Nominal exchange rates, commodity prices and central bank policy

Abstract

This thesis consists of three independent chapters on nominal exchange rates. The first chapter adds to the forward bias puzzle by noting that while the exchange rate of a small commodity-exporting economy can be closely tied to commodity prices, a portfolio of commodity futures exhibits little if any bias. This is demonstrated for Australia. Using a dependent economy model in which the exchange rate is a function of export prices, three potential explanations for the bias of exchange rate futures, but not commodity futures, are considered. Peso problems do not seem capable of explaining the puzzle. Monetary policy could explain some of the bias, though unlikely the full extent. Systematic expectation errors about the monetary process, while requiring strong assumptions, receive some empirical support from the behaviour of the exchange rate. The second chapter attempts to resolve the endogeneity of exchange rates and central bank intervention. Using a change in Reserve Bank of Australia intervention policy for identification, simulated GMM is used to estimate a model that includes the contemporaneous impact of intervention. Intervention is found to have an economically significant contemporaneous effect. A $US100m purchase of the domestic currency will appreciate the exchange rate by 1.35 to 1.81 per cent. Further, intervention is found to have the majority of its impact during the day in which it is conducted, with a smaller effect on subsequent days. Australian central bank intervention policy is confirmed to be characterised by leaning against the wind.

(cont.) The third chapter estimates the dependent economy model outlined in Chapter 1 for the Australian, Canadian and New Zealand dollars. The model provides a good representation of the exchange rates for all three countries up to 1995. In out-of-sample projections the Australian and Canadian models outperform a random walk. The New Zealand model breaks down during the Asian crisis. Commodity futures are used to construct forecasts of the Australian dollar, which at horizons of around one year are more accurate than no-change forecasts.