RDP 8711: Deviations from Purchasing Power Parity: The Australian Case 1. Introduction

In its simplest form the purchasing power parity (PPP) hypothesis states that, in equilibrium, international differences in the price of any bundle of goods will be constant when expressed in a common currency. That is, the roal exchange rate will be constant. In Australia, the real exchange rate, which is the inflation-adjusted nominal exchange rate, has shown an overall downward trend from the early 1970s. There have also been persistent deviations from that trend.

Purchasing power parity is usually associated with the notion of balance of payments equilibrium. In a sustainable long-run equilibrium the current account will be a constant proportion of GDP (for example zero) and, in many economic models, there is a unique value of the real exchange rate associated with that level.

Shocks can drive the real exchange rate away from its equilibrium level. To bring about a return to equilibrium, the nominal exchange rate may adjust, relative price levels may adjust, or there may be some mix of adjustments. Since prices tend to be more “sticky” than nominal exchange rates, the nature of the adjustment process may differ between exchange rate regimes.

Adler and Lehmann (1983) demonstrated that for many countries, over both fixed and flexible exchange rate periods, deviations from PPP evolve in a random fashion.[1] They show that there is no systematic tendency for exchange rates in “real” (i.e. inflation-adjusted) terms to revert to a constant equilibrium level (PPP) following a deviation from parity. The issue is of interest because of its implications for balance of payments adjustment. If the real exchange rate has no tendency to revert to an equilibrium level, then equally there may be no tendency for the balance of payments to settle down at its equilibrium level.

An alternative explanation of deviations from PPP is that the equilibrium value of the real exchange rate itself night be changing in response to shifts in economic “fundamentals”. In particular, a small economy with significant trade in commodities may be subject to sustained changes in its terms of trade consequent upon shifts in commodity prices. Such changes call for sustained shifts in the real exchange rate.

Such shifts in the real exchange are not a purely random process. They are partly predictable, given information about commodity price developments. Shocks may still drive the real exchange rate away from equilibrium. But there may be a tendency to revert back to an equilibrium level modified by any changes in the terms of trade during the intervening period.

In this paper, we derive a model of deviations from PPP that follow a random walk. Wo test the null hypothesis that the real exchange rate is best modelled as a random walk against two alternative hypotheses about the determinants of movements of the real exchange rate:

  • that the real exchange rate tends towards a constant long-run equilibrium level (long-run PPP); and
  • that the real exchange rate tends towards an equilibrium level which is itself a function of shifts in the terms of trade. (Long-run PPP does not hold, but for reasons that are at least partly predictable.)

The paper aims to provide evidence on whether movements of Australia's real exchange rate are a purely random process, or whether they are likely to revert to an equilibrium level – constant or variable. A secondary aim is to provide some insight into the role of the exchange rate regime in deciding this issue.

In the second section the behaviour of the real exchange rate in Australia from the early 1970s is examined. He distinguish between the periods before and after December 1983, when the Australian dollar was floated. In the third section a model is derived which is capable of explaining deviations from PPP as a random walk. Two alternative hypotheses against which the random walk model can be tested are considered: one ignores relative commodity price developments, while the other takes then into account. In the fourth section the random walk model is tested against the alternative hypotheses on Australian data, for the real bilateral exchange rate against the U.S. dollar and the real trade-weighted index. Quarterly and monthly observations and various lag lengths and sample periods are employed. In the fifth section some dynamics of commodity price influences on competitiveness under floating exchange rates are explored. Finally, in the sixth section, some concluding remarks are offered.

Footnote

Specifically, they assume that deviations from PPP are a “martingale process”. The expected value of the dependent variable at time t is the current value of the dependent variable. A random walk is a particular type of martingale that assumes independent and identically distributed errors. In fact, the econometric techniques employed by Adler and Lehmann make this latter assumption. Consequently, in the rest of this paper we shall refer to their model as the random walk model. Under the random walk hypothesis about deviations from PPP, changes in the real exchange rate should be serially independent. [1]