Labour Market Effects of Work- Sharing Agreements in Europe Discussion By

“Are Europeans Lazy? Or Americans Crazy?”
Portovenere, May 27th 2006
Labour Market Effects of WorkSharing Agreements in Europe
Giuseppe Nicoletti
• Title: is it an either/or issue?
• Discussion about employment effects of shorter workweek has long
history beginning in US Great Depression
• Objecting to the shorter (30 hours!) workweek proposed by the NIRA
in 1933, a member of the Cowles Commission (Roos, Econometrica
1935) pointed out many elements that are discussed in this report:
– Decreasing hours without increasing hourly wages is impossible
– Decreasing hours to stimulate employment necessarily results in higher
costs for firms and less employment, unless productivity hikes from less
fatigue are unrealistically high
– Decreasing hours in a coercive and inflexible way across industries fails
to recognize specificities and is unworkable
• No easy cross-country macro evidence of link between hours
worked and employment rates  so micro studies welcome
The merits of the report [1]
The theory chapter:
Provides simple tools that could help making the debate around this issue
more rational and clear (especially with social partners)
– Simulations, graphical analysis
Warning against simplistic views concerning the link between work-sharing
policies and employment
– Employment effects a priori ambiguous  empirical issue
But provides arguments that employment benefits of such policies can be
expected only under a range of special (and sometimes undesirable)
– Gains very unlikely under full wage compensation
– High gains from less fatigue and working time flexibility (low elasticity of labour
services to hours), high bargaining power, centralised bargaining, high market
power, strong focus of unions on employment, high preference for leisure
Thus, it provides useful leads for understanding role played by LM
institutions, product market competition and (intrinsic or policy-driven)
technology characteristics in determining employment effects of worksharing
The merits of the report [2]
The empirical chapters:
To my knowledge, first attempt to jointly review a range of national experiences from
a micro point of view
The country chapters are informative, although they are somewhat uneven
They bring out differences in approaches from which lessons can potentially be
drawn for future policy design
Coercive/statist vs negotiated approach to hours reduction (e.g. differences in flexibility,
gradualism, reversibility, public finance cost of shorter working time arrangements)
Hours vs alternation vs part-time schemes
Targeting (e.g. unemployed vs long-term unemployed in alternation schemes)
Moreover, some of the chapters provide original evidence on the hours reductionemployment link
In one case, they go beyond static “Keynesian” effects to look at longer-run and dynamic
effects (contribution to productivity slowdown, firm turnover)
Beyond employment, report also gives some (limited) evidence of welfare effects of
work-sharing policies, casting serious doubts on overall welfare benefits of hours
reduction in the French case
But is more positive on benefits of other work-sharing policies for beneficiaries
(career breaks and part-time)
General comments and suggestions [1]
Micro analyses give a partial account of the economic implications of work-sharing:
The analyses in the report are mostly short-run and over periods of negative output gaps. This
tends to favour “Keynesian”/”Malthusian” employment outcomes
By focusing on micro effects of policies they are partial equilibrium analyses and ignore the
implications for aggregate employment. With one exception, they do not explore dynamic
Also, focus on recent historical episodes is understandable, but deprives analysis of breadth:
what was the effect of cuts in hours in previous historical episodes (e.g. the 50s-70s)?
It is not clear that empirical analyses can distinguish well between effects of shorter
hours and effects of increased working time flexibility on employment
The analyses ignore the (indirect) negative effects on aggregate employment of
financing the subsidies that are granted to firms to mitigate the effects of wage
compensation for shorter hours
Complementary approaches include:
simulation of calibrated structural model (e.g. Gianella, 2006, Economie et Prévision)
impulse-response function and long-run solution of VECM (e.g. Altavilla et al., 2005, Journal of
Policy Modeling)
these results suggest that, when the three elements of work-sharing policies (shorter hours,
more flexibility, subsidies) are explicitly accounted for, long-run aggregate effects on
employment and (especially) potential output are unambiguously negative
General comments and suggestions [2]
Interesting hints in theory chapter concerning the influence of product market
competition and bargaining systems are only tenuously followed up in empirical
Why not controlling for market power in micro/sectoral analysis of employment effects?
The influence of bargaining arrangements or union objectives on the link between shorter
hours and outcomes is not explicitly investigated
Are concomitant effects of (changes in) other policies (unrelated to shorter hours)
adequately taken into account in empirical analysis?
Differences in differences approach only controls for other policies that affect equally all firms
But what about policy changes or policy interactions that may affect differently firms that adopt
shorter hours and other firms (e.g. EPL, entry regulations, taxation)?
In evaluating employment effects, there are a range of additional costs of work-sharing
policies that are difficult to be accounted for
Adjustment costs related to workplace reorganisation
Red tape related to administrative compliance and controls (especially in French case for
becoming eligible to subsidies)
Report does not mention the potential employment effects of another (perhaps more
widespread) work-sharing policy: early retirement schemes
General comments and suggestions [3]
Report provides only hints for explaining why work-sharing policies were
implemented in the first place. Yet, this would seem to be the most interesting
part of the story
For instance, a clear insider-outsider story seems to emerge in each chapter
– Faced with pressure of unemployment on wages, German and French insiders
may have opted for increasing total income (inclusive of the value of leisure)
through working hours reduction with full or partial wage offset
• This goes at the expense of outsiders (especially low skilled unemployed and workers in
non-unionised/non protected firms) who find it more difficult to keep or find a job
• Similar insider/outsider story on the firm side suggested by French results
– In Sweden, more employment-focused unions avoid shorter hours, but careerbreak programmes benefit mostly public sector employees with safe jobs and
seniority rules, with little gains for outsiders
• Taxpayers trade unemployment subsidies (for job matching and income support of
outsiders) against subsidies for improving welfare of insiders.
– Most likely, important wealth redistribution follows in all these cases
In the same spirit, other interesting issues not dealt with in the report are:
– What drove hours reductions in the 50s-60s?
– What is driving recent hours increases in GER? Have union objectives changed? If
so, why?
General comments and suggestions [4]
• In any case, it is difficult to fairly assess the welfare effects of worksharing policies without considering the market imperfections or
distortions that may have induced (or made necessary) these
policies in the first place
• For instance, the wish to produce services at home (e.g. in
continental European countries) is at least partially linked to
– Competitive conditions in product markets that affect the availability
and cost of market services
– Taxation policies, especially on second earners
– Externalities and social norms
• One could argue that a first best policy would be to eliminate some
of the distortions rather than adding to them through work-sharing
• Some of these factors are highlighted in the other report
Specific comments and suggestions [1]
• The empirical analysis of the French experience:
– Apart from Aubry I firms (in which a 10% hours reduction was legally
required), measurement errors in hours reductions can be great for
other kinds of arrangements
• Hence for non-Aubry I agreements the treatment can be heterogeneous
across firms, which may affect the analysis of employment effects
– Annualisation of working time was the counterpart of hours reduction
in 35 hours firms, but (as far as I understand) was not allowed in other
• Hence, it is difficult to distinguish between outcomes due to shorter hours
and outcomes due to increased working time flexibility (in other words the
treatment group differs from the control group by more than just hours
– Perhaps related to this, estimates of hourly productivity gains from
lower hours seem on the high side (see e.g. Cette’s lower estimates)
– Method used in table 4 should be spelled out more clearly (twostage?)
Specific comments and suggestions [1]
• The empirical analysis of the Swedish
– What about transaction/adjustment costs of
replacing/reintegrating workers?
– More information about kinds of jobs of absentees,
wages of replacement workers would be needed to
fully assess the programmes
– Is public sector profile of absentees related to more
long-term employment security and seniority wage
– No mention of fact that career break for over 50 is an
alternative early retirement route
– No mention of sickness leave policies
• Informative paper  diversity of approaches surveyed useful for
• Healthy warning against miraculous remedies to unemployment
• Micro approach fine but tends to underestimate negative effects
(productivity, indirect effects, additional costs)
• Important work-sharing policies overlooked (early retirement,
sickness leave)
• No attempt to explain empirically drivers of work-sharing policies
(institutions, political economy)
• No consideration of policy-induced distortions (taxation, labour and
product market regulation)