Empirical evaluation of rival theories of the business cycle : applications of structural var models to the New Zealand economy

Abstract

The objectives of this thesis are to carry out empirical tests for three rival theories of the business cycle using New Zealand macroeconomic data, to determine the most important shocks that lead to fluctuations in the main macroeconomic variables, and to outline stylised facts regarding comovements among these variables under different types of exogenous shocks. There are three rival business cycle theories covered in this thesis, namely the Real Business Cycle (RBC), the New Keynesian (NK), and the Monetary Business Cycle (MBC).

These theories are represented as economic models consisting of long run and/or short run relations, which are then used to construct long run and short run economic identifying restrictions that are embedded in empirical models. These models are estimated within the cointegrated SVAR framework using the sample period of 1975:1-1999:1. The cointegrated SVAR models are employed because of the non-stationarity found in all series used in this study, and of the fact that these series are cointegrated. The reduced form VAR contains four intervention dummy variables that are included based on important macroeconomic events in New Zealand. Without these dummy variables, the assumption of multivariate normality of the residuals is well rejected. The critical values for inferring the usual likelihood ratio statistic for determining cointegration rank under the presence of the intervention dummies are simulated. It is found that the cointegration rank is three, suggesting the presence of three long run relations. The cointegrated SVAR framework also enables consideration of the small, open nature of the New Zealand economy through weak exogeneity restrictions on the loading matrix, which are supported by the data.

The plausibility of each SVAR model is assessed according to the consistency of its estimated parameters, impulse response functions and forecast error variance decomposition with its theoretical underpinnings, as well as on results of the restriction tests. It is found that the RBC-SVAR model gains little support from the data. Several of the estimated long run parameters in this model are not consistent with the theory. Similarly, the MBC-SVAR model, at best, only obtains partial support from the data as the contemporaneous restrictions embedding its short run relations cannot be rejected while the long run restrictions are well rejected. In contrast, in the NK-SVAR model, all the estimated parameters have the signs as expected, dynamics of each variable as depicted by the impulse response functions are consistent with the NK theoretical explanations, and the contemporaneous restrictions embedding the short run relations of this model cannot be rejected. Therefore, the preferred model among the three competing models is the NK-SVAR model.

Some of the main findings from this preferred model are as follows. The demand side shocks are approximately as important as the technology (supply side) shocks. Real exchange rate shock is the most important source of fluctuations from the demand side of the economy. Defined as a shock to fiscal policy, this shock is important in driving fluctuations in real exchange rate, money demand and the domestic nominal interest rate, and output. Technology shocks are capable of explaining fluctuations in output, the real exchange rate, the domestic nominal interest rate and hours worked, and money demand. Another dominant source of fluctuations is the shock to employment, which in the NK-SVAR model is defined as a mark up shock arising from an imperfectly competitive market structure. This shock explains the variability of employment and money demand, output, and domestic nominal interest rate. A money demand shock is only capable of driving fluctuations in the demand for money, whereas a monetary policy (interest rate) shock is important in explaining variability in the interest rate and in money demand. External shocks, especially world real interest rate shocks, are found quite important in determining fluctuations in employment, money demand, the real exchange rate, and in output. Policy implications of these findings and suggestion for further research are discussed.