1 Introduction
Expansions and recessions in economic growth characterise all economies and have significant effects on employment, investment and economic welfare. During the past two decades New Zealand has experienced short-lived expansions and frequent recessions during the 1980s, a major recession from 1991 to 1993, and a long expansion during the 1990s that ended with the recession of 1998. There has been ongoing debate about the reasons for these expansions and recessions and attributed causes have ranged from the influence of international business cycles, overly aggressive monetary policy, volatile exchange rates, while the "Asian crisis” is a popular explanation for the 1998 recession.
The propagation impulse framework first introduced to economics by Frisch (1933) and Slutsky (1937) has come to dominate the analysis of expansions and recessions that characterise the evolution of economic growth. This approach, well illustrated by Blanchard and Watson’s (1986) analysis of US business cycles, conceives of fluctuations in economic activity as arising from impulses (shocks) that affect the economy through a complex dynamic propagation process. Therefore identification of shocks that precipitate expansions and recessions in economic growth, or business cycles, represents a major conceptual and statistical challenge. This complexity means that debates concerning the causes of booms and recessions and the contribution of macroeconomic policy are often not satisfactorily resolved.
The purpose of this paper is to develop a statistical model capable of identifying the major shocks to the New Zealand economy during the last two decades, to understand the dynamic response of the economy to these shocks and to measure their contribution to New Zealand expansions and recessions in economic growth. The model is a structural vector autoregressive (VAR) model of the New Zealand economy. Since its introduction by Sims (1980), VAR modelling has become a standard empirical method for evaluating the properties of macroeconomic systems and for tackling the challenges that have been set for this paper. This is illustrated for example in Blanchard’s (2000) recent review of the historical developments in theoretical and empirical macroeconomic modelling.
There are important geographical and industrial characteristics that must be captured in a VAR model of the New Zealand economy if that model is to be capable of identifying the major shocks and their propagation dynamics. These characteristics justify a more ambitious attempt to capture international trade and financial linkages than have been attempted in previous New Zealand models. New Zealand’s industrial structure is important in the way it influences the commodity structure of imported and exported goods. The industrial structure, with a relatively heavy weighting toward agricultural based production, combined with its geographical location and characteristics, justify attempting to capture the consequences of climatic conditions on the evolution of New Zealand’s real output growth and other macroeconomic variables.
Satisfactorily capturing these open economy and industrial features, while simultaneously understanding the way international and domestic sourced shocks impact on key New Zealand macroeconomic variables, poses several major modelling challenges. The choice of variables to include in the model has been informed by New Zealand’s characteristics and prior research. The model presented here has 13 variables, a large number for a VAR model. This is the largest structural VAR model of the New Zealand economy estimated to date.
There have been several important developments in model specification and estimation procedures that we have been able to draw on to estimate such a large VAR model. Included in these techniques are the block exogeneity procedures introduced by Cushman and Zha (1997) and Dungey and Pagan (2000) to identify international and domestic shocks and dynamic responses to these shocks in a small open economy. We have expanded on the number of international variables that have previously been included in New Zealand VAR models. These international variables are block exogenous and the model includes restrictions on contemporaneous and lagged variables. Given the interest in the relative importance of domestic interest rates and the exchange rate in generating expansions and recessions, the model includes a wider range of domestic financial variables than in previous models. Another novel contribution of this model is the inclusion and identification of the relative importance of domestic climatic conditions in generating business cycle fluctuations.
The model is used to identify the impact of foreign and domestic shocks on New Zealand macroeconomic variables including real GDP and prices, and to uncover the propagation process. Key conclusions to emerge from these model simulations include the following. International variables have been the dominant influences on fluctuations in New Zealand real GDP around its trend. Fluctuations in domestic climatic conditions have also been important, particularly in specific periods such as the 1998 recession.
In contrast to the attention they receive in public debate, shocks from the domestic exchange rate have been relatively unimportant. While the impact of monetary policy also features in public debate, this is treated in a companion paper (Buckle, Kim and McLellan, 2003). The interest rate shocks examined in this paper combine the effects of monetary policy and other influences on interest rates. Separation of these effects is essential before an evaluation of monetary policy is appropriate.
The remainder of the paper is structured as follows. Section 2 discusses in more depth the modelling procedure and the econometric issues involved in specifying the model and the restrictions required to identify the range of shocks that are of interest. The overall structure of the model and motivation for specific equation restrictions are discussed in Section 3. The next two sections discuss applications. The first set of applications discussed in Section 4 includes the simulation of international financial and trade shocks, domestic financial shocks and climate shocks. These are presented in the form of response multipliers or impulse response functions. The second set of applications includes historical decompositions (or accounting) of the contributions of these shocks toward booms and recessions in New Zealand’s real GDP. These results are reported in Section 5. Section 6 summarises key conclusions and discusses potential applications of the model.
