RDP 9105: Inflation in Australia: Causes, Inertia and Policy 1. Introduction

Australia's inflation rate has been high since the early 1970's, both in absolute terms and relative to the other developed market economies.^[1] Although the costs of this inflation have been difficult to quantify, most economists and policy makers agree that decreasing the rate of inflation ought to be one of the leading priorities for macroeconomic policy.

In this paper, we aim to analyze Australia's inflation rate from three perspectives. First, we examine the contributions of four factors – the growth rates of money, nominal wages and productivity, and world inflation – to inflation in the short-term. Second, we examine the contributions of these factors to the long-term (equilibrium) inflation rate. Third, we investigate how much inertia there is in the inflation rate, i.e. whether inflation can deviate from its equilibrium rate for a significant period of time. For comparative purposes, we similarly examine inflation in Australia's four major trading partners: Japan, the United States, the United Kingdom and New Zealand.

We intend particularly to examine the importance of aggregate wage growth in the determination of price inflation. The control of wage inflation has always assumed a great deal of importance in the making of Australian economic policy, often because of its perceived influence in restraining price inflation (Milbourne 1990). In recent years this argument has been used as one of the chief justifications for the centralized wages system. Recently, however, the costs of this system viz. the reduced flexibility of relative wages, have received greater recognition and some steps have been taken towards a deregulated wages system. Such a system, however, necessarily implies that policy makers will lose some (perhaps all) direct control over aggregate wage outcomes.

This would not necessarily be an adverse development, of course, provided goods and factor markets are competitive and clear continuously, i.e. prices in all markets are flexible. Under these circumstances, allocative efficiency suggests that the relative prices of various types of labour ought to be determined by relative scarcities, with the aggregate wage outcome of no more macroeconomic importance than, say, the aggregate peanut price outcome. Control of inflation can be achieved by control of an appropriate nominal quantity, such as the supply of money.

In practice, however, markets do not commonly reflect this competitive ideal. Wages are often determined by bargains between unions and firm. Relative wage outcomes, even when market determined, are often influenced by such irritating considerations as perceived fairness (Akerlof and Yellen 1990), while prices in imperfectly competitive markets will be determined as markups over costs (Blanchard and Fischer 1989, pp. 465–468). Under these circumstances, we have to ask whether a satisfactory outcome for price inflation can be delivered when goods and factor markets do not function perfectly and when policy makers have little or no control over the growth of aggregate wages.

Previous studies of Australian inflation have found wage growth to be an important determinant of price inflation. Carmichael (1974) constructs a model of inflation from micro-foundations, with a reduced form inflation equation estimated with quarterly data over the period 1960(1)–1973(3). This study finds labour productivity, world tradable prices and expected wage growth to all exert significant effects on inflation.

Nevile (1977) models the growth rate of the GNE deflator in Australia using annual data for the period of 1954/55 to 1973/74. He finds award wage growth, inflation expectations and excess demand to be significant determinants of the inflation rate.

Boehm (1984) investigates Granger causality between wages growth and inflation in Australia.^[2] Using quarterly data over the period 1954–82, Boehm finds Granger causation from wages to prices, but not vice versa. Alston and Chalfant (1987) extend Boehm's bivariate study by including money supply (M1) growth in their model and find lagged money supply growth to Granger-cause wage growth and inflation, and challenge Boehm's conclusion that wage growth causes price inflation. Their preferred interpretation of the data is that the causal links run from money to wages and money to prices with different time lags.

While the determinants of Australian inflation have been studied extensively, we are not aware of any previous work that has examined the degree of inertia in Australia's inflation rate. This stands in contrast to studies of inflation in other countries where the inertia issue has received some attention. Gordon (1985) finds lagged inflation to be a significant determinant of US inflation. In a recent study Gordon (1990) examines inflation in the US, UK, Japan, France and Germany over the period 1873–1987. He finds the emergence of considerable inertia in the last three decades for all countries except Japan.

The rest of the paper is organized as follows. In Section 2 we derive a structural model of the equilibrium inflation rate, and discuss dynamics which take into account possible inertia in the adjustment of prevailing inflation to its equilibrium rate. In Section 3 we examine the exogeneity of money and nominal wages. Evidence pertaining to the short-term determinants of inflation is presented in Section 4. In Section 5 we report the results from the estimation of our dynamic model of the equilibrium inflation rate and draw some implications for policy in Section 6.

Footnotes

Over the period 1973–1989 Australian and OECD average annual inflation rates (measured by the CPI) were 9.7 percent and 7.7 percent, respectively. In terms of GDP(GNP) deflators, they were 10.0 percent and 6.6 per cent respectively. (Sources: OECD Main Economic Indicators, Australian Bureau of Statistics Catalogue Nos. 6401.0 and 6442.0)
Carmichael (1990) provides an extensive overview of inflation in Australia during the 1980's. [1]

A variable x is said to Granger cause another variable y if, in a regression of y on lagged values of both x and y, the coefficients on those lagged values of x are jointly significantly different from zero. [2]