The table 6.1 shows unit root test results for the variables indicated in the ¬rst

column. The second column describes deterministic terms included in the test

regression: constant (c), seasonal dummies (s), linear trend (t), and shift and

impulse dummies according to the model c in Perron (1989) (P89c) allowing for

a change in mean and in the slope of a linear trend. The break point is given in

parentheses. Lags denotes the number of lags included in the test regression.

Column CV contains critical values. Three (two) asterisks denote signi¬cance

at the 1% (5%) level.

6.2 Money Demand 109

30

4.4

25

4.3

20

4.75+Y*E-2

Y

15

4.2

10

4.1

5

0 10 20 30 40 50 0 10 20 30 40 50

X X

1.5

1.5

1.4

1.4

1.3

1.3

Y

Y

1.2

1.2

1.1

1.1

0 10 20 30 40 50 0 10 20 30 40 50

X X

Figure 6.1: Data

Table 6.1: Unit Root Tests

Variable Deterministic Terms Lags Test Statistic 1/5/10% CV

“4.55 ——

mr c, t, s, P89c (98:3) 2 “4.75 / “4.44 / “4.18

“9.40 ———

y c, t, s, P89c (98:1) 0 “4.75 / “4.44 / “4.18

“9.46 ———

p c, t, s, P89c (98:1) 2 “4.75 / “4.44 / “4.18

“4.72 ———

r c, s 2 “3.57 / “2.92 / “2.60

Because the time series graphs show that there seem to be structural breaks

in real money, GNP and price level, we allow for the possibility of mean shift

and change in the slope of a linear trend in the augmented Dickey-Fuller test

regression. This corresponds to model (c) in Perron (1989), where the critical

values for this type of test are tabulated. In the unit root test for the interest

110 6 Money Demand Modelling

Figure 6.2: Fitted Trends for Real Money and Real GNP

rate, only a constant is considered. According to the test results, real money,

real GNP and price level are trend-stationary, that is they do not exhibit a

unit root, and the interest rate is also stationary. These results are quite stable

with respect to the lag length speci¬cation. The result of trend-stationarity

is also supported by visual inspection of a ¬tted trend and the corresponding

trend deviations, see ¬gure 6.2. In the case of real money, the change in the

slope of the linear trend is not signi¬cant.

Indonesian money demand is now estimated by OLS using the reduced form

equation 6.4 (sit are centered seasonal dummies, ds denotes a step dummy

switching from zero to one in the respective quarter, di is an impulse dummy

having value one only in the respective quarter, t-values in parantheses):

0.531 mrt’1 + 0.470 yt ’ 0.127 rt

mrt =

(6.79) (4.87) (’6.15)

’ 0.438 ’ 0.029 s1t ’ 0.034 s2t ’ 0.036 s3t

(’0.84) (’2.11) (’2.57) (’2.77)

+ 0.174 ds9802t + 0.217 di9801t + 0.112 ds9803t + u(6.10)

t

(3.54) (5.98) (5.02)

R2 = 0.987