RDP 8709: A Note on Aggregate Investment in Australia 1. Introduction

Investment is a fundamental determinant of long-run growth as well as an important component of short-run aggregate demand. However, as pointed out in Carmichael and Dews (1987), it is perhaps the least well explained macroeconomic aggregate. A recent theoretical contribution was made in an important paper by Fumio Hayashi (1982) who derived a theory of the investment decision of an optimising firm facing costs to adjusting its capital stock. His resulting model was very similar to Tobin's (1969) “q theory” of investment. The purpose of the current paper is to use an approach similar to Hayashi's to derive and estimate an aggregate investment equation for Australia with the aim of incorporating the equation into the McKibbin-Sachs Global (MSG) model of the Australian economy[1].

The paper proceeds as follows. A brief overview of approaches to modelling aggregate investment is given in Section 2. Section 3 summarises previous Australian studies. A model of the investment decision of an optimising firm is derived in Section 4. In testing this theory we recognise that some firms in the economy are unable to borrow and lend as assumed by the theory and, therefore, face a binding liquidity constraint. We also recognise that there are lags between the decision to invest and the appearance of productive capital. Both these phenomena are taken into account in deriving an aggregate investment function. This aggregated model is then estimated for quarterly Australian data over the period December 1966 to December 1986 and the results are presented in Section 5. A conclusion is presented in Section 6.

We find that the q theory performs poorly when conventional capital stock data are used because the assumptions implicit in the calculation of the capital stock data are not consistent with the assumptions of the cost of adjustment model used to derive the q theory. Once the capital stock data is adjusted to be consistent with the theory being tested, we find that the q theory performs quite well. This suggests that standard tests of q theory are biased against the theory. For plausible values of the cost of adjustment we find that a lower bound of 10 per cent of investent is determined by q.


See McKibbin (1987) for the derivation of the MSG model and McKibbin and Siegloff (1987) for the Australian extension. [1]