Note that we use the term “supply side” in a narrow sense to refer to the view that the
labor market indices noted in the text have the predominant effect on the unemploy-
ment trend. This is an influential group in the UK. Elsewhere, economists such as Phelps,
who emphasize the supply side, embrace a wider interpretation of supply side effects.
See Fitoussiet al.(2000).
The use of a reduced-form unemployment equation is, in part, a reaction to the well
known argument that the standard model of the labor market, such as that found in
Layardet al.(1991), has a wage equation which is not identified. See Manning (1993)
for a clear account of this. Hall and Henry (2006) argue that with non-stationary data
this lack of identification is generally not found.
This accords with the inclusion of commodity prices in the SVARs estimated in Stock
and Watson (2002) and Boivin and Giannoni (2003).
Results for longer samples are reported in Henry, Kirby and Riley (2007).
All equations in this sub-section use lnuas the dependent variable. For arguments
favoring this, see Nickell and Bell (1995).
All tests of cointegrating rank in this chapter use asymptotic tests. For small sample
corrections to these tests, see Johansen (2002).
See Wickens and Motto (2001) for a discussion of the treatment ofI( 0 )variables in the
VECM.
These can be nonlinear restrictions.
This account ignores the possibility of using restrictions on the short-run dynamics, as
our interest is in the long-run model only.
The structural model referred to here has unemployment and the remaining seven vari-
ables noted in the text as its contemporaneous variables, and not wages and prices. See
note 18 for further comment on this point.
Full data definitions are given in the appendix to this chapter.
Parsimony is based on the concept of a minimal cointegrating vector introduced by
Davidson (1998).
This treatment follows from the interpretation of equation (18.25) as a long-run
unemployment equation.
We have also conducted long solutions of the same order as other studies, and these show
the same sort of repeated escapes as reported by Sargent (1999) and McGough (2006),
for example.
Al-Eydet al.(2007) review some issues of the robustness of the findings with respect to
these settings.
Forv 2 tthe variance is scaled by the estimated variance of the acceleration in inflation
over the period.
Arguably, UK governments at this time did not subscribe to the view that there was a
trade-off. The adoption of a Medium Term Financial Strategy (MTFS) by the government
in 1979 included a central assumption of a vertical Phillips curve. In spite of this, the
model uses the assumption that government believes in a trade-off since, in practice, the
government may have continued to act as if there was a trade-off in using the threat of
higher unemployment consequences if “excessive” pay demands were accepted.
In this account, we are deliberately ignoring other transmission effects of oil price
changes on inflation. This is in line with our treatment of these real shocks as impinging
on unemployment only. But the partial nature of the account provided should be borne
in mind throughout what follows.
The 3.25% for this period was partly a forecast.
A related issue is the effects of technology shocks on the performance of monetary policy.
Recent US research has been directed at this issue too (see, for example, Gali, Lopez-Salido
and Valles, 2003).
For an alternative view, however, see Nordhaus (2007).
