Building the minimum wage: the distributional impact of Germany’s first sectoral minimum wage on wages and hours across different wage bargaining regimes

Abstract

The first minimum wage in Germany was introduced in 1997 for blue-collar workers in sub-sectors of the construction industry. In the setting of a natural experiment, blue-collar workers in neighboring 4-digit industries and white-collar workers are used as control groups for differences-in-differences-in-differences estimation based on linked employer–employee data. Estimation results reveal a sizable positive impact on mean wages in East Germany, but no significant effect in West Germany. Size and significance of effects are neither homogeneous across wage regimes (individual vs. collective contracts) nor across the distribution. The patterns suggest a compression in the lower part of the wage distribution and spillover effects to wages where the minimum is not binding, even in West Germany, where the bite of the MW was low. No effects on hours of work or substitution between workers of different qualification levels are found.

Appendix: Details on the data source

The GSES (Hafner and Lenz 2007) is an official micro database that every few years collects a cross section of data from establishments (Betrieb) with 10 or more employees. Establishments are legally bound to respond and non-response is low. At the employee level, the information supplied includes details on wages, hours worked, overtime, (payroll) taxes, education, job description, a rough classification of the tasks fulfilled in terms of intra-firm hierarchy, and time with the employer. At the establishment level, region, industry code, number of employees, fraction of blue- and white-collar workers, fraction of men and women, and participation in CAs are surveyed. The data do not contain any information on job quits.

I use two cross sections of the data from October 1995– 2001, restricting the sample to employees between 18 and 65 years of age, who are neither receiving vocational training nor participating in internships. The data allow for an accurate calculation of hourly wages as contracted hours are reported for every employee. ¹¹ Hours, according to contract, are used to compute hourly wages, since the variable on hours paid only exists for 70 % of observations in the sample. For estimation results, it does not make a difference if I use log gross hourly wage based on total working hours or log of monthly labor income as the dependent variable instead of hourly wages based on hours according to contract. Hourly wages calculated to be lower (higher) than €3 (€150) are not considered in the analysis for plausibility. The data discern between individual wage contracts, coverage by a CA, firm or establishment agreement. The variable on the difficulty of tasks fulfilled captures differences in qualification needed for the job and degree of responsibility; yet it is not possible to identify directly the pay scale as implemented in the CAs.

The sub-sectors in the construction industry that were subject to the MW legislation cover the main construction trade (4-digit industry codes 4521 through 4525 in the WZ93 classification). Plumbing, building installation, floor and wall covering, painting, and glazing constitute the control sectors (industry codes 4533, 4534, 4543, 4544). The remaining sub-sectors in the construction industry are not assigned to any of the two groups due to one of the following reasons: (1) their industry classification changed between 1995 and 2001 and conversion of the classification is not unambiguously possible; (2) a few other sector-specific MWs were introduced from 1997 onwards (see Section 3); Sectors that implemented their own MW rate in 1997 were excluded; (3) it is not certain whether they were treated or not.