A Pigovian Income Tax

In a shameless bid to join Greg Mankiw's Pigou Club, I've devised a Pigovian income tax. I first thought about this a couple weeks ago after Professor Mankiw proposed a new metric for computing average marginal tax rate. In a follow-up post a reader suggested an improvement to the metric and dubbed this the "Mankiw effective tax rate".

At that phrase, my mind wandered. I began to think about what income tax policy Professor Mankiw would prefer. Readers of his blog would instantly know that he would prefer a Pigovian income tax, but what exactly is that. Let's start with a definition

Pigovian tax: (n.) a tax levied to correct the negative externalities of a market activity

This begs the question: what are the negative externalities of our current income tax policy? Obviously, the answers to that question are debatable. Here are the two negative externalities that I'll try to address:

1. Free riding (The poorest people pay no income tax)
2. Greed (The rich do not "pay their fair share")

Free riding and greed are related. The tax burden in modern America falls almost exclusively on the high income earners. The top 50% of earners pay 97% of income taxes. The top 5%, pay 60%. We're already soaking the rich. The bottom 40% of wage earners pay no taxes. They're free riding.

A corollary to both free riding and greed is class warfare. If the tax rates of the low and high wage earners could be linked, we could address not only greed and free riding, but also class warfare. So, my Pigovian income tax would tightly couple the interests of high and low wage earners. I would accomplish this by making the tax rate for the top wage earners a multiple of the tax rate for the bottom wage earners. For example, if we chose a multiple of two and set the bottom tax rate to 20%, then the top tax rate would be 40%. Let's apply the 20% rate to the bottom quintile and the 40% rate to the top quintile.

That leaves the middle three quintiles up for grabs. Tax brackets could be divided up differently and probably would be in the real world. Still, quintiles offer a simple way to illustrate my point. If we used a typical progressive scheme, we would wind up with progressively higher tax rates, say, 25%, 30%, and 35% for the middle quintiles.

I would prefer a tax policy that incentivizes productivity, so I will examine a policy of declining tax rates, say, 12%, 6%, and 0%. Let's consider these tax rates, but replace the quintiles with household income ranges so we can see what this might look like with real numbers.

Income Range	Tax Rate	Revenue Est.
$0 to $20,000	20%	$51,401,950,000
$20,000 to $40,000	12%	$114,728,120,000
$40,000 to $62,000	6%	$155,023,760,000
$62,000 to $100,000	0%	$171,086,480,000
$100,000 and up	40%	$776,028,380,000
Total:		$1,268,268,690,000

Note 1: I assume 22.25 million households per quintile which is slightly higher than the 2007 number implied at answers.com.
Note 2: Quintile ranges are based on data from taxpolicycenter.org. I used the mean income data there to calculate my revenue estimates.
Note 3: The taxpolicycenter.org also implies that total income tax revenue is about $1.25 trillion.

Are you shocked at that 0% rate for the upper middle class bracket? It has a couple of nice side effects. First, it clearly marks the transition to the higher "greed" tax. Second, it reduces the burden of filing tax returns for a quintile that is likely to have numerous deductions. If you're at the upper end of that quintile, there's no reason to itemize unless you can muster almost $40k in deductions. This would also apply, though to a lesser extent, with the middle class bracket of 6%. In that case, filers would have to decide if the burden of documenting a deduction was worth the 6% savings that it would net.

This incentive structure is the inverse of what currently exists. Because each tax bracket is at a higher tax rate, the incentive to itemize increases as earnings increase. I suspect that discouraging itemization (as my proposal does) will reduce fraud and raise revenues.