tag:blogger.com,1999:blog-6976144462093297473.post950823068147849161..comments2024-03-26T19:10:00.791-04:00Comments on West Coast Stat Views (on Observational Epidemiology and more): Marginal utilityJosephhttp://www.blogger.com/profile/10760453165301871031noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-6976144462093297473.post-36305699390804279532011-09-25T16:42:56.659-04:002011-09-25T16:42:56.659-04:00I think there are several income "ranges"...I think there are several income "ranges" and thus a very non-linear curve.<br /><br />For the very poor (people without basic food ans shelter), I think efficiency of transfer is a secondary issue. Unless we are in danger of running out of resources, there is no excuse for a rich society not to help the poor (even if the transfer is $0.01 on each dollar of tax). <br /><br />Remember Hayek: “I have always said that I am in favor of a minimum income for every person in the country.” <br /><br />For the very wealthy, those who make more than they can possibly consume (or their heirs consume) then I am not overly sympathetic to losses of income. Here I am talking about incomes of $100 million per year plus. People whose incomes is 1000's of times greater than a median worker. Nobody reaches this level of income without leveraging the structure of society as an individual's productivity simply cannot exceed that of a medium city. <br /><br />It is the middle range (notice the tremendous layering of non-linearity here), I seem to assume that log transforms of income seem to be a good reflection of marginal utility. <br /><br />BTW, Frances, I love Worthwhile Canadian Initiative and the great set of posts that show up there.Josephhttps://www.blogger.com/profile/10760453165301871031noreply@blogger.comtag:blogger.com,1999:blog-6976144462093297473.post-21555986499915095872011-09-25T13:27:06.609-04:002011-09-25T13:27:06.609-04:00Joseph - thanks for the link.
Yes, you're ri...Joseph - thanks for the link. <br /><br />Yes, you're right, clearly the 100,000th dollar isn't worth as much as the 100th. But how much less is it worth? <br /><br />Here's another way of phrasing the question: suppose you were redistributing income from a highly paid CEO to a poor person. But you have to redistribute income with a leaky bucket - that is, every time you take a dollar from the CEO, you lose some amount in 'deadweight loss'. <br /><br />How much waste are your prepared to accept before it's no longer worth trying to redistribute income. E.g., is it worth taking $1 from the CEO if only $0.20 makes it to the poor person? If only $0.50 makes it? If only $0.01 makes it?Frances Woolleyhttps://www.blogger.com/profile/04289318268301647625noreply@blogger.comtag:blogger.com,1999:blog-6976144462093297473.post-38052360162856139812011-09-23T20:59:07.562-04:002011-09-23T20:59:07.562-04:00I also used to model income on the log scale. I a...I also used to model income on the log scale. I agree that it does not handle the skew but it is a much, much better approximation than linear. Defensible, even, and with some intuitive predictions. <br /><br />I think slightly greater than flat makes sense. I wonder how far we are from flat if you combine income taxes, FICA, state taxes and local taxes (the first of which is progresssive, the latter of which are regressive).Josephhttps://www.blogger.com/profile/10760453165301871031noreply@blogger.comtag:blogger.com,1999:blog-6976144462093297473.post-31727353352802285572011-09-23T20:36:56.718-04:002011-09-23T20:36:56.718-04:00My immediate thought is that income is linear on t...My immediate thought is that income is linear on the log scale. I frequently use a log transform when analyzing money data, and a 3% bonus should mean something to anybody on the income scale. This would imply that a flat rate is fair.<br /><br />But that would imply that incomes have a lognormal distribution. But I seem to recall that real income distributions are actually more skewed than that (at least I was trying to make income distributions out of lognormals and they didn't look quite right). If this is the case, then maybe even a flat rate is not fair enough.ralmondhttps://www.blogger.com/profile/05940072941068366047noreply@blogger.com