RDP 2007-09: Private Business Investment in Australia 5. Conclusions

The modelling work in this paper shows that it is possible to explain a sizeable proportion of the variation in aggregate Australian business investment using models that make economic sense. However, this has only been possible by introducing two innovations. The first has been to recognise the special nature of computing equipment and to exclude it when estimating models of equipment investment. The second has been to exclude outlying or influential observations from the estimation, recognising that regression models may not be able to explain all the extreme short-term movements in the data.

The modelling work makes two key points about the determinants of Australian business investment. First, there is a significant negative long-run relationship between the investment share of GDP and the cost of capital for all types of investment. In contrast, the traditional neoclassical long-run model – based on an inverse relationship between the capital-to-output ratio and the cost of capital – fails for Australian data over the sample periods considered.

Second, the other determinants of investment vary considerably across different types of investment. For equipment investment, measures of business confidence are important. For engineering investment, the terms of trade are a key driver of investment expenditures. Building investment is more difficult to explain, but it is responsive to the cost of capital, particularly when expected capital gains and losses are taken into account.