Tests of NZ equilibrium exchange rates hypothesis : A cointegration approach

Abstract

This study is an empirical testing of NZ equilibrium exchange rate hypotheses using the method of cointegration. This cointegration technique does not only overcome the problem of the non-stationarity of macroeconomic time series variables but also provides a convenient method for testing the economic hypotheses. In this study, two equilibrium exchange rate hypotheses proposed by Williamson (1985) have been tested.

The first hypothesis is known as current market equilibrium - ie, the equilibrium that would obtain if market agents had full knowledge of all relevant facts and reacted rationally to that knowledge. Instead of quantifying the current market equilibrium rate, this study examines whether the foreign exchange market is 'efficient' enough to produce the current market equilibrium rate. Then this equilibrium hypothesis is subjected to two efficiency tests.

First, we test the theory of covered interest parity hypothesis. Second, the predictive performance of forward market is examined. Our results provide strong evidence that the covered interest parity condition holds in the New Zealand foreign exchange market. However, our results fail to accept the forward market efficiency hypothesis. Given the above empirical results, it can be suggested that the market is not efficient enough to ensure that the current market equilibrium is achieved.

The alternative equilibrium exchange rate hypothesis is known as fundamental equilibrium. This equilibrium hypothesis suggests that the equilibrium real exchange rate is a function of a number of real variables. In this study, a three-sector general equilibrium model is presented and the real exchange rate is defined as the relative price of nontradables to importables (or the relative price of nontradables to exportables). The model suggests that the determinants of the equilibrium real exchange rate include the terms of trade, factor endowments, welfare levels and real balance of payments.

The empirical evidence rejects the null hypothesis that real exchange rates (RERs) have persistently deviated from their fundamental equilibrium values. As a result, an Error Correction Mechanism (ECM) model of RER determination is constructed. However, the error correction terms in the model indicate that the actual RERs in the period from 1985 to 1990 have indeed exhibited a behaviour different from what the fundamental variables suggested. Furthermore, the estimated model suggests that the rate of convergence to the equilibrium is very slow.

In order to examine the dynamic interactions among the variables and the stability of the system, the impulse analysis is conducted with the ECM model and the VAR in differences model. The results of this analysis suggest that the response of both models are similar. Finally, a forecasting experiment is carried out to compare the forecasting ability of the ECM and VAR in differences models. Results of the experiment confirm that the ECM model performs better at the longer horizon than the VAR in differences model.