Search This Blog

Thursday, April 20, 2006

Tax Cuts and Revenue Gains

A tax rate cut was enacted in the United States in 2003, yet tax revenues from capital gains were better than predicted. Many supply siders and Laffer curve enthusiasts are loving it. For example, Don Luskin wrote in the National Review:
[I]nstead of costing the government $27 billion in revenues, the tax cuts actually earned the government $26 billion extra.

CBO's estimate of the "cost" of the tax cut was virtually 180 degrees wrong. The Laffer curve lives!

Unfortunately, while the numbers Don refers to are correct, his analysis has some serious problems.

Don't get me wrong, I'd love it if Don's analysis was correct. I'm all for low tax rates because in the long run, due to the increased economic growth that coincides with lower taxes, everybody will be better off, rich and poor alike. Even the government will have more to spend in absolute terms.

There are several things in Don's analysis that have problems, but in this post I will focus on the biggest, yet simplest, problem: you can't compare predicted revenues with later actual revenues unless you have a high degree of confidence that the predictions are accurate. That simply isn't true for CBO predictions of revenues from taxes on capital gains.

First of all, the CBO knows that it has a terrible record predicting such revenues. On page 56 of "THE BUDGET AND ECONOMIC OUTLOOK: FISCAL YEARS 2001-2010" the CBO states:
Capital gains realizations, which are often considered relevant to the accuracy of forecasts, are notoriously difficult to predict. They constitute a relatively small percentage of individual tax receipts, however, and errors in forecasting them are unlikely to play a large role in errors in revenue forecasts.
In other words, they know that they're very bad at predicting capital gains revenues and they're not going to try to improve. And true to form, the CBO's predictions have been truly terrible. As and example, the following table shows the CBO's prediction of capital gains liabilities for 2001 through 2005 and the actual revenues:
Year   Predicted   Actual
-------------------------
2001: 129 61
2002: 95 47
2003: 54 47
2004: 46 71
2005: 58 81
The estimate for the predicted revenues was reported at the beginning of the year. The actual revenues were reported two years after that to allow the dust to settle, except for 2005 which was reported at the beginning of 2006. As you can see, being off by a factor of two is par for the course.

Therefore, you can assign very little meaning to the fact that revenues after the tax rate cuts were higher than predicted by the CBO.

3 comments:

Oroborous said...

The thing about the Laffer curve is that it only works if marginal tax rates are already high, and get cut substantially.

The effect is very small if fairly low taxes are reduced slightly, so there's some point at which it doesn't pay to reduce taxes - behaviors won't change.

Anonymous said...

If you look at the numbers in your table, you will see that the CBO errors are not random. They consistently overestimated revenues for higher capital gains rates (2001 & 2002)* and consistently underestimated revenues for lower capital gains rates (2004 & 2005).

Five data points does not prove a theory, but I bet the CBO model assumes that investor behavior does not change in response to changes in tax rate.

Oroborous' point that behaviors won't change in response to slight changes in low rates is, of course, true. Laffer pointed out in his original article that if rates are 0%, then revenue is zero, and if rates are 100% then revenue is also zero. The question is: at what rate is tax revenue maximized? My guess is that revenues will maximize between 10% and 20% rates because higher rates encourage people to make inefficient decisions to minimize taxes.

*2001 was taxed at Clinton rate (28% for individuals, 35% for corporations), Bush tax cut became fully effective with capital gains rate of 15% on May 5, 2003. Note the CBO predicted a 50% decrease in captial gains revenue with a 50% cut in rate, actual was a 20% decrease, followed by a 50% growth.

Bret said...

David Rothman:

It's fairly tedious to gather the predicted versus actual data, which is why I did it only for the five years, but a quick glance leads me to believe that the CBO has been way off, in both directions, even when there were minimal tax law changes between years. That's why they include their disclaimer.

There is so much complexity that it's extremely difficult to say which component, if any, of the change in capital gains revenues are due to the Laffer curve effect. For example, perhaps small business owners are now paying themselves lower salaries and routing more of the money to themselves via dividends and/or having the corporation buy back shares of their stock. That would increase capital gains taxes while lowering overall taxes. There are numerous effects like this that can't be separated out.

Lastly, in an upcoming post, I'll put forth why I think the Laffer curve is good for illustration, but not really applicable in the real world.