RDP 8609: The Performance of Exchange Rate Forecasts 1. Introduction

“The personal traits shared by most spot traders are often quite an explosive mixture … Their most elusive feature – and the one they are mostly paid for – is their “instinct” to pick the trends or swings in the market … Hand in hand with what could be called a trader's “second sight”, superstitution also appears to have a niche in the forex market … For example, one spot trader for inspiration peers at his dealing screen through a pair of black lace woman's panties.” Australian Finanacial Review
12 December 1985, p.17

Since the floating of the Australian dollar in December 1983 the domestic foreign exchange market has grown considerably. The last two and a half years have seen the number of licensed dealers increase from around 10 to over 70 and the volume of business transacted increase many fold. Paralleling this growth has been an increase in interest in the behaviour and operation of the domestic foreign exchange market. This growth has seen the market and its reaction to economic and political events become, in many eyes, one of the principal barometers of the performance of the Australian economy.

Despite the importance of the foreign exchange market the ability of its participants to provide reasonable forecasts of future exchange rates has been questioned widely. Headlines such as “Cain Lashes Forex ‘Whiz Kids’”,[1] “BT Charts by the Light of the Moon”[2] and “Scepticism of Currency Forecasts Grows”[3] and comments such as “predicting the exchange rate level of the $A even one week ahead is rapidly becoming a game of Russian roulette”[4] have appeared frequently in the financial press over the last twelve months.

Notwithstanding these headlines, and the theory of efficient markets which suggests that all available information regarding the future exchange rate is incorporated into the current exchange rate, foreign exchange dealers and corporate advisers have little choice but to continue making and publishing foreign exchange forecasts.

While the accuracy of say a one week ahead or one month ahead forecast may not significantly influence the profitability of a trader (or his/her employer) who opens and closes positions over a period of minutes, the accuracy of such forecasts may be important to the corporate customers of the forecaster. The increased volatility of the Australian dollar since the float has made the need for such forecasts, and the consequent decisions regarding the timing of foreign exchange transactions, of increased importance to many corporate treasurers. Therefore, a failure of the forecaster to provide reasonable forecasts should be of some concern to corporate treasurers.

Any analysis of the forecasting performance of various forecasters is made difficult by the lack of a relatively long, publicly available, and consistent series of forecasts. However, during 1985 the Australian Financial Review (AFR) surveyed various foreign exchange dealers about their expectations of exchange rate movements over a one week horizon. This paper uses these forecasts to throw some light on the above criticisms.

Section 2 of the paper discusses the nature of the AFR's survey while Section 3 defines a number of benchmark forecasting models against which the performance of the forecasts published in the Australian Financial Review are compared.[5] The results of this comparison for the $A/US$ exchange rate forecasts are presented in Section 4, while Section 5 compares the performance of these forecasts with that of forecasts of the US$/Yen exchange rate. Section 6 compares the performance of individual forecasters with a “no change” forecast model and with the “group mean” forecasts. The principal conclusions of the paper are drawn together in Section 7.


Australian Financial Review, 4 July 1986, p9. [1]

Australian Financial Review, 10 June 1986, p1. [2]

The Australian, 28 February 1986, Special Report, p3. [3]

Australian Financial Review, 21 November 1985, p1. [4]

These forecasts will be referred to as the “AFR survey forecasts”. [5]