Testing Hall's permanent income hypothesis for a developing country: the case of Fiji

Abstract

Hall (1978) has stimulated considerable controversy and empirical work on testing the validity of the permanent income hypothesis (PIH). Much of this work is on the developed countries. In the developing countries per capita incomes show larger fluctuations and for the majority, opportunities for intertemporal substitution are limited. This paper uses the extended framework of Campbell and Mankiw (1989) and finds that current consumption is determined by current income for more than two thirds of the consumers in Fiji.

Acknowledgement

I wish to thank my colleague Rup Singh for help with the data and for comments. However, responsibility for errors remains with me.

Notes

Since this is a well known result and the derivations are available in advanced textbooks, e.g., Romer (2001), there is no need for an elaboration here.

In the Campbell and Mankiw estimating σ turned out to be small and insignificant. Hence the justification for the simpler specification.

This proxy is well known and can be derived from the ISLM model. When money supply increases, LM shifts in the (short-term) nominal rate of interest. However, since more money means higher inflationary expectations, the nominal (long-term) rate of interest increases. Consequently, the spread between the short and long-term interest rates increases and thus it is a good proxy for the liquidity in the economy.

Consumption expenditure is on non-durables and durables. Data on these two components are not available for Fiji and even for some G7 countries in the Campbell and Mankiw (1989) study. Similarly data on disposable incomes are also hard to get and therefore Campbell and Mankiw have used per capita incomes in their estimates for the G7 countries. These approximations did not make any significant difference to their estimates. However, for Fiji the available per capita disposable income shall be used.

A common set of instrument variable, with an intercept, in all the first stage regressions are used and these are: Δ ln Y _{t −3}, Δ ln C _{t −3}, Δ ln CPI _{t −3}, R _{t −3} and a dummy variable for political coups of 1987 and 2000. This dummy is 1 during 1987, 1988, 1989, 2000 and 2001. The first political coup seems to have taken a slightly longer period to return the economy to normal conditions. Alternative values were also tried but yielded less impressive results. When the residuals from the instrument equations are regressed on the instruments, T × R ² s have a χ² distribution with (k − 1) degrees of freedom, where T is the number of observations and k is the number of parameters. The computed χ² is a test for the validity of the over-identifying restrictions implied in IV estimates of Equation 4. For the four first stage equations, the computed χ² s are: 9.8397 for output, 9.8075 for R, 7.2955 for r s and 6.3565 for rl. The 5% critical value is 11.1. Therefore, these over-identifying restrictions are not rejected.