Consumption Behaviour in Zambia: The Link to Poverty Alleviation?

K. Ludi: consumption behaviour in Zambia

8

All of the data used, apart from LN_GDP_CAPITA, is integrated of order one.⁷ Due to
the fact that GDP per capita is a ratio of two non-stationary series, it is regarded as
stationary, and thus is integrated of order zero.

4. Model and Estimation Results

4.1. Cointegration

Cointegration allows for non-stationary data series to be combined into a stationary series
through a linear combination. This represents the long-run relationship between the data
series. It was found in this case that real PCE in Zambia is dependent in the long run on
constant GDP, real short term lending rates and a dummy variable accounting for the
sudden and large decrease in PCE in 1992, largely as a result of the 30 per cent increase
in short-term interest rates in that period. It is expected using economic intuition that an
increase in national income will result in increased PCE and an increase in interest rates
should result in less PCE:

+  --

ln_cons_zk = f (ln_ gdp _ zk,r_lendrate,dum_92)                                (3)

1                1

( Lendrate₁₉₉₂ ^λ∣ 2 , √ tbill₁₉₉₃ ^λ∣ 3 ,
I--------------¹9⁹2-1 -1 and I-------¹9⁹31 -1.

^ Lendrate₁₉₉₀ J        ∣ tbill₁₉₉₀ J

See Appendix A for the unit root tests of all data used.

<<   <   6   7   8   9   10   11   12   >   >>