Journal Issue
Share
Article

United States of America: Selected Issues

Author(s):
International Monetary Fund
Published Date:
August 2003
Share
  • ShareShare
Show Summary Details

IV. Domestic and International Impact of U.S. Budget Policies1

1. The U.S. fiscal position has deteriorated significantly in recent years. In 2000, the Congressional Budget Office (CBO) projected surpluses in the range of 3 percent of GDP for the next ten years and for the federal debt to be nearly paid down by 2010. Since then, partly owing to the economic downturn, but mainly reflecting policy initiatives to boost spending and cut taxes, the budgetary balance has swung into substantial deficit. The fiscal deficit seems likely to reach around 4 percent of GDP in FY 2003 and FY 2004 and remain significant well into the future.

2. The turnaround in the fiscal situation—and the Administration’s call for further tax cuts—has revived the long-standing debate about the macroeconomic impact of fiscal policies. On the one side has been the view that tax cuts generate positive supply-side benefits sufficient to offset the adverse effects of higher fiscal deficits on interest rates and lower private investment (CEA, 2003). Others, however, have questioned the size of the supply-side benefits, and have argued that higher deficits would ultimately lower output.

3. The discussion in this chapter tends to support the view that budget deficits have adverse effects in the longer run, both domestically and abroad. In particular, model-based simulations on the Administration’s FY 2004 budget proposals, as well as a review of the recent crowding-out literature, suggest that recent U.S. fiscal policies would boost output in the short run, but larger deficits would tend to cause interest rates to rise above and output to fall below their baselines in the longer run.2 Moreover, empirical evidence suggests that higher levels of U.S. public debt would cause an increase in global interest rates, illustrating the potential for adverse spillovers from U.S. fiscal policies.

A. The Historical Experience and Simulations of the FY 2004 Budget Proposal

4. Previous episodes of large fiscal expansion in the United States raise questions about the effectiveness of fiscal stimulus. There are three recent cases when the federal fiscal balance fell by at least 1½ percent of GDP, in cyclically adjusted terms, over a two-year period (Figure 1).

Figure 1.Two-Year Changes in Structural Fiscal Variables

During 1965–67, structural outlays rose by 1.8 percent of GDP, mainly owing to military spending associated with the Vietnam war. Tax cuts caused structural revenues to fall by 1.1 percent of GDP between 1974–76 and by 2.3 percent of GDP during 1981–83. Although the tax cuts were associated with some acceleration in real GDP growth over each of the subsequent three years, real GDP growth declined, or remained essentially unchanged in each of the subsequent ten years (Table 1).

Table 1.Change in Real GDP Growth Before and After Large Fiscal Expansions(in percent)
1965–671974–761981–83
Post 3-year average less
pre 3-year average-2.11.73.3
Post 10-year average less
pre 10-year average-1.1-0.30.3
Source: Fund staff estimates.
Source: Fund staff estimates.

5. Simulations of the FY 2004 Budget proposal also suggest that the short-term stimulus would wane quickly. Table 2 summarizes the results of analyses of the FY 2004 Budget proposals based on macro-econometric models of the U.S. economy maintained by Macro-economic Advisors (MA) and Global Insight (GI). Both models predict that the proposals would have a significant positive effect on output growth over the next two years. However, the boost to aggregate demand would be more modest thereafter, owing to the crowding-out of private investment from higher real interest rates.3

Table 2.Estimates from Large-Scale Models(Real GDP growth, change from baseline; in percent)
Study by:CEA 1/CBO 2/MA 3/GI 4/HF 5/
Model used:n.a.MAGIMAGIGIAverage
20030.40.50.40.50.20.30.4
20041.11.31.31.00.80.61.0
2003–07 (avg.)0.20.61.30.00.10.10.4

6. Moreover, most analyses indicate that the budget would dampen output in the long term. In most macroeconomic models, the decline of public and national saving implied by the FY 2004 Budget proposals would cause higher real interest rates and lower capital accumulation—for example, in the MA model the effect is to lower labor productivity by about ½ percentage point in 2017, relative to the baseline scenario (Figure 2).

Figure 2.Impact of 2003 Tax Cuts on Labor Productivity

Source: Macroeconomic Advisers.

7. This result is confirmed by a CBO study that examines the budget proposals from the perspective of several alternative models. In a “textbook” neoclassical growth model, in which economic agents do not modify their behavior in response to expected future changes in policy, the budget would lower GDP ¾ percent below baseline during 2009–2013. Using life-cycle and infinite-horizon models, in which economic agents are forward-looking, the CBO shows that the budget proposals would only increase long-run output if the tax cuts are anticipated to be reversed in the future. In this case, households work and save more in order to be able to pay for future taxes, offsetting the crowding-out effect (Table 3).4 The CBO’s analysis also illustrates that, in an open economy context, net inflows of foreign capital can help offset the decline in national savings and alleviate crowding out.

Table 3.Estimates from Small-Scale Models(Average change in GDP from CBO’s baseline, in percent)
2004-20082009-2013
Textbook Growth Model-0.2-0.7
Closed Economy Life-Cycle Growth Model
Lower government consumption after 2013-0.3-1.5
Higher lump-sum taxes after 20130.50.3
Open Economy Life-Cycle Growth Model
Lower government consumption after 2013-0.6-0.5
Higher lump-sum taxes after 20130.30.6
Infinite Horizon Growth Model
Lower government consumption after 20130.2-0.6
Higher lump-sum taxes after 20130.91.4
Source: CBO (2003).
Source: CBO (2003).

B. Fiscal Deficits and Real Interest Rates

8. A key indicator of the extent to which budget policies risk crowding out private investments is their impact on interest rates. Consequently, most empirical analysis of crowding-out has focused on the connection between fiscal deficits and interest rates. This literature is summarized below.

9. Simulations of large-scale macro-econometric models generally indicate that budget deficits have a sizeable effect on interest rates. In these models, the size of crowding-out typically depends on the monetary policy reaction function, the interest rate sensitivity of investment, the openness of the economy, and on how expectations of future policies are modeled.5 In a recent survey, Gale and Orszag (2002) found that the average prediction of these types of models is that a 1 percentage point increase in the primary deficit-to-GDP ratio, caused by a tax cut, is followed by a 40 basis point increase in long-term interest rates after one year, and 60 basis point increase after ten years. This compares to an increase of 60 basis points and 130 basis points, respectively, if the same increase in the primary deficit is induced by higher government spending.

10. Econometric estimates of reduced-form models have often provided conflicting results on the relationship between fiscal deficits and interest rates. This likely reflects the difficulty that studies have faced in taking into account the extent to which long-term interest rates respond to expectations of future fiscal policies, rather than to the current policy stance.6 More recent papers (surveyed in Table 4) that have sought to address this issue have found a positive and significant impact of expected budget deficits on expected future interest rates—averaging 35 basis points for a one percentage point increase in the deficit-to-GDP ratio, roughly in line with the estimates of the large-scale models.7

Table 4.Survey of Selected Studies on the Impact of U.S. Deficits on Real Interest Rates
Crowding out

effect

(in bps) 1/
Interest rates consideredFiscal variableBusiness cycle

regressor
Laubach (2003)2310-year Treasury bond yield

expected over the next 5 years
CBO 5-year ahead forecastNo
"365-year Treasury bond yield expected over the next 5 yearsOMB 5-year ahead forecastNo
"910-year Treasury bond yieldCBO 5-year ahead forecastNo
Canzoneri,

Diba, Cumbi
60Slope of yield curve (10-year

note less 3-month bill)
CBO 5-year ahead forecastNo
"40Slope of yield curve (10-year

note less 3-month bill)
CBO 10-year ahead

forecast
No
Elmendorf (1993)49Change in 3-year Treasury

bond yield
DRI forecast of

deficit/GDP ratio
Unemployment

rate

Increase in interest rates caused by a 1 percent rise in the deficit/GDP ratio.

Increase in interest rates caused by a 1 percent rise in the deficit/GDP ratio.

11. The Administration’s FY 2004 Budget proposals were accompanied by estimates of much smaller effects of deficits on interest rates. These estimates were based on a neoclassical framework developed by Elmendorf and Mankiw (1999), in which real interest rates in the steady state equal the marginal productivity of capital, which in turn depends on the capital share of income and the income-to-capital ratio. Using historical averages for these parameters, and assuming that a one dollar increase in public debt reduces the long-run stock of capital by 60 cents, a 1 percentage point increase in the debt-to-GDP ratio leads to an increase of real interest rates of only around 2–3 basis points (CEA, 2003).8

12. However, these arguments do not provide significant comfort. For example, Laubach (2003) notes that the estimates above would be consistent with an increase of interest rates of approximately 15 basis points following a permanent one percentage point increase in the deficit-to-GDP ratio. Moreover, while the CEA’s analysis suggests a relatively modest interest rate effect, this is predicated on a substantial degree of crowding out. In the CEA’s example, a 5 percent of GDP increase in government debt would lower the capital stock by around 3 percent of GDP, which given estimates of the gross marginal productivity of capital of around 10 percent would be consistent with a permanent reduction in output of roughly one third of a percent.

C. International Implications of Higher U.S. Public Debt

13. The integration of capital markets over the last three decades suggests the possibility of important spillovers from U.S. fiscal policy to the rest of the world. Higher fiscal deficits and public debt in one country will tend to absorb global savings and might cause higher world interest rates. This proposition is examined below.

14. Country-specific real interest rates have tended to move together over the last three decades. Table 5 shows that the real interest rate correlations for industrialized countries are all positive and generally quite high, which some authors have argued suggests the existence of a “world” real interest rate.9Figure 3 shows the evolution of different proxies for this rate the unweighted average of the national rates, their GDP-weighted average, and a measure based on the first principal components of the national rates. Each of these indices exhibited significant increases during most of the 1980s, a period of rapid growth of world public debt, but declined over most of the next decade despite still-high levels of world public debt.

Table 5.Correlations of G-7 Real Interest Rates (1977–2002)1
CanadaGermanyU.K.JapanU.S.FranceItaly
Canada1
Germany0.71
United Kingdom0.60.41
Japan0.70.70.51
United States0.60.30.50.51
France0.60.50.60.70.51
Italy0.70.50.70.50.50.81
World0.80.80.70.80.60.80.8
Source: OECD

Interest rates are 12-month Euromarkets interest rates deflated by the same period CPI inflation rate. The world real interest rate is the simple average of national rates.

Source: OECD

Interest rates are 12-month Euromarkets interest rates deflated by the same period CPI inflation rate. The world real interest rate is the simple average of national rates.

Figure 3.World Government Net Debt to GDP Ratio and World Real Interest Rate

15. Several studies have suggested that “world” fiscal policy matters for the determination of national real interest rates. Net public debt is found to be a significant determinant of the “world” real interest rate in Helbling and Wescott (1995) and of national real interest rates in Orr and Conway (2002). Ford and Laxton (1999) estimate the impact of world government net debt and consumption on national real interest rates of selected industrialized countries. They find that a 1 percentage point increase in world net government debt raises real interest rates by around 20 basis points. The main results and estimation methodologies adopted by these three papers are reported in Table 6.

Table 6.Survey of Selected Studies on the Impact of World Fiscal Variables on Real Interest Rates
AuthorsCrowding out effect (in bps)Interest rates considered as dependent variableFiscal RegressorExpected InflationMethodology and period
Helbling-Wescott (1995)16-20World (GDP

weighted) 3-month

Treasury bill yield
World (GDP

weighted) gross

government debt to

GDP ratio
Exponential smoothingCointegration analysis

through dynamic OLS and

error correction model (1960-1993)
"37-50World (GDP weighted) 3-month Treasury bill yieldWorld (GDP weighted) net government debt to GDP ratio""
"10-13World (GDP weighted) 10 year Treasury bond yieldWorld (GDP weighted) gross government debt to GDP ratio""
"16-42World (GDP weighted) 10 year Treasury bond yieldWorld (GDP weighted) net government debt to GDP ratio""
Ford-Laxton (1999)17-23National 12-month

Euromarket

certificates of

deposits interest rate
World (GDP

weighted) net

government debt to

GDP ratio
Same year CPI inflation

rate
Pooled OLS and SURE time

series estimation, imposing

equality of coefficients across

countries (1977-1997)
Orr and Conway (2002)16National 10 year

government bond

yields
National net

government debt to

GDP ratios
Hodrick-Prescott filter

of CPI inflation rate
Error correction model

estimation, imposing equality

of long-term coefficients

across countries (1986:Q1

–2002:Q1, countries: Australia,

New Zealand, USA, UK,

Canada, Sweden, Germany).

Increase in interest rates caused by a 1 percent increase in the government debt/GDP ratio (in bps)

Increase in interest rates caused by a 1 percent increase in the government debt/GDP ratio (in bps)

16. The relationship between national real interest rates and world public debt is re-examined below. The sample comprises 11 industrialized countries (the G-7 countries plus Belgium, Netherlands, Switzerland and Denmark) over the period 1977–2002. The interest rates used are the 12-month Euro market interest rates on certificates of deposits, deflated by the same-period CPI inflation rate. Two approaches were used in this chapter. First, each country’s real interest rate was regressed by OLS on two world fiscal variables, namely, the net public debt-to-GDP ratio and the share of real GDP absorbed by government consumption and investment.10 Second, the data were pooled and the 11 equations were estimated as a system, imposing the constraint that the coefficients of the fiscal variables were the same across all countries.11 Instrumental variables were used to avoid potential biases stemming from the dependence of public debt on interest rates.12 The system estimates were derived using a Generalized Method of Moment (GMM) estimation methodology, which yields consistent and asymptotically normal estimators under relatively unrestrictive assumptions on the error term and regressors.

17. The regression results generally confirm that an increase in world public debt affects national real interest rates, but cannot rule out the existence of a break in the relationship over the 1990s. The OLS coefficients of the world fiscal variables have the right signs, but only in few cases are significant at a 5 percent level (Table 7). The results also indicate that augmenting the OLS regressions with the country-specific public debt-to-GDP ratios does little to improve the results, as this coefficient is rarely both significant and positive.

Table 7.OLS Regressions of Real Interest Rates on World Fiscal Variables 1/
Wu-Hausman test 5/National public debt to GDP
cWGND 2/WGA 3/DWGA 4/R Sq.DWWGNDWGA
Belgium-16.31-0.051.071.130.500.59-2.49-0.520.10
[0.20][0.34][0.04][0.38][0.00]
Canada-29.480.111.36-2.080.170.65-2.92-1.030.05
[0.02][0.04][0.01][0.16][0.32]
Switzerland-12.950.080.54-0.050.101.00-1.98-0.13-
[0.23][0.11][0.20][0.95]-
Germany-9.380.010.571.520.160.61-2.850.65-0.16
[0.36][0.82][0.15][0.24][0.05]
Denmark-48.70.102.291.290.480.59-1.410.140.07
[0.00][0.02][0.00][0.20][0.15]
United Kingdom-53.50.262.18-1.140.380.67-3.260.74-0.10
[0.00][0.00][0.00][0.36][0.00]
Japan-40.050.071.861.160.490.46-3.861.050.03
[0.00][0.29][0.00][0.35][0.18]
Netherlands-26.310.001.38-0.310.340.44-3.010.33-0.02
[0.04][0.91][0.01][0.86][0.77]
United States-25.200.061.21-2.320.140.31-1.05-0.63-0.21
[0.30][0.55][0.21][0.20][0.03]
France-67.100.232.890.480.420.63-1.52-0.26-0.03
[0.00][0.00][0.00][0.55][0.64]
Italy-71.950.273.02-1.690.450.89-1.62-0.990.14
[0.00][0.00][0.00][0.27][0.05]

Data are semiannual from 1977:1 to 2002:2; p-values from Newey-West heteroskedasticity and autocorrelation-consistent standard errors are reported in square brackets. The dependent variables are the national 12-months Euromarket interest rates deflated by the same period CPI inflation rate.

WGND is the GDP-weighted average of national net government debt ratio to GDP, with the only exclusion of Switzerland. GDP is converted using PPP exchange rates.

WGA is the GDP-weighted average of national real government absorption (consumption plus investment) as a share of GDP.

DWGA is the first difference of WGA.

t-statistics of the Hausman-Wu test for the exogeneity of WGND and WGA. The null hypothesis is exogeneity. The list of instruments used comprises the second and third lags of WGND and WGA. Critical values are from the standardized normal distribution (10 percent = ±1.28, and 5 percent = ±1.64).

Data are semiannual from 1977:1 to 2002:2; p-values from Newey-West heteroskedasticity and autocorrelation-consistent standard errors are reported in square brackets. The dependent variables are the national 12-months Euromarket interest rates deflated by the same period CPI inflation rate.

WGND is the GDP-weighted average of national net government debt ratio to GDP, with the only exclusion of Switzerland. GDP is converted using PPP exchange rates.

WGA is the GDP-weighted average of national real government absorption (consumption plus investment) as a share of GDP.

DWGA is the first difference of WGA.

t-statistics of the Hausman-Wu test for the exogeneity of WGND and WGA. The null hypothesis is exogeneity. The list of instruments used comprises the second and third lags of WGND and WGA. Critical values are from the standardized normal distribution (10 percent = ±1.28, and 5 percent = ±1.64).

18. The system estimates show that both world public debt and government absorption are significant determinants of national real interest rates. A 1 percentage point increase in the world government debt to GDP ratio induces an increase in national real interest rates of around 10 basis points over the 1997–2002 period (Table 8). This result is robust to the addition of other variables, such as those capturing the business cycle and monetary policy and inflation changes. Given the relatively scarce number of observations available, however, it is difficult to test for the stability of the coefficients over the period considered.13 Moreover, as most of the desirable properties of GMM estimators are only valid asymptotically, the point estimates should be taken with caution. With these caveats in mind, these estimates suggest that the 15 percentage point increase in the U.S. public debt ratio projected over the next decade by the CBO would lead to an average ½–1 percentage point increase in national real interest rates.

Table 8.Joint GMM Estimation of National Real Interest Rates Imposing Equality of Coefficients Across Equations 1/
WNGD0.120.090.150.150.160.13
[0.00][0.00][0.00][0.00][0.00][0.00]
WGA1.341.341.671.661.922.79
[0.00][0.00][0.00][0.00][0.00][0.00]
DWGA1.24-0.041.100.860.621.41
[0.00][0.88][0.00][0.00][0.05][0.00]
Euro Dummy 2/-1.47-0.68-0.94-0.88-0.75-0.32
[0.00][0.01][0.00][0.00][0.00][0.02]
UNE 3/0.47
[0.00]
DINFL 4/0.20
[0.00]
DIRS 5/0.14
[0.00]
LAF 6/0.17
[0.27]
PBAL 7/0.99
[0.00]
J-statistic 8/2.7510.800.110.117.235.58
[0.99][0.46][0.99][0.99][0.78][0.89]

Data are semiannual, from 1977:1 to 2002:2. Total system observations: 566. p-values are in square brackets. The Heteroskedasticity and Autocorrelation Consistent covariance matrix is estimated based on the Newey and West estimator. The vector of instruments comprises the second and third lags of WNGD and the other regressors.

Dummy for 1999:1–2002:2.

World (GDP-weighted) unemployment rate.

Change in world (GDP-weighted) CPI inflation rate.

Change in world (GDP-weighted) short term real interest rate (CPI inflation deflated).

World (GDP-weighted) labor force growth.

Change in world (GDP-weighted) primary balance.

Model specification test. The Null is that the model is well specified. Critical values are from a Chi-2 distribution with 11 degrees of freedom (equal to the number of overidentifying restrictions in the system).

Data are semiannual, from 1977:1 to 2002:2. Total system observations: 566. p-values are in square brackets. The Heteroskedasticity and Autocorrelation Consistent covariance matrix is estimated based on the Newey and West estimator. The vector of instruments comprises the second and third lags of WNGD and the other regressors.

Dummy for 1999:1–2002:2.

World (GDP-weighted) unemployment rate.

Change in world (GDP-weighted) CPI inflation rate.

Change in world (GDP-weighted) short term real interest rate (CPI inflation deflated).

World (GDP-weighted) labor force growth.

Change in world (GDP-weighted) primary balance.

Model specification test. The Null is that the model is well specified. Critical values are from a Chi-2 distribution with 11 degrees of freedom (equal to the number of overidentifying restrictions in the system).

References

    Beach, W.B., R.A.Rector, A.Goyburu, and N.J.Michel,2003, “The Economic and Fiscal Effects of the President’s Growth Package,” A Report of The Heritage Center for Data Analysis, CDA 03–05, Heritage Foundation.

    • Search Google Scholar
    • Export Citation

    CBO (Congressional Budget Office), 2003, An Analysis of the President’s Budgetary Proposals for Fiscal Year 2004, Washington, D.C.

    CEA (Council of Economic Advisors), 2003, Economic Report of the President, Washington, D.C.

    Canzoneri, M., R.Cumby, and B.Diba,2002, “Should Central Bank and the Federal Reserve be Concerned About Fiscal Policy?,” in: Federal Reserve Bank of Kansas City, Rethinking Stabilization Policy, Kansas City.

    • Search Google Scholar
    • Export Citation

    Elmendorf, D.W.,1993, “Actual Budget Deficits Expectations and Interest Rates,” Harvard Institute of Economic Research Discussion Paper No. 1639.

    • Search Google Scholar
    • Export Citation

    Elmendorf, D.W.,1996, “The Effect of Deficit-Reduction Laws on Real Interest Rates,” Board of Governors of the Federal Reserve System, Finance and Economic Discussion Paper Series No. 96/44.

    • Search Google Scholar
    • Export Citation

    Elmendorf, D.W., and N.G.Mankiw,1999, “Government Debt”, in: J.Taylor and M.Woodford (eds.), Handbook of Macroeconomics (vol. 1c)Amsterdam: Elsevier.

    • Search Google Scholar
    • Export Citation

    Evans, P.1987, “Do Budget Deficits Raise Nominal Interest Rates? Evidence from Six,” Journal of Monetary Economics, Vol. 20, No. 22, pp. 281–300.

    • Search Google Scholar
    • Export Citation

    Ford, R., and D.Laxton,1999, “World Public Debt and Real Interest Rates,” Oxford Review of Economic Policy, Vol. 15, No. 2, pp 77–94.

    • Search Google Scholar
    • Export Citation

    Gagnon, J. and M.Unferth,1995, “Is There a World Real Interest Rate?,” Journal of International Money and Finance, Vol. 14, No. 6, pp. 845–55

    • Search Google Scholar
    • Export Citation

    Gale, W., and P.Orszag,2002, “The Economic Effects Of Long-Term Fiscal Discipline,” Urban-Brookings Tax Policy Center Discussion Paper.

    • Search Google Scholar
    • Export Citation

    Helbling, T., and R.Wescott,1995, “The Global Interest Rate,” Staff Studies for the World Economic Outlook, IMF, Washington, D.C.

    Laubach, T.,2003, “New Evidence on the Interest Rate Effect of Budget Deficits and Debt,” Board of Governors of the Federal Reserve System, Finance and Economics Discussion Paper Series No. 2003–12.

    • Search Google Scholar
    • Export Citation

    Macroeconomic Advisers, 2003, “A Preliminary Analysis of the President’s Jobs and Growth Proposals,” mimeo.

    Mauskopf, E., and D.L.Reifschneider,1997, “Dynamic Scoring, Fiscal Policy, and the Short-Run Behavior of the Macroeconomy”, National Tax Journal, Vol. 50, No. 3, pp. 631–55.

    • Search Google Scholar
    • Export Citation

    Newport, P.,2003, “Bush Plan Boosts Short-Term U. S. Growth, but Adds to Deficits,” Global Insight, February.

    Orr, A., and P.Conway,2002, “The GIRM: A Global Interest Rate Model,” Westpac Institutional Bank Occasional Paper.

    Plosser, C.,1987, “Fiscal Policy and the Term Structure,” Journal of Monetary Economics, Vol. 20, No. 2, pp. 343–67.

1

Prepared by Roberto Cardarelli and Ayhan Kose.

2

The Staff report contains a detailed description of the Administration’s original FY 2004 Budget proposals and of the tax legislation that was passed in May 2003.

3

The studies employ different assumptions regarding the baseline scenario. Further, while the CBO study examines the implications of the entire FY 2004 Budget proposals, the other studies focus only on the Economic Growth Package. Finally, the study by the CEA does not report which model is used, but notes that “the particular values of the numerical estimates presented reflect judgments regarding the implementation of the proposals” (CEA, 2003)

4

In the textbook growth model, labor supply increases because of lower marginal tax rates, but output declines because higher government and private consumption crowds out capital accumulation. It is only when expectations of higher taxes after 2013 induce additional savings that the tax cuts have a positive impact on savings, investment and output (as in the two models with forward-looking agents). This effect is larger in an infinite-horizon model, since agents take into account the higher tax burden on their descendants. In all models, maximum effect is achieved if the future increase in taxation is through higher lump-sum taxes. Estimates assuming an increase in future marginal tax rates fall between those presented in Table 3.

5

For a brief description of some of these models in the context of a dynamic scoring analysis of fiscal policy measures (including two large-scale structural models of the U.S. economy used by the Federal Reserve), see Mauskopf and Reifschneider (1997).

6

Among the studies that find no statistically significant relationship between fiscal deficits and interest rate are the ones by Plosser (1987) and Evans (1987), which proxied expected fiscal deficits using forecasts from vector autoregressive models (VAR). However, the usefulness of this method to capture actual expectations is subject to a series of limitations (Elmendorf, 1993).

7

A caveat on these results is that the reduced-form relationship between expectations of future budget deficits and interest rates could be driven by changes in the expectations of output growth. However, Elmendorf (1996) shows that this relationship is robust to the explicit introduction of a variable capturing expectations on the future state of the business cycle.

8

The assumption made by CEA (2003) is that, while private savings do not respond at all to the increase in public debt, around a third of the decrease in national savings is offset by larger capital flows from abroad.

9

Using panel data techniques, Gagnon and Unferth (1995) show that national real interest rates do not exhibit persistent deviations from a common world interest rate, defined as the simple average of the rates of nine OECD countries. The only exception seems to be the United States, a result that the authors suggest may be reflecting the lower trade integration of this country with the rest of the world. On the correlations reported in Table 4, it should be noted that since 1999 the European countries that joined the Euro have essentially shared the same interest rate.

10

This captures the two channels through which fiscal policy is supposed to crowd out private investments, the “portfolio” channel (via higher public debt) and the “transaction” channel (via higher government spending). Following Ford and Laxton (1999), the change in real government consumption is also used as a regressor. As economic theory suggests that both the fiscal variables (expressed as a share of GDP) and the real interest rates are stationary, no attempt is made to estimate a long-run relationship between these variables using a cointegration approach.

11

This approach improves the efficiency of the estimators, if disturbances are correlated across countries, and also increases significantly the degrees of freedom, as it allows estimating the coefficients of the fiscal variables using a much larger number of observations.

12

The list of instruments consists of the lagged values of the world net government debt to GDP ratio, plus the other fiscal regressors which are taken as predetermined. The Wu-Hausman test reported in Table 7 supports this choice, as it failed to exclude the exogeneity of the world net public debt to GDP ratio in the interest rates OLS regressions, while it could not rule out the exogeneity of government consumption. This may reflect the fact that government consumption does not include interest paid on the stock of debt.

13

A Chow test on the stability of the coefficients in two equally sized sub-samples rejects the null of stability.

Other Resources Citing This Publication