RDP 8501: Neoclassical Theory and Australian Business Investment: A Reappraisal 5. Simulation Results

Three counterfactual simulations will serve to illustrate the properties of the model. We will consider in turn: 1) a sustained one percentage point increase in the bond and money rates; 2) a sustained 10 per cent increase in the price of consumption goods; 3) a sustained 1 per cent increase in employment.

For the purpose of the simulations, the model was linearised around sample means. This not only facilitates the simulation exercise, but it also guarantees that the simulation results are independent of the starting point. Furthermore, it was necessary to exogenise π_I (in the definition of r_K) as it proved to be a source of instability. The modelling of expectations is an extremely delicate matter, and it is clear that our model leaves ample room for improvements in this respect. With this change the model is stable. The simulation results are presented in terms of graphs for the main variables of interest: y_I, w_K, P_I, and r_K. The behaviour of the other variables is evoked in the text.

The effect of an increase in r_M and r_B is to reduce the demand for capital goods (by making ownership less attractive), and hence, the supply being inelastic in the short run, the price of capital falls. (The drop in the price of capital actually dampens the decrease in demand by raising, ceteris paribus, the return on capital.) As a result the production of new capital goods falls, that is investment declines. One complicating factor arises because the production of investment goods involves the use of capital services. As indicated by the negative sign of β₁, the drop in the price of investment goods favours capital at the expense of labour. w_K therefore increases, thereby lifting the return on capital. This further reduces the decline in demand, and hence the need for a fall in the price of capital goods. The bottom line is that the increase in interest rates has only a relatively small effect on investment once that allowance is made for endogenous adjustments in p_I, w_K and r_K. As indicated by the first panel of Figure 4, investment falls to a value approximately 2 per cent below control (after 5 quarters) before starting to rise again. The increase that takes place after the fifth quarter reflects to a large extent the effect of the falling capital stock on the output of investment goods (negative sign of α₃). In the long run, investment is only 0.1 per cent below control. The price of capital goods exhibits pretty much the same pattern. It is 3.3 per cent below control in the third quarter, but it gradually recovers as the supply of capital goods declines, and eventually it reaches a value about 1.8 per cent below control. The increase in the rate of return on capital and in its rental price take place fairly rapidly, and they are substantial: approximately 6 per cent for the latter, and over one percentage point for the former in the long run. The rise in interest rates is therefore more than offset by the increase in r_K. This result is explained by the drop in wealth that is caused by the fall in both x_K and p_I.

An increase in the price of consumption goods tends to lead to a shift of resources from the production of investment goods to the production of consumption goods, that is investment tends to decline. However, this reallocation of resources favours capital over labour: the rental price of capital rises, hence the price of capital must increase for portfolio equilibrium to be maintained. The increase in the price of capital goods partially offsets the rise in the price of consumption goods, so that the effect on the production of capital goods is much reduced. Investment is 2.3 per cent below control after two quarters. However, the movement rapidly reverses itself under the influence of the falling stock of capital which affects the output of investment goods in two different ways: directly (through α₃), and indirectly through further rises in the price of capital resulting from its increased scarcity. The increase in the price of capital goods actually rapidly exceeds the rise in the price of consumption goods, so that investment becomes larger than in the control solution. In the long run, investment is fractionally above control,^[35] w_K has gone up by approximately 6.6 per cent, while r_K has lost about two thirds of a percentage point.

A 1 per cent increase in employment has a powerful effect on investment since it favours the production of capital goods over the production of consumption goods. Moreoever, it results in an increase in the rental price of capital (since labour intensity increases). This translates itself in an increase in the price of capital goods (portfolio equilibrium obliging), which further stimulates investment. In the long run, the capital stock is 1.2 per cent above its control level, and investment exceeds its control solution by about 0.3 per cent. The return on capital is little affected by the shock,

To sum up the results from our simulations, it appears that Australian business investment is more sensitive to the level of activity (proxied here by employment) than to variations in interest rates and in the relative price of investment goods. This result is hardly surprising. Many researchers before us have failed to uncover significant links between interest rates and investment, and there seems to be widespread scepticism regarding their existence. It is noteworthy that the absence of any strong effect does not result from exceedingly small price and interest elasticities at the level of the individual behavioural relationships. Instead it is due to the interactions of a number of mechanisms which tend to neutralise the impact of external shocks. It is important therefore that variables such as the price of capital or its rental price be endogenised when assessing the effects of price or interest rate changes. Our results also suggest that tax measures aiming at making the use, ownership, or production of capital goods more attractive will have little lasting effects.^[36]

Footnotes

y_I and r_K return to control, while p_I and w_K increase by exactly 10 per cent in the long run if p and B are exogenously increased by the same percentage simultaneously with p_C. The simulation results are available on request. [35]

Simulation results of several taxation experiments are available on request. [36]