How effective are enforcement measures for compliance with the minimum wage? Evidence from Germany

Abstract

The extent of non-compliance with minimum wages is heavily debated, but little is known about the effectiveness of enforcement measures. Following the introduction of a national minimum wage in Germany in 2015, employers in a catalogue of industries deemed at high risk of non-compliance were subject to more stringent enforcement requirements, such as an obligation to record hours worked. Using national administrative employment data, in this study the authors exploit the variation in enforcement measures to analyze the effect on non-compliance. As an empirical strategy, they balance jobs from industries with stricter enforcement measures with jobs from other industries and apply difference-in-differences estimations. The evidence points to a small compliance-enhancing effect of the enforcement measures. The gains in compliance are not offset by more pronounced employment losses in those industries subject to stricter enforcement.

Keywords

Compliance difference-in-differences enforcement entropy balancing minimum wage

Introduction

In January 2015, a national minimum wage became effective in Germany, one of the hitherto few major industrial countries without one. Minimum wages are a contentious policy instrument and the specific effects of it are an empirical question (cf. e.g. Brown, 1999). While several studies assess the employment effects of Germany’s minimum wage introduction (e.g. Bossler and Gerner, 2019; Caliendo et al., 2017, 2018; Frentzen et al., 2018; Garloff, 2016), few studies have investigated compliance and enforcement.¹ This is surprising as the labor market effects of a minimum wage depend on enforcement and compliance. Indeed two studies of the minimum wage introduction in Germany (Bruttel et al., 2018; Caliendo et al., 2018) suggest non-compliance as an explanation for ‘the near absence of an effect on regular employment’ (Caliendo et al., 2018).² To understand the labor market effects of a minimum wage, it is therefore important to investigate enforcement and compliance. At least three broad categories of questions arise: (1) How large is the extent of non-compliance? (2) How effective is enforcement at increasing compliance? (3) How does enforcement affect employment? In the context of Germany’s introduction of a national minimum wage, this article provides evidence on all three questions but focuses on the second.

A major reason why the empirical literature is scarce is that the study of minimum wage compliance and enforcement faces challenges. First, due to its nature, non-compliance is subject to measurement error (e.g. Ronconi, 2010). Like the rest of the literature, we cannot in principle solve this issue. However, being able to use a very large administrative data set, we believe that the results contain information about the true relationship between enforcement and compliance. Second, compliance and enforcement are endogenous: enforcement should affect compliance, and compliance should affect enforcement (e.g. Ashenfelter and Smith, 1979). The approach we take to address this issue is to use sectoral variation in minimum wage enforcement stemming from preceding labor market regulations to analyze the effects of enforcement on compliance. Specifically, to curtail the endogeneity of compliance and enforcement, we employ a difference-in-difference strategy in which an entropy balancing algorithm matches jobs from industries with stronger enforcement to jobs from control industries with similar pre-2015 trends in wages.

To preview our results, we find that around 20% of wages that are below €2000 per month and that are covered by the minimum wage law fall short of the hourly minimum wage of €8.50, with considerable variation by industry. As for the current enforcement regime, some – but not all – estimates point to a slight reduction in non-compliance with the minimum wage. We did not see evidence that the rate of compliance rose as a result of the elimination of low-wage jobs.

The remainder of this article is organized as follows. In the next section, we summarize the related literature and highlight theoretical predictions regarding minimum wage compliance. In the third section, we describe the institutional background, with special emphasis on the institutional variation in enforcement that we will exploit empirically. The fourth section introduces our method and the data. The fifth presents the results of the empirical analysis in three steps: first, we establish that industries subject to stricter enforcement exhibited lower compliance before as well as after the minimum wage introduction; the core analysis consists in an assessment of the effect of enforcement measures on compliance; finally, we analyze if compliance gains are driven by employment reduction job terminations. The final section concludes.

Literature review

This article relates to several strands of literature: namely the literature on minimum wage compliance, evaluations of minimum wage effects and models linking enforcement and labor market outcomes. The following presentation focuses on the effects of the German minimum wage introduction, with references to international or general contributions.

Minimum wage compliance

Ashenfelter and Smith (1979) seems to be the first modern economics paper to draw a significant amount of attention to compliance with minimum wages. In the spirit of Allingham and Sandmo’s (1972) theoretical analysis of tax evasion, they laid out an economic model of firms’ minimum wage compliance decisions. Grenier (1982) and Chang and Ehrlich (1985) later criticized, modified, and generalized this model. In their theoretical analysis, Chang and Ehrlich (1985) conclude that the incentive for non-compliance increases with the minimum wage bite,³ and more so when the elasticity of labor demand is high (in absolute terms). Not surprisingly, the models also predict that compliance typically increases with the strength of enforcement. Besides their purely theoretical groundwork, Ashenfelter and Smith (1979) also suggested a (relative) measure of compliance that respects employment reductions as compliant behavior and intentionally excludes above-minimum wages from the basis of calculation. This proposed measure of compliance, too, has attracted criticism (Sellekaerts and Welch, 1984) and seldom been applied, with many studies measuring non-compliance by the number of covered employees paid below the minimum wage expressed as a proportion of the total covered workforce. Nevertheless, Ashenfelter and Smith’s contribution appears to have been instrumental in sparking a theoretical and, perhaps to a lesser extent, empirical literature on minimum wage compliance.⁴ We take an approach closer to Ashenfelter and Smith (1979) by analyzing the fraction of subminimum wage employments among lower wages only.

For Germany, the empirical literature on minimum wage compliance is still in its infancy. We are aware of four descriptive (German-language) reports (Günther and Frentzen, 2017; Pusch, 2018; Pusch and Seifert, 2017; Von Burauel et al., 2017). In absolute terms, the reported number of entitled employees paid below minimum wage ranges from 650,000 to 2.8 million employees. This variation demonstrates the difficulty to measure non-compliance accurately, but also shows that non-compliance is quantitatively meaningful even when looking at the most conservative numbers. Hence, it demonstrates that non-compliance with minimum wages is a significant issue for research.

The German Economic Institute (DIW) published non-compliance figures based on the German Socio-Economic Panel (SOEP; Von Burauel et al., 2017). According to their analysis, non-compliance is between 1.8 million and 2.8 million employees paid below the minimum wage.⁵ In a robustness check that allows for a measurement error of 10%, the number of entitled employees paid below the minimum wage shrinks to about 1 million. Using the same data set and focusing on contractual hours plus paid overtime, researchers of the WSI, a union-financed research institute, estimate 2.2 million non-compliant employments (Pusch, 2018).⁶ The absolute numbers correspond to between 7 and 11% of the covered employment contracts.

The German Statistical Office published numbers of employees below minimum wage based on the Structure of Earnings Survey. This employer survey includes official earning records of a well-defined sample of employees from within each of the selected workplaces (Günther and Frentzen, 2017). Excluding those employees not entitled to the minimum wage under the legislation, the authors report about 750,000 employees below minimum wage in 2015 and 650,000 in 2016. Two observations make the data source particularly reliable. First, it covers official earnings records, reported also to the social security administration. The data are highly accurate and can demonstrate plausible and distinct peaks in hourly wages after the minimum wage introduction. However, participation in the 2015 and 2016 surveys was voluntary, raising the concern of selective participation.

There is one study presenting recent non-quantitative evidence on enforcement activities by the customs authorities in Germany (Bosch et al., 2019). Apart from organizational issues, a major complaint are insufficient personnel resources for audits especially in the first two years after the minimum wage introduction.

Employment effects and other margins of adjustment

Compliance with the minimum wage is only one aspect of the minimum wage literature, with the bulk of the empirical literature focusing on employment effects and other margins of adjustment.⁷ From a theoretical perspective, the employment effect of a minimum wage can go in different directions, with classical models predicting employment losses (see e.g. Stigler, 1946) and monopsonistic models allowing for employment gains (see e.g. Ashenfelter et al., 2010; Bhaskar and To, 1999; Bhaskar et al., 2002; Manning, 2003, 2006; Stigler, 1946). The effects of enforcement and compliance on the employment impact of a minimum wage also depend on the competitiveness of the market (Basu et al., 2010; Chang and Ehrlich, 1985; Danziger, 2009, 2010; Yaniv, 2001). The available empirical studies of the impact of the German minimum wage find limited employment effects, at least in the short run. Caliendo et al. (2018), exploiting regional variation in the minimum wage bite, conclude that employment adjustment may not have exceeded 80,000 jobs across Germany. One explanation for the absence of major employment effects may be the absence of any discernible effect on reported monthly pay, as suggested in a companion paper (Caliendo et al., 2017). Instead, the authors find a significant working hours reduction, a potential channel to comply with the new law. Looking at labor demand adjustments in response to the German minimum wage introduction using firm data, Bossler and Gerner (2019) demonstrate that employment adjustments of German firms tend to increase with the wage effect of the minimum wage, following the argument of a downward sloping labor demand curve. They detect an employee-level wage effect of about 10% and a job termination probability of about 3%, in combination implying a labor demand elasticity of 0.3. Garloff (2016) finds no relation between the sectoral minimum wage bite and overall employment or unemployment growth, but an increase in the growth of regular employment at the expense of marginal employment.

Finally, it is worth noting that enforcement is typically costly and itself a policy choice (cf. e.g. Becker, 1968; Polinsky and Shavell, 2000; Stigler, 1970). One implication is that because of cost–benefit considerations, complete enforcement is economically rarely desirable (Polinsky and Shavell, 2000; Stigler, 1970). Another implication is that enforcement should be concentrated in areas where violations are most likely (Chang and Ehrlich, 1985; Ronconi, 2010). Indeed, already Ashenfelter and Smith (1979) reported evidence that enforcement efforts are ‘concentrated on sectors where violations are most likely to occur’. As we will see in the empirical analysis, this is also true for the German minimum wage.

Institutional background

Prior to the introduction of the national minimum wage in 2015, the only form of minimum wage setting on a statutory basis in Germany was at the industry level. Such minimum wages were introduced in particular through the Posted Workers Act of 1996. The regulation stipulates that collectively negotiated wages are declared binding for all employers of an industry, even for those that were not represented in the negotiations. Unions as well as the respective employers’ association and the Ministry of Labor and Social Affairs have to agree to such a declaration of an industry-specific minimum wage.⁸ These requirements only led to relatively few industry-specific minimum wages that were introduced, for example, in the main construction sector, for roofers and for hairdressers. Fueled by an increasing low-wage sector and wage inequality, the political support to introduce a national minimum wage increased over time. In contrast to sectoral minimum wages, the national minimum wage was specified by the government and enshrined in law. After the federal elections in 2013, the ruling coalition of the Christian Democratic Union and the Social Democratic Party agreed on the introduction of a national hourly minimum wage, following other major western economies such as the US, UK or France in that respect. The minimum wage was initially set at €8.50/hour and took effect on 1 January 2015. At the recommendation of the Minimum Wage Commission, constituted in equal parts of employer and employee representatives, it was raised to €8.84/hour in 2017, to €9.19/hour in 2019, and to €9.35/hour in 2020. The minimum wage applies to all employees, but with some exceptions, notably youths below 18 years of age, apprentices, certain interns and volunteers.⁹ During a transitional period of two years, any industry-specific minimum wages already agreed under the Posted Workers Act and other statutes were allowed to remain in force, irrespective of whether they were above or below the new statutory rate.¹⁰ Since 2017, industry-specific minimum wages are only applicable if they exceed the national minimum wage.

To enforce the minimum wage, the legislators made use of pre-existing laws and capacities of the inspection authorities. Arguably, the government did so to be able to introduce the minimum wage quickly and without major immediate investments in new enforcement capacities. Specifically, the 2014 Minimum Wage Act introduced the obligatory recording of hours of work for industries listed in § 2a of the Act to Combat Undeclared Work. The customs authorities, which had already been in charge of verifying compliance with industry-specific minimum wages and detecting illicit employment under this law, were also assigned the task to inspect compliance with the national minimum wage. In principle, the customs authorities can audit workplaces in any industry for minimum wage compliance; in practice, however, it has proven difficult to establish minimum wage non-compliance without an employer’s obligation to register and keep records of working hours. Although in 2015, the authorities prioritized information and awareness raising over enforcement, pegging minimum wage enforcement to pre-existing regulations resulted in systematically unequal treatment of workplaces in terms of the likelihood of detection in the event of non-compliance.

Accordingly, for the purposes of this study, we define treatment industries as those listed in § 2a of the Act to Combat Undeclared Work. These are the industries that are obligated to document and retain records of daily working hours and were hence the primary target of the inspection authorities. They comprise the construction sector, catering and hotels, passenger transportation, transportation logistics, exhibition industry, forestry, industrial cleaning, meat industry, prostitution and security services (SchwarzArbG, 2004). While excluding industries with industry-specific minimum wages under the Posted Workers Act (AEntG) and the temporary work industry, as these were already subject to inspections of their respective minimum wages, all remaining industries with a minor threat of inspections and without the requirement to record working hours serve as control industries.¹¹

Table 1 summarizes the numbers of treated and untreated industries using the German industry classification (WZ 2008) at the five-digit-level: 73 industries belong to the treatment group and 660 industries are assigned to the control group. In total, 106 industries already had been subject to enforcement measures before the minimum wage was introduced and therefore are excluded from the analysis. Employers in the treated industries are required to document working hours (jobs with wages below €2000 per month) whereas employers in control industries only have to record hours for minijob employments. Another important difference between the two groups emerges in the last column of Table 1: while firms in control industries hardly ever experience audits by the customs authorities, for firms in the treated industries the probability of audits is about 5% in the year 2016.

Table 1.

Industry variation in enforcement measures.

	Industries	Number of 5-digit industries (WZ08)	Compulsory recording of working hours	Likelihood of audits in 2016
Control group	All remaining industries	660	No (except minijob employment)	1.1%
Treatment group	Restaurants, hospitality, hotels, transportation and logistics, forestry, exhibitions and fairs	73	Yes	4.9%
Excluded group I (industry-specific minimum wage and included in the list of the Act to Combat Undeclared Work)	Main construction sector, roofing, electric trade, meat industry, cleaning, scaffolder, painting	25	Yes, already ahead of the mw introduction	6.1%
Excluded group II (industry-specific minimum wage)	Stonemasonry, social care, temporary agency work, further training industry	81	Yes, already ahead of the mw introduction

Sources: Definition of groups as in Bossler and Möller (2018). Number of establishments with full-time employees from the Establishment History Panel, 2016. Mindestlohnkommission (2018) for the number of audits at the establishment level.

Statutory provisions and enforcement activities will only affect compliance with the minimum wage if:

Sanctions are sufficiently high and

Enforcement is perceived as sufficiently likely.

As to the first condition, the penalties for non-compliance with the minimum wage law are statutorily similar for all employers and span several areas. Failure to pay the hourly minimum wage rate can attract a fine of up to €500,000. Violations of documentation, registration or examination requirements can be fined with up to €30,000. Moreover, fines exceeding €200 are documented in the central business register, which authorities must consult for public procurement contracts exceeding €30,000. Public administrations must temporarily exclude companies fined €2500 or more from public procurement biddings. Another class of penalties consists of damage claims. That is, following a successful claim by an employee at a Labor Court, employers are also liable to make up for any difference between the wage paid and the prevailing minimum wage rate. Even when an employee does not sue the employer, the social security institutions can demand restitution for unpaid social security contributions. In summary, violations can cause substantial fines as well as long run consequences such as exemptions from public procurement.

Condition 2 is the focus of our empirical analysis. A precondition for effective deterrence is that employers are aware of the enforcement. Survey data from IAB-QUEST provide some information in this respect. In this survey conducted by the Institute of Employment Research, about 300 establishments were asked about the additional administrative requirement associated with hours recordings in the course of the minimum wage introduction. The survey was conducted in early 2016, more than 12 months after the minimum wage introduction. To answer the question whether the administrative requirement had increased following the introduction of the minimum wage, employers had to choose among three response categories: (a) did not increase, (b) increased, and (c) heavily increased. The results in Table 2 show how the answers differ between treatment and control industries. The result clearly shows that employers in treatment industries are more likely to report an additional administrative requirement and are less likely to report that the administrative requirement did not change.

Table 2.

Additional administrative requirements associated with the minimum wage, ordered probit.

	Administrative requirements. . .
	Did not increase	Increased	Heavily increased
Treated (by stricter enforcement)	−0.111** (0.056)	−0.006(0.006)	0.117** (0.059)
Unconditional averages	0.311	0.314	0.376

Notes: Employers’ response to the question whether the minimum wage increased the administrative requirements at the respective establishment. Coefficients are partial effects from an ordered probit estimation on the three response categories indicated by column headings (did not increase, increased and heavily increased). Asterisks indicate significance levels: * 10%, ** 5% and *** 1%. Control variables for German states (16 categories), establishment size (7 categories), the share of minijobs and the share of female employees. Data source: IAB-QUEST Survey 2016, 322 establishments without industry-specific minimum wage and establishments with at least one employee receiving below €8.50 per hour of work ahead of the minimum wage introduction.

Summing up, we observe a big difference in the intensity of enforcement measures by industry. While fines for non-compliance can be quite substantial in all industries, the risk of detection for breaches of the law varies considerably. Moreover, employers in the industries with stricter enforcement activities are aware of these, at least in terms of increased administrative requirements associated with recording working hours. This awareness of enforcement measures could plausibly translate itself into differences in compliance with the minimum wage between industries with stricter enforcement and control industries. Before we present evidence on compliance, the next section describes our method and the data.

Method and data

In this section, we first describe our difference-in-differences strategy, which we combine with entropy balancing to identify the effects of the enforcement treatment.¹² The next subsection describes the data.

Method

The difference-in-difference regression comparing wages of jobs in treated industries with wages of jobs in control industries over time can be summarized as follows:

y_{i t} = \sum_{t = 2013}^{2016} t r e a t e d_{s} * y e a r_{t} * δ_{t} + \sum_{t = 2013}^{2016} y e a r_{t} * γ_{t} + t r e a t e d_{s} * α + φ_{i (s)} + ε_{i t}

(1)

where $y_{i t}$ captures the outcome of interest, indicating whether the wage in job i is below the (2015) minimum wage level. Before the introduction of the minimum wage, the outcome variable represents the ‘individual bite’ (afterwards it constitutes non-compliance).¹³ $δ_{t}$ captures the treatment effects for each year before and after the minimum wage introduction, where $t r e a t e d_{s}$ indicates jobs in treatment industries. $α$ captures time-constant differences between the treatment and control group.¹⁴ $γ_{t}$ includes common time effects for each year in the data and $φ_{i (s)}$ is the group effect which can be an industry effect or a worker-establishment fixed effect.

It is important to note that the basic difference-in-difference specification (1) is an assessment of the differential trends in the share of wages below the minimum wage level for the treatment and control group. Because the treatment group is not only affected by stricter enforcement measures, but also has a higher initial share of workers paid below minimum wage, the parameters $δ_{t}$ estimate the combined effect of the minimum wage introduction and the enforcement. To isolate the enforcement effect, we balance jobs in the treatment group with weighted jobs in the control group.

We use entropy balancing to reweight the control group.^15,16 It creates an artificial control group¹⁷ of jobs that closely resembles the jobs in the treatment group with respect to outcomes and various characteristics before the minimum wage introduction. Hence, we build a counterfactual of jobs allocated in the control industries for which we may plausibly assume that they are similarly affected by the minimum wage. The underlying assumption is that minimum wage induced wage adjustments are the same for the treatment group and the counterfactual, and therefore any remaining reactions to the introduction of the minimum wage should be due to their differential treatment with respect to enforcement.¹⁸ The enforcement effect is then identified by applying the entropy balancing weights in a weighted least squares estimation of equation (1).

It is not clear a priori which variables should be included in the balancing procedure. As a general guideline, we need variables that are good predictors of the probability that wages are below €1400 in the absence of a minimum wage. We choose to use both past outcomes and further control variables for balancing. To avoid endogeneity of covariates, we only use observations from the years preceding the minimum wage introduction. As job-level or individual-level outcomes are potentially volatile over time, we mainly use industry-level data, with one balancing model combining industry-level and employer- or job-level variables.

We estimate three balancing models with different sets of covariates:

1. The first model uses industry-level measures of the minimum wage bite and the industry wage structure:

- industry-level bite of the years 2012, 2013 and 2014 ( $b i t e_{s, t = 2012, 2013, 2014}$ ),

- industry-level average percentage deviation of wages from the minimum wage for wages below that level ( $p b i t e_{s, t = 2012, 2013, 2014}$ ),

- industry-level share of employees paid below €2000 per month, which is a central sample selection criterion in our analysis ( $s a m p l e_{s, t = 2012, 2013, 2014}$ ).

2. The second model uses the variables from model 1 and additional industry-level variables describing an industry’s employment and employer structure in the year 2013:

- industry-level share of women, share of skilled and high-skilled workers, share of foreigners,

- industry-level average age of workers, average age of establishments, average number of employees.

3. The third model uses the variables from model 1 and additionally job-level characteristics of workers and employers:

- female, skilled, high-skilled, foreign, age,

- establishment age, number of employees.

Data

The major data source is a 10% sample from the administrative employment data (Employee History; BeH), containing individual employment information from compulsory social security notifications, and available at the Institute for Employment Research (IAB). The large sample size of the data allows us to compare low-wage jobs in treated and non-treated industries, conditional on individual, establishment and industry characteristics. The wage information is very accurate, but comes with two major drawbacks: first, it refers to the reported period of employment; second, the data do not contain information on hours worked.

Our main analysis sample is a 10% sample of all workers in regular employment (with full social insurance) who had at least one job during the years 2012–2016. From this sample, we select all full-time employees in post on 30 June for the years 2012–2016 and aged between 13 and 65.¹⁹

Our treatment variable is defined by industry, as described in the last section. Jobs in industries listed in the Act to Combat Undeclared Work form the treatment group and jobs in industries that are exempted from obligatory working time recording form the control group. These industries are identified in our data using the five-digit-level of the industry classification of 2008 (see Table 1). Jobs from industries with pre-existing minimum wages in which obligatory recording of working times was required already before the introduction of the general minimum wage are excluded.²⁰

To operationalize the primary outcome, whether a wage paid was below or above the national minimum wage level, we follow Garloff (2016, 2017) in translating monthly wages to hourly wages. That is, based on an assumption of 165 hours worked per month, we define monthly wages below €1400 as falling below the minimum wage level.²¹

Our control variables are constructed at different levels: we use individual characteristics including gender, nationality, level of vocational qualification. We also control for establishment age, which is computed in years using the first appearance of an establishment in the administrative employment records, and for establishment size, coded as a categorical variable. We distinguish Western and East Germany with a binary indicator. Industry variables are computed using average values of job-related characteristics.

To check the sensitivity of our results with respect to the assumption of constant working hours across jobs, we use information from the Quarterly Earnings Survey run by the Federal Statistical Office (Statistisches Bundesamt [Destatis]). From these external data, we obtain the average number of hours worked in an industry, differentiated by region (West/East) and gender, in the second quarter of each year from 2012 to 2016.²² This allows us to generate an alternative measure of minimum wage bite/compliance.

In the last step of our empirical analysis, we look at monthly industry-level separation rates. They are defined as the number of terminated jobs in an industry per month, divided by the number of existing jobs existing in the same industry at the end of the preceding month, and are calculated separately for Western and East Germany.²³

Results

Establishing differences in compliance in treated and control industries

As a first step of our empirical analysis, we plot the fraction of workers paid below the minimum wage level (as in force since 2015) over the analysis period from 2012 to 2016 for treatment industries and the control group. Figure 1 illustrates the averages of $I [w a g e_{i t} < € 1400]$ I[wageit < €1400] for the two groups over time. Before the introduction of the minimum wage, the fraction of workers paid below the minimum wage level was about 6 percentage points higher in the treated industries compared with the control group. This illustrates that the industries subject to stricter enforcement measures were more impacted, on average, by the introduction of the statutory minimum wage. A downward trend for both groups indicates positive wage trends in the lower half of the wage distribution already before the introduction of the national minimum wage. A sharp drop between 2014 and 2015 reflects the effect of the minimum wage, which significantly decreased the fraction of workers paid below the respective threshold. In 2015 and 2016, the fraction of workers paid below €1400 per month is still at a higher level among the treated industries. This suggests that (on average) the targeting of industries based on the Act to Combat Undeclared Work has some justification; possibly narrowing the gap in compliance between treated and control industries. Among the sample of workers with monthly earnings below €2000, the calculated level of non-compliance of about 20% is remarkable.

Figure 1.

Fraction of workers paid below minimum wage before and after the reform.

To examine these trends in compliance statistically, including controls for industry and job matches, we estimate several versions of equation (1). The results are listed in Table 3. The outcome variable is either an indicator for monthly wages below €1400 (models 1–3) or an indicator for (imputed) hourly wages below €8.50 (models 4–6).

Table 3.

Effect of enforcement on wages below the minimum wage threshold.

	Dependent variable
	I[monthly wage < €1400/month]			I[hourly wage < €8.50/h]
	(1)	(2)	(3)	(4)	(5)	(6)
	OLS	Industry-FE	Match-FE	OLS	Industry-FE	Match-FE
Treated*D2013	−0.008***	−0.008***	−0.002	−0.006***	−0.006***	0.001
	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)
Treated*D2014	−0.012***	−0.014***	−0.003	−0.008***	−0.009***	0.008***
	(0.003)	(0.003)	(0.003)	(0.003)	(0.003)	(0.003)
Treated*D2015	−0.048***	−0.052***	−0.049***	−0.037***	−0.039***	−0.029***
	(0.003)	(0.003)	(0.004)	(0.003)	(0.003)	(0.004)
Treated*D2016	−0.049***	−0.055***	−0.052***	−0.043***	−0.047***	−0.039***
	(0.003)	(0.003)	(0.005)	(0.003)	(0.003)	(0.005)
Treated	0.059***	−	−	0.062***	−	−
	(0.003)			(0.003)
D2013	−0.016***	−0.018***	−0.033***	−0.010***	−0.012***	−0.029***
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
D2014	−0.027***	−0.030***	−0.066***	−0.022***	−0.025***	−0.064***
	(0.001)	(0.001)	(0.002)	(0.001)	(0.001)	(0.002)
D2015	−0.106***	−0.110***	−0.153***	−0.107***	−0.112***	−0.163***
	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)
D2016	−0.109***	−0.114***	−0.168***	−0.115***	−0.121***	−0.185***
	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)	(0.003)
Constant	0.291***	0.310***	0.337***	0.326***	0.347***	0.377***
	(0.002)	(0.001)	(0.001)	(0.002)	(0.001)	(0.001)
N	1,318,998	1,318,998	1,318,998	1,350,177	1,350,177	1,350,177

Notes: Reported coefficients are from linear difference-in-differences estimations. D2013 to D2016 are dummy variables for years of observation. Establishment-level clustered standard errors are in parentheses. Asterisks based on establishment-level clustered standard errors indicate significance levels: * p < 0.1, ** p < 0.05 and *** p < 0.01. FE = fixed effects. Data: BeH 2012–2016, analysis sample.

Columns 1 and 4 show the results of OLS regressions. In addition, we estimate two types of fixed effects models, controlling for industry (models 2 and 5) and alternatively for job matches (combined person–establishment effects, models 3 and 6). Deviating results in these fixed effects specifications would point to a greater role for heterogeneous industry effects or for idiosyncratic job matches. However, we observe only small differences between the fixed-effect models and the OLS results. For the difference in the evolution of the outcomes between treatment and control industries, the coefficients reflect the pattern observed already in Figure 1. The level difference of jobs paid below the minimum wage threshold already before the minimum wage introduction is roughly 6 percentage points (coefficients of ‘Treated’). The interaction effects indicate that this initial level difference diminishes over time, and the remaining gap represents the differential change of the outcomes in treatment and control industries in the period after the minimum wage introduction. The results are similar for the indicator based on hourly wages.

The results of Figure 1 and Table 3 show that the minimum wage has had a larger bite in the treatment group, the level of non-compliance is somewhat larger in these industries, and the wage adjustment was somewhat stronger. At this point of analysis, however, we cannot attribute the estimated changes in the outcomes to the interventions of the customs authorities in the treatment group. Rather, what we observe is a mix of the effects of stricter enforcement measures in the treatment group, a stronger effect of the minimum wage introduction on the treatment group due to higher shares of low-wage workers, and further differences between the two groups affecting the responses of firms and jobs to the minimum wage introduction.

Different effects of the minimum wage on the treatment and control industries are possibly due to differences in the structure of firms and in the characteristics of workers and jobs. Apart from the threat of enforcement activities, the propensity to keep wages below the minimum wage might be higher for small firms than for large ones because of the lack of internal control mechanisms (i.e. works councils or a trade union presence). Also for jobs where workers are easily replaced (and therefore may not dare to demand the legal wage) and generally for the least productive jobs, which are at risk of being abolished when higher wages are enforced, the risk of non-compliant wages increases. We therefore now move to an analysis of enforcement effects that controls for such differences, i.e. entropy-balanced difference-in-differences.

Effect of enforcement measures on non-compliance

In the second step of our empirical analysis, we aim at estimating an isolated effect of enforcement measures. To predict what would have happened to a job in the treated industries without the threat of being controlled by the customs authorities, by intuition we would compare this job to another job equally affected by the minimum wage but not subject to stricter enforcement. We build a counterfactual of jobs within the control industries for which it can be plausibly assumed that they are as affected as those in the treatment industries by the introduction of the minimum wage. As explained in the Method section, we use an entropy balancing algorithm to achieve this.

In order to give a visual impression of the enforcement effect and to demonstrate that the balancing procedure has been effective in equalizing the enforcement group and the control group in the period before the introduction of the minimum wage, Figure 2 shows the mean values of the weighted outcome variable obtained from the first set of balancing variables. The weighted control group closely resembles the treatment group in the years ahead of the minimum wage introduction, i.e. by putting a larger weight on jobs in more severely affected industries in the control group. This illustration shows that entropy balancing produces equal pre-introduction bite in treatment and control industries to support our assumption that the minimum wage induced adjustment in the enforcement and the control industries should be very similar. Hence, differences in the proportion of jobs with wages below the minimum wage level in the years 2015 and 2016, after the minimum wage introduction, can plausibly be attributed to their differential treatment in terms of enforcement. Figure 2 suggests that even for jobs in very similar industries, the likelihood to be paid below the minimum wage has decreased more if the worker is employed in an industry subject to stricter enforcement measures.

Figure 2.

Fraction of workers below minimum wage before and after the reform after balancing treatment and control group.

Regression techniques allow for a more precise estimate of the enforcement effect. Table 4 contains the results from estimating equation (1) using different sets of balancing variables. For each set of balanced data, we estimate a regression with industry-specific fixed effects and another regression with match-specific fixed effects. Columns (1) and (2) in Table 4 contain results from the first set of balanced data, columns (3) and (4) the results from the second and columns (5) and (6) results from the third set of balancing variables.

Table 4.

Effect of enforcement on wages below the minimum wage threshold, monthly wages.

	Dependent variable: I[monthly wage < €1400/month]
	(1)	(2)	(3)	(4)	(5)	(6)
	Balancing model 1		Balancing model 2		Balancing model 3
	Industry-FE	Match-FE	Industry-FE	Match-FE	Industry-FE	Match-FE
Treated*D2013	0.000	0.000	−0.000	−0.006	−0.003	−0.001
	(0.003)	(0.003)	(0.008)	(0.009)	(0.003)	(0.003)
Treated*D2014	−0.000	0.001	−0.001	−0.008	−0.006	−0.003
	(0.003)	(0.004)	(0.010)	(0.012)	(0.004)	(0.005)
Treated*D2015	−0.026***	−0.032***	−0.001	−0.024	−0.029***	−0.036***
	(0.004)	(0.006)	(0.011)	(0.016)	(0.005)	(0.007)
Treated*D2016	−0.028***	−0.035***	−0.005	−0.032**	−0.031***	−0.042***
	(0.004)	(0.006)	(0.011)	(0.016)	(0.005)	(0.007)
Treated	−	−	−	−	−	−
	−	−	−	−	−	−
D2013	−0.025***	−0.034***	−0.024***	−0.027***	−0.022***	−0.032***
	(0.002)	(0.002)	(0.008)	(0.009)	(0.003)	(0.003)
D2014	−0.044***	−0.067***	−0.043***	−0.057***	−0.038***	−0.063***
	(0.003)	(0.003)	(0.009)	(0.010)	(0.004)	(0.004)
D2015	−0.137***	−0.165***	−0.161***	−0.171***	−0.134***	−0.160***
	(0.003)	(0.004)	(0.010)	(0.014)	(0.004)	(0.006)
D2016	−0.144***	−0.178***	−0.166***	−0.178***	−0.141***	−0.170***
	(0.003)	(0.005)	(0.011)	(0.014)	(0.005)	(0.006)
Constant	0.368***	0.397***	0.441***	0.418***	0.391***	0.399***
	(0.006)	(0.019)	(0.013)	(0.040)	(0.007)	(0.023)
N	1,318,852	1,318,852	1,318,852	1,318,852	1,318,852	1,318,852

Notes: Reported coefficients are from linear difference-in-differences estimations. D2013 to D2016 are dummy variables for years of observation. Balancing variables are measures of the industry wage/minimum wage incidence 2012–2014 in models (1) and (2). In addition to these variables, models (3) and (4) use 2013 industry shares of women, foreigners, skill groups, worker age, age of firm and firm size, while models (5) and (6) use actual individual characteristics of workers and firms. Establishment-level clustered standard errors are in parentheses. Asterisks indicate significance levels: * p < 0.1, ** p < 0.05 and *** p < 0.01. Data: BeH 2012–2016, analysis sample.

The enforcement effects are given by the coefficients of the interaction terms for the years 2015 and 2016. For the years 2013 and 2014, the coefficients of the interaction terms in Table 4 are small and insignificant, confirming that, before the introduction of the minimum wage, treatment group and weighted control group are comparable with respect to the share of jobs with wages below €1400. For the years 2015 and 2016, the interactions are negative, indicating a slight compliance enhancing effect of the enforcement. When significant, the estimates for the two years range from −0.042 (last column in 2016) to ∓0.026 (column (1) in 2015), implying an enforcement effect of 2 to 4 percentage points. However, the estimates based on the second balancing model are smaller and turn insignificant in column (3) for both years and in column (4) for the year 2015.

To understand the differences in our results across specifications, Table 5 lists the most important control industries after applying the balancing weights. The individual weights result from the balancing procedure. The ‘industry weight’, reflecting the importance of an industry for the control group, is obtained by summing up the weights by industry and year and taking averages. Among the most important industries are dental practices, retail bakeries, and manufacturers of bread and pastries. These industries are not included in the treatment group, because they are not listed in the Act to Combat Undeclared Work, but are chosen by means of our balancing procedure as being comparable regarding the incidence of low wages and other industry and job characteristics before the introduction of the minimum wage.

Table 5.

Ten most important industries in the balanced control groups (by balancing model).

Rank	Classification number (WZ08)	Industry	Individual weight	Average number of jobs	Industry weight
	Balancing model 1
1	86230	Dental practices	1.46	6803.4	9958.9
2	47240	Retail bakeries, flour confectionery and sugar confectionery in specialized stores	1.81	2787.6	5057.6
3	10710	Manufacture of bread; manufacture of fresh pastries and cakes	0.56	7471.2	4153.5
4	86220	Medical specialist practices	0.57	6014.6	3436.6
5	86902	Activities of midwives and male midwives, physiotherapists and other paramedical practitioners in the field of massage, hydrotherapy, etc.	1.07	2837.2	3041.7
6	47761	Retail sale of flowers, plants, seeds, fertilizers in specialized stores	1.30	1563.0	2029.3
7	82200	Call centers	0.52	3617.6	1872.8
8	96022	Beauty parlors	4.19	411.8	1725.3
9	81291	Cleaning of means of transport	5.73	258.0	1477.6
10	86210	General medical practices	0.51	2511.0	1287.5
	Balancing model 2
1	45202	Washing, polishing, etc. of motor vehicles	22.07	478.2	10551.8
2	46320	Wholesale of meat and meat products	12.35	505.4	6240.9
3	81100	Janitor services	3.08	1965.8	6061.8
4	47820	Retail sale via stalls and markets of textiles, clothing and footwear	162.33	20.2	3279.1
5	81291	Cleaning of means of transport	11.60	258.0	2992.2
6	93210	Amusement parks and theme parks	4.45	591.2	2631.2
7	77210	Renting and leasing of recreational and sports goods	50.27	51.6	2593.8
8	47301	Retail sale on behalf of others of automotive fuel in specialized stores	1.21	2061.0	2493.8
9	47210	Retail sale of fruit and vegetables in specialized stores	6.20	345.2	2141.0
10	25620	Mechanics, not elsewhere classified	0.73	2295.6	1666.0
	Balancing model 3
1	86230	Dental practices	0.51	6803.4	5589.9
2	10710	Manufacture of bread; manufacture of fresh pastries and cakes	0.32	7471.2	4096.8
3	47240	Retail bakeries, flour confectionery and sugar confectionery in specialized stores	0.68	2787.6	3648.9
4	86902	Activities of midwives and male midwives, physiotherapists and other paramedical practitioners in the field of massage, hydrotherapy, etc.	0.67	2837.2	3087.0
5	45202	Car washes	3.07	478.2	2512.3
6	82200	Call centers	2.45	3617.6	2236.0
7	86220	General medical practices	0.34	6014.6	2190.0
8	47761	Retail sale of flowers, plants, seeds, fertilizers in specialized stores	2.90	1563.0	1910.4
9	47810	Retail sale via stalls and markets of food, drink and tobacco	19.53	173.8	1750.2
10	81291	Cleaning of means of transport	4.57	258.0	1703.8

Notes: ‘Individual weight’ is the weight of each individual job in an industry. ‘Average number of employees’ is the number of jobs in an industry, averaged over sample years. The ‘Industry weight’ is given by the product of the individual weight and the number of workers in the industry, and indicates the importance of the respective industry in the control group.

Remarkably, balancing models 1 and 3 largely coincide in the 10 most important control industries. By contrast, most of the important control industries from model 2 are not contained in the industry sets from the other two models. This is a first plausible explanation for differences in the presented enforcement effects in Table 4. Moreover, the individual weights in the control industries of balancing model 2 show large values in some cases, larger than those from model 1 and 3. This means that there are control industries consisting of a rather small number of jobs, which receive a large weight by our balancing procedure. This is a second reason for differences in the estimation results in Table 4 and it implies that effects are estimated less precisely.

We may note that the second balancing model uses industry-level characteristics of workers and employers in addition to the wage-structure related variables of the first model. The large weights of the second model suggest that using industry level variables to balance treatment and control industries both on the wage structure and on the employment structure might just not be feasible or comes at the price of assigning high weights to a few industries. So, for some or all the treated industries, the wage structure is accompanied by an employment and employer structure for which it is not easy to find matches within the group of control industries.²⁴

Heterogeneity in enforcement effects

An obvious reason for difficulties of finding controls could be heterogeneity of industries within the treatment group with some industries having idiosyncratic features. In these cases, using different control groups within the group of treatment industries is an alternative to the construction of just one single control group.

We estimate separate effects for each of the treatment industries, repeating the balancing procedure for each industry. Estimation is now performed at the three-digit-level of the NACE category, as otherwise the case numbers for some industries are too small. The 28 industries contained in the treatment group at the three-digit-level enabled us to estimate 28 separate enforcement effects. The effects are weighted by the number of jobs in each of these industries and displayed in Figure 3. The size of the enforcement effects is displayed on the x-axis and the range with relevant amounts of density is from −10 to +4 percentage points. This implies some positive estimates of the enforcement effect, with the mass however clearly in the region below zero. Based on these results and the results obtained before (Table 4), we may conclude that the enforcement effect is negative and not very large on average but may be as large as (minus) 10 percentage points in some individual industries.

Figure 3.

Industry-specific enforcement effect heterogeneities.

Evidence with imputed hourly wages

Table 6 replicates the main results of enforcement effects identified through entropy balancing, but uses an hourly wage measure to indicate individuals paid below the minimum wage level, where hours are imputed from an external survey, the German Quarterly Earnings Survey, on industry-specific working hours and earnings (see Data subsection). The estimated coefficients for the interaction terms of enforcement and time are in the range from −0.044 to −0.018 (when significant) in the years 2015 and 2016, showing slightly more variation than the results from Table 4. However, as in the estimates based on monthly wages, when using the weighted control group from the second balancing model, which accounts for pre-existing worker characteristics at the industry level, most of the coefficients are smaller and insignificant.

Table 6.

Effect of enforcement on wages below the minimum wage threshold, hourly wages.

	Dependent variable: I[hourly wage < €8.50/hour]
	(1)	(2)	(3)	(4)	(5)	(6)
	Balancing model 1		Balancing model 2		Balancing model 3
	Industry-FE	Job-FE	Industry-FE	Job-FE	Industry-FE	Job-FE
Treated*D2013	−0.000	−0.001	0.002	−0.005	−0.005	−0.009**
	(0.004)	(0.004)	(0.007)	(0.008)	(0.004)	(0.004)
Treated*D2014	−0.000	0.005	0.000	−0.001	−0.010*	−0.005
	(0.005)	(0.005)	(0.009)	(0.011)	(0.005)	(0.006)
Treated*D2015	−0.018***	−0.022***	0.006	−0.009	−0.026***	−0.030***
	(0.006)	(0.008)	(0.010)	(0.014)	(0.007)	(0.009)
Treated*D2016	−0.023***	−0.030***	−0.016	−0.037**	−0.033***	−0.044***
	(0.006)	(0.009)	(0.012)	(0.017)	(0.007)	(0.011)
Treated	−	−	−	−	−	−
	−	−	−	−	−	−
D2013	−0.018***	−0.025***	−0.019***	−0.022***	−0.013***	−0.017***
	(0.003)	(0.004)	(0.007)	(0.008)	(0.004)	(0.004)
D2014	−0.034***	−0.057***	−0.034***	−0.052***	−0.024***	−0.047***
	(0.004)	(0.005)	(0.009)	(0.011)	(0.005)	(0.005)
D2015	−0.133***	−0.166***	−0.156***	−0.179***	−0.125***	−0.157***
	(0.005)	(0.007)	(0.010)	(0.013)	(0.006)	(0.008)
D2016	−0.148***	−0.188***	−0.154***	−0.183***	−0.136***	−0.174***
	(0.006)	(0.008)	(0.012)	(0.016)	(0.007)	(0.009)
Constant	0.468***	0.445***	0.518***	0.453***	0.491***	0.441***
	(0.007)	(0.021)	(0.012)	(0.036)	(0.008)	(0.025)
N	1,350,062	1,350,062	1,350,062	1,350,062	1,350,062	1,350,062

Enforcement and separation rates

As noted in the discussion of the minimum wage literature, the observed gains in compliance in the treatment industries might be driven by employment effects if establishments responded by reducing the number of low-wage jobs rather than raising wages to the required minimum.²⁵ This could then be an indicator for the minimum wage level set ‘too high’ for some low-wage jobs to survive. Even if the literature suggests that employment effects are rather small on aggregate, there might be a stronger employment reaction in the treatment industries because establishments have to cope not only with the introduction of the minimum wage but possibly also with a higher administrative burden and/or stricter audits by the customs authorities.

In Table 7, we present results of regressions of the monthly separation rate for the period from 2013 to 2016, using indicators for years and calendar months as well as interactions of the yearly indicators with the enforcement variable as covariates. In columns 3 and 4, the results are based on balanced samples, with the balancing variables given by the balancing model 3 already used for the analysis of the non-compliance with the minimum wage. We would expect at least a temporary rise in the separation rate in all industries (reflected in the yearly indicators) or a comparably larger increase in the separation rate in the treatment industries (reflected in the interactions of years and enforcement indicator). However, we do not find any evidence of changes (albeit temporary) in the separation rate over time, neither for the control nor for the treatment industries.²⁶ We conclude that the elimination of jobs has not been an important adjustment strategy of establishments in our sample when confronted with the minimum wage introduction.

Table 7.

Enforcement measures and industry-level separation rates.

	Unbalanced sample		Balanced sample
	(1)	(2)	(3)	(4)
Treated	0.013***	0.007***	0.000	0.000
	(0.003)	(0.002)	(0.006)	(0.004)
D2014	−0.000	−0.000	0.002	0.002
	(0.001)	(0.001)	(0.001)	(0.001)
D2015	−0.001	−0.001	0.001	0.001
	(0.001)	(0.001)	(0.002)	(0.002)
D2016	−0.001	−0.001	−0.001	−0.001
	(0.001)	(0.001)	(0.002)	(0.002)
Treated*D2014	0.000	0.000	−0.002	−0.002
	(0.002)	(0.002)	(0.003)	(0.002)
Treated*D2015	−0.002	−0.002	−0.004	−0.004
	(0.002)	(0.002)	(0.003)	(0.003)
Treated*D2016	−0.002	−0.001	0.001	0.001
	(0.002)	(0.002)	(0.004)	(0.004)
Western Germany	0.010***	0.010***	0.011***	0.011***
	(0.001)	(0.001)	(0.003)	(0.003)
Female (share)	−	−0.009***	−	−0.004
		(0.003)		(0.009)
Foreign (share)	−	0.027	−	0.094***
		(0.027)		(0.025)
Age (mean)	−	−0.000	−	−0.000
		(0.000)		(0.001)
Skilled (share)	−	−0.065***	−	−0.025
		(0.023)		(0.017)
High-skilled (share)	−	−0.027	−	0.004
		(0.019)		(0.020)
Constant	0.022***	0.076***	0.038***	0.049*
	(0.001)	(0.024)	(0.007)	(0.027)
N	65,048	65,048	65,048	65,048

Notes: Results from industry-level OLS regressions, unbalanced samples and samples based on balancing model 3. Dependent variable: monthly separation rate for full-time jobs with initial monthly earnings of less than €2000. D2013 to D2016 are dummy variables for years of observation. Results for a set of calendar month controls are not reported. Establishment-level clustered standard errors are in parentheses. Asterisks indicate significance levels: * p < 0.1, ** p < 0.05 and *** p < 0.01. Data: BeH 2013–2016, analysis sample.

Discussion and conclusion

The present paper analyzes non-compliance with the newly introduced German minimum wage and the effectiveness of measures to enforce compliance. The minimum wage was set at €8.50 per hour of work at its introduction in 2015 and through 2016. Enforcement measures consist of an obligation to record working hours and unannounced audits by the German customs authorities. For practical reasons, enforcement varies by industry: industries listed in the Act to Combat Undeclared Work must record working hours and are substantially more likely to be audited while other industries hardly ever experience audits by the customs authorities. This creates a natural experiment allowing us to analyze the effectiveness of enforcement measures.

Our descriptive evidence shows that industries subject to stricter enforcement measures had a higher proportion of wages below €8.50 before the introduction of the national minimum wage. After the minimum wage introduction, in 2015 and in 2016, the gap between treatment and control industries narrowed, but jobs in the treatment industries were still more likely to be paid below the new mandatory floor. These two findings suggest that on average, authorities concentrated enforcement efforts on industries with a higher risk of non-compliance (Mindestlohnkommission, 2018).

When designing enforcement institutions, legislators should consider their efficiency (Polinsky and Shavell, 2000; Stigler, 1970). To a first approximation, the concentration of enforcement efforts on a group of industries in which non-compliance is most likely represents an enforcement implementation at moderate cost. On the other hand, it is not clear whether the resources spent on controlling compliance with the minimum wage are sufficient to be a credible threat for non-complying firms (see Table 1). Furthermore, an enforcement focus on specific industries creates some degree of arbitrariness because several control industries (e.g. bakeries, retailing and car washing) exhibit a level of non-compliance similar to the treatment industries.

To isolate the effects of enforcement measures from the wage and employment effects of the national minimum wage introduction, we employ an entropy balancing procedure. On a very detailed industry level and using a large individual data set, this allows us to compare the outcomes in the industries subject to stricter enforcement measures with the outcomes of appropriately weighted observations from control industries.

With respect to non-compliance, we find some evidence that the treatment industries did indeed experience a marked reduction in the share of jobs paid below minimum wage compared to the group of control industries. However, using different sets of variables in the entropy balancing, as well as different measures of compliance, the significant effect of enforcement on non-compliance does not hold in all specifications. When significant, the point estimates are small, in the range of 2 to 4 percentage points.

Difficulties in finding control observations may be due to particularities of some of the industries subject to stricter enforcement measures. By constructing separate control groups for each industry, we also estimate enforcement effects that are free to vary across industries. We get some positive estimates of the enforcement effect, with the job-level average effect continuing to be negative. For some jobs, the enforcement effect is estimated to be as large as minus 10 percentage points.

A possible shortcoming of our analysis is the lack of information on hourly wages. We repeat the main estimations with an alternative measure of non-compliance based on imputed working time information from an external survey. The set of estimates of the enforcement effects across models is similar to the results obtained when using the original wage variable.

Rather than adjusting wages, compliance may be achieved by eliminating the jobs in question. Similar to the non-compliance models, we estimate models of the monthly job separation rate. The lack of significant differences between treatment and control industries with respect to changes in the job separation rate suggests that job elimination is not the primary channel in which the enforcement measures affect compliance.

Are there any policy conclusions that can be derived from our results? Although the estimated effect is small, a straightforward request could be to devote more resources to the customs authorities. In fact, the customs’ statistics for the years 2017 and 2018 demonstrate an increase in the number of audits since 2017 and an extension of audits to further industries like taxis or call centers (Bundesregierung, 2019a, 2019b). It might become easier to detect non-compliance after a recent ruling of the European Court of Justice (2019), which obligates member states to ensure that employers document daily working times. In addition to increasing the number of audits, however, improvements in the process of sanctioning and prosecuting underpaying establishments seem to be necessary (Bosch et al., 2019). Furthermore, alternatives to legal controls and sanctions may be considered. As an innovative example, researchers from the German DIW propose a ‘Fair Pay’ certification for employers guaranteeing to document working times in an objective and transparent way (Fedorets et al., 2019).

Footnotes

Acknowledgements

We thank the Data and IT-Management Department of the IAB for data provision and we thank Lisa Feist for particularly helpful research assistance.

*

The opinions expressed in this article are those of the authors and do not necessarily reflect the views of the Inter-American Development Bank (IDB), its Board of Directors, or the countries they represent.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Mario Bossler

Notes

Author biographies

Mario Bossler is head of the working group Minimum Wages and a post-doctoral researcher at the Institute for Employment Research (IAB). His research interests focus on minimum wages and other labor market institutions. Among his recent publications is a minimum wage evaluation with HD Gerner, ‘Employment effects of the new German minimum wage: Evidence from establishment-level microdata’, ILR Review (2020), 73(5).

Ursula Jaenichen is a senior researcher at the Institute for Employment Research (IAB). She has worked on various topics of applied labor market analysis. More recently, her interest is in the effects of the minimum wage and in aspects of data quality in survey and administrative data.

Simeon Schächtele is a post-doctoral researcher at the Inter-American Development Bank. Applying impact evaluation methods, his research focuses on public, labor, and behavioral economics. Among his recent publications is ‘Tax responses at low taxable incomes: Evidence from Germany’, Fiscal Studies (2020), 41(2).

References

Abadie

Diamond

Hainmueller

(2010) Synthetic control methods for comparative case studies: Estimating the effect of California’s tobacco control program. Journal of the American Statistical Association 105(490): 493–505.

Allingham

Sandmo

(1972) Income tax evasion: A theoretical analysis. Journal of Public Economics 1(3–4): 323–38.

Ashenfelter

Smith

(1979) Compliance with the minimum wage law. Journal of Political Economy 87(2): 333–350.

Ashenfelter

Farber

Ransom

(2010) Labor market monopsony. Journal of Labor Economics 28(2): 203–210.

Basu

Chau

Kanbur

(2010) Turning a blind eye: Costly enforcement, credible commitment and minimum wage laws. Economic Journal 120(543): 244–269.

Becker

(1968) Crime and punishment: An economic approach. Journal of Political Economy 76(2): 169–217.

Bellmann

Bossler

Gerner

H-D

Hübler

(2018) Collective bargaining coverage, works councils and the new German minimum wage. Economic and Industrial Democracy: Epub ahead of print 23 March 2018. DOI: 10.1177/0143831X18762304

Bhaskar

(1999) Minimum wages for Ronald McDonald monopsonies: A theory of monopsonistic competition. Economic Journal 109: 190–203.

Bhaskar

Manning

(2002) Oligopsony and monopsonistic competition in labor markets. Journal of Economic Perspectives 16(2): 155–174.

10.

Bhorat

Kanbur

Mayet

(2012) Estimating the causal effect of enforcement on minimum wage compliance: The case of South Africa. Review of Development Economics 16(4): 608–623.

11.

Bhorat

Kanbur

Mayet

(2013) A note on measuring the depth of minimum wage violation. Review of Labour Economics and Industrial Relations 27(2): 192–197.

12.

Bhorat

Kanbur

Stanwix

(2015) Partial minimum wage compliance. IZA Journal of Labor and Development 4(1): 1–20.

13.

Bosch

Hüttenhoff

Weinkopf

(2019) Kontrolle von Mindestlöhnen. Berlin: Springer VS.

14.

Bossler

Gerner

H-D

(2019) Employment effects of the new German minimum wage: Evidence from establishment-level micro data. Industrial and Labor Relations Review 73(5): 1070–1094.

15.

Bossler

Möller

(2018) Auswirkungen des gesetzlichen Mindestlohns [Written hearing of the German Minimum Wage Commission]. IAB-Stellungnahme, 01/2018, Nuremberg.

16.

Brown

(1999) Minimum wages, employment, and the distribution of income. In: Ashenfelter

Card

(eds) Handbook of Labor Economics, Vol. 3, Part B. New York: Elsevier, pp. 2101–2163.

17.

Bruttel

Baumann

Dütsch

(2018) The new German statutory minimum wage in comparative perspective: Employment effects and other adjustment channels. European Journal of Industrial Relations 24(2): 145–162.

18.

Bundesregierung (2019a) Mindestlöhne – Kontrollen der Finanzkontrolle Schwarzarbeit im Jahr 2018. Bundestagsdrucksache 19/8830.

19.

Bundesregierung (2019b) Mindestlohnkontrollen in den Bundesländern. Bundestagsdrucksache 19/9573.

20.

Caliendo

Fedorets

Preuss

, et al. (2017) The short-term distributional effects of the German minimum wage reform. IZA-Discussion Paper No. 11246.

21.

Caliendo

Fedorets

Preuss

, et al. (2018) The short-run employment effects of the German minimum wage reform. Labour Economics 53: 46–62.

22.

Chang

Y-M

Ehrlich

(1985) On the economics of compliance with the minimum wage law. Journal of Political Economy 93(1): 84–91.

23.

Danziger

(2009) Noncompliance and the effects of the minimum wage on hours and welfare in competitive labor markets. Labour Economics 16(6): 625–630.

24.

Danziger

(2010) Endogenous monopsony and the perverse effect of the minimum wage in small firms. Labour Economics 1(1): 224–29.

25.

European Court of Justice (2019) Ruling from 14 May 2019, Case C-55/18. Available at: https://curia.europa.eu/jcms/upload/docs/application/pdf/2019-05/cp190061en.pdf

26.

Fedorets

Grabka

Schröder

(2019) Mindestlohn: Nach wie vor erhalten ihn viele anspruchsberechtigte Beschäftigte nicht. DIW-Wochenbericht 28/2019, Berlin.

27.

Frentzen

Beck

Stelzer

(2018) Beschäftigungswirkungen des Mindestlohns. Wirtschaft und Statistik 1: 35–51.

28.

Garloff

(2016) Side effects of the new German minimum wage on (un-)employment: First evidence from regional data. IAB Discussion Paper, No. 4.

29.

Garloff

(2017) Side effects of the introduction of the German minimum wage on employment and unemployment. Evidence from regional data. Update. Bundesministerium für Wirtschaft und Energie, Diskussionspapier 04, Berlin.

30.

Garnero

(2018) The dog that barks doesn’t bite: Coverage and compliance of sectoral minimum wages in Italy. IZA Journal of Labor Policy 7(1).

31.

Grenier

(1982) On compliance with the minimum wage law. Journal of Political Economy 90(1): 184–187.

32.

Günther

Frentzen

(2017) Verdiensterhebung 2016: Erhebung über die Wirkung des gesetzlichen Mindestlohns auf die Verdienste und Arbeitszeiten der abhängig Beschäftigten. Ergebnisbericht zum 28. April 2017. Im Auftrag des Bundesministeriums für Arbeit und Soziales (BMAS). Wiesbaden: Statistisches Bundesamt.

33.

Hainmüller

(2012) Entropy balancing for causal effects: A multivariate reweighting method to produce balanced samples in observational studies. Political Analysis 20(1): 25–46.

34.

Hainmüller

(2013) ebalance: A Stata package for entropy balancing. Journal of Statistical Software 54(7): 1–18.

35.

Lechner

(2010) The estimation of causal effects by difference-in-difference methods. Foundations and Trends in Econometrics 4(3): 165–224.

36.

Manning

(2003) Monopsony in Motion: Imperfect Competition in Labour Markets. Princeton, NJ and Oxford: Princeton University Press.

37.

Manning

(2006) A generalised model of monopsony. Economic Journal 116(508): 84–100.

38.

Marinakis

(2014) Incumplimiento con el salario mínimo en América Latina. Santiago: Organización Internacional del Trabajo.

39.

Metcalf

(2008) Why has the British national minimum wage had little or no impact on employment? Journal of Industrial Relations 50(3): 489–512.

40.

Mincer

(1976) Unemployment effects of minimum wages. Journal of Political Economy 84(4): 87–104.

41.

Mindestlohnkommission (2018) Zweiter Bericht zu den Auswirkungen des gesetzlichen Mindestlohns. Bericht der Mindestlohnkommission an die Bundesregierung nach § 9 Abs. 4 Mindestlohngesetz. Berlin.

42.

Polinsky

Shavell

(2000) The economic theory of public enforcement of law. Journal of Economic Literature 38(1): 45–76.

43.

Pusch

(2018) Bilanz des Mindestlohns: Deutliche Lohnerhöhungen, verringerte Armut, aber auch viele Umgehungen. Policy Brief WSI 19.

44.

Pusch

Seifert

(2017) Unzureichende Umsetzung des Mindestlohns bei Minijobbern. Wirtschaftsdienst 97(3): 187–191.

45.

Rani

Belser

Oelz

(2013) Minimum wage coverage and compliance in developing countries. International Labour Review 152(3): 381–410.

46.

Ronconi

(2010) Enforcement and compliance with labor regulations in Argentina. Industrial and Labor Relations Review 63(4): 719–736.

47.

SchwarzArbG (Schwarzarbeitsbekämpfungsgesetz) (2004) Act to Combat Undeclared Work and Unlawful Employment. Available at: www.gesetze-im-internet.de/englisch_schwarzarbg/englisch_schwarzarbg.html

48.

Sellekaerts

Welch

(1984) An econometric analysis of minimum wage noncompliance. Industrial Relations 23(2): 244–259.

49.

Statistisches Bundesamt (Destatis) (2020) Preise. Verbraucherpreisindizes für Deutschland. Lange Reihen ab 1948. April 2020. Available at: www.destatis.de/DE/Themen/Wirtschaft/Preise/Verbraucherpreisindex/Publikationen/Downloads-Verbraucherpreise/verbraucherpreisindex-lange-reihen-pdf-5611103.pdf

50.

Statistisches Bundesamt (Destatis) (various years) Verdienste und Arbeitskosten. Arbeitnehmerver-dienste. Fachserie 16, Reihe 2.1: Tables 4.1.2 and 4.1.3, 2012(2)–2016(2).

51.

Stigler

(1946) The economics of minimum wage legislation. American Economic Review 36(3): 358–365.

52.

Stigler

(1970) The optimum enforcement of laws. Journal of Political Economy 78(3): 526–536.

53.

Viollaz

(2018) Enforcement of labor market regulations: Heterogeneous compliance and adjustment across gender. IZA Journal of Labor Policy 7(2): 1–28.

54.

Von Burauel

Caliendo

Fedorets

, et al. (2017) Minimum wage not yet for everyone: On the compensation of eligible workers before and after the minimum wage reform from the perspective of employees. DIW Economic Bulletin 7(49): 509–522.

55.

Weil

(2005) Public enforcement/private monitoring: Evaluating a new approach to regulating the minimum wage. Industrial and Labor Relations Review 58(2): 238–257.

56.

Yani

(2001) Minimum wage noncompliance and the employment decision. Journal of Labor Economics 1(3): 596–603.