RDP 9808: What Moves Yields in Australia? 4. A Diversion: International Influences

The news in Table 1 all hits the screens during the domestic trading day, which is the focus of this paper. Australia, however, is highly integrated with the global financial system, being completely open to international capital flows: Australia has a deregulated financial system, a floating exchange rate and no exchange controls. Consequently, Australian financial markets are potentially exposed to the gamut of influences on international capital flows and to any news likely to move major overseas markets.

While the important overseas news from the United States and Europe is almost always released when Australian markets are closed, yields on Australian securities can respond directly to such news since the SFE's bill and bond futures contracts are traded overnight on SYCOM, the SFE's electronic trading system.[12] Thus, by executing trades on SYCOM, traders of the 10-year bond futures contract in Australia, for example, can (and do) react as soon as, say, US non-farm payrolls data appear on screens in New York. In the first instance, such a reaction is almost always a ‘reflex’ response to any move in US yields at the time of this news. More generally, however, the close links between the Australian and US economies, including the well-documented similarity in the respective business cycles, mean that US news may contain information fundamentally relevant for changes in Australian interest rates.

Tick-by-tick readings for bill and bond yields from overnight trading on SYCOM are available. But overnight trading lacks the breadth of participation of daytime trading and volumes on SYCOM are much lower than during the day. Accordingly, movements in Australian yields on SYCOM might be less reliable than those recorded during the day. For this reason, we have not made use of the high-frequency overnight readings but have adopted a different, somewhat less precise, approach to estimating the effect on domestic yields of major releases of US news.

This approach involves dividing the 24-hour period into two parts: the daylight session between 8.30 am and 4.30 pm when the floor of the SFE is open for trading, and the overnight session (from 4.30 pm to 8.30 am the next day) when any movement in yields would occur on SYCOM. We then estimate a pooled regression, with the absolute change in yields during both the daylight and overnight sessions regressed on two dummy variables (Equation 1 below). The first dummy variable takes a value of 1 during the daylight session on days when any of the announcements in Table 1 are made, and 0 otherwise (that is, it takes a value of 1 in the daylight session when there is ‘domestic’ news). The second dummy variable takes a value of 1 during the overnight session on days when US economic announcements are made, and 0 otherwise.

The overnight dummy variable is defined for the US economic announcements which Fleming and Remolona (1997) found to have an important effect on US bond prices. These are (in descending order of importance): non-farm payrolls, the Producer Price Index, Fed monetary policy announcements, retail sales, the Consumer Price Index, the National Association of Purchasing Managers' (NAPM) index, industrial production and GDP.[13]

We also include two synthetic variables in the regression to capture any relationship between the movement in yields during a particular trading session and the movement in yields in the preceding session. For example, although Australian yields respond to US news during the overnight session, there is sometimes a ‘second round’ effect when Australian markets open the next morning, as local traders assimilate the US news. Similarly, when markets in the US open, traders may respond to the movement in yields during the daylight session in Australia.


|Yns| is the absolute change in yields on day n for both the daylight and overnight trading sessions, s; where s1 = daylight session and s2 = overnight session.
S1 = |Yns2|−1 in the daylight session, 0 in the overnight session.
S2 = |Yns1|−1 in the overnight session, 0 in the daylight session.

In this section, the focus is on US economic announcements, since experience shows that they consistently affect global markets, whereas European statistical news has a more isolated incidence. Regular economic announcements from Asia often occur during the trading day in Australia but rarely had much effect on Australian markets in the period being reviewed. Over the period, there were 285 days when local announcements were made and 242 days when US announcements were made.

The important difference between the approach undertaken in this section and that in the following section is that the absolute value of daylight and overnight changes in Australian bill and bond yields are, respectively, regressed on a common dummy variable for any Australian announcement and a common dummy variable for any US announcement. In the next section, by contrast, a dummy variable is created to measure the separate effects of each item of local news in the five minutes after it is released.

While the approach adopted here has obvious deficiencies, it might, at least, provide some indication of the relative information content of domestic and US announcements for yields in Australia. The estimated effects of Australian and US announcements on volatility in Australian bill and bond yields are shown in Table 3.

Table 3: Impact of US and Australian Announcements on Yields in Australia(a)
  Bill yields
  Episode I Episode II Episode III Whole period
Australian news 2.7** 1.2* 2.5** 2.1**
US news 2.0** 1.8** 0.6* 1.5**
  Bond yields
  Episode I Episode II Episode III Whole period
Australian news 0.7 1.5** 1.6** 1.2**
US news 4.3** 2.5** 2.9** 3.3**

Notes: (a) The formal results from these regressions are contained in Table A1 of Appendix A.
* and ** denote significance at the 5% and 1% levels, respectively.

The estimates in Table 3 can be interpreted as the average absolute movement in yields, in terms of basis points, on days when fresh news arrived in the respective trading session, compared with days when no fresh news arrived. For example, over the whole period, Australian announcements were estimated to have induced an average absolute movement of 2.1 basis points in bill yields in Australia during the daylight trading session; similarly, US announcements were estimated to have induced an average absolute movement of 1.5 basis points overnight in bill yields in Australia. The corresponding estimates for bond yields were 1.2 basis points for Australian announcements and 3.3 basis points for US announcements. This implies that Australian announcements had a larger impact on bill yields in Australia than did US announcements, whereas US announcements had a larger impact on bond yields in Australia than did local announcements. The estimated coefficients on the synthetic variables (as shown in Appendix A) suggest some small ‘second round’ effects, on both bill and bond yields, from US news when the Australian market opens, and from Australian news when the US market opens.

Other features of Table 3 are:

  • the effect of Australian announcements on bill yields declined in the second episode, perhaps because of the period of stability in monetary policy in Australia, which encouraged expectations that policy would remain ‘on hold’. By contrast, the effect of Australian news on bond yields increased during this period – a period in which news about inflation was virtually pre-eminent; and
  • the size of the impact of US announcements on bond yields in Australia was largest in the first episode when US monetary policy was being tightened. It was also larger than the effect of Australian news in the other episodes, although the gap between the relevant coefficients was smaller. The insignificant effect of local news on bond yields in the first episode – and the large effect from US news – implies that US developments simply overwhelmed local news in the global bear market of 1994.

These results are consistent with the idea that domestic factors are more important than international influences in the consideration of monetary policy. On the other hand, the global bond market has become increasingly highly integrated and, especially in a world of universally low inflation, movements in bond yields internationally have become more tightly synchronised (see Kortian and O'Regan (1996) and Kortian and Thompson (1998)). As the US market is the largest and most liquid in the world, developments there inevitably set the tone elsewhere.

As noted, the estimates in Table 3 are crude and are perhaps most useful as an indication of the relative information content for the local fixed-interest market of the release of local as opposed to US news. They underscore a point, however, now more widely recognised about the significance of the US market for developments in Australian interest rates, especially bond yields. While in 1994 the effect of US developments on Australian yields may have been extreme, US news had a smaller, but still large influence on short- and, especially, long-term interest rates in Australia in the later periods of more orderly markets.

With this background, the remainder of this paper focuses on the impact of Australian news on bill and bond yields during the daylight trading session.


SYCOM is the Sydney Computerised Overnight Market, which is a computerised trading system open between 4.40 pm and 6.00 am Eastern Standard Time (7.00 am Eastern Daylight Saving Time) for trading in futures and options contracts on a number of financial instruments. [12]

The size of the announcement effect of individual items of US news on Australian yields would, presumably, be highly correlated with the effect on US yields. We exclude some financial information also found to have an effect on US bond yields, such as the announcement of results of some regular Treasury Auctions – partly because Fleming and Remolona (1997) find them to be of less importance than the run of economic news. [13]