Comments and Discussion

Eric M. Engen: Government debt and deficits matter. Although some politicians may argue that deficits do not affect the politics of public policy or voters' behavior, government borrowing does affect the economy. The disagreements among economists concerning government debt and deficits are primarily over the channels through which the effects occur and the magnitude of those effects.

In this paper, William Gale and Peter Orszag review past and expected future federal government borrowing and present a theoretical summary of some of the different ways in which government debt can affect the economy. They also contribute to the body of research on the economic effects of government debt by providing new empirical estimates of two channels through which these effects operate. First, they estimate the degree to which private domestic saving may rise coincidently with increased government borrowing. Second, they estimate the effect of federal government debt and deficits on interest rates. My discussion will focus first on their empirical analyses of the two channels and then turn to their more general discussion of the economic effects of government debt.

The first part of the authors' empirical analysis estimates both an aggregate consumption function and an Euler equation specification for changes in aggregate household consumption spending. In both specifications they estimate the impact of changes in debt-financed government taxes while controlling for government spending, (lagged) government debt, marginal tax rates on capital and labor, and other economic factors. This empirical analysis contributes to the literature that studies whether households are more Keynesian or more Ricardian in their reactions to government borrowing, that is, whether households view government bonds issued because of a tax cut as net wealth or as future tax obligations.

The authors' results, using several different empirical specifications, imply that households are neither purely Keynesian nor purely Ricardian. In most of their regressions they find that, in the short run, a statistically significant portion of a debt-financed tax cut is saved, but their estimates of that response cover a fairly broad range. In their aggregate consumption function specifications, with federal spending held constant, a one-dollar decrease in federal tax payments is estimated to increase consumption spending in the short run by 30 to 46 cents, implying that households offset 54 to 70 percent of the increase in federal borrowing by saving more. The Euler equation estimates yield a much broader range of results, which depend crucially on whether changes in consumption are measured in levels or as a ratio to net national product, whether controls for marginal tax rates are included, and whether explanatory variables are treated as endogenous. In these estimates, with federal spending held constant, a one-dollar decrease in federal tax payments is estimated to increase consumption spending in the short run in the range of 22 to 98 cents, implying that household saving offsets anywhere from 2 to 78 percent of the increase in federal borrowing.

Although Gale and Orszag state that their preferred estimates suggest that increases in private saving offset a much narrower 20 to 50 percent of the increase in federal borrowing from a tax cut, it is difficult a priori to rule out any of their specifications, and readers may have their own preferences.¹ Despite the authors' improvements over many previous econometric analyses of this issue, their broad range of estimates provides little assurance that this analysis advances the consensus on the short-run magnitude of this effect, even if we can safely rule out that households do not appear to be purely Keynesian or purely Ricardian. Moreover, as the authors note in discussing their results, these estimates do not address the potential long-run effect of deficit-financed tax cuts on private saving, which could differ from the short-run effect.

The second part of Gale and Orszag's empirical analysis investigates the effects of federal government debt and deficits on interest rates. Their empirical research here is similar to recent studies by Thomas Laubach and by Glenn Hubbard and myself.² In general, their estimates of the effect of the federal debt or deficit on the level of the...