The correlation now goes up to .11. [update: I cleaned up the chart a little. On the right side of the oval are Clinton tax years; the left are GWB tax years.]
[Second update. I added the three most recent years. Correlation now rises to .17!]
Sunday, March 08, 2009
The Oriole Way asks me to try GDP growth and lagged taxes
Pleased to oblige:
The correlation now goes up to .11. [update: I cleaned up the chart a little. On the right side of the oval are Clinton tax years; the left are GWB tax years.]
[Second update. I added the three most recent years. Correlation now rises to .17!]
The correlation now goes up to .11. [update: I cleaned up the chart a little. On the right side of the oval are Clinton tax years; the left are GWB tax years.]
[Second update. I added the three most recent years. Correlation now rises to .17!]
Thanks for putting these data together.
ReplyDeleteg=growth, t=tax rate. 1. g=a+bt+e and 2. t=A+Bg+E. Equation 1 is the effect of taxes on growth, 2 is tax policy responding to economic conditions. OLS of 1 is biased and inconsistent because t is correlated with e. I think you are proposing lag(t) as an instrument for t. But lag(t) will not be a valid instrument for t if e is serially correlated. Looking at the chart, it is pretty obvious that the errors are serially correlated.
Shocks unrelated to tax rates might work as instruments for g. Instruments for t are harder. Election results, maybe?
David, the point is not that higher tax rates cause growth, ceteris paribus. That would be difficult to believe. The point is that it is hard to make a case that small changes in marginal tax rates at the top of the income distribution would be economically catastrophic.
ReplyDeleteHi Richard.
ReplyDeleteBobbi here from the Hoyt Center. Hope you are doing well. This is just a test message. We are in the process of setting up Maury Seldin's blog and Ron Racster wanted to know some of the limitations, etc.
Best,
Bobbi
I agree that a small increase is unlikely to be catastrophic, and I know you are not claiming that b>0. Your chart suggests that b is basically zero for small changes in t - you and Steve Roth make an interesting case for this.
ReplyDeleteI believe that b<0, but b is hard to measure because of simultaneity and other issues. Raising t to 39.6% would be bad, but not the end of the world.
This all makes the following questions pop into my head:
ReplyDelete1. Is the data skewed at all by the higher rates of growth in GDP in the post WWII era and the 50s when the U.S. economy was less mature and therefore had more room for growth?
2. Would higher unused capacity and unemployment rates create room for greater GDP growth no matter what the tax rates?
3. When poor Asian countries began economic reform they began having very high growth rates. Did tax rates matter there? Was it just reflective of increased capital investment and a huge labor force that was readily available.
4. Come on now. I want answers.
Hi all,
ReplyDeleteHere's an interesting one for you.
An analysis of the current economic crisis we are all unfortunately facing but looked at from a slightly different perspective.
This analysis looks at past banking crises and how they have effected various aspects of the economy.
It is titled The Banking Crisis - Where are we now? (follow the link should you be interested) and has particularly interesting points about how the previous banking crises has effected assets including property prices.