The Five Dumbest Things on Wall Street This Week

1. I Dream of the Gini Index

We at the Five Dumbest Things Research Lab hate to go all antiacademic on you, but here's a little advice: The next time you see a statistic in the newspaper, don't believe it. It's wrong.

OK, OK. That's overstating our case a little. It's not necessarily wrong. But it's not right, either.

Exhibit A: The 2002 household income figures released last Friday by the U.S. Census Bureau.

The takeaway from the report, as you may have read in The Wall Street Journal's Monday account, was that the poverty level rose, but income inequality didn't, because rich folk's income took a beating, too.

But something further down in the write-up caught our eye. "Difficulties in recording seven-figure incomes," reported the Journal, might have resulted in underreported income among the wealthiest Americans.

In other words, the rich may be richer.

That's odd, we thought. People pay a lot of attention to these annual income-disparity figures. How come no one's getting worked up about inaccurate data from such a key segment of the surveyed population? This can't be true.

We called up Edward Welniak, chief of the Census Bureau's income survey, to check.

What's Mine Is Mine

Indeed, there are difficulties with high-income data, Welniak told us.

Here's why: Starting with the 1993 numbers, the bureau's staff -- which interviews a sample of 78,000 households for the income survey either in person or over the phone -- has been entering people's responses directly into portable or desktop PCs. As part of the survey, respondents are asked to report how much money they made the previous year from numerous sources -- stuff like the job held the longest, interest and dividends.

And here's the catch: In each category, the highest dollar amount one can enter is $999,999.

So let's say a Census employee had dropped by the $15,000-umbrella-stand-festooned apartment of ousted Tyco TYC Chairman Dennis Kozlowski in 2001. And let's say the then-executive wanted to report the $50 million or so in undisclosed compensation the Securities and Exchange Commission says he received in 2000.