RDP 9803: Forward-Looking Behaviour and Credibility: Some Evidence and Implications for Policy 1. Introduction

It is universally accepted that expectations play a crucial role in the way the economy works, but there is considerable dispute about how expectations are formed. In particular, there is debate about whether people anticipate future changes in policy when they form their expectations, or whether they simply extrapolate what has happened in the recent past into the future. This paper analyses some issues related to the way expectations are formed and examines a number of policy implications. The first half of the paper examines inflation forecast errors and the information people use to form their expectations of the future. The second half explores some implications for monetary policy.

In Section 2, we assess three pieces of information about expectations processes. We start by examining the ‘rationality’ of inflation expectations as measured by surveys of financial market economists and households. These survey measures are found to be inconsistent with rational expectations, although the inconsistency is smaller for financial market participants than for households. We also examine the relationship between actual inflation and households' expectations, and show that they interact with each other, such that expectations help predict actual inflation and vice versa. Current beliefs about future inflation have an effect on future inflation, but these expectations are also affected by past inflation.

The characterisation of expectations probably differs across sectors in the economy, with the financial markets, for example, typically regarded as more explicitly forward looking. We look, therefore, at how perceptions in the foreign exchange market of future monetary policy have evolved over time. If inflation is higher than expected, financial markets will expect higher future real interest rates and an appreciation of the exchange rate (all else given) if they believe that the central bank will not accommodate higher inflation. We show that in recent years inflation surprises have on average induced a movement of the Australian dollar in the same direction as the inflation surprise, whereas the response was more ambiguous during the 1980s. This is consistent with participants in the foreign exchange market forming their expectations on the basis of a model of the economy in which the Reserve Bank will not accommodate inflation shocks.

In Section 3, we use the simple data-consistent model of the Australian economy set out in de Brouwer and O'Regan (1997) to examine some implications of forward-looking behaviour and credibility. We look, first, at forward-looking monetary policy. Forward-looking policy reduces the variability of inflation around target and output around potential considerably more than when policy reacts only to current inflation and output statistics. We find that a rule based on four-quarter ahead forecasts of inflation and the output gap outperforms rules based on other forecast horizons, given that policy-makers use forecasts which are consistent with the policy rule (rather than assuming, for example, that the nominal interest rate is constant over the forecast horizon).

The effect of shocks also depends on the way in which expectations are formed. To illustrate this, we set out a model which embodies two extremes of how price, wage and exchange rate expectations can be formed. One case is that people form their expectations of the future based on the structure of the model, including how the central bank sets monetary policy. The other case is that their expectations of the future are simply extrapolations from the recent past. A more realistic model is probably some combination of these two cases. Consistent with earlier work, our results show that shocks are less destabilising when people know the structure of the economy and the central bank's reaction function, rather than when they rely only on data from the recent past. Moreover, when groups of people in the economy form their expectations about inflation in different ways, the ex ante real interest rates that influence activity and the real exchange rate can differ from each other. This can induce oscillations or overshooting in the exchange rate, depending on the particular mix of expectations, with implications for the variability of inflation and output.

Given that expectations which are formed only on the basis of past data tend to increase the variability of inflation and output in response to shocks, we examine whether policy-makers can compensate for this if they set interest rates ‘optimally’. For a given set of shocks and using a common basic economic structure, we find that optimal policy cannot fully compensate for the greater variability produced when people take insufficient account of the structure of the economy and future monetary policy in forming their expectations. In this sense, policy-makers have to work with the economic structure around them.