384
International Food and Agribusiness Management Review Vol. 2/No. 3/4/2001
Table 1. Results of the Augmented Dickey Fuller test for Pp et Pr in the
North and in the South
|
Variables |
t-stat (f) |
Lags |
Critical Values of the Test | ||
|
1% |
5% |
10% | |||
|
Pp North |
21.50 |
4 |
23.457 |
22.872 |
22.572 |
|
Pr North |
21.63 |
6 |
23.457 |
22.872 |
22.572 |
|
Pp South |
22.24 |
5 |
23.457 |
22.872 |
22.572 |
|
Pr South |
22.25 |
5 |
23.457 |
22.872 |
22.572 |
Note: The number of lags is chosen according to information criteria (Akaike and Schwarz) so that there is no autocorrelation in the
residuals.
Price Transmission and Symmetry in the North
As Pp and Pr are not cointegrated, we first test for Granger (1969) causality to
determine which price series is the dependent variable to be used in (2)
(Monhanty, Peterson, Kruse, 1995). The series to be included in the test of
Granger causality have to be stationary. However, Sims et al. (1990) demonstrate
that causality tests can be conducted on series which are not stationary in levels
but are stationary in differences, by estimating a vector auto-regressive represen-
tation of the first-differenced series. The results of the ADF test conducted on the
first differences of Pp and Pr are presented in Table 3 and show that the null
hypothesis of stationarity is rejected for both series at one percentage significance.
The results of the test for causality2 show that in the North, DPp causes DPr. DPr
thus becomes the dependent variable in (2). The variables are then segmented
following the Houck procedure (Table 5).
Model (2) is estimated using the least-squares method (Monhanty, Peterson,
and Kruse, 1995). Lags are included in the model until there are no more
significant effects (Kinnucan and Forker, 1987). After two lags, there is no more
significant effect between the two variables. It implies that price adjustments at
the retail level are occurring from one to two weeks after a shock emerging at the
production level.
Focusing on Table 6, note first that the R2 is 0.96, which means that nearly all
the fluctuations of the retail prices of pork in Ha Noi are explained by fluctuations
in the live hog production price in rural areas supplying the city. Second, the
Table 2. Test for Cointegration between Pp et Pr using Johansen Method
|
Trace Statistic |
Trace (0.90) |
H0 |
lmaxStatistic |
lmax (0.90) |
lmax (0.95) | |
|
North | ||||||
|
r50 |
15.62 |
17.79 |
19.00 |
13.56 |
13.75 |
15.67 |
|
r51 |
2.05 |
7.50 |
9.138 |
7.52 |
9.24 |
9.24 |
|
South | ||||||
|
r50 |
43.86** |
17.79 |
19.00 |
39.95** |
13.75 |
15.67 |
|
r51 |
3.88 |
7.50 |
9.138 |
3.88 |
9.24 |
9.24 |
Note: ** 5 significant at 5%.