Decomposing the Ins and Outs of Cyclical Unemployment
| Date | 01 December 2016 |
| DOI | http://doi.org/10.1111/obes.12133 |
| Published date | 01 December 2016 |
853
©2016 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 78, 6 (2016) 0305–9049
doi: 10.1111/obes.12133
Decomposing the Ins and Outs of Cyclical
Unemployment*
Ronald Bachmann† and Mathias Sinning†,‡
†RWI and IZA, Hohenzollernstrasse 1-3, 45128 Essen, Germany (e-mail: ronald.bachmann
@rwi-essen.de)
‡Crawford School of Public Policy, Australian National University, JG Crawford Building
#132, Lennox Crossing, Canberra ACT 2601, Australia (e-mail: mathias.sinning@anu.edu.au)
Abstract
This paper analyses the contribution of the composition of the pool of employed and un-
employed individuals to labour market dynamics in different phases of the business cycle.
Using individual-level data from the Current Population Survey (CPS), we decompose
differences in employment status transition rates between upswings and downturns into
explained and unexplained parts. We find that the duration of unemployment contributes
to explaining unemployment outflowsto employment and observe that its initially positive
contribution turns negative in deep recessions. Composition effects play an important role
for unemployment outflows to non-participation but dampen the cyclicality of unemploy-
ment inflows from employment.
I. Introduction
Starting with the contribution of Shimer (2012), the cyclicality of the US labour market
has attracted a great deal of attention recently. The main question in this debate concerns
the relevance of the inflows into and the outflows from unemployment, which has typically
been addressed by the analysis of aggregate time series of labour market transitions.1
We contribute to the debate by exploiting the micro information available at the individual
worker levelto study the underlying factors of unemployment inflows and outflows, paying
particular attention to the contribution of the duration of unemployment.
The empirical analysis performed in this paper is descriptivein nature. Weuse individual-
level longitudinal data from the Current Population Survey (CPS) for the time period
JEL Classification numbers: J63, J64, J21, E24
*Wethank Emanuel M ¨onch,Alfredo Paloyo, two anonymous referees and participants of meetings of the European
Association of Labour Economists, the European Economic Association, the European Society for Population Eco-
nomics, the Royal Economic Society and the Scottish Economic Society for valuablecomments and suggestions and
gratefully acknowledge the support of the LeibnizAssociation and the Australian Research Council (DP150104247).
1While earlier studies found inflows from employment into unemploymentto be the decisive factor for the cycli-
cality of unemployment (e.g. Darby, Haltiwanger and Plant, 1986) and later studies found a more important role
for outflows from unemployment to employment(Hall, 2005; Shimer, 2012), more recent articles have established a
relatively balanced role for inflows into and outflowsout of unemployment (e.g. Yashiv, 2008; Elsby, Michaels and
Solon, 2009; Fujita and Ramey, 2009, Elsby, Hobijn and Sahin, 2010).
854 Bulletin
February 1976–October 2009 to study the determinants of transition rates from employ-
ment to unemployment, from unemployment to employment, and from unemployment to
non-participation, respectively. We apply a Blinder-Oaxaca decomposition to decompose
the estimated transition rates between economic upswings and downturns into a part that
is due to ‘composition effects’ (i.e. differences in observed characteristics that describe
the socioeconomic and demographic composition of employed and unemployed popula-
tions) and a part that may be attributed to differences in coefficients associated with these
characteristics and differences in sample-specific intercepts, which may be considered as
‘unexplained’. This approach allows us to derive several important stylized facts about
cyclical upswings and downturns since February 1976. We also compare specific booms
and busts to describe the ‘big picture’ that holds for most boom-bust cycles, and to point
out peculiarities of specific recessions.
A strand of the economic literature that is closely related to our analysis focuses on the
duration of unemployment, either with respect to trends over the last decades (e.g. Abra-
ham and Shimer, 2002; Portugal, 2007) or regarding the latest recession (e.g. Aaronson,
Mazumder and Schechter, 2010; Kroft et al., 2014). Our analysis is closest to Baker (1992)
who uses CPS data in order to examine the (cyclical) determinants of the expected duration
of unemployment of different worker groups as they enter unemployment. He concludes
that composition effects did not play an important role for unemployment duration during
the 1980s. Consequently,aggregate unemployment duration (and, concurrently,the outflow
rate out of unemployment) was driven by increases in unemployment duration occurring
across most demographic groups.
Another strand of the literature that is related to our analysis examines inflows into
and outflows out of unemployment. In their seminal papers, Darby, Haltiwanger and Plant
(1985); Darby et al. (1986) examine the role of the level and the composition of inflows
and outflows for the cyclical variation of the stock of unemployment. Using time-series
techniques for worker flows constructed from individual-level data, they find that the com-
position of unemployment inflows is an important determinant of unemployment over the
cycle. Also analysing unemployment flows, Shimer (2012) shows that composition effects
do not play an important role for unemployment outflows.
The importance of the participation margin has recently been stressed by Elsby et al.
(2011), Kroft et al. (2014) and Elsby, Hobijn and Sahin (2015). Kroft et al. (2014), who
analyse CPS data for the time period 2002–07 and calibrate a matching model, argue that
transitions to (and from) non-participation may play an important role for the rise in long-
term unemployment and the observed outward shift of the Beveridge curve after 2008.
Elsby et al. (2015) find that the participation margin accounts for around one-third of
unemployment fluctuations. They also argue that the share of individuals with high labour
market attachment in the pool of the unemployed increases during recessions–afactalso
established by Mueller (2012).This reduces the transition probability from unemployment
to non-participation in times of recession.
We contribute to these studies of the US labour market by exploiting available micro
data and by quantifying the importance of composition effects for the cyclicality of labour
market flows within a multivariate context. Previous studies have mainly focused on the
use of time-series techniques, thereby losing at least some of the heterogeneity present
in micro data. We apply the Blinder–Oaxaca decomposition method, which has become
©2016 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial