The gender wage gap, adjusted for several characteristics, shows some variance over time. In this research, we explore the underlying causes of such instability. We find that during periods of fast job destruction, differences in earning between men and women tend to increase up to 20%. However, we did not find a similar reduction during fast hiring episodes.
1. When the opportunity knocks
large structural shocks and the gender wage gap
Joanna Tyrowicz Lucas van der Velde
FAME|GRAPE, IAAEU Warsaw School of Economics
University of Warsaw and IZA FAME|GRAPE
Our contribution
* Link labor market flows and GWG
* Comparable estimates of adjusted GWG for 15
countries (1600+ databases) experiencing large
structural shocks
Findings
* Adjusted GWG increased by up to 20% during episodes of intensive job destruction
* Effects are assymetric: episodes of intensive job creation relate only weakly to lower adjusted GWG
* Exploit the discontinuity between participants prior to the shocks and subsequent entrants → cohorts
active before transition experienced stronger effects
Measuring adjusted GWG across countries experiencing reallocation shocks
Method:
* Nopo (2008) decomposition: compare the comparables
(within common support)
* Control for age, education, marital status and urban
location + occupation.
Sources: Structure of Earnings Survey, Labor force Surveys, Longi-
tudinal Monitoring Surveys, Living standards measurement survey,
International Social Survey Program
Estimates of the GWG:
0
.5
1
1.5
2
2.5
Density
−.5 0 .5 1
Raw gender wage gap
Raw GWG
0
.5
1
1.5
2
2.5
Density
−.5 0 .5 1
Adjusted gender wage gap
Adjusted GWG
Cohorts: Born before 1965 Born after 1965
Episodes of reallocation
Labor market flows
Hirings =
FlowN→E + FlowEi→Ej
Et−1
Separations =
FlowE→N + FlowEi→Ej
Et−1
E employment, N non-employment and i, j sectors (i = j)
Structural change: SOE ↓ + services ↑
OutflowsOLD =
FlowEOLD→N + FlowEOLD→E
Et−1,OLD
InflowNEW =
Hiringst,NEW
n
i Hiringst,NEW
NEW: service sector or private firms
OLD: manufacturing or state-owned enterprises (SOE)
Episodes of fast reallocation (=shocks)
Definition follows Hausmann et al (2005)
Episode =
1 if flowt > 80th
pctile
and flowt > 1.5 ∗ flowt−1
0 otherwise
Source: Life in Transition Survey (EBRD): individual worker
flows 1989-2006 (retrospective survey)
Results: vulnerable workers (women) hurt by shocks
adjusted GWGc,s,t = βnLnEpisodeflow,c,t−n + θt + θc×s + c,s,t
c country, t year, n lag, s data source, θt & θc×s are time and country×source F.E
flow denotes the type of labor market adjustment
Adjusted gender wage gap and labor market shocks
[Each cell represents a separate regression for a given lag/flow. All flows (and episodes) measured for men]
Estimates of Hirings Separations Outflows from Inflows to
βn SOE Manufacturing Private Services
Cohorts active prior to transition (born before 1965)
Episodeflow,t−1 -0.01 0.04*** 0.05*** 0.03*** 0.03 0.01
(0.02) (0.01) (0.02) (0.01) (0.03) (0.02)
Episodeflow,t−2 0.01 0.02** 0.04*** 0.04*** -0.03 -0.00
(0.02) (0.01) (0.02) (0.01) (0.03) (0.01)
Episodeflow,t−3 0.01 0.02 0.04*** 0.04*** -0.05*** -0.02
(0.02) (0.01) (0.02) (0.01) (0.02) (0.02)
Cohorts entering during transition (born after 1965)
Episodeflow,t−1 0.01 0.00 -0.02 -0.00 -0.03 -0.06
(0.02) (0.02) (0.02) (0.02) (0.03) (0.05)
Episodeflow,t−2 0.01 -0.00 -0.01 -0.01 -0.00 -0.03
(0.02) (0.02) (0.02) (0.02) (0.02) (0.02)
Episodeflow,t−3 0.00 0.01 -0.02 -0.01 -0.01 -0.00
(0.02) (0.02) (0.01) (0.02) (0.01) (0.03)
Notes Columns identify flow types, while t − n indicates whether country experienced at least one episode over t − n
periods. Standard errors (in parentheses) clustered at the country year level. *, **, *** indicate significance at the 15%,
10% and 1% level. Observations are weighted by the inverse standard deviation of the estimated adjusted GWG.
Discussion
The results are robust to a number of checks (e.g. reestimate GWG with different controls, wider variety
of flows measures, alternative cohort breakdown, etc.)
→ Unjustified (adjusted) wage inequality grows in periods of large shocks, women (as a disadvantaged
group) present in every country, data is scarce for ethnic, racial or religious minorities, migrants, etc
→ Why are effects concentrated among older cohorts?
1. Alternative adjustment channel is through employment: younger women faced more entry barriers
(related paper: Tyrowicz, van der Velde and Goraus, 2018, Social Science Research)
2. Wage - stability trade-off limits mobility of women more
Acknowledgements
Earlier versions of this paper received valu-
able comments from participants of ESNIE, IEA,
EACES, AIEL and WIEM. S. Estrin, T. Mickiewicz,
K. Staehr, G. Roland, A. Agrawal, D. Tomaskovic-
Devey, I. Magda, and K. Goraus-Tanska offered
stimulating insights. At early stage of collecting
the data we benefited greatly from GDN RRC 12
implemented by CERGE-EI. This study was sup-
ported by a grant from the National Science Cen-
tre, UMO-2012/05/E/HS4/01510. The remaining
errors are ours.