Responding to a Critique of the Taxes and Growth Model
December 4, 2017
A recent piece on The Washington Post’s Wonkblog makes several claims about the Tax Foundation’s Taxes and Growth Model that omit important information about the model’s economic assumptions and present important economic debates with too little nuance or context.
We welcome critiques of our model and, contrary to the implication in the article, strive to make the methodology and underlying assumptions publicly available to further the economic modeling debate. Despite sending nearly 1,500 words answering the writer’s various questions, the article fails to provide a fair assessment of how the TAG model works. Below we provide clarifications and additional information to counter some of the assertions in the piece.
Compares scores of different tax bills: The article says that the TAG Model finds “a growth effect that is four to 12 times bigger than in mainstream models.” The growth numbers cited in the piece are from the Tax Foundation’s score of the Senate bill as introduced, which the author then compares to scores of the amended bill passed by the Senate Finance Committee, as produced by the Tax Policy Center, Penn-Wharton, and Joint Committee on Taxation. The legislation as passed by the committee contains substantive differences that would certainly have altered our scoring of the proposal. The author did not ask us for our thoughts on how the revised Senate bill would have fared under our model. We certainly would have provided that information if asked.
Clarifying how our model treats the U.S. as a “small, open economy”: The article criticizes our model for treating the U.S. as a small open economy instead of a large one. The fixation is on the terminology, rather than its meaning. A “small open economy” can either be jargon shorthand for a country which has only a small weighting in the average of world interest rates, or, in a different sense, whose incremental borrowing has little impact on world interest rates. One should not confuse the two meanings. In calculating the U.S role in making up an average world interest rate, we are obviously a large country. But the latter issue, how added U.S. borrowing would raise interest rates at the margin, is what matters for economic forecasting. It is not the size of the economy that matters, it is the size of the borrowing compared to the amount of saving in the world.
Too little analysis on the “crowd out” effect: A main reason for the larger growth effects in our model than others is how the model treats an increase in the federal deficit. The models used by TPC, JCT, and Penn-Wharton assume that an increase in government borrowing due to higher deficits will “crowd out” savings that would have instead gone toward private sector activity. This is an important discussion and many economists are now revisiting the issue. This Wall Street Journal piece from October 2016 goes into great detail on the current debate. We’ve written about the crowd out impact on many occasions, including this post comparing our results to those found in the Penn-Wharton model. Unfortunately, this Wonkblog article dedicates only one paragraph to the crowd out discussion. We feel that this economic assumption deserves a much more robust debate, particularly since the piece says that our model is “outside the mainstream” and “seems to come up with some ad hoc justifications for why things that normally don’t go together should here.”
We estimate that borrowing by the U.S. government due to the tax bill will rise by about $1 trillion over the next decade. The World Bank projects that global saving over the period will total about $250 trillion. Adding $1 trillion to the demand for funds would be only four-tenths of a percent of the total, which would be expected to raise global interest rates by no more than the same ratio, or about two basis points on a 5% bond. An increase in interest rates from 5% to 5.02% would have no measurable effect on the expansion of the U.S. economy. (Basic arithmetic dictates that bond prices and interest rates move in opposite directions. Northwestern University Professor Robert Eisner taught us that an equal but opposite percentage change in bond prices and interest rates would “make room” in the amount of desired asset holdings for the added debt to be issued and accepted.)
The amount of borrowing attached to the congressional tax bills is a drop in the bucket, and too small to move global interest rates. In that sense, we can and should treat the added borrowing as being done by a small open economy. Yes, the United States is bigger than Ireland, as the article notes. Yet the effect of the borrowing on global interest rates depends on the amount of debt issued, not on where or by whom the bonds are sold, provided the issuers are of equal creditworthiness. In fact, one might expect less rise in interest rates on U.S. debt than on Irish debt if both countries tried to borrow an additional trillion dollars.
Modeling the estate tax: The Tax Foundation recently responded in great detail to a critique of how our model treats changes to the estate tax, among other things. This post provides information on the discussion, which we believe comes down to different opinions on proper economic assumptions. Again, as with crowd out, the Wonkblog piece only includes one paragraph on this point focused on the “small, open economy” discussion noted above. Again, far too little information is included to present a balanced discussion, despite the explanations that were provided to the writer but not included.
For these reasons and others, we’re disappointed not that an article critiques the TAG Model or that other economists have differing opinions, but the one-sided way in which these issues were presented. Not only were we not given a chance to fully respond to the accusations listed, but the piece omitted details and analysis that we believe are crucial to an informed debate about taxes and the economy.
Better Understanding the TAG Model’s Assumptions
Our model is based on “comparative statics” analysis. It seeks to predict how much larger or smaller the economy would be under an alternative tax system. The prediction is relative to whatever else the economy might be doing over the period. The change in the ultimate level of output can be determined by noting how much the proposed tax change alters the profitability of investment or the reward to labor, compared to what they now are.
However, we do not predict the precise dollar level of output in the long run, because we do not presume to predict what the economy would do under current law. No one knows that for sure, and any claim to the contrary is false precision. Consequently, we do not create our own “baseline” to set the year to year level of output going forward. For illustration, we use the CBO baseline forecast, but only because Congress finds that convenient for its work. Readers may pick their own baseline assumptions, with our blessing, if they do not like the CBO projection.
Nor do we attempt to predict the exact quarter-to-quarter path of the adjustment to the alternative level after the tax change. We have noted that most historical instances of major tax changes have been followed by a return to normal levels of compensation for investment within a decade, and sometimes sooner. Consequently, we show a standardized geometric adjustment path over ten years. A very big reform might take the full decade, or even longer. A small change might take less time. We hope to research the historical experience in more depth over time. Right now, our interest is in the ultimate, permanent change in the level of activity, not the transition. It would be nice to be able to predict the transition path with precision, but we think there are far too many factors affecting the economy in the short run for any such predictions to be possible or useful.
Some commentators have suggested that they “do not understand” our model. We have been more transparent than most in posting our equations and the sources of our data, and our model’s calculations of the cost of capital and its effect on output are squarely in the mainstream. It may be that the professed confusion is not about how our model works, but amazement that we do not follow the procedures and assumptions popular in other, older models of the economy. The reason is that we have studied the older methods, and find them wanting.
For example, models that try to predict future behavior by the Federal Reserve, and assign to it an influence over the real economy after a decade or more, are simply unsupportable. They are decided by an ever-changing group of individuals who may react differently under different circumstances. Furthermore, as the Chicago School demonstrated beyond any doubt as long ago as the 1960s and 1970s, the Fed’s long-term influence is primarily on the price level, not on real output. The Federal Reserve can, and often does, create short-run disruptions in the real economy, but these wear off over time as people come to adjust to the resulting changes in the price level. The “Phillips Curve,” a device that asserts that inflation can permanently reduce unemployment, or that growth causes inflation, is dead in respectable academic circles, although it is still found at the heart of the Federal Reserve’s economic model. Realistically, the Fed’s effect on the real economy a decade or more from now is unknowable, and irrelevant as to how a tax change affects long-run economic output and income. That is why we do not follow the practice at the Joint Tax Committee of generating a range of outcomes assuming a range of Federal Reserve behaviors.
Some models of the economy assume that any expansion of capital formation in the United States is either entirely or largely limited to the use of U.S. saving, and that government borrowing crowds out U.S. investment nearly dollar-for-dollar. Such models that assume that the United States is cut off from the saving and capital markets of the rest of the world are behind the times, and far off base. In the 1950s, Europe and Japan were still struggling to recover from World War II, and the Eastern Bloc and China effectively cut off from global markets. The United States was the dominant factor in the free world economy. More capital flowed out of the United States to aid the growth of Europe and Latin America than flowed in. U.S. economists could get good results predicting the effects of U.S. domestic policy on the U.S. domestic while ignoring the rest of the world. Those days are long past. Capital flows are now bidirectional, the U.S. goods and capital markets are far more integrated into the world system, and the Eastern bloc and China are now significant parts of the global economy.