RDP 9808: What Moves Yields in Australia? 2. Explaining Market Movements

The efficient market hypothesis suggests that asset prices reflect all available information and react only to unanticipated news that might be expected to affect fundamental asset values.^[2] Investors and portfolio managers make decisions on the basis of expectations about performance over the term of their investments, and constantly seek new information that confirms, or challenges, existing assessments. The steady stream of news about the economy and financial conditions provides investors with a basis for comparing expectations against reality as it unfolds. Since policy settings are also likely to influence economic performance, policy announcements themselves are likely to affect returns on assets, especially financial assets.

The arrival of new information can affect different markets in different ways. Fleming and Remolona (1997) note that the effect of new information on share prices has been difficult to quantify in most studies. They consider that this is because information about macroeconomic developments can have ambiguous implications for share prices: an upward revision to expected real activity, for example, would raise the expected cash flow for a stock (via increased dividend payments), thereby increasing its present value; but it would also increase the discount rate, which would tend to have an offsetting effect. The net effect on share prices would depend on whether the cash flow effect dominated the discount rate effect, or vice versa. Hardouvelis (1988) argues that some information, for example about inflation, could, in the short term, have similarly ambiguous effects on exchange rates.^[3]

By contrast, cash flows from fixed-interest securities (coupon payments) are set in nominal terms, and are, therefore, independent of revisions to expectations about economic activity. Upward revisions to expected economic growth (which led to higher inflationary expectations) would be fully, and unambiguously, reflected as a fall in the price of a fixed-interest asset via an increase in the discount rate. Thus, ‘announcement effects’ from unexpected macroeconomic news might be more readily identified in fixed-interest markets than in share markets or in foreign-exchange markets.

Fleming and Remolona survey the evidence from a number of studies of announcement effects on the US bond market; some of these studies are based on daily data, with announcements measured exclusively in terms of their surprise element, i.e. deviations of published estimates from market expectations (measured in a variety of ways). The announcements most frequently cited as significant for bond prices in the US have been those for the money supply, industrial production, the Producer Price Index, the Consumer Price Index, the unemployment rate and non-farm payroll employment.

The evidence also points to a ‘flavour-of-the-month’ effect; i.e. some announcements seem to be in fashion for a time, before losing the market's attention. For example, without pre-empting the later discussion, the market reaction to the release of current account data in Australia has varied over the years. This may be more than a mere change of fashion; it may reflect genuine policy content such as a change in the monetary policy regime or better understanding of the policy regime. In the US, money supply announcements have declined in significance in studies conducted since the mid 1980s, a period in which the importance of financial aggregates as a guide to monetary policy progressively diminished.

Studies using intra-day price data have found a wider range of announcements to be significant, including announcements of changes in monetary policy by the Fed (Fleming and Remolona 1997), a practice introduced in 1994.^[4] Fleming and Remolona find that market participants distinguish between announcements which contain inherently different information, irrespective of any surprise content, although they find that the magnitude of any surprise enhances the estimates of ‘announcement effects’ on bond prices. Urich and Wachtel (1981) had earlier concluded that survey data of market expectations (of the kind Fleming and Remolona use) were biased, or did not contain all available information. When expectations were modelled as an ARIMA process they found a stronger effect from the unanticipated element of news. We, following Fleming and Remolona, base our study on survey information because market participants are more likely to respond on the basis of their in-house economists' views, or from consensus forecasts, rather than from a more mechanical approach such as an ARIMA process.

Footnotes

There are three versions of the efficient market hypothesis: weak; semi-strong; and strong. These versions differ by their notions of what is meant by the term ‘all available information’. For more detailed definitions of these versions, see Bodie, Kane and Markus (1993). [2]

In Australia, de Brouwer and Ellis (1998) suggest that inflation surprises tend to lead to an appreciation of the exchange rate. This conclusion is based on estimating long-term behavioural relationships rather than on the search for virtually instantaneous market reactions – the approach in the current paper and earlier work from which it draws. [3]

This practice was introduced in Australia in 1990, as discussed in Battellino, Broadbent and Lowe (1997), who also discuss market reactions to monetary policy announcements. [4]