Czech FirmLevel Bargaining and Wages

1
Czech Firm-Level Bargaining and Wages

• EVIDENCE FROM MATCHED EMPLOYER-EMPLOYEE DATA

tepán Jurajda
CNB September 14, 2005
2
OUTLINE

• DATA
• COVERAGE
• ANALYSIS OF WAGE LEVEL (mean and median)
• at individual level
• at firm level

3
1. DATA AND MEASUREMENT

• ISPV (MPSVMoL, Trexima)
• entrepreneurial sector only
• firms with over 10 employees
• Data have lower sampling rates for smaller firms.
  Weighting matters for firms with under 250
  employees.
• Hourly wages for 1Q 2004 in over 2500 firms.
• Add 4Q 2003 indicator of firm-level collective
  agreement (with wage stipulations) based on firm
  answers in ISPV and a follow-up firm telephone
  survey 2000 firms have CA i.e. 24 missing (more
  likely to be smaller firms, foreign or private
  firms, in hotel/restaurant, retail and
  construction).
• Add 4Q2003 indicator of higher-level agreements
  (industry-level) based on MoL extensions 450
  firms, of which 250 have the firm-level CA
  indicator available.

4
2. UNION COVERAGE57 (74) 85 of employees
are covered by firm-level coll. agreement (CA)
based on full (limited) no weighting. Combining
the firm- and higher-level CAs, the coverage is
70 (80) 90. Even if we suppose that all
missing-CA-indicator firms have no CA, 44 (61)
75 of employees are still covered by firm
collective agreements. gt A safe bet is
that union coverage is over 50 (70) before
(after) extensions.
5
2. COVERAGE of firm-level CA in firms over 250
employees (unweighted)
6
3a. INDIVIDUAL WAGE ANALYSIS

• The weighted average hourly wage is a bit higher
  in firms with firm CA (117 vs. 115CZK). The
  weighted median wage is 10 (8 CZK) higher with
  firm CA (101 vs. 83 CZK).
• Excluding top 1 of wages (gt451), covered avg.
  wages are 8 higher (103 vs. 95). Excluding
  managers (ISCOlt20) avg. (median) covered wages
  are 6 (17) higher.
• Average log-wage trade-union premium of 10 is
  not affected by worker demographics, but drops to
  0 after controlling for firm size and 2-digit
  industry.
• There is no unconditional log-wage premium to
  having a higher-level CA, but controlling for
  worker and/or firm characteristics, there is a
  5 gap compared to no-CA firms.
• Covered firms pay substantially more to most of
  their workforce, but not their managers.
• But this wage gap disappears once we compare
  firms of similar size and industry.
• In firms with over 250 employees (limited
  weightingfull weighting)
• the avg. wage is 3 CZK lower and the median wage
  is 4 CZK higher in covered firms
• unconditional log-wage premium to firm-level CA
  is zero but it is -20 for higher-level CA
  however, conditional log-wage gaps are small and
  mostly insignificant.
• Conditional firm-level-CA log-wage premium in
  large firms
• for men with college -0.23, high-school
  -0.14, apprenticeship 0, primary 0
• for women with college -0.17, high-school
  -0.10, apprenticeship 0, primary 0
• Within firm type, those firms that have a CA pay
  less to highly productive (male) workers.

7
3a. BY INDUSTRYWages in Large Firms Effect of
Firm-Level CA (ignoring higher CAs)
8
3a. BY INDUSTRYWages in All Firms Effect of
Firm-Level CA (ignoring higher CAs)
9
3. Things to do

• Of course, the descriptive comparisons provided
  up to now do not identify the causal effect of
  trade union coverage on wages.
• What more can one do with the available
  cross-sectional data?
• INDIVIDUAL WAGE ANALYSIS
• Run quantile regressions
• Is there an effect on gender wage gaps? (no)
• Is predictive power of occupation dummies
  stronger in unionized firms? (no)
• Etc.
• FIRM-LEVEL ANALYSIS
• Control for propensity score (matching) instead
  of relying on regressions, DiNardo and Lee (2004)
• Card and de la Rica (2004) Spanish wages about
  10 percent higher with firm-specific contracts
  compared to industry agreements.