Comparing Bonus Expensing Estimates of the Joint Committee on Taxation and the Tax Foundation
November 2, 2015
The Joint Committee on Taxation (JCT) has recently released its budget score of the House-passed bill to allow the permanent expensing of half of equipment spending by businesses. Fifty percent expensing means that half of the cost of a machine or vehicle may be deducted as a business expense in the year it is purchased, while the rest of the cost must be written off over the tax life of the asset.
In accord with the request of the House Ways and Means Committee, the JCT has provided an estimate of the economic benefits of expensing, along with a revenue estimate which takes the economic gains, and associated revenue gains, into account. Accordingly, JCT provided both the static revenue estimate, before economic adjustments, and the dynamic revenue cost, after economic adjustments. This is a very helpful presentation, and provides Congress with important information about the tax change.
The Tax Foundation has also issued an analysis of the economic gains and revenue cost of the expensing provision. We find that expensing yields about two to three times the economic gains as the JCT estimates over the budget window. Although our revenue estimate on a static basis is fairly close to that of the JCT over the period, our dynamic cost estimate is well below the JCT number, due mainly to our higher growth forecast.
Both models predict that the expensing provision will reduce the cost of obtaining and using equipment, and that it will raise worker productivity, wages, and employment. The JCT estimates that the increase in economic output will average about 0.2 percent over the budget window, and that the gains in wages and employment will be correspondingly small. The capital stock would rise by about 1.2 percent in the last half of the budget window. Nonetheless, by 2025, the added growth would be restoring over 40 percent of the static revenue loss from the tax reduction, cutting the JCT’s projected 2025 revenue loss from -$10.3 billion to -$5.9 billion.
The Tax Foundation estimate is for 1.2 percent additional economic output by the end of the budget window, after all adjustments, with a correspondingly higher rise in wages and employment. This would correspond to about a 0.55 percent higher average output over the period. The capital stock is projected to increase by 3.3 percent, almost three times as much as the JCT estimate. Toward the end of the period, annual revenues would actually be ahead of the baseline. Why the differences?
The reason for the dynamic economic and revenue differences between our two estimates stems mainly from differences in our economic models. One difference is that the JCT MEG model (Macro-Economic Growth model) assumes that only about 40 percent of businesses are currently profitable enough to take advantage of the provision. The TAG model assume 70 percent are able to use it. As the economy recovers from the great recession, more firms will use up their loss carry-forwards, and more firms will be in a position to use the expensing provision. We think that our assumption is in line with a normally functioning economy in the long term. In addition, the JCT MEG model contains several other features that act to dampen the investment stemming from a reduction in the taxation of capital.
First, the MEG model arrives at its forecast increase in investment indirectly, determining how the tax change alters the incentive to save, and deriving how much added capital can be sustained over time by the added saving. Its saving “elasticity” of 0.25 with respect to the after-tax return on capital investment suggests a lack of available savings to fund additional equipment purchases, and so it assumes only a small rise in the stock of equipment.
Second, although the MEG model is said to be an “open” economic model that accounts for international trade and capital movements, it may not adequately allow for substantial shifts in world saving from one county’s financial instruments to another’s.
Third, in some of its runs, it appears to assume that the added federal debt will in and of itself crowd out private borrowing or raise interest rates far more than the historical record suggests, or that the Federal Reserve will fight any economic expansion by tightening credit or driving up interest rates deliberately. This would particularly hit any portion of investment funded through debt.
By contrast, the Tax Foundation TAG model (Taxes and Growth model) calculates how much additional capital would become profitable under the revised tax rules, in much the same manner as businesses would consider whether the added investment was worth pursuing. It then assumes that the saving will be found to enable the investment to occur. This may happen gradually, but it will happen within a reasonable amount of time.
The TAG model assumes the amount of incremental investment is financed through added equity, which benefits from bonus expensing. It assumes steady Federal Reserve policy. It assumes no major increase in market interest rates beyond that which reflects the higher yield on plant and equipment due to the tax cut. It assumes a frictionless flow of saving across national borders. These account for the chief differences in the forecasts. The TAG model also has a slightly higher labor response to higher wages than the MEG model (a labor supply elasticity of 0.3 versus .19 in the MEG model’s best case), but that should account for less than 10 percent of the difference in the forecasts.
There are several sources of additional saving to support a higher stock of capital in the United States, even if the federal deficit rises. First, individuals and businesses we may consume less and save more out of the existing level of national income. MEG assumes a small amount of added private saving from these sources. In fact, the immediate effect of a rise in depreciation allowances and a reduction in business taxes is a dollar for dollar increase in business saving, until it is spent on investment or distributed to shareholders.
Second, we may save more due to the rise in wage and investment income that is generated by the growth in the economy. Note that the initial investment in new capital has to be financed with a new source of funds, but future replacement of the new investment is funded by its earnings. In fact, businesses will only invest in new projects that are expected to return enough to replace themselves. Since the TAG model predicts more income growth than the MEG model, it assumes more business and individual saving from these sources. Third, people may choose to work longer to earn more money to take advantage of the improved investment opportunities.
Fourth, and most importantly, more of the $20 trillion in annual world saving may be reallocated toward the United States. U.S. savers and lenders acquire hundreds of billions of dollars of foreign assets each year, and foreign savers and lender buy hundreds of billions of dollars of U.S. financial assets each year. There are also major cross border purchases of physical assets (direct business investment). A modest shift in the destination of existing world saving could finance a huge expansion of the U.S. capital stock, provided it becomes attractive to do so. The MEG model does not appear to be sufficiently open to the world economy to allow for such an influx.
As one historical example of this effect, the United States reduced tax rates and accelerated depreciation allowances in 1981, and began to restore the stability of the U.S. dollar by ending the 1970s inflation. Instead of lending to Latin America, Europe, and Asia, U.S. banks lent more at home. Lending by U.S. banks to foreign borrowers dropped by $100 billion (over 80 percent!) between 1982 and 1984, funding the tax cuts and a revival of U.S. business investment by keeping more of home-generated saving inside the United States. Later in the decade, increased foreign investment in the United States contributed further to the expansion.
For these reasons, the Tax Foundation TAG model projects a larger growth in the capital stock, labor productivity, wages, and jobs from a permanent extension of partial expensing than does the JCT MEG model, as well as more recovery of the initial static revenue loss. If the expensing provision becomes law, we look forward to comparing the real world results with these and other models’ forecasts. As we test and compare different economic and tax models over time, we can provide the Congress with improved information about the economic and budgetary effects of tax and spending changes. The JCT’s contribution to the discussion is vital, and we are grateful for their analysis and experimentation with economic analysis of tax changes.