Assessing the Empirical Relevance of Labour Frictions to Business Cycle Fluctuations
| Published date | 01 June 2018 |
| Date | 01 June 2018 |
| DOI | http://doi.org/10.1111/obes.12215 |
| Author | João Madeira |
554
©2017 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 80, 3 (2018) 0305–9049
doi: 10.1111/obes.12215
Assessing the Empirical Relevance of Labour Frictions
to Business Cycle Fluctuations*
Jo ˜
ao Madeira†
†Department of Economics and Related Studies, University of York, York YO10 5DD, UK
(e-mail: joao.madeira@york.ac.uk)
Abstract
This paper describes a dynamic stochastic general equilibrium model augmented with
labour frictions, namely: indivisible labour, predetermined employment and adjustment
costs. This improves the fit to the data as shown by a higher log marginal likelihood and
closer match to key business cycle statistics. The labour frictions introduced are relevant
for model dynamics and economic policy: the effect of total factor productivity shocks on
most macroeconomic variables is substantially mitigated; fiscal policy leads to a greater
crowding out of private sector activity and monetary policy has a lower impact on output.
Labour frictions also provide a better match to impulse response functions from vector
autoregressive models.
I. Introduction
The understanding of the role of labour markets to business cycle dynamics has for long
been viewed as a key question in macroeconomics. Keynes (1936) argued that a failure of
the labour market to clear was essential to the understanding of the Great Depression in
the 1930s. Labour markets was also at the core of Friedman’s (1968) work whichidentified
informational problems as preventing labour markets from clearing at the natural rate of
unemployment.This proved to be invaluableto making sense of the emergence of stagflation
in the 1970s.
Despite this, for the most part, labour market rigidities have deserved relative little at-
tention by modern researchers working in dynamic stochastic general equilibrium (DSGE)
models for the purpose of the analysis of inflation and output cyclical fluctuations. Also,
those labour rigidities which have been most recently explored in the literature imply a
departure from Walrasian markets. These include search and matching frictions of work-
ers to jobs (Walsh, 2005; Trigari, 2009), efficiency wages (Alexopoulos, 2004; Danthine
and Kurmann, 2004), imperfect competition in labour supply (Zanetti, 2007) and wage
stickiness (Erceg, Henderson and Levin, 2000). Of these only imperfect competition in
JEL Classification numbers: E20, E24, E30, E31, E32.
*I am thankful to Simon Gilchrist, Fran¸cois Gourio, Robert King, Alberto Ortiz, Francesco Zanetti and two
anonymous referees for useful suggestions. All errors are my own.
The relevance of labour frictions 555
labour supply and wage stickiness have become widely adopted by DSGE researchers,
being present in several of the most commonly used references such as Smets and Wouters
(2003, 2007) and Christiano, Eichenbaum and Evans (2005). While search and matching
frictions have certainly proved insightful in the analysis of labour market flows (see e.g.
Yashiv, 2006) its importance to improving the fit of DSGE models to the aggregate time
series data still remains to be proven (see Shimer, 2005; Krause and Lubik, 2007; Krause,
Lopez-Salido and Lubik, 2008; Lubik, 2009).
In this paper, I introduce labour frictions which do not imply a departure from Walrasian
labour markets into an otherwise standard New Keynesian(NK) model similar to Smets and
Wouters(2007).1In particular, I assume labour to be indivisible (labour cannot be supplied
in continuous units, households are constrained to be either in a straight time shift, a straight
time and overtime shift or unemployed),predetermined straight time employment numbers
(in which case, firms adjust overtime employment to respond to unexpected shocks) and
convex labour adjustment costs. Therefore, similarly to Blanchard and Gal´ı (2010), the
model allows for unemployment,has elements common to those found in search and match
models (which typically also assume indivisible labour and predetermined employment),
but abstracts from other less essential ingredients (e.g. vacancies are filled immediately
which is not the case in standard formulations of search and match models).
The labour rigidities introduced are motivated by empirical evidence and legislation
requirements: (i) Studies summarized in Hamermesh (1993), show that the lag in adjust-
ing employment demand to be three to six months and that hours per worker adjust more
rapidly than employment(Elsby, Hobijn and Sahin, 2010, show that the pattern of a quicker
adjustment in hours relative to employment also occurred in the 2007 recession);2(ii) Em-
pirical studies at the micro level indicate that labour adjustment costs are quite significant
(see Hamermesh and Pfann, 1996), with some suggesting they amount to as much as one
year payroll for the average worker (notice that this excludes costs which are harder to
measure, such as disruptions to production from changing the number of employees); (iii)
The Fair Labor Standards Act (FLSA) overtime pay provisions cover more than 80% of
workers and overall compliance rates are around 90% (seeTrejo, 1991).3
The model is estimated with Bayesian methods (as in Smets andWouters, 2007, Gertler,
Sala andTrigari, 2008) which has become the most popular approach in macroeconometrics
(see Fern´andez-Villaverde, 2009). This approach is based on the likelihood which uses all
the information in a sample and provides a useful tool for model comparison that embodies
1The Smets and Wouters (2007) model includes most real rigidities known to be empirically relevant for DSGE
models. Firm-specific production factors, as in Woodford(2005), are among the few frictions absent from the model.
Foran empirical assessment of fir m-specific employmentsee Madeira (2014) and for firm specific capital see Madeira
(2015).
2Furthermore, Hansen and Sargent (1988), using a vector autoregressive (VAR) approach find that overtime
employmentappears to adjust more rapidly than full time employment to output innovations. This indicates that many
firms are likely to be constrained in the short run in adjusting their total employment and resort to overtime work in
order to respond to unexpected fluctuations. Other recent VAR studies confirm that employment and unemployment
respond little (if at all) on impact to demand shocks (see Monacelli, Perotti and Trigari,2010; Br ¨uckner and Pappa,
2012).
3As noted by Trejo (1991), it is surprising howlittle attention is given to overtime pay by economists. Only 11%
of workers earn the minimum wage or below, whereas about 25% of all employees work overtime during a typical
week (a fraction that has remained quite stable, see the references inTrejo, 1993).
©2017 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial