July 2004

Does Federal Government Debt Affect Interest Rates?

The recent resurgence of federal budget deficits has rekindled debate about the effects of government debt on interest rates. While the effects of government debt on the economy can operate through a number of different channels, many of the recent concerns about federal borrowing have focused on the potential interest rate effect. Higher interest rates caused by expanding government debt can reduce business investment spending, inhibit interest-sensitive household spending, and decrease the value of assets held by households, thus indirectly dampening spending. Despite a substantial body of empirical analysis, the answer based on the past two decades of economic research is mixed.  AEI's Eric M. Engen and other economic experts reexamined the effect of federal government debt and interest rates at a July 8 conference.

Eric Engen
AEI

Past papers studying the relationship between government debt and interest rates have found mixed results, depending on the measures of interest rates and debt and econometric specifications they used.  Our paper uses a standard economic model to create benchmark empirical estimates of what the relationship is and then compares estimates across different measures of debt and real interest rates.

In a standard economic model of crowding out, increase in government debt would lead to an increase in real interest rates, though this effect might be dampened by increased private and foreign saving.  This model would suggest an empirical specification where government debt is related to real interest rates, though alternatively, a Keynesian model might suggest a relationship between the deficit and real interest rates.  We calculate from our standard economic model that an increase in the federal debt of 1 percent of GDP would lead to an increase in interest rates of 2.4 basis points, given no offset from private and foreign saving.  

There are several key facts about the market for debt that are relevant to understanding the effect of federal debt on interest rates:  First, the share of federal government debt as a share of all nonfinancial domestic debt is currently 18 percent, an historically low figure.  Also, international financial market integration has made the relevant market larger; foreign holdings of treasury debt has risen from 5 percent to 35 percent over the past thirty-five years, and real interest rates across countries have converged.  There is not a high correlation between real interest rates and federal debt outstanding for a given country; for example, Italy has a low real interest rate despite having a high level of debt.

For our empirical estimates, we use three different specifications: one estimates the effect of expected debt on forward-looking estimates of real interest rates, one estimates the relationship between expected debt and current real interest rates, and one estimates the relationship between current levels of debt and current real interest rates.  We also use other variables, including real GDP growth, real oil prices, a military spending dummy, and the equity premium.  We get lower estimates than Federal Reserve economist Thomas Laubach did in his study because we include these other variables.

We estimate that the effect of an increase in expected federal debt by 1 percent of GDP results in an increase in current and forward-looking interest rates of 3 basis points, an effect that we found to be statistically significant.  The effect of current federal debt on current interest rates was found to be statistically insignificant.

We are not saying that deficits do not matter.  The effect on interest rates is still significant: deficits have an impact on the economy beyond their effect on interest rates, and deficits cannot be run in perpetuity.  More importantly, how deficits are reduced does matter if the level of government spending ends up being inefficiently high or low. 

Peter Orszag
Brookings Institution

I want to make two basic points: one, that the focus of this paper is quite narrow, and there is a broader effect of fiscal imbalances beyond the effect on interest rates; and two, that beyond the rhetoric, there is a surprising degree of convergence of the estimates of the effect of deficits on the interest rates.

The key question regarding the effect of deficits on future national income is to what extent deficits reduce national saving.  As long as national saving declines, future national income will decline.  This can happen even if, as a result of increased foreign investment, interest rates do not rise-in this case, a greater portion of future returns to national investment will go to foreigners, thus still decreasing national income.

Engen and Hubbard argue that the theory says that debt--not deficits--is linked to interest rates.  I would contend that even neoclassical models could imply that deficits and the level of interest rates are linked.  Reassuringly, though, Engen and Hubbard's estimates based on the level of debt are consistent with other estimates based on deficits.

One would anticipate a relationship between long-term rates and expected deficits.  Though the literature seems to be mixed, when one considers only the results regarding expected deficits and interest rates, the result seems to be fairly consistently a positive relationship.  Though official and private forecasts are correlated with investor expectations of future deficits, they probably introduce a downward bias to the estimates found in past papers, which range about 30 to 60 basis points per 1 percent increase in sustained projected deficit as a percentage of GDP.  These estimates can be reconciled with the Engen-Hubbard result; a sustained increase in deficits by 1 percent of GDP would increase the debt by 10 percent and increase interest rates by 30 basis points in ten years.
 
In conclusion, the Engen and Hubbard results are consistent with results from past papers.  Deficits will reduce future national income if the world is not Ricardian.  So, for example, when considering the 2001 and 2003 tax cuts, one has to weigh the positive effect of tax cuts against the negative effect from the fact that they are deficit-financed and reduce future national income.  Most studies I have seen have found that the net effect of the tax cuts has been negative.

Darrel Cohen
Federal Reserve Board

Based on analytic-type arguments, the real interest rate only rises slightly with increased government debt, even if government debt crowds out real capital dollar-for-dollar. To put it another way, a 1-percent increase in the debt-to-GDP ratio, or a $100 billion rise in debt, only raises interest rates 2 to 3 basis points.

All analytic results flow from a stylized fact of the U.S. economy that over long periods of time, capital income is a fairly constant share of GDP [about one-third]. The analytic argument is then that since output per unit of capital cannot change much with GDP, the Marginal Product of Capital--the interest rate--changes even less. While this econometric result supports the analytic reasoning presently, it is not true for all cases. The results can vary according to where we are in the production function.

Prior literature has shown significant negative, zero, and positive impacts of fiscal policy on interest rates. The reason for these variations is that the results are sensitive to many variables, including the measurement of the fiscal variable, the presence of a lag-dependent variable in the regression, the expectations of the fiscal variable, control for government purchases, and open economy controls. Because of these variables, official fiscal forecasts beyond one or two years are generally inaccurate and cannot be used effectively.

Taking the Engen-Hubbard model and its discussion literally, a large change in debt-to-GDP ratio has little effect on interest rates but will have a sizeable effect on sustainable per-capita consumption. However, the Engen-Hubbard model is not the best to use for short- or intermediate-term analysis. In these periods, prior literature and the Laubach analysis show the Engen-Hubbard results to be less than robust.

AEI interns Philo Davidson and Bo Harrison prepared this summary.