Appendix 1. Derivation of Interest Rate-Growth Differential
Appendix 2. Financial Crises and Dynamics of the IRGD
Arellano, M. and O. Bover (1995) “Another Look at the Instrumental Variables Estimation of Error-Components Models,” Journal of Econometrics 68, 29–51.
Baldacci, E., G. Callegari, D. Coady, D. Ding, M. Kumar, P. Tommasino, and J. Woo (2010) “Public Expenditures on Social Programs and Household Consumption in China,” IMF Working Paper WP/10/69.
Bartolini, L., and C. Cottarelli (1994) “Government Ponzi Games and the sustainability of Public Deficits under Uncertainty,” Ricerche Economiche 48, 1-22.
Beck, T., R. Levine, and N. Loayza (2000), “Finance and the sources of growth,” Journal of Financial Economics 58 (1-2), 261-300.
Beck, T., A. Demirgüç-Kunt, and R. Levine (2000), “A New Database on Financial Development and Structure,” World Bank Economic Review 14, 597-605.
Blanchard, O., J., C. Chouraqui, R.P. Hagemann, and N. Sartor (1990) “The Sustainability of Fiscal Policy: New Answers to an Old Question,” OECD Economic Studies No 15, Autumn (http://www.oecd.org/dataoecd/50/21/34288870.pdf).
Blundell, R. and S. Bond (1998) “Initial Conditions and Moment Restrictions in Dynamic Panel Data Models,” Journal of Econometrics 87(1), 115–43.
Caballero, R., E. Farhi, and P. Gourinchas (2008) “An Equilibrium Model of Global Imbalances and Low Interest Rates,” American Economic Review, 98, 358-393.
Chinn, M. and H. Ito (2008) “A New Measure of Financial Openness,” Journal of Comparative Policy Analysis, 10, 309-322 (update in July 2010).
Edwards, S. (1996) “Why Are Latin America’s Saving Rates So Low: An International Comparative Analysis,” Journal of Development Economics, Vol. 51.
Escolano, J. (2010) “A Practical Guide to Public Debt Dynamics, Fiscal Sustainability, and Cyclical Adjustment of Budgetary Aggregates,” IMF Technical Notes and Manuals (http://www.imf.org/external/pubs/ft/tnm/2010/tnm1002.pdf).
King, R.G. and R. Levine (1993) “Finance and Growth: Schumpeter Might be Right,” Quarterly Journal of Economics 108 (3), 717–737.
King, R.G. and S.T. Rebelo (1993) “Transitional Dynamics and Economic Growth in the Neoclassical Model,” American Economic Review 83(4).
Kumar, M. and J. Woo (2010), “Public Debt and Growth,” IMF Working Paper 10/174. (http://www.imf.org/external/pubs/ft/wp/2010/wp10174.pdf)
Reinhart,C. and K. Rogoff (2009), This is Different: Eight Centuries of Financial Folly (Princeton University Press, Princeton).
Spaventa, L. (1987) “The Growth of Public Debt: Sustainability, Fiscal Rules, and Monetary Rules,” Staff Papers, International Monetary Fund, Vol. 34, No. 2.
The authors would like to thank Olivier Blanchard, Carlo Cottarelli, Marcello Estevao, Phil Gerson, Matthew Jones, Manmohan Kumar, Gian Maria Milesi-Ferretti, Rodrigo Valdes, Seokgil Park and participants in various IMF seminars or presentations for helpful comments and discussions. Petra Dacheva and Raquel Gomez Sirera provided excellent research assistance.
Except when otherwise indicated, the IRGD is computed as the differential between the effective interest rate (actual interest payments divided by the debt stock at the end of the previous year) and the growth rate of nominal GDP, divided by the latter plus one. It is immaterial whether the interest and growth rates are both measured in nominal or real terms. Also, the interest rate is adjusted for the change in the domestic currency value of foreign currency-denominated debt due to exchange rate changes. This measure best approximates the IRGD factor relevant for debt dynamics. When data availability does not allow computation of the effective interest rate paid, a market benchmark government rate is used. See Appendix 1 for details on the derivation of IRGD.
Even advanced economy data compiled on a reasonably consistent basis were scarce until relatively recently. In this regard, the AMECO database of the European Commission represented an important step forward, as it includes effective average interest rates on government debt as well as other variables relevant to the debt dynamics. However, at this time, AMECO only covers EU economies and the largest of the non-EU advanced economies.
A similar value results from averaging broader OECD samples for 1991-2008 (Escolano 2010) and for the 1980s (Blanchard et al. 1990). The IRGD, however, was negative for many advanced economies during the 1970s and in some earlier periods (Reinhart and Sbrancia (2011)).
For the purposes of this analysis, we consider advanced economies the OECD members in 1990, except for Turkey, to eliminate those which are currently considered advanced economies but they may have been in transition towards a balanced growth path during a significant part of the sample period (such as Korea and new EMU members). Also, we dropped about one third of the low income countries in our sample because concessional debt was a substantial proportion of their debt (above 50 percent of public and publicly guaranteed external debt as reported in World Bank’s Global Development Finance database).
The volatility is measured by the standard deviation of the IRGD in each country, and the persistence is measured by the first-order autoregressive AR(1) coefficient.
For G-7 economies, the IRGD has averaged 2 percentage points during 1980-2009.
Incidentally, simulations for catch-up economies with realistically calibrated parameters tend to produce, not only higher real interest rates than in advanced economies, but also much higher IRGDs than for advanced economies and hence strongly positive (King and Rebelo (1993)).
The country sample size in Figure 3 changes (increases) over time due to data availability. Also, the growthadjusted interest rate represented in this figure is the simple difference between the real interest rate and the real growth of GDP, both as a differential with respect to the corresponding average for the G-7. This allows the additive decomposition of this differential with the G-7 between the contributions of the real interest rate differential and the real growth rate differential.
Consistently with estimates in previous sections, effective interest rates are computed as the government’s actual interest cost divided by the stock of debt at the end of the previous year, adjusted for currency valuation effects.
The dynamic panel regression is estimated by employing a system GMM approach, which uses suitable lagged levels and lagged first differences of the regressors as their instruments in order to mitigate the endogeneity problem (Arellano and Bover (1995); Blundell and Bond (1998)). The results from different estimation methods such as OLS or 5-year panel regressions are quite similar. Although results for longer time periods such as 1980-2008 are remarkably similar and often stronger, we focus on the period of 1999-2008 (post-Asian crisis period) because panel data are significantly unbalanced for longer periods with many missing observations especially for emerging and developing economies.
The effective interest cost of public debt also depends importantly on debt structure, debt management policies, taxation and other institutional factors, but there are no consistent data available in these features for a large number of countries.
An underdeveloped financial system can lead to higher savings (Baldacci et al. (2010); Caballero et al. (2008)). In a growing economy where the desired consumption bundle shifts towards durable goods, inability to borrow against future income streams could lead households to save more in order to self-finance their purchases. However, Edwards (1996) argues that financial deepening induces higher saving by creating more sophisticated financial systems.
The two first principal components account for 70 percent of total variance in the original variables. Since the number of observations available for the financial liberalization index is less than half of those of other financial repression measures, we did not include the financial liberalization index in the principal components analysis. However, the results do not change appreciably if we include it at the cost of fewer observations.
We performed two standard tests of the validity of the instruments—Arellano and Bover (1995) and Blundell and Bond (1998)—which were both passed. The first is a Hansen J-test of over-identifying restrictions, which tests the overall validity of the instruments by analyzing the sample analog of the moment conditions used in the estimation process. We cannot reject the null hypothesis that the full set of orthogonality conditions are valid (for example, p-value=0.29 for the regression in Column 1). The second test examines the hypothesis that the error term εit is not serially correlated. We use an Arellano-Bond test for autocorrelation, and find that we cannot reject the null hypothesis of no second-order serial correlation in the first-differenced error terms (p-value=0.68 for the regression in Column 1).
Panel unit root tests reject the presence of unit roots in the differential in the data. The country fixed-effects are correlated with the lagged dependent variables in the autoregressive model. The order of bias is 1/T (Nickell, 1981), which is small in this data. Judson and Owen (1999) also show that the bias of the least squares dummy variable (LSDV) estimators is approximately 2-3 percent on the lagged dependent variable and less than 1 percent on other regressors for a panel of size N=100 and T=30 and low persistence.
Dt is set equal to 1 for only one period and assumed to be zero otherwise. Data are from Reinhart and Rogoff (2009).
The coefficients of the first two lagged terms of the crisis dummy variable are statistically significant at the 1-5% level. Similarly, the coefficients of the crisis dummy and its lagged terms in the regressions for banking and sovereign debt crises are mostly significant at the conventional level.
The same estimation is also repeated only for the big five banking crises identified by Reinhart and Rogoff (2008) and four additional recent episodes (Laeven and Valencia (2010)): Spain (1977), Norway (1987), Finland (1991), Sweden (1991), Japan (1992), United States (2007), United Kingdom (2007), Ireland (2008) and Iceland (2008) (not shown to save space).
Even in the absence of a debt crisis, high levels of sovereign debt are significantly negatively associated with subsequent growth and positively with sovereign bond yields (Kumar and Woo (2010); Baldacci and Kumar (2010), respectively), which in turn raises the IRGD. After controlling for other variables such as initial income per capita, inflation, trade openness, indicators of financial repression and capital market openness, the public debt is significantly and positively associated with the IRGD in a panel regression for our sample of countries for various periods (not reported to save space).