30.06.2013 Views

TESTING INTERNATIONAL PRICE TRANSMISSION UNDER ...

TESTING INTERNATIONAL PRICE TRANSMISSION UNDER ...

TESTING INTERNATIONAL PRICE TRANSMISSION UNDER ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Empirical Analysis: Cointegration Models Accounting for Policy Regime Changes<br />

to the disequilibria between the French and the intervention price (see previous<br />

paragraph).<br />

5.2.2.3 Model 4: both adjustment coefficients and cointegration parameters can<br />

vary<br />

Finally, in Model 4, both adjustment coefficients and the cointegrating<br />

relationship were allowed to differ according to the policy regime in place. In<br />

equation 5.5, the disequilibrium term zt has been calculated as in Model 3, and<br />

then has been multiplied by the policy regime dummy like in Model 2.<br />

90<br />

t = α regt<br />

1 z t 1 + α − regt<br />

1 z t 1 + ∑ − k 1<br />

4 * − * − 5 * ( 1 − ) * −<br />

i=<br />

1<br />

∆p Γ ∆p + ε<br />

i<br />

t−i<br />

t<br />

(5.5)<br />

What happens is that, if regt-1 = 1, the adjustment coefficients to the LOP<br />

holding with the US price are given by α4; if regt-1 = 0, then the adjustment<br />

coefficients to the LOP holding with the intervention price are given by α5.<br />

The empirical estimates reported in table 5.9 79 show that the adjustment<br />

coefficients of both the US and the EU prices behave consistently with Model 2,<br />

though slightly higher in absolute value.<br />

α 4<br />

α 5<br />

LM test<br />

OV test z 1,t-1, z 2,t-1<br />

ARCH (12)<br />

Table 5.9 Model 4 estimates<br />

∆swfr ∆hrw Cointegration vector<br />

-0.023<br />

(0.026)<br />

-0.260***<br />

(0.049)<br />

0.233<br />

(p-value 0.999)<br />

28.815<br />

(p- value 0)<br />

1.814<br />

(p-value 0.999)<br />

0.021<br />

(0.031)<br />

-0.069<br />

(0.060)<br />

0.537<br />

(p-value 0.938)<br />

28.096<br />

(p-value 0.005)<br />

Standard errors are reported in parenthesis<br />

LM: LM test with null hypothesis of no-autocorrelation<br />

OV: Omitted Variable test (χ 2 )<br />

* 10% significance; ** 5% significance; *** 1% significance<br />

swfr = 1.<br />

400*<br />

* * + 0.<br />

707*<br />

* * wref − 0.<br />

022*<br />

* * reg wref<br />

79 Autocorrelation was removed by adding the 12 th differentiated lag. Monthly dummies were selected with<br />

specification tests.<br />

t<br />

( 0.<br />

205)<br />

( 0.<br />

040)<br />

t<br />

( 0.<br />

002)<br />

t<br />

t

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!