Transcript Mayes1

The Relation Between Inflation
and Regional Unemployment and
Sectoral Income Growth
Dispersion:
Evidence From EU Countries
David G Mayes
Bank of Finland
HYPOTHESES
• The Phillips curve is not linear so there is a
problem of aggregation across regions/sectors
– At lower levels of unemployment/postive output gaps
the effect on inflation may be much stronger
• The dispersion across labour markets may
matter because it is the tightest labour markets
that tend to drive the inflationary process
NOT A NEW ISSUE
• Brechling (1973); Archibald (1969);
Phillips (1958); Lipsey (1960)
'If one wishes to predict the rate of change
of money wage rates, it is necessary to
know not only the level of unemployment
but also its distribution between the
various markets of the economy.'
• Laxton et al. (1995, 1999)
SOURCES OF ASYMMETRY
• In the labour market – unions
– Real and nominal wages sticky downwards?
– Capacity constraints
– Mayes and Buxton (1986)
• In the growth process
– ‘loops’ (Lipsey, Phillips)
– Okun curve (Harris and Silverstone, 2001)
• In policy
– Fiscal policy
– Monetary policy (Mayes and Virén, various)
THE MODEL
• Hybrid New Keynesian
pt = α1ut + α2sdt + α3pt-1 + α4pet + vt
• Threshold model (Tang, Teräsvirta)
pt = α11u-t + α12u+t + α2sdt + α3pt-1 + α4pet + vt
• Backward-looking with shocks
pt = α11u-t + α12u+t + α2sdt + α3pt-1 + α4pmt
+ α4taxt + vt
• (VAR in u, p and pe)
THE DATA
• Unbalanced panel of the 15 ‘old’ EU countries
1974-2004 – annual data – (no Luxembourg)
• Regional data on unemployment from Regio
database 248 regions (no Portugal or Greece)
• 6 sectors: agriculture; construction; rest of
industry; private non-financial services;
financial services; public services (no Ireland)
• Output gap (HP) (also use output growth)
• Expectations – published OECD forecasts
RESULTS 1
• Inflation positively related to unemployment
dispersion
• Expected inflation more important than lagged
inflation
• Inflation falls over the period so may not be
well specified (GMM OK)
• Feedbacks indicated by VAR – expected
inflation tends to increase unemplyment and
actual inflation to decrease it
RESULTS 2
• Clear non-linearity – greater impact in low
unemployment positive output gap periods
p = .797pe + .325p-1 - .031u|u>u* - .101u|u<u* + .053sd
(13.24) (8.43)
(2.31)
(3.78)
(1.62)
• Dispersion right sign but significance weak
• Adding import prices and VAT changes as shock
variables removes much of the effect of dispersion
• Backward-looking expectations reinstate role of
dispersion
CONCLUSIONS
• Nonlinearities large enough to matter in
EU
• Unemployment dispersion suggests a
labour market effect
• But other sources of asymmetry also at
work
• Does euro area monetary policy