30 C. Bencivenga, G. Sargenti, and R.L. D’Ecclesia
Ta b le 4 .Engle and Granger cointegration test
Dep. variable Indep. variable β tβ p-value
Elect Gas 0.514 23.90 0.00
Gas Elect 0.516 0.00
Elect Oil 0.732 25.33 0.00
Oil Elect 0.394 0.08
Oil Gas 0.432 29.03 0.00
Gas Oil 0.805 0.00
tβare thet-statistics for the coefficientsβin equation (5). The last column reports thep-values
for the unit root tests on the regression residuals.
Cointegration tests per se do not focus on the economically interesting parametersα,
β,φandθ[3]. The ECM highlights that the deviations from the long-run cointegrating
relationship are corrected gradually through a series of partial short-run adjustments.
In the long run equilibrium the error correction term will be equal to zero. However, if
the variablesy 1 ,tandy 2 ,tdeviate from the long-run equilibrium, the error correction
term will be different from zero and each variable adjusts to restore the equilibrium
relation whose speed of adjustment is represented byθ.
The results reported in Table 5 highlight no significative value for coefficientφ
in any cases. Therefore we apply an ECM using a different lag for the independent
variable.
Ta b le 5 .Estimated speed of adjustment parameters for the ECM
Dep. variable Indep. variable φ tφ p-value θ tθ p-value
Elect Gas 0.010 0.150 0.880 − 0. 452 − 21. 54 0.00
Elect Oil − 0. 427 − 1. 059 0.289 − 0. 461 − 21. 71 0.00
Gas Oil 0.028 0.189 0.849 − 0. 053 − 6. 553 0.00
For electricity and gas, significative coefficientsφandθare found (φ= 0 .25,
θ= 0 .46) with a lag of two days, indicating that in addition to a long-run relationship a
short-run influence exists among the two series. For the pair electricity/oil, considering
the independent variable with a five-day lag, a significative coefficient,φ= 0 .68 (9%
level), is found whereasθ= 0 .452; also in this case, the price adjustment in the short
run is detected with a lag of five days. For the pair gas/oil a significative coefficient
φis found (φ= 0 .29) at the 5% level with a lag of six days.θis equal to 0.05,
showing that the speed adjustment to the long-run equilibrium is particularly low.
The presence of a short-run relationship among the various commodities may also be
explained by the fact that the analysis refers to European markets where deregulation
has not been completely performed yet. Some of the markets still experience market
power and in this context the oil prices may still represent the leading actor for the