Printer Friendly

Interest rates, expected future deficits and tax proposals.



In several recent publications Paul Evans [1985 1987 1988a 1988b! presents evidence suggesting a negative wealth effect in response to federal deficits. Evans [1987! investigates the impact future deficits have on current interest rates. While he argues that expectations of future deficits are based in part on past history, he also recognizes that economic agents could "...observe the legislative process and assess how likely Congress is to enact new tax legislation appreciably affecting future budget deficits" [1987, 45!. Evans assumes that legislative information is both relevant and available but does not include it as a regressor. Instead, he looks to the error term and tries to unpack the effect of tax legislation. A more straightforward method for testing this hypothesis is to simply include a tax legislation variable in the estimated equation.


The reduced form equation to be tested is [[Delta!I.sub.nt! = f ([[Delta!I.sub.t-j!, [[Delta!G.sub.t-i, [[Delta!M.sub.t-i, [[Delta!T.sub.t-i!) where I is the nominal interest rate on an n period bond, G is government purchases of goods and services, D is the federal deficit variable and is simply the difference between federal outlays and federal receipts, M is the money stock (M1), T is a variable that reflects tax legislation currently being considered by Congress, j = 1, 2, 3 and i = 0, 1, 2, 3. By incorporating the tax variable into the relevant information set, the impact of tax legislation, if there is one, may be detected. (1)

The tax variable is constructed in the following manner. MAjor federal tax proposals enacted during the estimation period are taken from Pechman [1987!. The proposed revenue impact figures that were used by Congress while these bills were being considered are included in Congressional Quarterly Almanac along with a description of each tax proposal. This information is treated as the best guess as to the economic impact of these tax proposals and is the information available to the market. It is these estimated revenue impact figures that are included in the tax variable (T).

Many of the tax proposals during the sample period were multi-year and their estimated revenue impacts extended over that time period. In addition, some tax proposals overlapped, so that their individual impacts must be aggregated. As an example, the Excise Tax Reduction Act of 1965 provided a series of cuts over three and one-half years. The estimated impact for 1966 (the second stage of the act) was that revenue would fall by $1.6 billion, or $.4 billion per quarter. At the same time, the Tax Adjustment Act of 1966 sought to increase taxes to dampen an inflationary trend brought on by the financing of the Vietnam War. Revenues were expected to increase by $1.13 billion during the second quarter of 1966 and by $1.2 billion per quar ter over the 1967 fiscal year. The combined effect of these two proposals during 1966 is illustrated in Table I.

The ta x variable is the expected change in the deficit for the succeeding quarter. In 1966: I the deficit was +0.6 billion and, ceteris paribus, could be expected to fall $.73b the next quarter due to imminent tax changes. The tax variable gives an indication of the future path that the deficit will take. (2)


Table II shows the results obtained from the above equation using quarterly data from 1959:I-1984:IV. (3) Both a short- and long-term interest rate are used in this study. (4) The short-term rate (S) is the yield on newly issued taxable three-month U.S. Government securities, and the long-term rate (L) is the yield on U.S. Treasury bonds due or callable in ten years or more. The lag lengths used, two and four lags, coincide with those used by Evans [1987!, as does the differenced data. And finally, Evans argues that the sums of the coefficients of individual variables provide the most meaningful information consequently those are presented.

The results presented confirm Evans's preliminary findings. Many of the coefficient sums are insignificant. Additional results of note include the following. The coefficient sum on the deficit variable is both negative (as is Evans's) and statistically significant in all cases, suggesting that larger deficits are indeed associated with falling interest rates. This clearly provides no support for conventional macroeconomic theory.

The tax variable also supports the negative wealth effect. Tax cut proposals are included as negative numbers, so a negative wealth effect would imply a positive coefficient all of the sums on the tax variable are positive, and two of the four sums are significant at the 10 percent level. This is consistent with the impact of deficits. A tax cut would lead people to expect higher deficits and falling interest rates (given the above results). If economic agents expect future short-term rates to fall, then via the term structure of interest rates, current long-term rates will also fall. If bonds of different maturities are substitutes, then current short-term rates should also fall as investors substitute the relatively cheaper asset.

One of the more interesting results of both Evan's study and this one is the negative relationship between deficits and interest rates. Barro [1987! provides a possible explanation for these unusual results, given that most of the tax changes used in this study involve income taxes. A cut in today's tax revenue, which implies a higher deficit tomorrow and, according to Ricardian Equivalence, a higher tax rate in the future, will motivate people to work more today and less in the future. This increase in work effort will lead to an excess supply of goods and, consequently, a fall in the interest rate. Richard Sweeney [1988! provides another possible explanation. He argues that introducing uncertainty about taxes leads to a reduction in aggregate demand. An issuance of bonds is, therefore, contractionary and interest rates must fall to restore equilibrium. Fremling and Lott [1989! give yet a third possible theoretical justification. By including deadweight losses associated with income tax changes, they argue that rational individuals will save not only for future additional taxes, but also for the associated deadweight losses. It is possible, therefore, for the additional saving to exceed the deficit and lead to falling interest rates.

The money variable is insignificant three out of four times. The consistently positive coefficient indicates that any liquidity effect is short-lived and that the income and price effects quickly dominate. A more interesting puzzle is the government purchases-interest rate correlation. The coefficient is significantly only once, but it is consistently negative, implying that increases in government purchases, if they have any effect at all, actually lower short-and long-term interest rates. Barro's above argument might serve to explain these unusual results. If an increase in government spending is accompanied by an expectation of rising income tax rates, people might substitute work today for work tomorrow, increasing supply and reducing interest rates.


To be sure, the results presented in this paper do not conform to traditional macroeconomic theory. For this reason alone, the negative relationship between budget deficits or government purchases and interest rates should be an important topic for further analysis.

(1.) The government spending variable, the money stock and the deficit are all taken from National Income and Product Accounts data. These variables along with the tax variable are converted into real terms by dividing by the GNP deflator. In addition, all regressors, except the lagged interest rates, are scaled by natural real GNP on the assumption of homogeneity. Both the GNP deflator and natural real GNP are taken from Gordon [1990!.

(2.) It is certainly possible to construct the tax variable in alternative ways. One would be to represent the various stages of a tax proposal in a matrix rather than as a single vector. The problem here is that the information available on estimated revenue impacts is inconsistent sometimes one year is estimated, sometimes five years. Aggregating the various stages of a tax proposal (in net present value) could also be done although the problem just mentioned still exists. Treating the tax variable as a single vector capturing next quar ter's expected effect seems as appropriate as any other method and eliminates some obvious problems by simply cutting off the estimated impact after the current quarter. Finally, the tax legislation variable is assumed to be used by the market immediately following House passage and through the quarter before actual implementation.

(3.) The sample period begins with 1959:I due to the Fed's policy of pegging interest rates between roughly 1930 and 1950.

(4.) Both series are from the Board of Governors of the Federal Reserve System and are taken from The Survey of Current Business.


Barro, Robert J. Macroeconomics. New York: Wiley, 1987.

Congressional Quarterly Almanac. Washington D.C.: Congressional Quar terly, Inc., 1959-84.

Evans, Paul. "Do Large Deficits Produce High Interest Rates?" American Economic Review, March 1985, 68-87.

_____. "Interest Rates and Expected Future Budget Deficits in the United States." Journal of political Economy February 1987, 34-58.

_____. "Are Consumers Ricardian? Evidence for the United States." Journal of political Economy, October 1988a, 983-1004.

_____. "Are Government Bonds Net Wealth?" Economic Inquiry, October 1988b, 551-66.

Fremling, Gertrud M. and John R. Lott. "Deadweight Losses and the Saving Response to a Deficit." Economic Inquiry, January 1989, 117-29.

Gordon, Robert J. Macroeconomics. Glenview, Illinois: Scott, Foresman, 1990.

Pechman, Joseph A. Federal Tax Policy. Washington, D.C. The Brookings Institution, 1987.

Sweeney, Richard J. Wealth Effects and Monetary Theory. New York: Blackwell, 1988.

U.S. Department of Commerce Bureau of Economic Analysis. Business Statistics: 1986. Washington D.C.: U.S. Government Printing Office, 1987.
COPYRIGHT 1991 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Johnson, Colleen F.
Publication:Economic Inquiry
Date:Jan 1, 1991
Previous Article:In search of Giffen behavior.
Next Article:Extending the equation of exchange: dual equations, for output and input.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters