IMPACT OF GOVERNMENT BOND ISSUES ON THE TERM STRUCTURE OF INTEREST RATES : THE AUSTRALIAN EXPERIENCE

Abstract

With large increases in the budget deficits of many industrialised countries over the last decade or so, the effects of fiscal policy on economic variables has become a major issue for debate among economists and laypeople alike. This paper investigates the hypothesis that increases in bond-financed government deficits lead to increases in the term structure of interest rates in the economy. The existing body of literature concentrates on the US economy, and uses very simple econometric techniques. This paper models a small, open economy (Australia) in a general equilibrium framework and uses advanced econometric techniques. Mussa's (1980) model of current account determination is adapted to include a monetary sector as well as a real sector. Asset demand functions are modelled using the portfolio balance approach (Branson and Henderson 1985), because capital is assumed to be very mobile while assets are imperfectly substitutable. In addition, the model assumes expectations are rational and covered interest parity prevails (but uncovered interest parity does not). Central bank intervention in foreign exchange markets is incorporated into the model. The final general equilibrium model comprises a system of seven equations with seven endogenous variables, two expectational variables and eight predetermined variables. The data extends from the float of the currency in December 1983 to the September quarter 1994. This system of equations is estimated using generalised two step two stage least squares (Nijman and Palm (1991)) because the presence of future variables raised concerns about autocorrelation among the residuals and many of the instrumental variables are predetermined rather than strictly exogenous. Under these circumstances, Cumby, Huizinga and Obstfeld's (1983) generalised method of moments estimator is asymptotically efficient in a class of instrumental variable estimators, and is computationally more accessible than a full-information estimator. However, instead of using next period's variable values as a proxy for current expectations about the future, auxiliary equations are obtained for each of the expectational variables, following a procedure suggested by Nijman and Palm (1991 ). These auxiliary regressions are projected over the range of the study and used as proxies for the future expectations. This method improves the efficiency of Cumby et al’s (1983) technique. The results of the study suggest that the short and long term interest rates in Australia are negatively correlated with the size of the budget deficit. These results are robust and significant. On the basis of this analysis, the hypothesis that higher budget deficits raise interest rates is rejected. This has important policy implications, the main one being that crowding out effects of budget deficits are not significant. It should be noted that the way this study has been undertaken does not allow scope for any comment to be made about whether or not Ricardian equivalence holds in the Australian economy. The absence of a positive correlation between the budget deficit and domestic interest rates does not imply that higher sustained budget deficits have no adverse consequences. The effects of sustained government dissaving have not been covered in this paper.