Democracy at work: A Study of the 2008 French Union ... · 1 End of the quasi monopoly given to the...

Democracy at work: A Study of the 2008

French Union Representativity Reform

Philippe Askenazy1 and Thomas Breda2

1Paris School of Economics (PSE), CNRS, and ENS

2Paris School of Economics (PSE), CNRS, and IPP

November 2016

Main changes introduced by the Law

1 End of the quasi monopoly given to the French historicalunions

• All unions with more than two years and respectingrepublican values can now apply for firm-level recognition

2 Electoral requirement for recognition in firms: at least 10%of votes during professional elections

• Whereas before the law, historical unions only had to find aworker willing to be a representative (de jurerepresentativeness)

3 Recognition at sectoral and national level depends on firmand worplace-level election results

⇒ The law introduces more democratic rules for appointingunion representatives

Main changes introduced by the Law

1 End of the quasi monopoly given to the French historicalunions

• All unions with more than two years and respectingrepublican values can now apply for firm-level recognition

2 Electoral requirement for recognition in firms: at least 10%of votes during professional elections

• Whereas before the law, historical unions only had to find aworker willing to be a representative (de jurerepresentativeness)

3 Recognition at sectoral and national level depends on firmand worplace-level election results

⇒ The law introduces more democratic rules for appointingunion representatives

What we do

• Study the effect of the 2008 law on “social capital”:

• Satisfaction and trust towards unions from both employersand employees

• Unionisation• Redistribution of welfare within the firm (who benefited the

most)• Conflicts

• Wages and other economic or financial outcomes: not inthis paper (data not available yet)

What we do

• Study the effect of the 2008 law on “social capital”:

• Satisfaction and trust towards unions from both employersand employees

• Unionisation• Redistribution of welfare within the firm (who benefited the

most)• Conflicts

• Wages and other economic or financial outcomes: not inthis paper (data not available yet)

Motivation (1)

• One of the rare policy evaluations regarding the institutionalrules governing trade unions’ actions

• Such policies are rare and hard to evaluate• Most studies of bargaining institutions are cross-country

comparisons• The 2008 French policy is a major change• It was implemented gradually and can be evaluated

• Organization of bargaining should matter for rent-sharingand trust

• Rent-sharing matters for inequalities• Corporate governance has changed dramatically, workers’

governance less so• Trust matters for productivity (and well-being)

Motivation (1)

governance less so

• Trust matters for productivity (and well-being)

Motivation (1)

governance less so• Trust matters for productivity (and well-being)

Motivation (2)

• Why are employment relations conflictual in some countriesand not others?

• Just a matter of culture?

• Should we try to change the culture directly?

• Alternatively, can institutions play a role and how fast?

Cooperation in labor-employer relations

Figure 1: Cooperation in labor-employer relations in selectedcountries

Source: World Economic Forum - The Global Competitiveness Index Historical Dataset.Note: France is 131 out of 151 countries.

External validity?

• Some external validity outside the firm:• Informative on the relationships between institutions,

culture and social capital (e.g. trust)

• Informative on the effects of democracy

• Contrast with the existing evidence• Which is mostly cross-country• With difficult identification

External validity?

culture and social capital (e.g. trust)• Informative on the effects of democracy

External validity?

• Contrast with the existing evidence• Which is mostly cross-country

• With difficult identification

External validity?

Roadmap

1 Institutional settings

2 Empirical strategy

3 Data

4 Main results

5 Robustness checks

6 Conclusion

Institutional details3 types of workers’ representation

1 Consultation, collective voice• 2 councils: work council, health and safety council• Members are elected every 2, 3 or 4 years and can be non

unionized• First election ballot only for unions

2 Individual workers’ voice• Elected workers’ delegates, first elec. ballot only for unions

3 Bargaining• Done by union reps: mandatory once a year when they are

present• Bargain mostly wages and working conditions• There can be several unions, and a few reps per union

(depending on firm/workplace size)• No direct election: only need 10% at work council elections

1st ballot

The 2008 reform of social democracy

• Mostly affects bargaining:

• Union reps now need to have collected at least 10% ofvotes on their names during professional elections firstballot

• All unions older than 2 years and respecting republicanvalues can have candidates at professional elections

Method (1)• New law starts to apply at the first election following its

promulgation

• Professional elections occur within each firm/workplace withmore than 10 employees according to a pre-definedfrequency

• 4 years unless sectoral or firm-level agreement reduce it to3 or 2 years.

• Very hard to delay next elections• Can bargain over next mandate length, not current one• Can extend current mandate a little bit, but heavy process

⇒ Election dates around 2008-2009 only depend on formerelection dates, and can be considered as quasi-random withrespect to the new law

• At least in firms that are old and large enough.

⇒ Identification is based on a regression discontinuity design

promulgation• Professional elections occur within each firm/workplace with

more than 10 employees according to a pre-definedfrequency

Method (2)We run equations of the type:

yj2011 = f0(Dj) + β1(Dj≥1jan2009) + f1(Dj)1(Dj≥1jan2009) + εj

• yj2011 is outcome of interest measured in early 2011

• Dj is the election date

• β is the local effect of the reform• f0 and f1 are functions capturing the effect of the distance

to the election on the outcome• Distance between survey and election is likely to affect

several outcomes (unionisation, conflicts, trust, etc.)• We may assume the effect is similar on both sides of the

cut-off

• Estimation• Polynomials of increasing order• Local linear with endogenous bandwidth (Cross-validation

method or Calonico, Cattaneo, and Titiunik selector)

Method (2)We run equations of the type:

yj2011 = f0(Dj) + β1(Dj≥1jan2009) + f1(Dj)1(Dj≥1jan2009) + εj

• yj2011 is outcome of interest measured in early 2011

• Dj is the election date

• β is the local effect of the reform• f0 and f1 are functions capturing the effect of the distance

to the election on the outcome• Distance between survey and election is likely to affect

several outcomes (unionisation, conflicts, trust, etc.)• We may assume the effect is similar on both sides of the

cut-off

• Estimation• Polynomials of increasing order• Local linear with endogenous bandwidth (Cross-validation

method or Calonico, Cattaneo, and Titiunik selector)

Timing of the reform• New law promulgated on August 20th 2008

• No subsequent legal order (Decret)• A ministerial order (Circulaire d’application) precising

practical details• Dated November 13th 2008• Published December 30th 2008

• Old system applies if elections’ first ballot preparatorymeeting is before August 21th 2008

• This meeting must be at least 30 days before first ballot• Election date must be published at least 45 days before

election first round• Usually negociations start beforehand

• Most elections before November 2008 are likely to be underold system, those in November and December are uncertain.

⇒ We set the cut-off date to be the 1st of January 2009 andperform robustness checks with cutoff on 1st of November2008.

1 REPONSE survey in 2011• Employment relations in ≈ 4000 workplaces of 10+

employees• Face-to-face interviews with employers• When elections took place, last election year given (≈ 2000

workplaces)• Questionnaire sent to ≈ 8000 randomly drawn workers in

those workplaces

2 Administrative data on elections for the period 2009-2012• First time it is used• We got first access 2 weeks ago• Only type of election and date of current and past election

for workplaces in REPONSE• No election results available so far

1 REPONSE survey in 2011• Employment relations in ≈ 4000 workplaces of 10+

employees• Face-to-face interviews with employers• When elections took place, last election year given (≈ 2000

workplaces)• Questionnaire sent to ≈ 8000 randomly drawn workers in

those workplaces

2 Administrative data on elections for the period 2009-2012• First time it is used• We got first access 2 weeks ago• Only type of election and date of current and past election

for workplaces in REPONSE• No election results available so far

Empirical analysis

2 approaches:

1 Use administrative data to retrieve exact date of mostrecent election before REPONSE survey

• Administrative data may be incomplete• Not so easy: several elections, but not all of them matter

for appointment of union reps.• Exact date of interview in REPONSE survey unknown

(between nov 2010 and May 2011)• We do cross-checking with year of election in REPONSE

2 Only use REPONSE survey• Self-declared year of election• Average outcomes by year of election, look for jumps

between 2008 and 2009• Run placebo with the 2004 REPONSE survey

Empirical analysis

2 approaches:

1 Use administrative data to retrieve exact date of mostrecent election before REPONSE survey

• Administrative data may be incomplete• Not so easy: several elections, but not all of them matter

for appointment of union reps.• Exact date of interview in REPONSE survey unknown

(between nov 2010 and May 2011)• We do cross-checking with year of election in REPONSE

2 Only use REPONSE survey• Self-declared year of election• Average outcomes by year of election, look for jumps

between 2008 and 2009• Run placebo with the 2004 REPONSE survey

Results: Length of time during two elections

Figure 2: Number of months between two consecutive elections

0 12 24 36 48 60Time between two elections in months

Note: The figure represents the length of time (in months) between any election during the period 2009-2012 andthe preceeding election. Partial elections have been removed.

Distribution of election dates

Figure 3: Distribution of election dates during the period of interest

1jan06 1jan07 1jan08 1jan09 1jan10 1jan11Date of last relevant election

Note: The figure represents the distribution of dates for the latest election that matters for union reps. appointmentbefore the REPONSE survey was done in early 2011.Source: Our own computations from administrative data (MARS) matched with REPONSE data.

Distribution of election dates

Figure 4: Distribution of election dates during the period ofinterest: zoom on 2008 and 2009

1jan06 1jan07 1jan08 1jan09 1jan10 1jan11Date of last relevant election

Note: The figure represents the distribution of the latest election that matters for union reps. appointment beforethe REPONSE survey was done in early 2011.Source: Our own computations from administrative data (MARS) matched with REPONSE data.

Results (part 1): Effects on workers’ representation

and unionization

1 Probability to have a union representative

2 Union membership

3 Probability to have a workers’ delegate

4 Probability to have a work council

5 Probability to have more than 4 unions

6 Probability to have an non-entranched union

Results: Presence of a union rep in the workplace

Figure 5: Election dates and presence of a union representative in2011

1jan07 1jan08 1jan09 1jan10 1jan11

16 bins. Left: N = 509, Right: N = 1728

Results: Percentage of union members in the

workplace

Figure 8: Election dates and percentage of union members in 20110

Results: Presence of a worker rep in the workplace

Figure 9: Election dates and presence of a worker representative in2011

Results: Presence of a work council in the

workplace

Figure 10: Election dates and presence of a work council in 20110

Results: Probability to have more than four unions

in the workplace

Figure 11: Election dates and probability to have more than fourunions in 2011

Results: Probability to have a non-entranched

union in the worplace

Figure 12: Election dates and probability to have a non-entranchedunion in 2011

Results: Probability to have a non-entranched

union in the worplace

Figure 13: Election dates and probability to have a non-entranchedunion in 2011

Results (part 2): Employers’ views on unions and

workers’ representatives

• Four questions asked to employers and employees on unions(in their firm if present):

1 “Unions play an irreplaceable role in representing workers”2 “Unions are helping (providing services to) workers”3 “Unions consider and defend their own interests before those of workers”4 “Unions are disturbing/troubling/hampering the good functionning of firms’

activities”

• Four questions asked to employers and employees onworkers’ representatives (in their firm if present):

1 “Workers’ representatives convey workers’ aspirations and claims”2 “During bargaining, workers’ representatives take into account firms’

economic possibilities/potential”3 “During bargaining, workers’ representatives influence the decision-making

of firm management”4 “In our firm/workplace, workers are able to defend their interests directly”

Results (part 2): Employers’ views on unions and

workers’ representatives

• Four questions asked to employers and employees on unions(in their firm if present):

1 “Unions play an irreplaceable role in representing workers”2 “Unions are helping (providing services to) workers”3 “Unions consider and defend their own interests before those of workers”4 “Unions are disturbing/troubling/hampering the good functionning of firms’

activities”

• Four questions asked to employers and employees onworkers’ representatives (in their firm if present):

1 “Workers’ representatives convey workers’ aspirations and claims”2 “During bargaining, workers’ representatives take into account firms’

economic possibilities/potential”3 “During bargaining, workers’ representatives influence the decision-making

of firm management”4 “In our firm/workplace, workers are able to defend their interests directly”

Results: Employers’ views

• 5 possible answers: Strongly agree, Slightly agree, Slighlydisagree, Strongly disagree, Do not know

• An additional question only for employers:• “In general, what is today the representativity of trade

unions?”

• We build standardized indexes• with the four questions on unions• with the four questions on workers’ representatives

• The fifth question only for employers is considered separetely

unions?”

Figure 14: Election dates and employers’ views on unions in theirfirm (standardized index)

-.4-.3

-.2-.1

Figure 15: Election dates and employers’ views on unions in theirfirm (standardized index)

-.4-.3

-.2-.1

Figure 16: Election dates and employers’ view on unions’representativity in general

-.3-.2

Figure 17: Employers have a positive view on unions’representativity in general

Figure 18: Employers have a very negative view on unions’representativity in general

Figure 19: Election dates and employers’ views on workers’representatives in their firm (standardized index)

-.4-.3

-.2-.1

Results: Workers’ views

Figure 20: Election dates and workers’ views on unions in theirfirm (standardized index)

-.4-.3

-.2-.1

32 bins. Left: N = 471 workplaces, Right: N = 1618 workplaces

Results: Workers’ views

Figure 20: Election dates and workers’ views on unions in theirfirm (standardized index)

-.4-.3

-.2-.1

Workers’ views: Women only

Figure 21: Women’s views on unions in their firm (standardizedindex)

-.4-.3

-.2-.1

Workers’ views: Young only

Figure 22: Younger workers (less than median 42) views on unionsin their firm (standardized index)

-.4-.3

-.2-.1

Workers’ views: Skilled

Figure 23: Skilled workers’ (college or more) views on unions intheir firm (standardized index)

-.4-.3

-.2-.1

Workers’ views: Executives

Figure 24: Executives’ views on unions in their firm (standardizedindex)

-.4-.3

-.2-.1

Convergence in workers’ trust towards unions?

Table 1: Value of trust towards unions index for differentcategories of workers

Women 0.11Men -0.07Younger workers -0.08Older workers 0.08Skilled workers -0.05Unskilled workers 0.03Executives -0.21Non executives 0.06

Estimation results

• Effect of the reform on employer views:• 30 to 45% of a standard deviation• Very robust

• One third-degree polynomial (similar on both side of thecut-off)

• One third-degree polynomial on the left and a differentslope on the right

• Coefficients very stable but get non significant• Local linear estimates highly significant as well

• Effect of the reform on employees’ views:• Around 10 to 15% of a standard deviation• Less robust:

• Not significant with the pure RDD approach• Significant when doing year-by-year comparisons• Significant on some subgroups

Robustness checks

• Analysis based on the year of election in REPONSE

• Placebos based on REPONSE survey in 2004

Employers’ views on unions in their firm

(standardized index)

Figure 25: Employers’ views on unions in their firm (standardizedindex) by year of election

-.3-.2

2007 2008 2009 2010 2011Year of election

Mean95% CI

Note: year of election given by respondant in 2011 REPONSE survey

Conclusion

• Positive effects of the reform on “social capital”,unionization and union coverage

• Larger effects for groups of workers the least represented byhistorical unions

• Reduce the variance in workers’ views

• Take aways:• Institutions can have a rapid effect on trust• More social democracy is desirable• Inspiring reform for European countries with limited

elecoral requirements for bargaining

Future research

• Long-term effect of the law versus local short-term effects

• Conflicts and social climate

• Effect of the reform on employment and economic outcomes

• Using all French firms• Get exhaustive MARS data• Match with French employer-employee wage data

(DADS+BRN)• Match with Workers’ flows data (DMMO)

Future research

• Long-term effect of the law versus local short-term effects

• Conflicts and social climate

• Effect of the reform on employment and economic outcomes

• Using all French firms• Get exhaustive MARS data• Match with French employer-employee wage data

(DADS+BRN)• Match with Workers’ flows data (DMMO)

General research agenda: theoretical motivation

• The labor market generates rents• specific human and physical capital (holdups), matching

complementarities, matching frictions, pure moving cost,etc.

• No good normative theory on how those rents should beshared

• Limited understanding of the determinants of “bargainingpower”: capital and workers governance are likely to matter

⇒ Positive unions’ effects on wages may not be harmful⇒ Useful to focus on unions’ governance and organization

General research agenda: empirical motivation

• Striking secular trends:• Tremendous changes in corporate governance

• CEO pay, disclosure rules, independant board members

• Parallel surge in research in corporate governance (newtools)

• Decline in unionisation and research on unions• Wage inequalities and capital share increased

⇒ Positive unions’ effects on wages may even be desirable⇒ Useful to focus on unions’ governance and organization

• Decline in unionisation and research on unions

• Wage inequalities and capital share increased

Figure 26: Agree (slightly or strongly) that unions are irreplaceable0

Figure 27: Agree (slightly or strongly) that unions are irreplaceable0

Figure 28: Agree (slightly or strongly) that unions are helpingworkers

Figure 29: Disagree (slightly or strongly) that unions serve theirinterests before workers’

Figure 30: Disagree (slightly or strongly) that unions trouble thegood functionning of the firm

Democracy at work: A Study of the 2008 French Union ... · 1 End of the quasi monopoly given to the...

Documents

Transcript of Democracy at work: A Study of the 2008 French Union ... · 1 End of the quasi monopoly given to the...