LIES, DAMNED LIES, & STATISTICS
I was reading Poop Throwing Monkey
, and much to my surprise I noted the following statement:
"Even under the Bush Tax Cuts, the rich still keep paying more" together with a statistical table that purports to support the claim that the share of total federal tax liabilities by income quintile was rising, even with the tax cut. The relevant part of it is reproduced below:
Highest Quintile Share in 1979-1990: 56.80%
Highest Quintile Share in 1991-2000: 62.57%
Highest Quintile Share in 2001: 65.30%
has the full scoop." FTM tells us. I have to admit, this seems pretty counter-intuitive to me, given the fat juicy tax cuts that got passed in 2001 (and beyond). So, out of curiosity I follow the link to the blog in question. I find that what he has done to get his results is the following:
"As you can see I've averaged the periods 1970-1990 and 1990-2000, so these can be compared with 2001, the first year in which traces of the Bush tax cuts appear."
I check the link to the Congressional Budget Office Paper in question, Effective Federal Tax Rates: 1979-2001
and from there looked at Table 1B: Shares of Federal Tax Liabilities
This is what I find:
Highest Quintile Share in 2000: 66.7%
Highest Quintile Share in 2001: 65.3%
How exactly does this support the thesis that the share of Federal tax for the wealthiest went down?
Econopundit prides himself on being an economist. Since he slams Paul Krugman for being a "so-called" economist, he must be pretty good himself to make that judgement. Yet, why would an economist make such a glaring error as being interested in averages, when what economists are really interested in are the margins?
To put it another way: What I want to know is how tax liability shares changed from the last year of the Clinton administration to the first year of the Bush administration. That is the point, right? So what is the relevance of taking a TEN year average when I want to know the marginal
impact of a tax cutting decision.
Econopundit uses the same technique in looking at the very rich
. Needless to say, it suffers from the same issues, which a cursory glance at the appropriate Table information would make instantly clear:
Top 10% in 2000: 52.2% in 2001: 50.0%
Top 5% in 2000: 41.4% in 2001: 38.5%
Top 1% in 2000: 25.6% in 2001: 22.7%
What this seems to be saying, when I compare it with the information above is that for the Highest Quintile as a whole, tax liabilities went down by 1.4% of the total liabilities. If you happened to be really wealthy, it went down by 2.9% of the total liabilities. That seems like a big deal to me, even if it isn't to Econopundit.
Now, this is not to say that the share of tax liabilities couldn't rise. They could. Just not in a way that I think Econopundit might be comfortable to broadcast. If, as is looking to be the case, the wealthy are owners of capital, one would expect that in an up-turn (the jobless recovery we have been having over the last couple of years), that we could expect that if the wealthy were compensated at a relatively greater rate than those in the lower qunitiles, then the share of the taxes paid by the wealthy should get bigger. If they are getting a bigger slice of the pie relative to others, this makes sense. This will work against the trend of lowering rates for the wealthy. It might be interesting to contemplate that in 2003, nominal GDP rose twice as fast as pre-tax employee compensation. That implies that the lion's share of income and wealth generated in that year went to the unfortunates at the top of the food chain.