Treasury
Publication

# Appendix 3 Consumption equation

log(COND)=C0101+C0102*log(RINCOME)+C0103*log(RWEALTH)-

RLPCON+C104*DUMMY

The dependent variable used in the regression is the logarithm of real consumption expenditure (log(COND)+RLPCON).

Table A3.1  Estimation of a long-run consumption equation
Sample period 87:2-01:4 87:2-96:4 87:2-01:4 87:2-01:4
C0103=(1-C0102) C0103=(1-C0102) C0103=(1-C0102) C0102=1 , C0103=0
log(RINCOME) 0.982
(0.0519)
0.563
(0.0990)
0.665
(0.0844)
1
(na)
log(RWEALTH)   1-C401 1-C401
DUMMY1     0.0614
(0.01381)

Constant -0.004
(0.1586)
-1.216
(0.3070)
-0.993
(0.2616)
0.0503
(0.0040)
Cointegration tests
Dickey-Fuller **   **
Augmented Dickey-Fuller **   ** **

1 Note that the variable takes the value 1 if t > 1996:4 and 0 otherwise. This dummy variable is to capture a structural break in the consumption behaviour such as the effects of greater access to credit.

Table A3.2  Unit root tests (Levels)
Lag length p ADF test statistic
log(COND)+RLPCON 4 -2.605
log(RINCOME) 2 -2.439
log(RWEALTH) 0 -2.266
Table A3.3 Unit root tests (First Differences)
Lag length p ADF test statistic
log(COND)+RLPCON 2 -2.996**
ln(RINCOME) 0 -9.800*
ln(RWEALTH) 1 -2.997**

* significantly different from 0 at 1% level.

** significantly different from 0 at 5% level.

The critical values are based on MacKinnon[8] critical values for unit root tests.

#### Notes

• [8]See Mackinnon, J. G. (1991) “Critical values for cointegration tests,” Chapter 13 in R. F. Engle and C. W. J. Granger (eds), Long-run economic relationships: readings in cointegration, Oxford University Press.
Page top