RDP 9704: Financial Aggregates as Conditioning Information for Australian Output and Inflation 2. Background

Estrella and Mishkin (1996) suggest three potential roles for the financial aggregates in the conduct of monetary policy: as information variables, as indicators of policy, and as instruments of policy. These roles require successively stronger and more stable relationships between the aggregates and the final goals of monetary policy in order to perform satisfactorily. The authors find little evidence of the required relationships in the United States and in Germany, suggesting that the monetary aggregates are not good indicators of the stance of policy and have little value as information variables.

In Australia during the late 1970s and early 1980s, financial aggregates, and particularly M3, were a primary focus of monetary policy. As was the case in many countries, there were explicit target rates of growth for money, based on the idea that there was an underlying, stable relationship between the aggregates and the objectives of policy, namely real output growth and inflation. This use of financial aggregates would align with the Estrella and Mishkin idea of financial aggregates as instruments of policy.

Since the mid 1980s, the policy role of financial aggregates, in Australia as in many other industrialised countries, has been de-emphasised. In Australia, changes in the regulation of financial intermediaries and various innovations of financial products altered the perceived relationship of financial aggregates with real output and inflation. Currently, monetary policy in Australia is implemented by direct changes to the short-term rate of interest, with financial aggregates having become just another of the many information variables used in formulating monetary policy.[1]

Tallman and Chandra (1996) review the literature on investigations of whether information on financial aggregates helps predict real output growth or inflation in Australia. The general conclusion from the literature is that there is little evidence supporting financial aggregates as independent explanators of subsequent real output growth and inflation for Australia, and such evidence appears to weaken further as the sample is extended. The implication is that while the financial aggregates might provide some corroborating evidence regarding future developments, they provide little if any information that is not contained in other variables.

In this paper, we investigate further the findings in Tallman and Chandra (1996) and the results in the literature. We focus on the information content of financial aggregates in Australia in two distinct ways. First, we employ a technique introduced by Roberds and Whiteman (1992).[2] In their paper, they examine the observed decline in the predictive value of monetary data for real output and the price level in the United States using VAR techniques. They notice a substantial decline in the information content of specific aggregates for explaining the behaviour of policy targets after the apparent change in Federal Reserve operating procedures in the third quarter of 1979. An innovation in their paper is the technique that allows comparison of conditional versus unconditional forecast accuracy, expanded upon below. We apply this technique to Australian data as an additional method of evaluating the information value of financial aggregate data for policy. The experiment essentially compares the in-sample forecast errors of an unconditional forecast in which each variable in the VAR must be forecast, with the in-sample forecast errors from a conditional forecast that assumes perfect knowledge of the financial aggregate measures eight quarters into the future.

The second method we employ examines whether adding financial aggregate variables to restricted reduced-form single-equation models of real output growth and inflation improves the explanatory power of those single-equation models. These models are re-estimations of existing, rigorously fitted models of Australian output growth (Gruen and Shuetrim 1994) and inflation (de Brouwer and Ericsson 1995). To examine the information content in financial aggregates for these measures, we employ tests of the marginal significance of the aggregates in these single-equation models.


Astley and Haldane (1997) suggest that data for policy analysis in general can either contain incremental information that is not available in other sources or may simply corroborate what other indicators reflect. [1]

Roberds and Whiteman (1992) investigate whether monetary aggregates in the United States are useful for explaining real output and price-level fluctuations using a VAR methodology. The prediction exercise was a key test of the relationships in the US data between output, the price level and chosen monetary aggregates. [2]