Talk:Kaldor–Hicks efficiency

I do not feel that many lay folks would fully understand this article. I would love for the article to be made more accessible. -- anonymous

"A Kaldor-Hicks improvement is any alternative that increases the economic value of social resources": this strikes me as a strange definition, which makes a number of (perhaps questionable?) assumptions about what "economic value" means, about the possibility of a coherent notion of "the economic value of social resources" simpliciter, etc. In fact a Kaldor-Hicks improvement is a much more formally defined notion, as defined in the body of the article.

If no one convinces me otherwise in the next few days, and I don't forget, I'm going to replace the first paragraph with something along the lines of "Kaldor-Hicks efficiency (named for Nicholas Kaldor and John Hicks) is a type of economic efficiency that captures some of the intuitive appeal of Pareto efficiency, while having less stringent criteria and therefore being applicable in more circumstances." Or something like that. -- Orbst 18:18, 13 Jun 2005 (UTC)

Analysis has been excellent. But the Marxist critique should be highlighted.

looks like a mistake

quote from article: "While all Kaldor-Hicks efficient situations are Pareto efficient, the reverse is not true. Conversely, though every Pareto improvement is a Kaldor-Hicks improvement, most Kaldor-Hicks improvements are not Pareto improvements."

I think it should be: "While all Pareto efficient situations are Kaldor-Hicks efficient, the reverse is not true."

I think you are incorrect. There are plenty of situations where a Pareto efficient allocation is not Kaldor-Hicks efficient, indeed that is why the compensation principle was invented in the first place. If I value a good more than you but I don't have anything that you want enough for us to be able to trade, then this allocation is Pareto efficient (redistributing the good anyway would make me better off, but at the cost of making you worse off). The compensation principle holds, however, that since I value the good more than you do, redistribution should take place, because I could potentially compensate you for your loss. Only after the redistribution has taken place will the allocation be Kaldor-Hicks efficient. It will also still be Pareto efficient as when no Kaldor-Hicks improvements are possible, neither are Pareto improvements.

Unless you guys disagree, I'm changing it back. RoyalTS

Criticism

This:

"One problem with the K-H criterion is that it will conclude that any change that results in an increase in income will lead to Pareto optimality. Further, this Pareto optimality will result, no matter what the income distribution consequences of the change are."

is wrong. Kaldor-Hicks states that, under a Kaldor-Hicks improvement, a Pareto improvement could be achieved given the proper transfers, not that it will be achieved.

Also, this:

"At a more technical level, various versions of the K-H criteria lack desirable formal properties. For instance, Tibor Scitovsky demonstrated that the Kaldor criterion alone is not symmetric: it's possible to have a situation where an outcome A is an improvement (according to the Kaldor criterion) over outcome B, but B is also an improvement over A. The combined Kaldor-Hicks criterion does not have this problem, but it can be non-transitive (A may be an improvement over B, and B over C, but A may not be an improvement over C)."

is wrong as well. Kaldor-Hicks is "transitive" (if the net benefits are higher, then it is preferable to lower net benefits), and I don't know who had the idea that it is not.Jackson744 17:17, 3 March 2006 (UTC)

Methinks you are wrong, at the very least on your last point. This is hardly my area of expertise, but a quick search through the journals reveals several articles that speak of this intransitivity (take this one, for example: http://www.springerlink.com/link.asp?id=vew2t8j6d6tdfaqw ). The asymmetry of either one of the criteria is also well-demonstrated, so I'm reverting this part of your edit.

As for your second change, I don't know, because the way the sentence was phrased originally is kind of confusing. If I'm not mistaken the only point that's made is that Kaldor-Hicks pays no attention to the distribution of income, just to aggregate income. Please correct me if I'm wrong.

Furthermore, I do not understand why you reverted my change concerning all Kaldor-Hicks efficient situations being Pareto efficient, but not vice versa. I've already laid out the rationale for this, so I would really appreciate it if you could explain why you think this is wrong. RoyalTS

First of all, concerning the issue of transitivity, Kaldor-Hicks would suggest that a policy with higher net benefits is preferable to a policy with lower net benefits, and is indifferent between two policies with equal net benefits. There is absolutely no room for intransitivity, because the very criteria of K-H preclude this possibility. Your link does not work, and based on the fact that I (and I hold a graduate-level degree in economics) know this to be a false critique of K-H, I am going to revert.

Secondly, the reaqson why I reverted your change is the fact that all Kaldor-Hicks improvements are not Pareto improvements. You have it backwards. All Pareto improvements are Kaldor-Hicks improvements. A Kaldor-Hicks improvement implies that the net social benefit has increased, while Pareto efficiency implies that at least some are better off, while nobody is worse off. Therefore, K-H is necessary for Pareto efficiency, but not sufficient.

The only criticism I've heard of K-H is that it ignores the distributional impacts of policies, which is why I left the criticism which I did.Jackson744 17:54, 5 March 2006 (UTC)

If you know this to be a false critique of K-H, would you mind sharing a reference to a paper that shows this to be the case? The link I provided to the paper that says otherwise works fine here, so maybe you can give it another shot.

I am 100% in agreement with the way you put the relationship between Kaldor-Hicks and the Pareto criterion in this discussion, but the article does not reflect this. It says "While all Pareto efficient situations are Kaldor-Hicks efficient, the reverse is not true." If Kaldor-Hicks efficient means no more Kaldor-Hicks improvements are possible, then a Pareto efficient allocation is not necessarily Kaldor-Hicks efficient. As I said, this is the reason for having the criterion in the first place. In contrast, all Kaldor-Hicks efficient allocations are automatically Pareto-optimal, so the statement has it exactly backward. Notice that the sentence following the one I quoted is exactly right.

Thirdly, you removed the criticism as regards the distribution of income. If you do agree with it, why don't you put it back in? Moreover, I see that the sentence about the Scitovsky criterion got lost as well. Would you care to justify? RoyalTS 20:17, 5 March 2006 (UTC)

What I've removed from the criticisms section was, one, something which was inaccurate (it claimed that any Kaldor-Hicks improvement would lead to a Pareto improvement, which is not true at all). The portion of the criticisms which I have left deals with the issue of distribution, and thus, I felt that I would not need to add any additional information. I've also removed the part about transitivity, which I believe to be false. As I've explained, I've never heard such a contention, and I've explained why I don't see how it could be the case, but if you or anybody could explain to me why it would be the case in a language that is accessible to the average reader, then I'll agree that it should be included. At any rate, I, holding a Master's degree in economics, have never heard any such contention, and if a person with a Masters in economics hasn't heard of such an objection, it's unlikely to be relevant to the typical encyclopedia reader even if it is true. I think it's unlikely that it really needs to be included.

As for the other contention, you've got things backwards. Pareto improvements are, by definition, Kaldor-Hicks improvements, but Kaldor-Hicks improvements need not be Pareto improvements. As I've said, if at least some people are better off and nobody is worse off (Pareto improvement), then the net social benefit is positive, and it's also a Kaldor-Hicks improvement. If, on the other hand, the net social benefit is positive (Kaldor-Hicks improvement), then it need not be the case that sopme people are not worse off (Pareto improvement). In fact, I think I may add an image to the article soon to clear up the issue.Jackson744 23:28, 5 March 2006 (UTC)

No, what you have removed is not only what you said you removed, but all criticism except for one paragraph. You removed loads of stuff which was completely accurate. Moreover, what is left does not at all deal with the issue of distribution, but instead deals with the cardinality of the social welfare functions based on K-H. Moreover, I fail to see anything wrong with the sentence about the Scitovsky criterion. Please look at the changelog and reread what you deleted!

You previously claimed that you KNEW the transitivity criticism to be false, now you're saying that you have never heard of such a claim which is quite a different thing. As I said, a 2 minute search with Google Scholar reveals several articles in peer-reviewed journals which speak of just this intransitivity and I have linked to one, are you really simply going to cite your credentials to support your view? I'm more than willing to accept your edit if you can back it up with some evidence, but so far I have yet to see any. And just because you have never heard of such a contention does not mean it should not be included in the article. I doubt the majority of Master students in economics ever deal with game theory as in depth as it is explained in Wikipedia, I would not even be surprised if a substantial part of Bachelor students have never heard the term "Kaldor-Hicks", let alone its formal properties...

As regards the K-H vs. Pareto criterion, please read my previous comments carefully. I never disagreed with what you have been saying in this discussion. I know that Pareto IMPROVEMENTS are K-H improvements, but not vice versa. But this is not what the article says: "While all Pareto EFFICIENT situations are Kaldor-Hicks EFFICIENT, the reverse is not true". It does not speak of IMPROVEMENTS, it speaks of SITUATIONS. And in this case, a situation that is Kaldor-Hicks efficient is also Pareto efficient, precisely because if there are no more K-H improvements left this automatically means neither are Pareto improvements per your explanation. Conversely, the reverse is not true as should be self-evident.

I like the idea of a picture. What were you thinking about? RoyalTS 23:56, 5 March 2006 (UTC)

Here's my stance on the whole transitivity issue: I own the textbook "Cost-Benefit Analysis" by Boardman, Greenberg et. al., which I used for both undergrauate and graduate work. I've studied CBA pretty extensively, and I work with it regularly. In neither the book nor any of the classes (which did both extensively cover Kaldor-Hicks), nor my use of CBA since, have I ever encountered this objection. Now I'm not completely unwilling to believe that it may be a valid criticism - as I stated, if it can be explained to me in such a manner as counter my intuition on the matter (which I've stated above), then I will believe that it's valid. However, it was not explained whatsoever in the article before. Why is it intrasitive - give an example of why it might not be, because it seems to me that a policy in which society is better off is preferred to one in which a society is at the status quo, and would be indifferent between two policies under which society is equally well off. I'll even be willing, in that case, to say that it should go into the article, although I'm questionable to what utility it could possibly be for 99% of readers who might want to use the article. But I don't believe that the fact that something is on google scholar alone should merit its inclusion. For example, googling, "'Kaldor-Hicksintransitive'" nets about 80 sites, many of which do not discuss the issue you've raised. The same goes with "nontransitive" and "transitivity". There just aren't that many hits, and most of the hits are sites in which the two words merely appear together coincidentally. Certainly, the sites which do appear do not constitute any sort of consensus on the matter, and coupled with the fact that I've never heard this objection, I really question its merit. Even so, it did not explain the objection, which I think it must do to merit inclusion. It merely confuses if it isn't properly explained. So why is it intransitive?

As for the issue of Kaldor-Hicks "situations", you're right - any Kaldor-Hicks efficient allocation must be Pareto efficient, that's my mistake. I think I misread the article because you talk about improvements in the same spot in which you talk about efficiency. I'll try to make that a little clearer.

As for the picture, I was thinking of including a map in 2D space of the utility of two individuals, showing Pareto and Potential Pareto frontiers, as well as all possible Pareto and Kaldor-Hicks improvements.Jackson744 02:24, 6 March 2006 (UTC)

Far be it for me to claim that I actually understand the technical details of this matter, but a more thorough search at JSTOR reveals this article which formally deals with the transitivity issue: http://links.jstor.org/sici?sici=0020-6598%28197810%2919%3A3%3C547%3ATNWE1%3E2.0.CO%3B2-S (hope you have access, otherwise tell me and I'll send you the paper). Given the fact that the proof for the existence of such problems is so formal, I tend to agree with you that explaining it properly will not be interesting to 99% of people. So shall we simply mention it in one sentence and refer to the paper for details?

I can't quite picture the 2D map you propose yet, but the idea of illustrating the difference between Pareto and K-H criterion seems worthwhile. Do you mean a sort of Edgeworth box?

I think the exact phrasing of the section about the difference is still suboptimal, but at least it's correct now. Maybe I'll give it a shot sometime. Feel free to revert if you don't like it. RoyalTS 05:02, 6 March 2006 (UTC)

OK, I've added the picture. Tell me what you think. Also, I'm OK with adding a sentence about the intransitivity with a link to the paper. Was there anything else you think needs to be added/changed? I've also added in the part about the intransitivity in the criticisms section.Jackson744 16:03, 6 March 2006 (UTC)

Wait a sec, it's still wrong: "Since any Kaldor-Hicks efficient allocation maximizes social welfare, it must necessarily be the case that any Kaldor-Hicks efficient allocation is also Pareto efficient. However, while all Pareto efficient allocations are Kaldor-Hicks efficient, the reverse is not true." The second part says exactly the opposite of the first. I'm reverting to the original phrasing because it is both correct and nicely phrased. Unless we're going to expand on our description of WHY this is so, I think we should leave it like this.

Moreover "Tibor Scitovsky', having shown the intrasitivity of the Kaldor criterion, suggested combining the two, into what we now call the Kaldor-Hicks criteria." is not factually correct. What Scitovsky showed were reversals, not intransitivity.

I also cannot find in the document you linked (nice explanations, BTW!) that it was Paul Samuelson who discovered the intransitivity, if I understand it correctly, it was Gorman who discovered them, Samuelson proposed a solution. I'm reverting back to the original. Give it another look, it really is a quite sensible, easily understandable explanation. And inclluding the link you added gives people that want to read more a good reference. I'm not too familiar with Wikipedia rules, but shouldn't this link be in a Reference section at the bottom of the page?

I like the picture. For someone with some basic training in econ, this should be easily understandable! RoyalTS 21:14, 6 March 2006 (UTC)

Well, I think it's OK, but the lines which you reverted about K-H efficient "situations" and "improvements" needs to be modified to make it less confusing. The reason why I edited it in the first place is that I though it was stating that K-H improvements are Pareto improvements, due to the fact that it's too scrunched up. As for the transitivity portion, I also think that it's a bit confusing, and am going to attempt to re-phrase it to make it a bit more accessible. But otherwise, it's OK.Jackson744 21:27, 6 March 2006 (UTC)