Mind the Output Gap: The Disconnect of Growth and Inflation during Recessions and Convex Phillips Curves in the Euro Area
| DOI | http://doi.org/10.1111/obes.12291 |
| Author | Marco Gross,Willi Semmler |
| Published date | 01 August 2019 |
| Date | 01 August 2019 |
817
©2018 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 81, 4 (2019) 0305–9049
doi: 10.1111/obes.12291
Mind the Output Gap: The Disconnect of Growth
and Inflation during Recessions and Convex Phillips
Curves in the EuroArea*
Marco Gross† and Willi Semmler‡
†International Monetary Fund, 700 19th St NW, Washington, DC 20431, USA (email:
mgross@imf.org)
‡The New School, New York, Albert and Vera List Academic Center, University of Bielefeld,
6 East 16th Street, Bielefeld, NY 10003, USA (email: semmlerw@newschool.edu)
Abstract
We develop a theoretical model that features a business cycle-dependent relation between
output, price inflation and inflation expectations, augmenting the model bySvensson (1997)
with a nonlinear Phillips curve that reflects the rationale underlying the capacity constraint
theory (Macklem, 1997). The theoretical model motivates our empirical assessment, based
on a regime-switching Phillips curve and a regime-switching monetary structural VAR,
employing different filter-based, semi-structural model-based and Bayesian factor model-
implied output gaps. The analysis confirms the presence of a convex relationship between
inflation and the output gap, meaning that the coefficient in the Phillips curve on the output
gap recurringly increases during times of expansion and abates during recessions. Sign-
restricted monetary policy shocks based on a regime-switching monetary SVAR reveal
that expansionary monetary policy induces less pressure on inflation at times of weak as
opposed to strong growth; thereby rationalizing relatively stronger expansionary policy,
including unconventional volume-based policy, during times of deep recession. A further
augmented model shows that an effective euro exchange rate shock, too, implies busi-
ness cycle state-dependent responses, with more upward pressure on prices arising from
unexpected currency depreciation at times of expansion than during recession phases.
I. Introduction
This paper aims at examining the dynamics of price inflation and their relation to the
business cycle, a subject that is a well known one for macroeconomists. Starting from the
JEL Classification numbers: E31; E42; E52; E58.
*This work has benefited from useful discussions with Elena Bobeica, Inˆes Cabral, Matteo Cicarelli, V´ıtor
Constˆancio, Marek Jarocinski, Michele Lenza, Carlos Montes-Galdon, Dieter Nautz, and Chiara Osbat. The work
presented in this paper has served as an input to a speech held by the European Central Bank’s (ECB’s) for-
mer Vice President, V´ıtor Constancio, at the Jackson Hole Economic Policy Symposium at the Federal Re-
serve Bank of Kansas City in August 2015: https://www.ecb.europa.eu/press/key/date/2015/
html/sp150829.en.html. This paper should not be reported as representing the views of the International
Monetary Fund (IMF) or the European Central Bank (ECB). The views are those of the authors.
818 Bulletin
original work by Phillips (1958) there has been a long-lasting interest in the topic; from the
late 1960s during which Friedman (1968) and Phelps (1967) criticized the Phillips curve,
claiming that nominal variables cannot influence real variables, which led in the 1970s to
the development of the expectation augmented Phillips curve, as a result of its inability
to explain that inflation and unemployment rose simultaneously in the 1970s (in the face
of the oil price shock). The question as to why inflation rates are so persistently low and
difficult to stimulate in the aftermath of the global financial crisis ranks high, for obvious
reasons, also on the agenda of major central banks around the world.1
The different Phillips curve specifications that have evolved and been examined in
the literature can be categorized in three groups: the Traditional (or New Classical), the
New Keynesian and the Hybrid Phillips curve. In the New Classical form of the curve,
inflation is a function of lagged expected inflation and a contemporaneous measure of
excess demand. The underlying theoretical work by Phelps (1967) suggests that current
and lagged expected inflation shall move one-to-one.The parameter on excess demand, the
measure of marginal cost, indicates the degree to which prices are flexible, with a higher
coefficient implying less sticky prices. Roberts (1997) suggests that sticky price models,
as the one developed by Calvo (1983), shall imply that the inflation process should have
a forward-looking component, which led to an alternative specification in which current
inflation is related to currently expected future inflation, along again with a measure of
excess demand. In this New Keynesian specification, lagged inflation may only play a role
through its interaction with expected inflation at time tand it is not explicitly incorporated
in the model. Finally, in the hybrid Phillips curve equation structure inflation depends on
currently expected future inflation as well as lagged realized price changes, along with
contemporaneous economic slack, a theoretical model for which has been developed by
Gali and Gertler (1999). The underlying assumption is that not all firms reset prices in a
forward-looking manner. Some firms may not get the chance to adjust prices optimally
and rather use simple rules as a function of historic aggregate price behaviour (partial
indexation). Non-optimizing firms set prices to an average price level observed over recent
history and this renders inflation dynamics to some extent forward- and backward-looking.
The focus of our paper is on the potential for the relation of inflation and output dynamics
to be nonlinear, which rested in the back of economists’ minds since long. Phillips in
his original work already revealed a convex relationship, based on wage inflation and
unemploymentand it is since then dubbed a Phillips curve in f act, and not line. Evans(1986)
notes that eight of nine textbooks he reviewed at the time featured short-run aggregate
supply curves that were convex. As usefully summarized, for example, in Dupasquier and
Ricketts (1998), there are several (at least up to five) major theoretical frameworks that give
rise to some nonlinearity, either of direct or indirect nature with respect to the sensitivity
of inflation to some measure of economic slack.
The capacity constraint theory (Clark, Laxton and Rose, 1995; Macklem, 1997) starts
from the rationale that firms have spare capacity during recession times, thus are able to
1See, for example, the agenda of the Jackson Hole Symposium hosted by the Federal Reserve Bank of
Kansas City in August 2015: https://www.kansascityfed.org/publications/research/escp/
symposiums/escp-2015, with a speech related to inflation dynamics, including a possible role for nonline-
arities, by the ECB’s Vice President at the time,V´ıtor Constˆancio (https://www.ecb.europa.eu/press/
key/date/2015/html/sp150829.en.html).
©2018 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
Convex Phillips curves in the euro area 819
satisfy additional demand should it increase, with little or no incentive in this case to raise
prices. Only when firms movecloser to their capacity constraint, during boom times, would
they face more of an incentive to raise prices, as they would be less able to satisfy demand
by increasing production. This rationale implies convexity with regard to the coefficient
on the measure of slack, which would itself be an increasing function of the level of slack
(if the latter is measured in output space).
Second, the costly adjustment model (Ball, Mankiw and Romer, 1988; Ball and Mankiw,
1994) starts from the assertion that prices are not fully flexible due to the presence of menu
costs. The more firms that decide to change their prices, the more responsive wouldthe ag-
gregate price level become to demand shocks. As inflation rises, aggregate demand shocks
will have less of an effect on output and more on prices. An implication of this theory is
that the convexity of the coefficient on the measure of slack arises via its dependence on
the level of inflation.
The third theory starts from the assumed presence of downward nominal wage/price
rigidities (Fisher, 1989; Akerlof, Dickens and Perry, 1996). The rationale is that workers
naturally are more reluctant to accept a decrease in their wages than an increase.The effects
of nominal wage floors is thought to be more likely to be relevant at low inflation rates
(during recession times) because at higher levels of inflation it becomes less likely that
nominal wage cuts are required for a given decline in real wages. Hence, according to this
theory, the convexity of the coefficient on the measure of economic slack is again with
respect to the level of inflation, though in this case only at times of excess supply; unlike
under the costly adjustment model where the convexity with respect to inflation holds also
during times of excess demand.
A fourth theory implying a nonlinear trade-off between inflation and output comes
under the header of monopolistic competition (Stiglitz, 1984, 1997, Eisner, 1997). In a
monopolistically competitive economy, or oligopolistic markets respectively, firms are
expected to lower prices relatively swiftly to undercut rivals and not lose market share.
During boom times that come along with rising inflation, the same is assumed. This theory
implies a nonlinear dependence on the measure of slack directly, just as the capacity
constraint model, yet of the opposite shape; it implies concavity in the relation between
inflation and economic slack.
Fifth, the signal extraction model (Lucas, 1972, 1973) suggests that the slope of the
curve shall depend on the volatility of aggregate demand and supply shocks. If aggregate
prices are volatile (more volatile during high inflation regimes for instance) then it is less
easy for economic agents to infer whether price changes are of relative or aggregate nature
and hence more of a change would be attributed to aggregate price shocks. Disinflation
during recessions, coupled with lower inflation volatility, would imply a more pronounced
reaction of output. In comparison to the implication of the capacity constraint model, the
convexity of the coefficient on slack is therefore not on slack but via the dependence on
the volatility of inflation. However, that makes this theory different compared to the others.
To the extent that the level and variance of inflation are positively correlated (Kiley, 2000,
2007), a convexity may nonetheless effectively result in an empirical model with a state
dependence of the slack coefficient on the level of inflation.
There is a quite comprehensive set of studies that address nonlinearities in the Phillips
curve. Table 1 in Gross and Semmler (2017) summarizes the papers that address nonlin-
©2018 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial