382 The Long Swings Puzzle
...There is a fundamental problem: How do we come to our a priori knowl-
edge? Most macroeconomists expect empirical evidence to be relevant to our
understanding of the world. But if that evidence only can be viewed through
totalizinga prioritheory, then it cannot be used to revise the theory.
...The Marshallian approach is archaeological. We have some clues that a sys-
tematic structure lies behind the complexities of economic reality. The problem
is how to lay this structure bare. To dig down to find the foundations, modifying
and adapting our theoretical understanding as new facts accumulate, becoming
ever more confident in our grasp of the super structure, but never quite sure
that we have reached the lowest level of the structure.For example, the significant finding of two shocks rather than one and the rejec-
tion of long-run price homogeneity are two examples of important information in
the data, signaling the need to dig deeper in order to understand more. By taking
this information in the data seriously instead of just ignoring it, we have been able
to uncover more structure, and thus to improve our understanding, as demon-
strated in Frydmanet al. (2008) and Johansenet al.(2008). Needless to say, the
need to dig deeper does not stop here.
Notes
- Note that thepppterm is also the (logarithm) of the real exchange rate. We prefer to use
the labelpppin this chapter because we are adopting a parity perspective and also because
we do not model the real exchange rate in terms of so-called real fundamentals. - Johansen (2006a) demonstrated that valid inference on steady-state values requires more
than 5,000 observations if the model contains a near unit root of 0.998. - German CPI has been additively mean corrected for the reunification in 1991:1 prior to
the VAR analysis. - Note, however, that this diagnostic check is only reliable in a VAR model with a correct
lag length. A VAR model with too many lags will often generate complex pairs of large
(albeit insignificant) roots in the characteristic polynomial (Nielsen and Nielsen, 2006). - When the data areI( 2 ), price homogeneity ofβ′xtis a necessary but not sufficient
condition, as will be discussed in section 8.8. - When the lagk>2, there would also be lagged acceleration rates,^2 xt− 1 , to concentrate
out. - At this stage,pwill be left unrestricted in the model.
- A linear trend inα′⊥ 1 xtwould imply that the inflation rate, say, is allowed to grow with
a linear trend, and thus prices with a quadratic trend. It would be hard to argue for such
a specification, except possibly as a local approximation. - From a statistical point of view, an equivalent transformation would be achieved by
replacingp 1 withp 2.
References
Balassa, B. (1964) The purchasing parity doctrine: a reappraisal.Journal of Political Economy
72 , 584–96.
Cheung, Y.W. and K.S. Lai (1993) Long-run purchasing power parity during the recent float.
Journal of International Economics 34 , 181–92.