work out some costs, and it requires value judgments to figure out these costs right.
So instead of banging on about how “such a value judgment would not have been appropriate for an independent study such as ours” Dr Nana should admit they made a different set of value judgments and defend them.
I mean, for gods sake, if the two sides involved spent more time clarifying the differences in their value judgments and less time scoring points in the media (or delivering “smack downs” as Dr Crampton eloquently describes it), then we might have a clearer idea of what the policy relevant issues are.
“His criticism of rational choice is nonsensical”
Arg, I wouldn’t go that far.
I would say that BERL is presuming some failure of standard “rational choice”. Expectations/beliefs may be out of whack, time-inconsistency might be at play, or we might believe that people are lemmings (my least preferred interpretation).
When writing a newspaper article the writer can’t go into detail regarding the exact mechanism for the presumed failure. However, they can point out the failure exists – which is what Dr Nana did.
“Matt and Rauparaha loves to bang on about this rubbish all the time”
Link, link, link:
http://www.tvhe.co.nz/2008/11/20/frog-next-attack-on-economics/
Value judgments allow us to make conclusions. In theory a general framework might not require value judgments. So if BERL was only framing the issue rather than costing it they could claim to be (relatively) value free.
Overall, I agree that a better article would have been one that pitted value judgments against value judgments sure. But you seem to have implicitly weighed up these value judgments in your language by calling peoples assumptions “wild value judgments”. If you are going to attack the value judgments you should discuss the evidence and provide alternatives 😛
Part of the problem is that BERL never explain the “exact mechanism for the presumed failure.”
@Paul Walker
Ahhh, I haven’t read the BERL report – only the newspaper article the Hand linked to. Did they not describe the mechanisms which were influencing expectations or the potential for time inconsistent behaviour with an addictive substance in the report?
@Matt: Adrian Slack has come back with these kinds of arguments in response to criticism, but there isn’t really much in the report backing it up. Instead, there’s just the assumption that harmful consumption is by definition irrational. And, if it’s irrational, it by assumption has zero gross benefit rather than benefits less than costs by some amount.
I’m sure that some of this kind of thing was at the back of Slack’s mind when he was writing the report, but the report really doesn’t go into it.
Moreover, since costs are all of the costs for drinking for folks drinking above the threshold, plus the costs that normal drinkers bear if the adverse world state obtains, they’re counting a lot of costs that are really balanced off by benefits for folks where the adverse world state does not obtain. For the latter, it’s ex post regret that’s defining things for them, not necessarily ex ante irrationality.
@Matt Nolan
Matt, I have read the report and nearly all of the substantive discussion on BERL’s view of consumer is on page 173. I would read it, keeping in mind that around 2/3rds of their headline figure depends on that page’s contents. Its the outright baldness and outlandishness of assumptions on that page that gets me every time I read it.
Consider for example the disconnection between the discussion “…individual decisions are not necessarily made on a rational basis” and the methodology which follows from it: “We assume that it is irrational to drink alcohol to a harmful level and that harmful alcohol use has zero private benefit.” Non sequitur, anyone?
After all the criticism, BERL still has not justified this assumption, amongst many others. Why are value judgments even relevant with reasoning like this? That’s what the debate should be focussed on. BERL’s resort to values is pure red herring. Don’t fall for it.
I can’t think of anything anywhere in the report that comes close to contemplating expectations or time consistent behaviour. BERL does not think seriously about their model of the consumer at all – on which their results largely depend.
@Hand: what wild value judgments have I made? I’ve tried damned hard to avoid making value judgments in the costings. Nothing we do hinges on strong rationality assumptions; in arguing for external costs only, we require only that, across the set of all people who either are harmful drinkers or who have a probability of having a harmful outcome, the excess costs (ie costs above benefits for greater than optimal consumption) are roughly matched by consumer surplus on the earlier portion of consumption. Is this a wilder assumption that all harmful drinkers and anyone who realizes an adverse outcome never gets any benefit from any drinking? I think our approach is pretty standard in the econ literature and isn’t terribly “wild”.
Moreover, how does “value judgments” require assumptions of zero substitutability between capital and labour, zero ability to replace a missing worker and zero cohort differences between harmful drinkers in the counterfactual and the population at large when there’s established literature on these kinds of things saying the opposite?
My comment with a link is stuck in moderation so I’ll re-post sans hyperlink:
@Matt Nolan
Matt, I have read the report and nearly all of the substantive discussion on BERL’s view of consumer is on page 173. I would read it, keeping in mind that around 2/3rds of their headline figure depends on that page’s contents. Its the outright baldness and outlandishness of assumptions on that page that gets me every time I read it.
Consider for example the disconnection between the discussion “…individual decisions are not necessarily made on a rational basis” and the methodology which follows from it: “We assume that it is irrational to drink alcohol to a harmful level and that harmful alcohol use has zero private benefit.” Non sequitur, anyone?
After all the criticism, BERL still has not justified this assumption, amongst many others. Why are value judgments even relevant with reasoning like this? That’s what the debate should be focussed on. BERL’s resort to values is pure red herring. Don’t fall for it.
I can’t think of anything anywhere in the report that comes close to contemplating expectations or time consistent behaviour. BERL does not think seriously about their model of the consumer at all – on which their results largely depend.
@Eric Crampton
“Adrian Slack has come back with these kinds of arguments in response to criticism, but there isn’t really much in the report backing it up”
In terms of my comment, I was complaining about the Hand being too extreme in his critique of assumptions (namely saying that the criticism was “nonsensical” – as we COULD use things like time inconsistency to justify ex-ante issues involving alcohol.
Generally something should only be nonsense if it is completely false – not because we disagree with it (and note that I do disagree with the way Dr Nana framed the issue in his article – I just think it is still able to be defended).
@ben
“Matt, I have read the report and nearly all of the substantive discussion on BERL’s view of consumer is on page 173. I would read it, keeping in mind that around 2/3rds of their headline figure depends on that page’s contents. Its the outright baldness and outlandishness of assumptions on that page that gets me every time I read it.”
Interesting. I like it when major assumptions can be limited to a single page – if only I could write like that 😀
“After all the criticism, BERL still has not justified this assumption, amongst many others. Why are value judgments even relevant with reasoning like this?”
The presumption of individual behaviour is a value judgment. If they haven’t justified it then there is definitely a hole there I guess.
Like the Hand said we can’t get costs without value judgments. The debate should be on the validity of the value judgments not (as you said) “red herrings”.
Of course, again, I haven’t read anything so I’m not even going to attempt debating the value judgments. My prior assumption is that both Eric and BERL have used value judgments they believe are appropriate in appropriate ways and I’ll leave them to debate the validity of these assumptions – as they have been 😛
My gut supports the idea that the total costs from alcohol are a bit lower, but that could be because I’m a economic rationalist and a rampant alcoholic 😀
@Matt Nolan
Matt, to be honest I have no sympathy at all with your view here. The presumption of individual behaviour (actually I’m not sure I understand what you mean by that) and the idea that costs exist as a product of a value judgment is really in the realm of philosophy. Possibly a very interesting and useful discussion, but remember BERL was asked to produce an estimate of economic costs. And the economic framework takes these fundamentals as given. BERL should not feel entitled to materially deviate from that framework and cite their personal values after the fact as authoritative. That isn’t how economics works. If that was how it works, then it adds nothing because absolutely anything goes. In that case the Ministry ought to be asking for their money back.
I don’t see how BERL’s report can be defended by asking, “what is cost?” That was given when MoH asked for economic costs. Revisiting that now adds nothing.
@Matt: I get worried when the defense of particular numbers shifts with the critique. For example, BERL argued for counting the total product of labour as a cost rather than its marginal product on the basis of “value added”; we checked the lit and found reasonable measures of “value added” were only 5-10% over and above wages; we counted this part as an external cost (which even that you sensibly critiqued as being too high). After that, BERL shifted the goalposts: “Oh, we didn’t really mean value added, we meant inter-worker complementarities”. It’s all been ex post justification of shonky work. Yeah, you could build a story about how individual time inconsistency could lead to taxes making the alcoholic better off. But then you’d need to put in the hard yards adequately demonstrating that all of the harms they tabulate rely on this kind of time inconsistency story.
Note also that the page 173 bit, laying it all out, is buried amidst reams of tables of sensitivity analysis. Exactly where you’d bury something you wanted nobody to notice. Section 9 “Appendix – method and calculation detail” runs pages 88-130. Then it’s section 10 “Additional tables”. That runs through page 172. Then there’s the single page confession at page 173 “11.1: Harmful drug use and irrational consumption decisions”, then p. 174 Appendix Table 33 (more sensitivity analysis), then page 175 where they note that a survey from GPs reveals that less than 1% of patients visiting doctors from 05-07 were identified as having an alcohol or other drug use problem: not their presumed 1/6 of all adults noted earlier.
BERL has not been debating the validity of assumptions; instead, they’ve obfuscated about what assumptions they were allowed to use in the RFP (“oh, we couldn’t count benefits, outside of remit, maybe somebody should pay us another $130K to look at benefits”), which assumptions they’ve actually used (private costs count as social because of an assumption of zero gross benefits, p. 173), and ignored the extent to which the latter materially affects their conclusions. Further, they’ve mischaracterised our approach as requiring perfect rationality always and everywhere, despite repeated refutation, and accused us of supporting murder.
Maybe I am just ignorant here since I defnitely have read nothing about all this, but the “cost” of something is just adding up all the costs, to get “net costs” you need make some assumption about benefits people recieve, isn’t this where value judgements come into play? If you are talking about “net costs” then there has to be a value judgement doesn’t there?
Is the difference between BERL and Crampton what they assume about off-setting benefits?
Again, excuse my ignorance, being busy at work means I don’t get to see much more than the “he said/she said” being played out in the media.
Agnito: that’s the biggest difference, but there are other ones. Like the multiplier, etc. I have done my best to summarize the argument on my blog….
@ben
“BERL was asked to produce an estimate of economic costs”
I thought their task was to estimate total costs, surely these include a pile of value ladden social costs as well?
Revisting the idea of what a specific cost is is the very essence of what we should be doing – in order to understand what costs are relevant for policy and why.
From what I’ve seen Eric has said, ok even if we ignore the benefits (which he also says we shouldn’t when forming policy) the costs that have been estimated are too high. BERL replied by saying the difference in costs come from different value judgments – and as a result it is these issues that should be discussed.
@Eric Crampton
“I get worried when the defense of particular numbers shifts with the critique”
This is a very good point. I completely agree.
“BERL has not been debating the validity of assumptions; instead, they’ve obfuscated about what assumptions they were allowed to use in the RFP”
This seems like the essential point. The argument should be about “which assumptions seem better” at this point – and if it is true that they are avoiding this issue then I find that disappointing.
When I looked at Dr Nana’s article today I was quite disappointed by the attacks on rationality, and from what I’ve seen on your blog today your estimates don’t seem based on a super human rational agent.
Personally I would think that there is scope for the estimation of costs to differ – but what would be really cool is if the differing value judgments that defined the difference between the points of view could be distilled to a half page pdf.
As I’ve said, I personally find the lower cost scenario more “believable” – but this could be personal bias thanks to the fact that I’ve trained as an economist and drink excessively 😛
If you want to count any “irrational” consumption as being consistent with zero gross benefits, but want to correct the other problems, you get a number about 40% smaller than BERL’s headline. If you want to assume that, across the set of all harmful consumers, consumer surplus roughly matches the excess private costs incurred by harmful consumers, in which case only external costs matter, then you get a number about 95% smaller than BERL’s headline.
I can buy having a number that includes a big whack of private costs if you never call that social costs or net costs and you never call it welfare and if you jump up and down and scream if anybody tries to use the number as something usefully compared with tax revenues because it’s manifestly wrong to use a gross cost measure in comparison with pigovean tax revenues.
I cannot buy having a number that includes all those private costs when you’ve specifically assumed that there are no private benefits and when you’re more than happy for Geoff Palmer to cite the difference between a huge gross costs number and excise tax revenues as being something that indicates taxes are too low.
It’s is total obfuscation to claim that benefits were outside of remit after having claimed that benefits are equal to zero and hinged the inclusion of private costs on that.
From p.173:
So, the method they’re following says we cannot count private costs if there are offsetting private benefits.
Starting the argument for inclusion of the excess costs induced by irrationally-high consumption levels. I don’t see how irrationality is necessary to worry about externalities, but the rest of the paragraph makes clear that they’re really worried about internalities. Ok. But then we find
Again, I wouldn’t have a problem if they were trying to estimate net excess costs on the basis of excess irrational consumption having net costs to be weighed against prior consumer surplus. But then they wave their hands about benefits being outside of scope. If that’s the case, counting only external costs, or at least having a side tabulation of external costs and highlighting that as the important bit, is the appropriate next step I figure.
@Matt Nolan
Matt I like your point about comparing assumptions to see which are more reasonable. But why the focus on these value judgments? I mean, what is it about BERL’s cost study that makes value judgments front and center of your attention? Can’t the same be asked of every single economic study?
My point is that the personal values of the author and the reader is not what makes an economic analysis interesting. It is the quality of thinking and the fit with reality that does this. Great economic papers are not great because people with the best values (whatever that means) like them. Its because there is some objective sense in which they add to understanding.
BERL’s report is poor because of its poor fit with reality. In fact it is positively misleading in many places. While I recognise that reasonable people will disagree, there is also a point at which an argument can be recognised as faulty without reference to any values. This can be established by reference to logic or to empirical evidence, for example. Personal values are a poor substitute for these alternative means of finding truth. As I noted in an example above, BERL’s p. 173 premise does not support the conclusion which follows.
Thanks for the info- it turns out Microsoft’s bing really is making an impact, it seems to me your blog is getting a lot of search engine traffic from bing- have you found this to be the case?