On occassion, a study is released that is so obviously flawed that it boggles the mind that anyone would repeat it, much less generate it. Such is the case with the "Generosity Index" which was released by the The Massachusetts Catalogue for Philanthropy on November 5th. Two days later, the Wall Street Journal posted an item on its online Opinion Journal at http://www.opinionjournal.com/taste/?id=110004271 that mentions the index and concludes:
As OpinionJournal.com reader Gabriel Openshaw pointed out to us, the resulting index shows that the top 20 states all went for George W. Bush in the 2000 election--while 15 of the 20 least generous went for Al Gore. Maybe, he suggests, the difference is that those in red states are more generous with their own money while those in blue states are more likely to be generous with other people's money.
The Rush Limbaugh site picked up the story the same day, posting links to the OpinionJournal article and the index.
The 2001 IRS data that was used to calculate the index can be found at Rhttp://www.catalogueforphilanthropy.org/cfp/generosity_index/2003_Generosity_Index.xls. As can be seen from the data in the xls file, the index was calculated as follows:
First, the total AGI (Adjusted Gross Income) for all the returns from each state is divided by the total number of returns from that state to get the average AGI. The states are then ranked from the highest to the lowest average AGI to give a "Having Rank" from the best (1) to the worst (50).
Secondly, the total charitable contributions claimed via itemizing are divided by the total number of returns that itemized such deductions for each state. The states are then ranked from the highest to the lowerst average deduction to give a "Giving Rank" from the best (1) to the worst (50).
Finally, the "Giving Rank" is subtracted from the "Having Rank" to give the "Rank Relation". The states are then ranked from the highest to the lowest "Rank Relation" to give a "Generosity Index" from the best (1) to the worst (50). For example, Mississippi had the lowest average AGI ("Having Index" of 50) but the sixth highest average deduction ("Giving Index" of 6), giving it a "Rank Relation" of 44 (50 minus 6). This was the largest "Rank Relation" giving it the best "Generosity Index" of 1.
Well, the first major flaw with this method is that it is comparing apples and oranges. It is comparing the average AGI of all taxpayers in a state to the average charitable deductions claimed by only those who itemized such deductions. This favors states that have a large number of taxpayers with low AGIs who do not itemize. Such taxpayers will pull down the average AGI (and the "Having Rank") but will have absolutely no effect on the charitable deductions claimed (and the "Giving Rank"). The aforementioned xls file shows the percentage of the total number of taxpayers in each state who claimed charitable deductions. As can be seen, of the twenty states with the best generosity indices, all but 4 had LESS than 30% of all taxpayers claim such deductions. Of the twenty states with the worst generosity indices, however, all but 3 had MORE than 30% of all taxpayers claim such deductions. Coincidence?
The second major flaw in this method is subtracting the "Giving Rank" from the "Having Rank" to get the "Ranks Relation". This favors the states with low average AGIs. For example, suppose that all taxpayers in the state with the highest AGI gave all of their money to charity. They would likely earn a "Giving Rank" of 1. Subtracting that from their "Having Rank" of 1, however, would give them a "Ranks Relation" of 0. This would give them a mere average Generosity Index despite the fact that they had given all of their money away. Likwise, suppose that Mississippi had given nothing to charity. That would have earned them a "Giving Rank" of 50 which, subtracted from their "Having Rank of 50, would give the same Generosity Index of 0. In short, the Generosity Index penalizes "rich" states. It's likewise no coincidence that, of the twenty states with the best generosity indices, all but 4 have "Having Ranks" LARGER than 25. Of the twenty states with the worst generosity indices, however, all but 3 have "Having Ranks" SMALLER than 25.
The most direct way to calculate a generosity index would be to divide the average charitable contribution by the average AGI of all taxpayers in that state. This would require some estimation of the contributions made by taxpayers that could not be claimed. Like all averages, of course, this index would still have its limitations. But it would be much more accurate than this deeply flawed system that favors poor states with high disparity of income. I would hope that those publications that carried this story were simply not aware of any these problems with its methodology. Assuming that, however, might be overly generous.