Close

This page had been redirected to a new URL, please update any bookmarks.

Font Size: A A A

macroblog

« Just how out of line are house prices? | Main | "Secret loans" that were not so secret »

May 19, 2011

The long and short (runs) of tax reform

In an earlier part of my career, I spent a fair amount of time thinking about fiscal policy issues. The evolution of my responsibilities eventually moved my attentions in a somewhat different direction, but you never really forget your first research love. With questions of debt, government spending, and taxes at the top of the news, it isn't hard for my old fondness for the topic to reemerge.

So, in that context, I had a somewhat nostalgic response to an item at Angry Bear, written by contributor Dan Crawford. In essence, the Crawford post formally (through statistical analysis) asks the question "How is GDP growth related to marginal tax rates (that is, the tax rate applied to your last dollar of income)?" More specifically, Crawford analyzes how gross domestic product (GDP) growth next year is related to the marginal tax rate faced by the average individual and the marginal tax rate faced by the highest-income taxpayers.

I don't intend to quibble with the specifics of that experiment per se but rather highlight an aspect of taxation and tax reform that I think should not be forgotten. That is, the short run is no place for a decent discussion of tax policy to be hanging out. To say that more formally, the largest effects of tax policy accrue over time, and it is probably not a good idea to be too focused on the immediate—say, next year's—effects of any given policy or change in policy.

The following chart is based on tax reform experiments in a paper I co-authored over a decade ago with Alan Auerbach, Larry Kotlikoff, Kent Smetters, and Jan Walliser. (Note that the chart does not appear in the paper, but I created it using data from the paper—a publicly available version of the paper can be found here.) The chart depicts the cumulative percentage increases in national income that would be realized (in our model) in the years following three different tax reforms.

The three experiments depicted in this chart were as follows:

"[The clean income tax] replaces the progressive taxation of wage income with a single rate that is also applied to capital income. In addition, the clean income tax eliminates the major federal tax-base reductions including the standard deduction, personal and dependent exemptions, itemized deductions, the deductibility of state income taxes at the federal level, and preferential tax treatment of fringe benefits...


"Our flat tax experiment modifies the clean consumption tax by including a standard deduction of $9500. In addition, housing wealth, which equals about half of the capital stock, is entirely exempt from taxation."


Parenthetically, the "clean consumption tax"

"...differs from the clean income tax by including full expensing of investment expenditures. This produces a consumption-tax structure. Formally, we specify the system as a combination of a labor-income tax and a business cash-flow tax."


Finally,

"Our [flat tax with transition relief] experiment adds transition relief to the flat tax by extending pre-reform depreciation rules for capital in place at the time of the tax reform."


Here's what I want to emphasize in all of this. If the change in policy you might be considering involves a reduction in effective marginal tax rates (implemented via a combination of changes in statutory rates and adjustments in deductions and exemptions), the approach taken by Crawford in his Angry Bear piece is probably acceptable. The clean income tax reform is in the spirit of Crawford's calculations, and in our results the long-run impact on output is realized almost immediately. If, however, the tax reform involves changing the tax base in a fundamental way (in both versions of our flat tax experiments the base shifts from income to consumption), then the ultimate effects are felt only gradually. In our flat tax experiments, the longer-run effects on income are in the neighborhood of three times as large as the near-term effects.

All of the experiments described were done under the assumption of revenue neutrality, so questions of the right policy for budget balancing exercises weren't explicitly addressed. (Nor is it the nature of the experiment contemplated in the Angry Bear post.) Nonetheless, they do suggest that deficit reduction exercises that involve changes in tax rates and the tax base will have differential effects over time, and realizing the full benefits of tax reform may require a modicum of patience.

Note: A user-friendly description of the paper that the chart above is based on appeared in an Economic Commentary article published by the Cleveland Fed.

Update: Though the item at Angry Bear was posted by Dan Crawford, Mike Kimel actually wrote it. I apologize for the mistake and draw your attention to Kimel's follow-up post.


Dave Altig By Dave Altig
senior vice president and research director at the Atlanta Fed

May 19, 2011 in Deficits, Fiscal Policy, Taxes | Permalink

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d8341c834f53ef01538e959d51970b

Listed below are links to blogs that reference The long and short (runs) of tax reform:

Comments

Dan Crawford posted the analysis. Mike Kimmel wrote it. Just thought I should mention it.

Posted by: kharris | May 20, 2011 at 10:39 AM

So when does your theoretical chart reflect the dramatic plunge that has occurred in real nations with flat taxes (e.g. Iceland)? Must be in year 14, right?

Posted by: Devin | May 20, 2011 at 11:36 AM

Minor correction: Dan posted it, but Mike Kimel actually wrote it.

Posted by: Ken Houghton | May 20, 2011 at 12:20 PM

Dave,

Hi. Thanks for mentioning the post. I modified it to take into account some of your comments. (New version here: http://www.angrybearblog.com/2011/05/tax-rates-and-economic-growth-over-ten.html)

Now I use the tax rates in any given year to explain annualized growth rates over the subsequent ten years. Additionally, I've included quadratic forms of both the top and bottom rates, which allows me to compute growth maximizing rates at both ends of the scale.

It turns out that the growth maximizing top marginal rate isn't much changed from the first post (i.e., 67%), but the optimal bottom marginal rate is zero. I think that indicates that using historical US data at least, a flat tax is a bad idea, as the top and bottom marginal rates would be identical if rates were flat.

With respect, given the choice between the outcomes predicted by a simulation, and historical outcomes, I'd go with the historical outcomes.

Best regards.

Posted by: Mike Kimel | May 24, 2011 at 07:23 PM

Dave, I would be interested is learning more as to why this is the case...i.e. why would a change into a flat tax have minimal short-term effects (yet meaningful long-term effects)?

Posted by: Chris | May 26, 2011 at 07:45 PM

Post a comment

Comments are moderated and will not appear until the moderator has approved them.

If you have a TypeKey or TypePad account, please Sign in