“Buffett Rule” Would Cause Marginal Tax Rate of 90%
February 13, 2012
Correction: This blog post is based on a misreading of the bill, and is now known to be incorrect. The Buffett Rule as proposed by Sen. Whitehouse would not cause marginal rates of 90%. See this post for more details. I will leave the original post below, but retract all the claims I make in it.
The Obama administration released today its formal budget proposal for 2013. Included in this proposal are a variety of deficit reduction measures, of which tax increases form a significant part. Thankfully, the “Buffett Rule” – a new minimum tax for taxpayers earning more than $1 million, requiring them to pay at least 30% of their AGI (after charitable donations) regardless of any other deductions or preferential rates, is not formally part of the budget proposal, but rather a guiding principle for future tax reform. For this budget, the Buffett rule is best thought of as a policy goal rather than a specific proposal, and that’s a good thing because any attempt to literally implement the Buffett Rule would be disastrous. Nevertheless, Sen. Sheldon Whitehouse (D-RI) recently did just that.
Among the many problems with the Buffett rule (as Sen. Whitehouse proposes it) is the way it “phases in.” For sensible reasons, the minimum tax requirement doesn’t suddenly activate as soon as a taxpayer reaches $1 million in income – if it did, there’d be a huge incentive for taxpayers to keep their income under the threshold, as going above it causes a significant decrease in after-tax income. Things like this are known as “cliffs” in the tax code, and policymakers generally try to avoid them. The proposed solution is to have the rule gradually phase-in over the $1 million to $2 million range. The minimum required rate starts at 0% at $1 million dollars and rises to 30% at $2 million in a straight line. For example, a taxpayer making $1.6 million dollars is 60% of the way from $1 million to $2 million, so his or her minimum rate would be 30% times 60%, or 18%. In practice, taxpayers in this income range tend to pay effective rates of around 25%, so the Buffett rule would typically affect only taxpayers whose incomes are high enough to require at least that minimum rate (about $1.8 million.)
The problem with this formula is that a taxpayer inside the phase-in range who owes the minimum tax is subjected an extraordinarily high marginal tax rate, which is the rate on his or her “top dollar” of income, or the tax on any additional income earned. Consider a taxpayer who earns exactly $1.8 million and is subject to the new minimum tax. In this case, the minimum is equal to 80% x 30% x $1.8M: $432,000. Next, assume he gets a $10,000 bonus, so that his income is now $1.81 million. The minimum is now 81% x 30% x $1.81M: $439,830. An income increase of $10,000 has led to a tax increase of $7,830 ($439,830 – $432,000) – in other words, the marginal rate on that additional $10,000 is 78.3%. It can be shown mathematically that the marginal rate in the phase-in range is exactly equal to:
60% x (Adjusted Gross Income/$1,000,000) – 30% 
The rate gets as high as 90% at the upper end of the phase-in range before dropping back down to 30% for as long as the taxpayer is affected by the minimum tax. Combine this with state and local income taxes, as well as the new 3.8% tax on investment income beginning in 2013 and the marginal tax rate on investment income could plausibly exceed 100% within a narrow range. In other words, taxpayers in this range would have an incentive to lose money. (Update: it occurs to me that the 3.8% surtax, despite being technically implemented as a hospital insurance payroll tax, would still apply towards the minimum, so there’s no way it would be added on top of the Buffett Rule marginal rate. However, a handful of states have marginal tax rates of 10 to 11%, so exceeding 100% wouldn’t be out of the question.)
This state of affairs is almost certain to produce hugely distortionary effects, and there’s no easy way that a direct, literal implementation of the Buffett Rule can get around this problem unless it uses such a long phase-in range that it barely affects anyone at all.