350 The Long Swings Puzzle
8.9.3 The estimated driving forces 378
8.9.4 What did we gain from theI( 2 )analysis? 379
8.10 Conclusions 380
8.1 Introduction
International macroeconomics is known for a number of empirical puzzles, the
most notable among them being the “PPP (purchasing power parity) puzzle,” which
is closely related to the “long swings puzzle” and the “exchange disconnect puzzle”
(Rogoff, 1996). These puzzles are all related to the pronounced persistence away
from equilibrium states that have been observed in many real exchange rates during
periods of currency float. Among these, the Dmk–$ rate in the post-Bretton Woods
period is one of the more extreme cases.
One important purpose of this chapter is to demonstrate how the cointegrated
vector autoregressive (CVAR) approach (Johansen, 1995; Juselius, 2006) can be used
to uncover structures in the data that ultimately may help to explain theoretically
the forces underlying such persistent movements in the data. The CVAR approach
starts from a general unrestricted VAR model that gives a good characterization of
the raw data. It then tests down until a parsimonious representation of the data
with as much economic content as possible has been achieved. When properly
applied, the CVAR is able to extract valuable information about the dynamics of
the pulling and pushing forces in the data without distorting this information.
This entails the identification of stationary relationships between non-stationary
variables, interpretable as long-run equilibrium states, and the dynamic adjustment
of the system to deviations from these states. It also entails the identification of
the transitory and permanent shocks that have affected the variables and the short-
and long-run impacts of these shocks.
For the results to be reliable, the statistical properties of the model have, however,
to be taken seriously. This implies adequately controlling for reforms, interven-
tions, regime changes, etc., that are often part of the data-generating mechanism.
The reunification of East and West Germany is an example of such an important
event. The approach also entails the untying of any transformation of the vari-
ables, such as the real exchange rate transformation, imposed from the outset on
the data. Such transformations, common in empirical economics, often seriously
distort signals in the data that otherwise might help to uncover precisely those
empirical regularities which give a clue to the underlying reasons for the puzzling
behavior.
The weight of the empirical analysis is on characterizing data within the broad
framework of a theory model. To facilitate the interpretation of the empirical
results, the chapter argues that it is essential first to translate the underlying
assumptions of the theoretical model into hypotheses on the pulling and push-
ing forces of the VAR model (Juselius and Johansen, 2006; Juselius, 2006; Juselius
and Franchi, 2007). A careful formulation of such a scenario is indispensable for