heterogeneity across countries, even after the introduction of the euro, and also that
the Friedman-Ball link appears to have weakened or even broken down in a number
of cases in individual member states after the creation of EMU.
Our analysis extends theirs in several ways. First, as already mentioned, we focus on
the euro area as a whole, consistently with the ECB’s mandate. Second, we model
inflation as a function of the unemployment rate rather than past inflation only, thus
establishing a link between the present study and the literature on the Phillips curve in
Europe. Third, we investigate empirically the relationship between inflation
uncertainty and inflation in a bivariate VAR framework instead of a single-equation
model.
More specifically, we take a two-step approach as in Grier and Perry (1998) and
Caporale and Kontonikas (2009), namely we first estimate AR-GARCH models with
time-varying parameters to generate a measure of inflation uncertainty, and then
estimate the relationship between inflation and inflation uncertainty in a bivariate
VAR context. The model we use in the first step regresses current inflation on lagged
inflation and lagged unemployment. The advantages in terms of forecast accuracy
deriving from including the unemployment rate as a measure of real economic activity
in a model for inflation are discussed by Stock and Watson (1999) and Amisano and
Giacomini (2007).
The paper is structured as follows. Section 2 outlines the methodology. Section 3
describes the data and the empirical results. Section 4 offers some concluding
remarks, highlighting in particular the policy implications of our findings.
2. Econometric Framework
The GARCH models typically used in the literature have the drawback that they do
not take into account the fact that short-run and long-run inflation uncertainty might
be very different and affect inflation expectations in different ways.
As emphasised by Evans (1991), agents’ temporal decisions are more likely to be
affected by the conditional variance of short-run movements in inflation, whilst
intertemporal decisions might be based mainly on changes in the conditional variance
of long-term inflation. Moreover, one should distinguish between “structural
uncertainty” (associated with the randomness in the time-varying parameters of the
inflation process, and representing the propagation mechanism), which might
originate, for instance, from unanticipated monetary policy changes, and “impulse
uncertainty” (associated with the shocks hitting the conditional variance, which are
propagated through the parameters of the inflation process).~
The econometric framework suggested by Evans (1991), and also used by Berument
et al. (2005) and Caporale and Kontonikas (2009), has the advantage of yielding
estimates of the various types of uncertainty discussed above, and is adopted here as
well. More specifically, inflation is specified as a k-th order autoregressive process,
AR(k), and is also a function of unemployment, the parameters being time-varying
and the residuals following a GARCH(1,1) process. The model is the following: