Palgrave Handbook of Econometrics: Applied Econometrics

Anindya Banerjee and Martin Wagner 697

Table 13.9 Banerjee and Carrion-i-Silvestre (2007) cointegration test results

Model Cross-sectionally

independent

Panel test

t-ratio

Cross-sectionally dependent

Homogeneous break dates Heterogeneous break dates

Idiosyn. Break date ˆr rˆ 1 Idiosyn. rˆ ˆr 1

1 −22.596 −5.009 1999.02 6 1 −2.744 6 2

2 −26.500 −6.409 2002.11 6 1 −2.933 6 1

3 −25.018 −5.816 2002.03 6 1

4 −22.523 −5.201 2000.03 6 1 −3.060 6 3

5 −24.677 −6.189 2002.08 6 1 −2.239 6 2

6 −23.498 −6.353 2002.10 6 1

It is clear that in each case the tests overwhelmingly reject the null hypothesis of
no cointegration. The reported breaks all occur in the neighborhood of the intro-
duction of the euro in 1999 or the beginning of its strong appreciation in 2002. A
refinement of our tests to allow for multiple breaks would perhaps allow us to detect
both these “regime” changes, although it may be argued that the time dimension
of the panel will not permit such detailed discrimination. Finally, if one were to
think of Model 4 as the most plausible choice, there is very clear evidence, under
every configuration, for a long-run relationship with a structural break in 2000.
There are a number of further issues that may be considered, including the
magnitude of the pass-through coefficient and its change in response to the new
monetary arrangements (or exchange rate movements). On the whole, allowing
for a structural break in the relationship, we find that ERPT generally increases in
the vicinity of the introduction of the euro. This may be the effect of stabilization
in the monetary regime, leading to less noisy exchange rate behavior. Thus actual
changes in the exchange rate may be perceived as more permanent and based on
macro-fundamentals and exporters may therefore be more willing to pass these
on to prices. An alternative explanation relates to the effect of the appreciation of
the euro. In a world with a depreciating euro, exporters to the euro-area would be
expected to hold back from passing through exchange rate changes to the price
(since this would lead to their becoming more uncompetitive relative to the local
producers in the euro-area). An appreciating currency means, however, that dollar
prices become cheaper in the intra-euro market, leading producers to shift away
from local currency pricing. Passing through more of their dollar price, to maintain
their revenue in dollars, would still not erode their ability to keep an edge on the
market and compete with local products. This asymmetrical response to currency
appreciation versus its depreciation could explain the higher pass-through follow-
ing the likely changes in regime identified above. All these results are tabulated in
great detail in de Bandt, Banerjee and Kozluk (2008) and are also available from us
on request.
Having studied testing for cointegration in some detail, we move next to a con-
sideration of methods of estimating cointegrating vectors, both in single equations
and in systems.