Republicans - All Wrong, All the Time, Pt 7

Brad Delong announts Ed Lazear as his stupidest man of the day, (item 9 at the link) noting that Matt Yglesias and Peter Orszag  called him out for misrepresenting this simple fact about the 2009 budget: it was devised by the Bush administration in 2008.  But, hey, it a WSJ article, so honesty is not a requirement, and reality is irrelevant.

But, for the current director of the Office of Management and Budget to have to correct the Chairman of the former administration's Council of Economic Advisers on how the budget is done seems a bit over the top, even in that context.




In Lazear's long, rambling essey he also says this:

My analysis of data from 1950 to the present shows that periods with high tax-to-GDP ratios exhibit much slower economic growth than lower tax ratio periods. The GDP growth in high tax years (defined as years during which the ratio of tax-to-GDP was above 18%, the 60-year average) was about 1.5 percentage points lower than the growth rate in low-tax years.

Actually, my analysis shows a 1.75% difference. But doesn't Lazear's statement suggest a strongly bimodal distribution? Let's have a look.  This cloud graph shows the reality. Data from The Tax Policy Center.

Do you see anything striking here?  The various blue dots indicate results with a Democrat in the White House.  Red dots are similarly Republican.  Each President has his own symbol, so you can see within any administration a geat deal of data scatter.

My reaction at this point is big GDP whoop!  The data plot looks like a cloud, with some outliers.  But we can dig deeper.  The next graph shows the grand averages of the data sets.  The yellow vertical line is the 18% Tax/GDP line to which Lazear ascribes so much importance.  The heavy pink line is the average GDP growth from 1950 to 2006, 3.44%.  Placed around it are lines above and below by one standard deviation.  Nine points lie more than one Std Dev below the mean, 8 lie more than one Std Dev above the mean, 39 inside the box, or 68.42%.  There is nothing remarkable about viewing this as a single data set, and you really have to squint to make it anything else.  I've connected the points from the Truman administration.  Doing this for all the presidents just leads to a big mess.

The green line is the least squares regression best fit.  Sure enough, the slope is negative, and it passes close to where the mean lines cross.  Unfortunately for Lazear, though, the correlation coefficient is only .4946.  Anything below .50 is considered a weak correlation.


Not graphed is this gem of mathematical analysis.  The average DPG growth of the "low tax" years is 4.17%.  The average growth of the "high tax years" is 2.41%.  But the standard deviation of the "high tax" set is 2.34%.  for the "low tax" set it is 1.91.  So, the mean of each set is within one standard deviation of the other set.  By my reckoning, this cloud constitutes a single data set.  The division is meaningless, and misleading.  In other words, a lie.

Lezear continues:

High taxes are clearly bad for the U.S. economy. For example, were we to tax above the 18% tax-to-GDP ratio over the next 25 years, GDP per capita in 2035 would be about 50% less than if we were to tax below the 18% ratio. A 50% per capita GDP differential is about as large as the difference between the U.S. and Greece today.

The truth is, Lazear is so wrong on so many levels, it's nothing short of stunning.

1) There is a single, well behaved, randomly distributed data set.
2) The correlation between Taxes/GDP and GDP Growth is negative, but weak.
3) Lezear makes the implicit assumption that correlation is causality.  This is the classic logical fallacy known as Post Hoc Ergo Propter Hoc, or, A precedes B, therefore A causes B.  But there is no reason to conclude (or even posit) that taxes/GDP should be the cause of anything.
4) Think about this. The number Taxes/GDP is a ratio.  It can be increased by having taxes receipts be high, or by having GDP be low.  In the absence of any real correlation, this number is a data artifact, with no practical utility.
5) Lazear draws from all this the conclusion that, "High taxes are clearly bad for the U.S. economy."  A naked assertion that is not supported by these data, and, in fact cannot be validated by ANY data.
6) Even worse, he states this, "For example, were we to tax above the 18% tax-to-GDP ratio over the next 25 years, GDP per capita in 2035 would be about 50% less than if we were to tax below the 18% ratio," performing a thoroughly invalid predictive mathematical calculation.  This is intellectual sloppiness of an exceedingly high order. 
7) Maybe I read this into it, but it seems clear that part of his premise is that a "tax and spend," program is characteristically a Democratic policy, and that Democrats can therefore be expected to decrease GDP growth over time.  This is the opposite of the facts.  As the graph shows, the below average portion of the GDP data set is densely populated with Republicans; and, except for one point from the Carter administration, Republicans own the negative numbers.  In fact the single worst number on the graph, at -1.94% belongs to Saint Ronnie.

And this clown teaches Economics at Stanford!

The lying Repugniucant wrongitude - it burns.