>Doug:
>
>on page 32 of the working paper by Kennickell and Woodburn you
>mentioned on your web page, it says:
>
>"The consistent weights show a statistically significant increase in
>the share of household net worth held by the wealthiest 1/2 percent of
>households from 1992 to 1995, driven in large part by a rise in their
>share of personal businesses. However, looking more broadly over time,
>the standard error of the 1989 estimate is so large that one cannot
>reject the hypothesis that the 1995 and 1989 figures are the same."
>
>any comments from the 'stats maven'? how much can one trust these
>kinds of numbers to display trends such as the wealth of the top 1/2%
>vs the bottom 90%? and what are the estimated error bars on these
>numbers? i didnt find them in a quick read of the Kennickell paper.
I'm no math whiz, so I'll have to defer to those who know better than I on the technical questions. My guess is that measuring the very rich is such a dicey task that it's hard to tell trends from noise.
The Survey of Consumer Finances does go out of its way to find rich folks. Bob Pollin told me that an acquaintance of his was doing Beverly Hills for the survey, and she tried swapping recipes with her targets in order to get them to agree to cooperate. It's not be perfect, but it's probably about as good as social surveys can get.
I use Kennickell's estimates rather than Ed Wolff's mainly as a defense against audience doubts; if you tell the orthodox that something comes from the Fed, they tend to swallow their objections.
Doug