RDP 8808: Consumption and Permanent Income: The Australian Case 4. Conclusion

The analysis of the previous sections has presented evidence which provides reasonable support for the permanent income hypothesis, providing that allowance is made for a proportion of consumption which appears to be better explained by current income. This finding is in line with the U.S. studies by Hayashi, and Campbell and Mankiw, as well as a wide range of other studies which find that the permanent income hypothesis is not easily rejected.

Given the finding that the permanent income hypothesis appears to have explanatory power, it is worth considering the implications for some aspects of macroeconomic policy. In this concluding section, we look at the importance of stock market fluctuations for private consumption and the effect on consumption of changes in interest rates.

The effect on activity of a fall in share prices operates via a fall in the value of wealth. There are two possible explanations as to why consumption in both Australia and overseas did not appear to fall in response to the share market crash. The first explanation is that consumption did not rise with the booming share market preceeding the crash because the capital gains were not realised gains and therefore were not treated as part of wealth given agents understood the share market boom was a bubble with finite probability of bursting. The subsequent crash was discounted in the same way. This is not consistent with the theory since even if capital gains are not realised, as argued in Edey (1988), they should still be built into consumption decisions.

A second explanation is consistent with the theory. Assuming that only unconstrained consumers hold equities, the impact on consumption of a fall in share prices would be the proportion of wealth held as equities adjusted by the rate of time preference and the share of these consumers in aggregate consumption. Assume (as was the approximate effect in Australia in October 1987) a fall in share prices reduces the market value of private wealth by $100 billion. Using the parameter estimates in row 1, Table 6, this implies that, ceteris paribus, consumption should fall by approximately $1.5 billion (i.e. 0.6 *.025 *100), or 1 per cent of consumption in 1987.

That this fall in consumption apparently did not occur after October 1987 does not refute the model. Another implication of the life-cycle model is that movements in interest rates can have strong effects through the discounting of future income streams. A temporary fall in interest rates will have only a small effect, but a fall in short rates which is sustained and expected to be sustained so that it shows up in long rates will have a large effect. As an example, if long interest rates fall by half a percentage point from 13 to 12.5 per cent, that is a fall of 3.8 per cent in interest rates, human wealth would rise by 3.8 per cent. If human wealth is 80 per cent of total wealth, the life cycle model would imply an increase in consumption of about 3 per cent. In the case where only 50 per cent of consumption expenditures are based on the permanent income model, we would expect consumption to rise by 1 1/2 per cent. However, longer term interest rates fell by more than half of a percentage point in the six months after October 1987. Thus, the effect of the stockmarket crash on wealth could easily have been more than offset by the general trend of falling long interest rates.

The results of this paper suggest that the permanent income hypothesis explains a significant proportion of aggregate consumption expenditure in Australia. Perhaps the surprising aspect of the results is the robustness of this conclusion under the alternative tests we performed.