11 Conclusions
This paper has discussed a number of likelihood ratio tests in the I(2) model. Using
these procedures we have been able to investigate the empirical regularities behind
the long swings in the Dmk/$ rate. This has been done by structuring the data
according to different levels of persistence using the I(2) model. We have argued that
allowing for I(2) trends is important for the econometric analysis of macroeconomic
data To ignore such trends when they are present in the data is likely to impede a full
understanding of the data. Moreover, the I(2) framework enabled us to present some
empirical regularities in characterizing the long swings properties of real and nominal
exchange rates.
The estimated four stochastic trends summarizing the different levels of persistence
seemed primarily to derive from shocks to the interest rates. Thus, many of the basic
assumptions behind standard rational expectations models (such as the PPP and Fisher
parity) are incompatible with the empirical evidence. The sluggish adjustment of
relative goods prices and nominal exchange rates to the ppp is also incompatible with
standard rational expectations models. The fact that price inflation was found to be
’purely’ adjusting, whereas nominal exchange rates showed no long-run feed-back is
also in conflict with the assumptions of such models. In striking contrast, these results
accord well with the IKE monetary model in FG.
To conclude, we find the general-to-specific approach of a cointegrated VAR model
to be potentially very important as a way of making abductive inference in economics
(Hoover, 2006). This is because it allows us to systematically search for an econometric
model that is as simple as possible (but not more so) without distorting some of the
information in the data. Thus, this approach should allow us to address the question:
if the standard theory is too simple, what then? In contrast, the specific-to-general
approach is designed to replicate the favorite theory model and is, therefore, likely to
have a built-in theory bias (see Juselius and Franchi, 2007).
12 References
Bacchiocchi, E. and L. Fanelli (2005), Testing the purchasing power parity through
I(2) cointegration techniques. Journal of Applied Econometrics 20, 749-770.
Boswijk, H. P. (2000), Mixed Normality and Ancillarity in I(2) Systems. Econo-
metric Theory 16, 878-904.
Bec, F. and A. Rahbek (2004), Vector Equilibrium Correction Models with Non-
Linear Discontinuous Adjustments. Econometrics Journal 7, 628-651.
Dennis, J., Johansen, S. and K. Juselius (2005), CATS for RATS: Manual to Coin-
tegration Analysis of Time Series. Estima, Illinois.
Dornbusch, R.(1976), Expectations and Exchange Rate Dynamics, Journal of Po-
litical Economy, December, 1161-1174.
Dornbusch, R. and J.A. Frankel (1988), The Flexible Exchange Rate System: Ex-
perience and Alternatives, in S. Borner, ed., International Finance and Trade, London:
25