EMPLOYEE AUTONOMY AND THE WITHIN-FIRM GENDER WAGE GAP: THE CASE OF TRUST-BASED WORK TIME
1. EMPLOYEE AUTONOMY AND THE
WITHIN-FIRM GENDER WAGE GAP:
THE CASE OF TRUST-BASED WORK TIME
2019 GFP Annual Conference - Sydney, Australia
Steffen Viete
21 June, 2019
2. BACKGROUND
2
Technological change requires complementary investments in organizational
practices to affect firm performance (e.g. Bresnahan et al., 2002, Brynjolfsson et al., 2018)
Information technology (IT) is becoming mobile and detached from location
dissolving the temporal and spatial boundaries of work (e.g. Kossek and Michel, 2010)
This facilitates new organizational practices which allow employees to decide
where and when to engage in work-related tasks
There is evidence that firms enjoy higher returns from using mobile IT when
they provide workplace flexibility through formal work policies (Viete & Erdsiek, 2018)
3. BACKGROUND
WORKPLACE FLEXIBILITY IS WIDELY EXPECTED TO FOSTER
GENDER EQUALITY IN THE LABOR MARKET
3
Academia (e.g. Goldin, 2014 AER)
Disproportionate reward for
working long and particular hours in
many occupations
Women‘s social roles still associated
with higher non-work obligations
Change in the structure of work
"must involve a reduction in the
dependence of remuneration on
particular segments of time"
Public Debate
4. THIS PAPER
4
Research question:
What is the effect of the successful provision of maximum autonomy over the
structure of the workday on the gender wage gap (GWG) within the firm?
Effect is ambiguous from a theoretical perspective:
Negative, if dependence of remuneration on working particular segments of
time is reduced (Goldin, 2014)
Positive, if it results in higher compensating wage differentials for women
(Rosen, 1986)
5. TRUST-BASED WORK TIME (TBW)
5
Variants of TBW first appeared in Germany during the 1990s
Defining features:
• Abandonment of formal record of working hours
• Employee is merely evaluated by work output
• Delegation of autonomy to decide start- and endpoint, continuity and
duration of the workday
• Reduces the value placed on working specific hours and on the continuity of
work
Working arrangement with the highest degree of formal employee
autonomy in Germany (e.g. Wingen, 2004)
6. TRUST-BASED WORK TIME (TBW)
EMPIRICAL EVIDENCE
6
Beckmann et al. (2017, JEBO): TBW increases effort via intrinsic motivation
Godart et al. (2017, ILR): TBW adoption associated with higher innovation
Source: IAB Establishment Panel, own calculations.
7. 7
DATA
LINKED EMPLOYER-EMPLOYEE DATA - LIAB
IAB Establishment Panel
• Yearly representative business survey
• Ca. 19,000 establishments in Germany
• Since 2004 biannual question whether or not TBW is implemented
Employee Data (Employment History – BeH)
• German Employment Statistics Register based on individual employees‘
social insurance records
• Contains basic characteristics, e.g. gender, age, qualifications
• Gross daily earnings including overtime premiums and allowances
8. 8
DATA
OUTCOME: OBSERVED WITHIN-FIRM GENDER WAGE GAP
Average log differences of gross earnings 𝑤𝑤 between individuals 𝑖𝑖 of different
gender within each firm 𝑗𝑗 :
Average observed intra-firm GWG in LIAB
Data 2004-2012
Observed intra-firm GWG in LIAB Data -
Distribution in 2008
9. 9
METHODOLOGY
SAMPLE SELECTION
Research Design: ‘Treatment’ Methodology (e.g. Görg et al., 2016; Huber et al., 2016)
• Effect of introducing TBW between wave 2006 and 2008
• Retain all firms which did not use TBW in pre-treatment periods (2004-2006)
• Comparison of the within-firm GWG between adopter and non-adopter
Treatment:
Adoption of TBW
Control:
No adoption of TBW
pre-treatment post-treatment
2004 2006 2008 2010
adoption
10. 10
METHODOLOGY
SELECTION ISSUES
1. Non-random adoption of TBW by firms
Firms selectively rather than randomly introduce TBW
→ e.g. well-performing firms can afford to offer the amenity of TBW as well as
promote gender equality
Only compare ‘twin’ firms, which are observationally equivalent (Conditional
Difference-in-Differences, e.g. Heckman et al., 1998)
2. Non-random selection of individuals into firms w/o TBW
Women have higher preferences for workplace flexibility (Mas and Pallais, 2017)
→ more able women might select into firms with TBW in place
Compute outcome from individuals who were employed prior to the adoption of
TBW (e.g. Huber et al., 2016)
11. RESULTS
EFFECT ON THE WITHIN-FIRM GWG
11
Negative effect on the raw intra-firm GWG over three years after the adoption
(Panel A)
Result holds with slightly smaller magnitude among employees with pre-
adoption tenure (Panel B)
Note: Average Treatment Effects (ATE) with 90% confidence bands. N = 1290 with 131 treated and 1159 non-treated firms.
12. RESULTS
EFFECT ON WAGE LEVELS OF INCUMBENT EMPLOYEES
12
Reduction in GWG driven by absolute wage gains of women
Note: Average Treatment Effects (ATE) with 90% confidence bands. N = 1290 with 131 treated and 1159 non-treated firms.
13. RESULTS
EFFECT ON THE SHARE OF MALE EMPLOYEES
13
No significant effect on the gender composition of employees
Note: Average Treatment Effects (ATE) with 90% confidence bands. N = 1290 with 131 treated and 1159 non-treated firms.
14. POTENTIAL CHANNELS
14
1. Adjustments in the labor input (working time) within the same job
• Greater ability to structure the workday → facilitates adjustment of
(contractual) working hours
Estimate effect on the share of women in part-time contracts
2. Changes in the types of tasks and jobs performed
• Higher independence and autonomy under TBW might enable individuals
to perform different tasks
Estimate effect on share of women in positions with different job
requirements (based on KldB 2010)
15. POTENTIAL CHANNELS
SKILL REQUIREMENTS OF THE CURRENT JOB
15
Distinguishes four requirement levels
Intended to reflect the degree of complexity of an occupation
Requirement level Example within same occupation
1: Unskilled or semi-skilled activities Health and nursing care helper 81301
2: Specialist activities Nurse 81302
3: Complex specialist activities Specialist nurse 81313
4: Highly complex activities General practitioner 81404
Source: Paulus et al. (2013).
16. POTENTIAL CHANNELS
ADJUSTMENT OF WORKING HOURS OR TASKS PERFORMED?
16
No effect on the share of women in part-time contracts
Net increase (decrease) in the share of women in jobs with high (medium) skill
requirements
Year
% incumbent women in:
part-time
un-/semi-
skilled
tasks
skilled
tasks
complex
tasks
highly
complex tasks
2006 0.022 -0.004 0.058 -0.050 -0.004
2008 0.013 -0.001 -0.018*** 0.004 0.014***
2009 0.003 0.002 -0.020*** 0.003 0.015***
2010 0.016 0.006 -0.034*** 0.007 0.021***
adoption
Note: Average Treatment Effects (ATE). * p<0.10, ** p<0.05, *** p<0.01. N = 1290 with 131 treated and 1159 non-treated firms.
17. CONCLUSION
17
Main findings:
Current technological change favors decentralized work policies
Flexibility in the form of trust-based work is not gender neutral and can lead to
absolute wage gains for women
Organizational change that seems to enable women to perform different tasks
Limitations / future research:
Effect heterogeneity across individuals due to the nature of tasks and demands
in terms of self-reliance very likely
Policy and managerial implications:
Debate on how policy should promote / regulate alternative working practices
to support work-life-balance and equal opportunities
Working time autonomy appears to be a relevant measure
18. THANK YOU
18
Steffen Viete
Research Department “Digital Economy”
viete@zew.de
Tel: +49 (0)621-1235-359
ZEW – Leibniz Centre for European Economic Research
L 7,1
68161 Mannheim, Germany
Internet: http://www.zew.de
19. REFERENCES
19
Beckmann, M., T. Cornelissen, and M. Kräkel (2017). “Self-managed Working Time and Employee Effort: Theory and Evidence”. Journal of
Economic Behavior & Organization, 133: 285–302.
BMFSFJ (2013). Unternehmensmonitor Familienfreundlichkeit 2013. Bundesministerium für Familie, Senioren, Frauen und Jugend (BMFSFJ).
Bresnahan, T. F., E. Brynjolfsson, and L. M. Hitt (2002). “Information Technology, Workplace Organization, and the Demand for Skilled Labor:
Firm-Level Evidence”. Quarterly Journal of Economics, 117(1): 339–376.
Brynjolfsson, E., Rock, D., & Syverson, C. 2018. The Productivity J-Curve: How Intangibles Complement General Purpose Technologies,
National Bureau of Economic Research Working Paper No. 25148.
Cain Miller, C., "How to Close a Gender Gap: Let Employees Control Their Schedules", The New York Times, February 7, 2017.
Card, D., A. R. Cardoso, and P. Kline (2016). “Bargaining, Sorting, and the Gender Wage Gap: Quantifying the Impact of Firms on the Relative
Pay of Women”. Quarterly Journal of Economics, 131(2): 633–686.
Godart, O. N., H. Görg, and A. Hanley (2017). “Trust-based Work Time and Innovation: Evidence from Firm-Level Data”. Industrial and Labor
Relations Review, 70(4): 894–918.
Goldin, C. (2014). “A Grand Gender Convergence: Its last Chapter”. American Economic Review, 104(4): 1091–1119.
Goldin, C. and L. F. Katz (2016). “A Most Egalitarian Profession: Pharmacy and the Evolution of a Family-friendly Occupation”. Journal of
Labor Economics, 34(3): 705–746.
Heckman, J. J., H. Ichimura, J. Smith, and P. Todd (1998). “Characterizing Selection Bias Using Experimental Data”. Econometrica, 66(5):
1017–1098.
20. REFERENCES
20
Huber, M., M. Lechner, and C. Wunsch (2016). “The Effect of Firms’ Phased Retirement Policies on the Labor Market Outcomes of Their
Employees”. Industrial and Labor Relations Review, 69(5): 1216–1248.
Kossek, E. E. and J. S. Michel (2010). “Flexible Work Schedules”. In: Handbook of Industrial and Organizational Psychology: Building and
Developing the Organization. Ed. by S. Zedeck. Vol. 1. NE, Wash.: American Psychological Association, pp. 535–572.
Mas, A. and A. Pallais (2017). “Valuing Alternative Work Arrangements”. American Economic Review, 107(12): 3722–3759.
OECD (2019), Gender wage gap (indicator). doi: 10.1787/7cee77aa-en (Accessed on 11 June 2019)
Paulus,W., Matthes, B., et al. (2013). The German classification of occupations 2010: Structure,
coding and conversion table. FDZ-Methodenreport, 08/2013.
Rosen, S. (1986), `The theory of equalizing differences', Handbook of labor economics 1, 641-692.
Viete, S. and Erdsiek, D. (2018). Trust-based work time and the productivity effects of mobile information technologies in the workplace.
ZEW Discussion Paper, 18–013.
Wingen, S. (2004). Vertrauensarbeitszeit: Neue Entwicklung Gesellschaftlicher Arbeitszeitstrukturen. Bremerhaven: Wirtschaftsverlag NW,
Verlag für neue Wissenschaft.
22. BACKGROUND
THE GENDER EARNINGS GAP IS A PERSISTENT STYLIZED FACT
22
Most economic studies focus on individual
characteristics or take a market-based
perspective to explaining the gender wage
gap (GWG)
In contrast, legislation in many countries
focuses on the firm as a central agent in
establishing gender equality (Card et al.,
2016):
• US: Equal Pay Act of 1963
• UK: Equality Act 2010 Regulations 2017
• GER: Transparency of Remuneration Act 2017
What are relevant firm policies to address
gender pay differentials?1,5
13,8
34,6
0 10 20 30 40
ROU
CRI
ITA
TUR
NOR
NZL
HUN
FRA
MEX
ESP
OECD
NLD
AUS
CHE
DEU
GBR
USA
CHL
JPN
EST
KOR
Gender Wage Gap (GWG) for selected
countries
Source: OECD (2019)
23. 23
METHODOLOGY
SELECTION ISSUE NR. 1 – BALANCING OF COVARIATES
Common Support: Ensure overlap of distribution of characteristics by excluding
tails of the propensity score distribution
unmatched matched
Variable Treated Control Treated Control
log(num. employees) 4.88 4.18*** 4.88 4.93
% high skilled empl. 0.76 0.74 0.76 0.76
% male empl. 0.56 0.64*** 0.56 0.59
flextime 0.60 0.56 0.60 0.61
parental leave 0.21 0.14** 0.21 0.18
reorganization 0.40 0.26** 0.40 0.40
log(working hours) 3.66 3.67* 3.66 3.66
log(wage rate)(a) 4.27 4.15*** 4.27 4.23
% female managers (a) 0.12 0.14 0.12 0.11
average age 41.49 42.33** 41.49 41.73
Notes: Mean characteristics of firms by treatment status for the trimmed sample. t-test for equality in means between the groups. ***
Mean difference significant at 1%, ** significant at 5%, * significant at 10%. Matching procedure includes additional covariates. Match on 2
nearest neighbors with replacement. (a) log(wage rate), % female managers and age structure excluded form the matching process.
Balancing of observable characteristics