RDP 9607: Towards an Understanding of Australia's Co-Movement with Foreign Business Cycles 5. The Propagation of Business Cycles

Nicolas de Roos and Bill Russell

November 1996

It was argued above that the correlation between Australian and foreign business cycles may be partly explained by the correlation between share markets. But this explanation requires that foreign and domestic activity variables respond to their respective share markets in similar ways. This section looks at evidence for that proposition by looking at sharemarket correlations with three variables – investment, consumption and output – in Australia and the US. [27]

Table 8 shows the correlation between detrended Australian investment and lags in the detrended Australian real share price, at an aggregate level and for the mining and manufacturing sectors.[28] We see that investment lags the real share price with a peak in the correlation occurring with a lag of between three and seven quarters, depending on the industry and the level of aggregation in investment. Also in Table 8 are similar correlations for the US using aggregate investment. These show a similar pattern of correlations and lags to those found for Australia, except in two respects. First, the correlations are lower for the US; and second, the peak lags in Australia are 1 quarter longer. The low correlation may be because the US share market is a proxy for a ‘world’ hare market and, therefore, focuses less on the US.

Table 8: Investment Gap and Real Share Prices for Australia and the US: Correlations and Associated Lags
  Australia United States
  Aggregate investment Mining Manufact Aggregate investment
Lags in
real share
price		 Total
 		 Plant and equipment Non-dwelling construction Plant and equipment Plant and equipment Total
 		 Plant and equipment
 
0 0.23 0.01 0.13 −0.20 0.37 0.15 0.06
1 0.40 0.18 0.22 −0.12 0.48 0.25 0.10
2 0.56 0.38 0.33 −0.04 0.58 0.27 0.14
3 0.62 0.45 0.41 0.06 0.57 0.25 0.17
4 0.62 0.48 0.45 0.19 0.60 0.19 0.16
5 0.55 0.44 0.46 0.32 0.59 0.03 0.09
6 0.45 0.39 0.44 0.38 0.53 −0.09 0.00
7 0.30 0.24 0.40 0.41 0.47 −0.17 −0.07
8 0.15 0.08 0.33 0.35 0.4 −0.22 −0.12
9 0.02 −0.05 0.24 0.28 0.32 −0.22 −0.17
10 −0.13 −0.17 0.09 0.24 0.29 −0.18 −0.17

Notes: Shading identifies the lag with the highest correlation. The ‘gap’ is the difference between the log level of the variable and a linear time trend. It is approximately the percentage deviation from trend of the variable. For Australia, the mining real share price index uses the ‘All Mining’ accumulation share price index deflated by the mining sector investment deflator. Similarly, the manufacturing real share price uses the ‘All Industrials’ accumulation share price index deflated by the manufacturing investment deflator.

An alternative way to display the peak correlations is shown in Figures 8 to 11 where the peak correlation coefficients between the variables are shown with the number of lags at which that peak occurs. The arrows in the figures point to the lagging variable in the correlation.[29] We observe that the Australian and US real share prices are highly correlated, with the Australian share market lagging by 1 quarter. In Figures 8a and 8b we also see that lags between the real share price and investment, and between investment and GDP, are similar in the US and Australia. Therefore, if the real share price is influencing investment within each country, and if the Australian and US real share prices are closely related, then it is likely that the business cycles will be correlated.

Figure 8a: Real Share Price, Total Investment Gap and GDP Gap
Peak correlations and associated lags
Figure 8a: Real Share Price, Total Investment Gap and GDP Gap
Figure 8b: Real Share Price and Plant Equipment Investment Gap
Peak correlations and associated lags
Figure 8b: Real Share Price and Plant Equipment Investment Gap

Similarly, Figure 9 shows the correlation between the real share price and consumption. This correlation diagram displays greater divergence between the correlation and lag structures of the two countries than was the case with the investment diagrams. The lags between the real share price and consumption are considerably longer in Australia than in the US. Consequently, US and Australian consumption are not highly correlated and there is a long lag in the peak correlation. It would appear that the interaction between the share market and consumption is not as well defined in Australia and is unlikely to explain a large proportion of the correlation in business cycles.

Figure 9: Real Share Price, Consumption Gap and GDP Gap
Peak correlations and associated lags
Figure 9: Real Share Price, Consumption Gap and GDP Gap

Finally, Figure 10 shows the correlation at the aggregate level between the real share price and GDP for the US and Australia. The peak lags between the respective share markets and GDP are the same, consistent with the high correlation between Australian and US GDP.

Figure 10: Real Share Price and GDP Gap
Peak correlations and associated lags
Figure 10: Real Share Price and GDP Gap

A common feature of these peak correlation diagrams is that the lags emanate from the US share market. If the correlation in business cycles was due to simultaneous world-wide shocks to the real economy, it would be unlikely for the share market to regularly lead the shocks by the number of periods shown in the figures. This casts doubt on the proposition that the correlation of business cycles is attributable to simultaneous world-wide shocks to the real economy. Even if the shocks impact on share markets first, we would expect contemporaneous correlations between the Australian and US share markets. It appears, therefore, that a case can be made that the US share market ‘drives’ the Australian share market and, because of the similar domestic responses, this contributes to the correlation in business cycles.

Footnotes

This analysis does not imply causation. Instead, it is a consistency check of our more general analysis which implies a particular pattern of correlations between a number of the variables. [27]

The consumption, investment and output variables in this section have all been logged and detrended using a linear trend. The ‘gap’ between the actual level and the linear trend is approximately the percentage deviation from trend of the variable. [28]

For example, the maximum correlation of the Australian real share price and US real share price is 0.65 which occurs when the Australian real share price lags the US real share price by one quarter. When no arrow head is shown, the peak correlation occurs contemporaneously. [29]