The Treasury

Global Navigation

Personal tools

5  Future work

As all models are an abstraction of the real world, significant differences can arise between models. The initial aim of this paper was to identify the key differences in the dynamic properties of FPS and NZTM. An obvious next step is to look at the properties of actual New Zealand economic data and data driven models (such as SVARs), to determine which of the two models (or a combination of both) more closely represents the underlying structure of the New Zealand economy.

We are interested in exploring this further. For example, recent empirical work has been undertaken using a SVAR approach to model the New Zealand business cycle by Buckle et al. (2002). This model, which is more data driven than either NZTM or FPS, may provide a useful benchmark for further comparison work: for example, we could subject the SVAR to similar shocks and compare the impulse responses from the SVAR with the responses in this paper. Another technique for comparing models to the data is to undertake stochastic simulations (where the model is subjected to a battery of shocks similar to those seen over history), and compare the variability, co-movement and persistence of variables like exchange rate and exports in the model simulations to actual history. Furthermore, we could look at some of the specific drivers of the two models. For example, investigating the strength of the relationship between world commodity prices and the real exchange rate should help inform the external price transmission mechanism of the economy.

However, we are not sure that further comparison of FPS and NZTM to the data will lead to definitive conclusions about the true properties of the New Zealand economy. Significant parts of both models have been calibrated precisely because some empirical relationships are very hard to pin down, particularly in economies like New Zealand which have been subject to a lot of structural change. For example, it is difficult to generate strong empirical evidence about the short-term dynamics of the exchange rate, and even harder to determine whether the medium-term dynamics of the exchange rate are related to variables like net foreign assets (which is one key point of difference between the two models). While the increasing length of New Zealand's post reform dataset makes empirical analysis increasingly useful, we think macroeconomic modelling in New Zealand (and all other countries) will continue to be involved significant measures of theory and judgement.

Page top