34
* *

A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

A�ectedness and anticipatory behavior

in response to the new German minimum wage:

the role of industrial relations

Lutz Bellmann∗

Mario Bossler†

Hans-Dieter Gerner‡

Olaf Hübler�

June 2015

∗Institute for Employment Research, Friedrich-Alexander-University of Erlangen-

Nuremberg, and IZA.†Corresponding Author: Institute for Employment Research, Department Estab-

lishments and Employment, Regensburger Str. 104, 90478 Nuremberg, e-mail:

[email protected], phone: +49-911-179-3043.‡University of Applied Sciences Koblenz and Institute for Employment Research.�Leibniz Universität Hannover, IZA and IAB.

1

Page 2: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Abstract

In Germany, a new minimum wage of e8.50 per hour of work came into

force on 1 January 2015 . The minimum wage is largely binding and followed

a long lasting policy discussion, making anticipatory adjustments likely. Us-

ing the IAB Establishment Panel, we assess the role of industrial relations

in this respect. Descriptive regressions show that �rms with a works coun-

cil and collective bargaining are less likely a�ected by the new minimum

wage. When studying reactions ahead of the minimum wage introduction,

we observe that anticipatory wage adjustments are particularly emphasized

at establishments with collective bargaining. Looking at other adjustments

related to industrial relations, we �nd that the a�ected establishments' col-

lective bargaining coverage decreases. We also �nd that employers seek �exi-

bility and try partially to compensate the additional minimum wage costs by

using overtime work and working time accounts. Finally, a�ected employers

tried to increase productivity by the use of further training. [150 words]

2

Page 3: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

1 Motivation

Employees in Germany face a decreasing coverage in collective bargaining

(e.g. Addison et al. 2011). Figure 1 shows this trend in a descriptive time

series for the period from 1996 to 2014. While in the late 1990s more than

60 per cent of the work force was covered by collective bargaining, this rate

fell to about 40 per cent in recent years. A similar trend can be shown for

the fraction of employees that is represented by a works council, which fell

from about 40 per cent to 35 per cent within the same period. Together,

these patterns show a severe deterioration in fundamental institutions of the

industrial relations in Germany.

Figure 1: Time series of industrial relations' coverage in Germany

0

20

40

60

Cov

erag

e in

per

cen

t

1996 1998 2000 2002 2004 2006 2008 2010 2012 2014Year

Industry-level collective bargainingFirm-level bargainingWorks council

Notes: Yearly average coverage weighted by establishments size, which allows an interpreta-

tion in terms of covered employees.

Source: IAB Establishment Panel, 1996-2014.

Following the political discussion on the decreasing collective bargaining

coverage and the expansion of the low wage sector, the new German mini-

mum wage of e8.50 per hour of work was introduced on 1 January 2015. The

o�cial name of the legislation �Tarifautonomiestärkungsgesetz� depicts that

3

Page 4: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

the law intends to strengthen industrial relations in Germany. Theoretically,

the minimum wage could in�uence the bargaining over wages and strengthen

the position of unions. This is the case if the minimum wage creates a higher

fall-back position. A higher fall-back a�ects Nash-bargaining outcomes but

also strengthens the employees' position in the right-to-manage model, which

corresponds to the German collective bargaining process.

Many research questions are implied by the relation between industrial re-

lations and the new minimum wage. These include whether the minimum

wage is rather a complementary or a substitutive institution to the industrial

relations in Germany. Other open questions include how the a�ectedness and

anticipatory wage adjustments relate to the coverage by industrial relations.

It is also an open question to what extent establishments reacted with �rm-

level policies already ahead of 1 January 2015 and how industrial relations

relate to such anticipatory adjustments.

The German Council of Economic Experts (�Sachverständigenrat�) already

expected the minimum wage to dampen the economic development in 2014

(Sachverständigenrat 2014). This expectation received a wide public echo

when the German chancellor Angela Merkel expressed her disagreement with

this prediction. However, since the minimum wage legislation followed a

lengthy policy discussion, anticipatory reactions become likely and we shall

see that about 11 per cent of the establishments in our sample reported to

have adjusted wages already a few month ahead of 1 January 2015. We

also look at other potential adjustments and provide �rst evidence on antic-

ipatory policies in response to the minimum wage introduction and propose

several outcomes which are in the scope of industrial relations.

First, we should stress that collective bargaining coverage of establishments

could be a�ected by two opposing mechanisms. On the one hand, establish-

ments may have an incentive to joint collective agreements, because the new

law includes an exemption that existing bargaining agreements are allowed

to undercut the minimum wage until the end of 2016. However, this bene�t

only holds for two years and therefore adjustment costs might exceed the

associated bene�ts. On the other hand, employers may have a tendency to

leave collective bargaining as the minimum wage sets an alternative lower

4

Page 5: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

limit on wages and makes the bargaining over wages in such a�ected wage

groups redundant.

Our second scope is on adjustments concerning working time arrangements

such as overtime hours and working time accounts. The possibility of over-

time work may be introduced to increase the �exibility of the given labor

inputs. Working time accounts are also an instrument for the �exible use

of contracted working time, but for both employers and employees. How-

ever, working time accounts may also be introduced to track the employees'

working time, which is a way to check compliance to the minimum wage

legislation in a world where wages are paid in monthly pay checks.

Finally, we analyze adjustments in the provision of further training. This

is of particular interest since classical economic theory predicts wages to be

determined by the value of marginal product. Establishments may therefore

have an incentive to provide further training as an instrument to increase

the productivity of the respective workers which in turn could justify the

minimum wage.

The article is structured as follows: Section 2 describes the data source and

the sample for the analysis. Section 3 shows how the a�ectedness of es-

tablishments di�ers for those with collective bargaining and works council.

Section 4 presents results on the relation between anticipatory wage setting

and industrial relations. Section 5 evaluates other anticipatory reactions

ahead of the minimum wage introduction which include bargaining cover-

age, working time arrangements, and the provision of further training. In

Section 6, we present some robustness checks including the estimation of

non-linear speci�cations and alternate de�nitions of the treatment group.

Section 7 concludes.

2 Data

The data source of the analysis is the IAB Establishment Panel, which is a

large annual survey on general �rm policies and personnel developments in

Germany. The IAB Establishment Panel includes about 15,000 observations

each year since 1993. The survey's gross population units are all establish-

5

Page 6: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

ments located in Germany with at least one employee liable to social security.

The sample selection is representative for German states (�Bundesländer�),

industries, and establishment size. The interviews are conducted face-to-

face by professional interviewers, which ensures a high data quality and a

response rate of, on average 83 per cent. More comprehensive data descrip-

tions of the IAB Establishment Panel can be found in Ellguth et al. (2014)

or Fischer et al. (2009).

The 2014 cross-section of the IAB Establishment Panel contains informa-

tion on the establishments' a�ectedness by the minimum wage. The survey

includes information on the extensive a�ectedness by asking whether the re-

spective establishment has employees with an hourly wage below e8.50. But

it also includes information on the intensive margin counting the number of

currently a�ected employees with an hourly wage below e8.50. Finally, the

survey asks establishments whether wages were already adjusted within the

past 12 month, and hence, in anticipation of the minimum wage introduc-

tion.1

An unique establishment identi�er allows tracking establishments over time

if the respective establishments continue to participate in the survey. For

research questions addressing anticipatory reactions relative to 2013, the es-

tablishment identi�er allows to trace back establishments while using the

2014 a�ectedness by the minimum wage.

The descriptive statistics in Table 1 present a �rst snapshot of the analysis

sample comprising of all private sector establishments in 2014. We observe

9,321 establishments, of which 1,582 are a�ected by the minimum wage (ex-

tensive margin) and 1,019 establishments report to have already adjusted

wages in anticipation of the legislation. Within a�ected establishments the

intensive margin of a�ected employees is on average 39 per cent.

Looking at the coverage by industrial relations, 39 per cent of all establish-

ments have a works council and 32 per cent participate in collective bargain-

ing. Additionally, another 23 per cent of establishments reports an orien-

1The question on anticipatory wage setting covers the time period between the German

general election in fall 2013 and the period of the survey, which was conducted from June

to September 2014.

6

Page 7: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

tation at existing bargaining agreements without an explicit commitment.

Similarly, the incidence of over-time hours, the use of working time accounts,

and the provision of further training are well represented in the data.

When looking at anticipatory adjustments with respect to collective bar-

gaining coverage, overtime hours, working time arrangements, and further

training, we trace back the 9,321 establishments in the data and compare

these hypothesized workplace policies with the previous year, which is 2013.

In this year we do not expect an anticipatory behavior.

3 A�ectedness by the minimum wage

Before analyzing anticipatory behavior in response to the minimum wage, we

assess the a�ectedness of establishments. The a�ectedness is an important

concept as it di�erentiates a treated group of establishments from a unaf-

fected control group. It also gives an insight to what extent the minimum

wage is binding. We expect that a�ected establishments react in response to

the minimum wage since they have to adjust wages to comply with the law,

but they may also adjust other establishment-level measures to cope with

the increased wage costs. We also expect that establishments are more likely

to react to the minimum wage if a relevant fraction of employees is a�ected

within these establishments.

Most of the existing empirical evidence on the a�ectedness by the new min-

imum wage in Germany is based on employee data. Brenke and Müller

(2013), Kalina and Weinkopf (2013), Heumer et al. (2013), Falck et al.

(2013) and Knabe et al. (2014) predict the individual a�ectedness from pre-

vious waves of the Socio-Economic Panel 2012 to simulate potential e�ects

on employment. Their results reveal a higher a�ectedness in the Eastern

states compared with the West. The a�ectedness of employees in the East

is between 20 per cent according to the study of Knabe et al. (2014) and 32

per cent according to the studies conducted by Kalina and Weinkopf (2013)

and Heumer et al. (2013). By contrast, in West Germany between 13 per

cent according to Knabe et al (2014) and 18 per cent according to Kalina

and Weinkopf (2013) are a�ected.

7

Page 8: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

The IAB Establishement Panel allows a comprehensive assessment of the

establishment-level a�ectedness. We believe that the establishment level is

particularly relevant in this respect since the increased wage costs are faced

by employers. Employers are also the ultimate actor to decide over potential

adjustments in employment. A �rst descriptive overview using the IAB Es-

tablishment Panel is presented in a report by Bellmann et al. (2015). 12 per

cent of all establishments report to employ at least one employee who earned

less than e8.50 per hour. An overlapping 7 per cent of the establishments

reported that they already adjusted wages in anticipation.2 In correspon-

dence with the descriptives in Table 1, they also show a large a�ectedness in

the intensive margin, making reactions to the minimum wage likely.

As we are especially interested in the role of industrial relations, we want

to know whether �rms with works councils and collective bargaining are af-

fected di�erently by the minimum wage. We expect a lower a�ectedness

if a �rm is covered by collective bargaining since such �rms are in general

characterized by higher wages (e.g. Addison et al. 2014). The role of works

councils may be ambivalent. On the one hand, works councils are primarily

representatives of the core work force and less for the marginal work force.

As the latter are a�ected by minimum wage (Falck et al. 2013), we can

expect that works councils have not been a major player in favour of the

minimum wage, especially if employers try to re-allocate the additional costs

by the minimum wage to relatively higher paid employees. On the other

hand, works councils are known to have general positive impacts on pro-

ductivity and wages (Addison et al. 2010, Ellguth et al. 2014, Hübler and

Jirjahn 2003, Müller 2012). In this respect the a�ectedness by the minimum

wage could be lower in establishments with a works council.

Table 2 presents partial e�ects of the establishment-level a�ectedness on in-

dustrial relations from OLS and probit estimations. Within the multivariate

regression models we account for the correlation between the establishments'

a�ectedness by the minimum wage and the sector, regional a�liation as well

as other covariates. This controls for potential selectivity of works councils

2The discrepancy with our descriptive �gure that 11 per cent adjusted wages in an-

ticipation is caused by the exclusion of the public sector.

8

Page 9: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

with respect to sectoral and regional a�liation as well as establishment size.

The �rst two columns show negative and signi�cant partial e�ects of bar-

gaining, orientation at collective bargaining, and the existence of a works

council. The columns three and four add anticipating establishments3 to the

group of a�ected �rms. Anticipating establishments either have originally

been a�ected or are indirectly a�ected and therefore should be added to the

group of a�ected establishments.4 The results in columns three and four

remain unchanged. The industrial relations still show signi�cant negative

conditional correlations with the a�ectedness by the minimum wage.

Our interpretation is that unions are known to push for higher wages in the

bargaining process and also works councils are known to have an impact on

productivity and wages. These are potential reasons why the minimum wage

a�ectedness of such establishments is relatively lower.

In Table 3, the dependent variable di�erentiates a�ected establishments by

their proportion of a�ected employees. The OLS and fractional probit esti-

mations reveal highly signi�cant and negative partial e�ects. Again, these

partial e�ects are estimated controlling for a large set of potentially con-

founding variables. Thus, our hypothesis of a negative e�ect of unions rep-

resentation through both collective bargaining and works councils is corrob-

orated.

4 Anticipatory wage adjustments

11 per cent of the establishments in our sample report to have adjusted wages

ahead of the minimum wage introduction - see Table 1. But how does this

behavior relate to the existence of works councils and collective bargaining

coverage?

Although works councils have no direct in�uence on wages, previous research

mostly con�rmed a positive e�ect (Addison et al. 2010, Hübler and Jirjahn

2003). This could explain why works councils also press for wage adjust-

3Based on the question whether the establishments have adjusted wages in anticipation

of the minimum wage.4For the remainder of the analyses this is how we de�ne the treatment group. Robust-

ness check towards this de�nition are presented in Section 6.1.

9

Page 10: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

ments in anticipation of the minimum wage. They may also monitor the

minimum wage introduction at their respective workplaces and should be

interested in a quick compliance after the minimum wage introduction was

announced.

We would also expect a positive e�ect of collective bargaining on anticipa-

tory wage adjustments. This is especially the case if an establishment or an

industry has entered a new wage agreement in 2014, because unions would

insist that such new wage agreements do not fall below e8.50. However, col-

lectively bargained wages could also make use of the respective clause in the

minimum wage law that adjustments can be postponed until 31 December

2016 and therefore a reversed e�ect is possible.

Finally, there might be a reputation aspect to the discussion on anticipatory

wage adjustments. Establishments for which collective bargaining agree-

ments are binding or which orient themselves to those agreements and/or

establishments with a works council might regard these institutions to signal

a fair employer. This reputation as a fair employer could be endangered if

they wait until the last possible point in time before adjusting wages. It

is a much better signal to accomplish to the new legislation �voluntarily�.

However, the converse situation is also possible: Establishments without

works councils and without binding tari� agreements might have the need

to demonstrate that they act as fair employers. By increasing wages in an-

ticipation they could show a commitment towards higher wages and that

works councils and collective bargaining are not required to set fair wages.

The OLS and probit estimates of the e�ects of the industrial relation vari-

ables on anticipatory behavior are presented in Table 4. The e�ects of the

existence of works councils are negative but small and not signi�cant. Thus,

the hypothesis that union representation through works councils leads to

anticipatory wage adjustment is not approved. However, it remains open

whether this can be driven by establishments without works councils, which

may have an incentive to adjust wages early themselves. This could be

the case if such establishments without works council are confronted with a

higher adjustment volume, of which the costs can be reduced by spreading

the adjustment over time. Moreover, these establishments could be inter-

10

Page 11: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

ested to demonstrate that they can be regarded as fair employers - even

without a works council.

The results di�er for collective bargaining coverage. The e�ects of collective

bargaining and orientation at collective bargaining are highly signi�cant and

positive. This contradicts the possibility to delay wage adjustments for col-

lectively bargained wages until 31 December 2016.

Since we �nd a low a�ectedness and anticipatory wage adjustments at �rms

with collective bargaining, this is a clear indication that collective bargaining

has an impact on the lower part of the wage distribution. This corresponds to

fact that collective bargaining is observed across di�erent sectors, including

high-wage as well as low-wage industries (cf. Bispinck 2015).

5 Further anticipatory adjustments

While investigations on minimum wages are usually focussed on the wage

distribution and employment e�ects for a period after the installation of

minimum wages (e.g. Brown 1999, Card/Krueger 1995, Neumark/Wascher

2007), other adjustments and moreover adjustments ahead of the minimum

wage introduction are possible. Such reactions in anticipation of the intro-

duction, which are usually not detected are in the focus of the remaining

analyses.

Although the German Chancellor Angela Merkel was critical towards the

expectation that the minimum wage could dampen the economic activity

before the law came into force, our investigations show that anticipatory ad-

justments are a real phenomenon. A�ected �rms have not only conducted

wage adjustments as described in the last section. Further reactions are

plausible as employers seek for possibilities to balance the additional costs

of minimum wages and aspire strategies to escape the e�ects of a minimum

wage. The e�ects of minimum wages on the internal operation of �rms has

been left as a mostly unexamined black box even though the costs of mini-

mum wages are absorbed through a wide range of channels of adjustments

(Hirsch et al. 2015). In the public discussion in Germany, the reduction

of voluntary social bene�ts, higher work intensity by higher planned perfor-

11

Page 12: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

mance or other measures like further training, the substitution of time wage

by piece wage, more unpaid overtime hours and transformation of dependent

employment into pseudo self-employed activities were stressed as potential

measures. Pro�t sharing instead of exclusive wage payments, wage cuts of

higher paid employees and adoption of company-level pacts in combination

with collective opening clauses are further imaginable adjustments. In the

latter case, �rms have the chance to cut the wages below collective agree-

ments. The legislation is not conclusive in this aspect (Mindestlohngesetz

2014).

While many anticipatory reactions ahead of the minimum wage were possi-

ble, we concentrate on three questions, namely, whether establishments have

adjusted their collective bargaining coverage, working time arrangement and

further training. Before we present the empirical results, some hypotheses

to the expected e�ects are shortly discussed.

5.1 Hypotheses about anticipatory adjustments

We expect that the e�ects of the minimum wage announcement on reactions

to these three items di�er both between �rms with and without a works coun-

cil, and with and without collective bargaining agreement. Furthermore, the

in�uence in �rms that are characterized by both works council and collective

bargaining agreement should be di�erent. We follow this idea from another

context. In the discussion about e�ects on wages and labor productivity, the

literature detected di�erences with respect to works councils and collective

bargaining (Hübler/Jirjahn 2003, Gerlach/Meyer 2010).

From a theoretical perspective, an increasing and decreasing tari� commit-

ment - our �rst hypothesis - is possible when an establishment is a�ected

by the minimum wage. The minimum wage legislation allows an exemp-

tion from the e8.50 rule, namely existing bargaining agreements with wages

below the threshold do not need to be changed. This may have been an

incentive for some �rms to adopt collective bargaining ahead to the mini-

mum wage introduction. As this bene�t for �rms only holds for two years

the opposite e�ect may be dominant. Firms that pay low wages may have a

12

Page 13: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

preference to leave collective bargaining as the minimum wage makes wage

bargaining for such wage groups redundant. Firms with a works council and

collective bargaining could be less interested to leave the bargaining cover-

age. They pay lower wages than �rms with a works council but not covered

by collective bargaining (Hübler/Jirjahn 2003, p.478). There could also be a

reverse e�ect whereby collective bargaining strengthens the e�ectiveness of

performance-enhancing work practices from negotiations between the man-

agement and the works council (Jirjahn 2014).

Our second hypothesis of possible anticipatory adjustments addresses work-

ing time arrangements and has two dimensions. First, the working hours

per week can be extended or reduced. E. g. Stewart and Swa�eld (2008)

�nd that the introduction of the minimum wage reduced the basic hours of

low-wage workers by about one to two hours per week in the UK. Second, the

use of working time accounts can be changed in reaction to the announced

minimum wage. Here, we have to take into consideration that existing work-

ing time accounts can delay the date up to one year that the minimum wage

is e�ective (Mindestlohngesetz 2014, �2). This delay is in favor of the main-

tenance or introduction of working time accounts before the minimum wage

is enforced. We expect an increased use of overtime work if these additional

hours are unpaid or in combination with working time accounts when no

overtime premium has to be paid. Of course, works councils try to avoid

these disadvantages for the employees. Paid overtime is better practicable

under the control of works councils. Otherwise employees are reluctant to

work more than the bargained working time. They fear that the employer is

unfair at the implementation and arrangement of overtime.

Our third anticipatory adjustment hypothesis looks at whether �rms attempt

to circumvent the cost increases of the minimum wage by a higher produc-

tivity. This could justify the minimum wage economically when wages corre-

spond to the marginal productivity. This higher productivity can be achieved

by a higher planned performance or by further training if this causes no ad-

ditional wage increases. No information on planned performance is available

in our data set. Therefore, we restrict the analysis to further training and

analyze whether a�ected establishments intensi�ed further training activities

13

Page 14: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

ahead of the minimum wage introduction on 1 January 2015. The investi-

gation is not restricted to a�ected employees because we cannot split our

sample in this respect. However, it makes sense also to consider una�ected

employees as the wage costs of a�ected workers can also be compensated by

a higher productivity of other workers. From the literature it is known that

further training is more prevalent at �rms with a works council (Backes-

Gellner et al. 1997, Stegmeier 2012, Kriechel et al. 2014). Hence, one can

expect that this is also a crucial factor for the �rms' behavior in response to

the minimum wage also because works councils are known to raise produc-

tivity (Hübler 2015).

5.2 Empirical approach

For the empirical investigations of anticipatory adjustments with respect to

the minimum wage introduction we make use of the panel structure of two

time periods (t=2014; t-1=2013) and estimate for establishments i=1,...,N

∆yi = Aiδ + ∆x′iβ + ∆εi, (1)

where

Ai =

{1 if i would be a�ected by minimum wage in 2014

0 otherwise.

In equation (1), yi is the outcome variable of interest, A de�nes the broad

de�nition of a�ected establishments, including those who reported to have

adjusted wages in anticipation. δ is the e�ect of interest describing whether

a�ected establishments changed outcome y in anticipation of the minimum

wage, x includes a wide range of control variables. They are measured by

∆xi = xit − xi,t−1. Analogously, the error term is determined by ∆εi =

εit − εi,t−1, so that the time invariant establishment component is elimi-

nated. This is not a before and after regression model.5

The speci�cation in �rst di�erences comes at several advantages. First, it

5In a standard before-and-after regression model ∆Ai would be used instead of Ai,

because in 2013 most of the �rms also had employees with wages below e8.50.

14

Page 15: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

corresponds with our intuition to analyze changes in y. Second, it controls

for a �xed establishment-speci�c e�ect based on a balanced panel, which

both ensures that changes in y are in fact caused by changes in economic

behavior but not in the sample composition. Third, it requires only a short

panel of two periods which deals with attrition in the survey. Nevertheless,

we conduct placebo regressions to assess whether the results re�ect a general

underlying trend in the outcome variable. Fourth, it allows for simple statis-

tical inference because only one error component per establishment remains

in the estimation.

5.3 Results

5.3.1 Commitment to the collective bargaining agreement

Our estimates in Table 5 show that establishments a�ected by the new Ger-

man minimum wage tend to leave the bargaining coverage (BC=1, if cover-

age; BC=0 otherwise). The endogenous variable is the change of a dummy

(∆BC), which is a classi�ed variable on three levels

∆BCi =

−1 if i leaves BC = 1 from 2013 to 2014

0 if i does not change the BC status

+1 if i enters the status BC = 1 in 2014.

(2)

The results in Table 5 reveal a weakly signi�cant e�ect. This is the case in

the simple two variable model but also when control variables are incorpo-

rated. This supports the hypothesis of the substitution of coverage by the

minimum wage. We cannot rule out that some �rms react in an opposite

manner, i.e., that they join collective bargaining. Another explanation could

be that the reaction will be intensi�ed when the minimum wage is already

adopted. This is because the opposite force, that bargaining agreements are

allowed to undercut minimum wages until 31 December 2016, becomes less

e�ective and the incentive to abuse this exemptions becomes less attractive.

One could argue that the negative coe�cient expresses a general tendency

of shrinking coverage. In order to test this we have made an estimate with

an alternative period, namely, 2012-2013 instead of 2013-2014. When the

15

Page 16: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

decreasing trend of collective bargaining is crucial, the in�uence of the min-

imum wage a�ectedness should also be e�ective in the placebo regression.

In other words, the apparent anticipation e�ect of a�ected plants would in

fact be a time e�ect. However, we �nd a positive and insigni�cant placebo

e�ect. Further, we should notice that in Table 5 no speci�c works council

e�ect could be detected as columns (4) and (5) demonstrate.

5.3.2 Working time arrangements

The results in Tables 6 and 7 reveal e�ects with respect to anticipatory

working time adjustments ahead of the minimum wage introduction. The

estimates of both considered aspects show in the same direction. A�ected

�rms have a higher probability of overtime work compared with una�ected

�rms and adopt working time accounts in anticipation of the minimum wage.

The latter e�ect accords with the hypothesis in section 5.1 given the speci�c

rules of minimum wages under working time account arrangements. The ef-

fects are signi�cant in the baseline models with and without control variables,

and again, the a�ectedness coe�cient of the placebo regression is insigni�-

cant and has an opposite sign. This might be a hint that �rms attempt to

compensate the prospective higher wage costs by lower overtime premiums.

We should note that �rms without a works council tend to overtime work

and to introduce working time accounts but not those with a works council.

This seems plausible. A works council can hinder non-payment of overtime

premium by working time accounts. This reduces the incentive to introduce

working time accounts as a cost reducing response to the minimum wage.

5.3.3 Further training

In Table 8, the in�uence of a�ectedness (A) on further training (FT ) is esti-

mated, or more precisely, on the change in the provision of further training

(∆FT ). This variable is again speci�ed as a classi�ed variable with three

distinct values as in equation 5.3.1. The results show that a�ectedness (A)

leads to more further training. The intention is obvious. Via training the

labor productivity shall be increased until the value of the marginal product

16

Page 17: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

of labor equals the wage. Then the higher wage costs by minimum wage

are optimal given pro�t maximization in a fully competitive market. How-

ever, a low demand for unskilled labor due to the minimum wage may also

create an incentive for the a�ected workers to accumulate human capital

(Cahuc/Michel 1996).

The e�ect in the placebo regression is again insigni�cant. The most interest-

ing and surprising results in Table 8 are in columns (4) and (5). Firms with a

works council do not react with further training while �rms without a works

council extend their training activities when a�ected by the minimum wage.

We would expect the opposite that works councils support training activities

as this is also con�rmed by the empirical investigations. Related to mini-

mum wages works councils might be reluctant if training and in consequence

higher productivity is not combined with an additional wage increase. If

in this respect the training is a managerial intention, �rms without works

councils have better chances to assert such changes.

6 Robustness checks

In what follows we consider two dimensions of robustness checks. First,

the a�ected group is modi�ed. Second, an alternative estimation method is

applied. We estimate di�erence-in-di�erences probit models instead of the

linear speci�cation and evaluate whether signs and size of the coe�cients

alter.

6.1 Alternative de�nitions of the a�ected group

So far, the a�ected group consists of establishments that still have employ-

ees with a wage below e8.50 per hour of work and establishments that have

adjusted wages in response to the announced minimum wage legislation. A

priori, it is not clear, whether both groups behave in the same way. It is

possible that establishments, which have adjusted wages, do not react with

adjustments in the other hypothesized measures. Unions, works councils

and employees may have ensured that wage adjustments were anticipated

17

Page 18: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

and that other detrimental adjustments to the workers are hindered.

We carry out the same estimations as in columns (2) of Tables 5-8 with two

di�erent de�nitions of a�ected �rms. The �rst group only consists of �rms

which paid wages of less than e8.50 in 2014, without adding anticipating

plants. The second group excludes all establishments that adjusted wages

in anticipation. The results are presented in Table 9. We �nd robust results

with respect to the sign of the coe�cients for both alternative de�nitions

compared with columns (2) in Tables 5-8. However, the signi�cance varies.

Especially when using the second alternative de�nition, the estimates be-

come less signi�cant except for the adjustments concerning working time

accounts. The relatively smaller sample size might be one reason. However,

the hypothesis that �rms with anticipating wage adjustments do not show

further anticipatory adjustments is not con�rmed.

6.2 Non-linear di�erence-in-di�erences speci�cations

The di�erence-in-di�erences method (DiD) is a popular approach in empir-

ical economics for estimating causal e�ects that can easily be determined

(Lechner 2011). E.g. Card and Krueger (1994) have applied this in the

context of minimum wages. As in equation (1) time invariant in�uences are

eliminated. For panel data DiD and �xed e�ects estimators are very similar

and under speci�c assumptions even identical. Due to the non-linear nature

of the dependent variables it is not e�cient to use a linear approach as we

have done in Tables 5-9. The endogenous variables are dummy variables.

The corresponding nonlinear di�erence-in-di�erences model of a two-period

speci�cation is

yit = G(x′it · β + Ai · β1 + Y 2014 · β2 + Ai · Y 2014 · δ), (3)

where i = 1, ..., N , t = 2013, 2014 and

Y 2014 =

{1 if observation stems from the year 2014

0 otherwise.

18

Page 19: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

This means, the dependent variable yit is a nonlinear function G(·) of the

index of independent variables. The marginal e�ects of the interaction term

are computed following Norton, Wang and Ai (2004) by

∆2G(·)∆Ai · ∆Y 2014

= Φ(x′β+β1+β2+δ)−Φ(x′β+β1)−Φ(x′β+β2)+Φ(x′β). (4)

For the probit case, Φ(·) is the standard normal cumulative distribution.

Clustered standard errors are determined by a block clustered bootstrap pro-

cedure. The estimates are presented in Table 10, where only the coe�cient

and the standard error of the interaction term are displayed. In comparison

with Tables 5-8 the results are robust.

7 Conclusion

The introduction of a general minimum wage in Germany since 1 January

2015 is a large-scale experiment with uncertain outcomes. It was announced

by the government in October 2013. Among other things, the law was mo-

tivated by the declining industrial relation coverage in Germany. Minimum

wages can be regarded as a wage increasing instrument like collective bar-

gaining and wage councils. But it is not clear whether minimum wages cause

di�erent or additional intra-�rm adjustments. Despite the necessity for ex-

post evaluations of employment and wage e�ects of the minimum wage, an-

ticipatory adjustments ahead of minimum wages are under-investigated and

addressed in our analysis.

We apply the IAB Establishment Panel, which includes information about

the establishment-level a�ectedness and anticipatory wage adjustments. The

a�ectedness ahead of the minimum wage introduction in Germany varies.

While only 12 per cent of the establishments are a�ected, within these es-

tablishments on average 39 per cent of the employees had a wage of less than

e8.50 per hour of work in 2014. Firms with a works council and collective

bargaining are less a�ected than others. Furthermore, we �nd that compa-

nies with strong industrial relations tend to early wage adjustment prior to

the o�cial start of the minimum wage. Furthermore, �rms a�ected by the

19

Page 20: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

minimum wage extend or introduce overtime, adopt working time accounts

and expand further training. In contrast to the anticipatory wage adjust-

ments, establishments with weak industrial relations are mostly responsible

for such adjustments. Together these results corroborate the hypothesis that

a�ected �rms try to compensate the higher wage costs by counteractive mea-

sures.

A priori it was not clear whether industrial relations are strengthened by the

introduction of the minimum wage. Our results rather point at the opposite.

A�ected establishments rather leave than join collective bargaining and the

institutions indicating strong industrial relations do not have a mitigating

impact on the observed anticipatory adjustments with respect to working

time and training.

Further research should be devoted to the reasons for anticipatory behavior:

Our analysis cannot verify whether anticipatory adjustments are due to con-

cerns about the establishment's reputation and also not about the intensity

of the adjustments. While our study shows that anticipatory behavior is as-

sociated to the minimum wage, future research should uncover whether such

changes persist after the introduction of the minimum wage. Furthermore,

it is unclear whether adjustments concerning collective bargaining, working

time arrangements, and further training a�ect potential consequences of the

minimum wages such as e�ects on employment.

20

Page 21: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

References

Addison, J. T., Teixeira, P. and T. Zwick (2010), German works councils and the

anatomy of wages, Industrial and Labor Relations Review 63:2, 247-270.

Addison, J. T., Bryson, A., Teixeira, P. and A. Pahnke (2011), Slip Sliding Away:

Further Union Decline in Germany and Britain, Scottish Journal of Po-

litical Economy 58:4, 490-518.

Addison, J. T., Teixeira, P., Evers, K. and L. Bellmann (2014), Indicative and

updated estimates of the collective bargaining premium in Germany,

Industrial Relations 53:1, 125-156.

Backes-Gellner, U. Frick, B. and D. Sadowski (1997), Codetermination and Per-

sonnel Policies of German Firms: the In�uence of Works Councils on

Turnover and Further Training, The International Journal of Human Re-

source Management 8:3, 328-347.

Bellmann, L., Bossler, M., Gerner, H.-D. and O. Hübler (2015), Reichweite des

Mindestlohns in deutschen Betrieben, IAB-Kurzbericht 6/2015.

Bispinck, R. (2015), Tarifpolitischer Jahresbericht 2014: Zwischen Mindestlohn

und Tarifeinheit, http://www.boeckler.de/pdf/ p_ta_jb_2014.pdf, last

retrived on 17 February 2015.

Brenke, K. and K. U. Müller (2013), Gesetzlicher Mindestlohn: Kein verteilungs-

politisches Allheilmittel. DIW Wochenbericht, 80:39, 3-17.

Brown, C. (1999), Minimum Wages, Employment, and the Distribution of In-

come, in: O. Ashenfelter and D. Card (eds.), Handbook of Labor Eco-

nomics, Vol. 3, 2101-2163, Amsterdam: Elsevier.

Cahuc, P. and P. Michel (1996), Minimum wage unemployment and growth, Eu-

ropean Economic Review 40, 1463-1482.

Card, D. and A. B. Krueger (1994), Minimum Wages and Employment: A Case-

study of the Fast-Food Industry in New Jersey and Pennsylvania. Ameri-

can Economic Review 84, 772-793.

21

Page 22: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Card, D. and A. B. Krueger (1995), Myth and Measurement: the New Economics

of the Minimum Wage, Princeton: University Press.

Ellguth, P., Gerner, H.-D. and J. Stegmaier (2014), Wage e�ects of works coun-

cils and opening clauses * the German case Wage e�ects of works councils

and opening clauses * the German case, Economic and Industrial Democ-

racy, 35:1, 95-113.

Ellguth, P., Kohaut, S. and I. Möller (2014), The IAB Establishment Panel -

methodological essentials and data quality, Journal for Labour Market

Research, 47:1-2, 27-41.

Falck, O., Knabe, A., Mazat, A. and S. Wiederhold (2013), Mindestlohn in

Deutschland: Wie viele sind betro�en? ifo Schnelldienst 66:24, 68-73.

Fischer, G., Janik, F., Müller, D. and A. Schmucker (2009), The IAB Establish-

ment Panel * things users should know. Journal of Applied Social Science

Studies, 129:1, 133-148.

Gerlach, K. and W. Meyer (2010), Wage E�ects of Works Councils and Collec-

tive Agreements in Germany: The Anatomy of Wages and Productivity,

in: C. Brown, B. Eichengreen and M. Reich (eds.), Labor in the Era of

Globalization, Cambridge: University Press, 295-316.

Heumer, M., Lesch, H. and C. Schröder (2013), Mindestlohn, Einkommensvertei-

lung und Armutsrisiko. IW-Trends 1.

Hirsch, B. T., Kaufman, B. E. and T. Zelenska (2015), MinimumWage Channels

of Adjustment, Industrial Relations 54:2, 199-239.

Hübler, O. (2015), Do Works Councils Raise or Lower Firm Productivity? IZA

World of Labor 137.

Hübler, O. and U. Jirjahn (2003), Works Councils and Collective Bargaining in

Germany: The Impact on Productivity and Wages, Scottish Journal of

Political Economy 50, 1-21.

22

Page 23: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Jirjahn, U. (2014), Works Councils and Collective Bargaining in Germany: A Sim-

ple but Crucial Theoretical Extension, Research Papers in Economics No.

13/2014, Universität Trier.

Kalina, T. and C. Weinkopf (2013), Niedriglohnbeschäftigung 2011. IAQ Report,

2013-01.

Knabe, A., Schöb, R. and M. Thum (2014), Der �ächendeckende Mindestlohn.

Perspektiven der Wirtschaftspolitik 15, 133-157.

Kriechel, B., Muehlemann, S., Pfeifer, H. and M. Schütte (2014), Works Coun-

cils, Collective Bargaining, and Apprenticeship Training - Evidence From

German Firms, Industrial Relations 53, 199-222.

Lechner, M. (2011), The Estimation of Causal E�ects by Di�erence-in-Di�erence

Methods, Foundations and Trends in Econometrics 4:3, 165-224.

Mindestlohngesetz - MiLoG (2014), Gesetz zur Regelung eines allgemeinen Min-

destlohns (Mindestlohngesetz vom 11. August 2014 (BGBL. I S. 1348)).

Müller, S. (2012), Works Councils and Establishment Productivity, Industrial and

Labor Relations Review 65:4, 880-898.

Neumark, D. and W. Wascher (2007), Minimum Wages and Employment, Foun-

dations and Trends in Microeconomics 3, 1-182.

Norton, E. C., Wang, H. and C. Ai (2004), Computing interaction e�ects and

standard errors in logit and probit models, The Stata Journal 4:2, 154-167.

Sachverständigenrat (2014), Mehr Vertrauen in Marktprozesse, Jahresgutachten

14/15, p. 8, http://www.sachverstaendigenrat-wirtschaft.de/�leadmin/

dateiablage/gutachten/jg201415/JG14_ges.pdf, last retrived on 17 June

2015.

Stegmeier, J. (2012), E�ects of Works Councils on Firm-Provided Further Train-

ing in Germany, British Journal of Industrial Relations 50:4, 667-689.

23

Page 24: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Stewart, M. B. and J. K. Swa�eld (2008), The Other Margin: Do Minimum

Wages Cause Working Hours Adjustments for Low-wage Workers? Eco-

nomica 75, 148-167.

24

Page 25: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Table Appendix

Table 1: Descriptive summary of the analysis sample

(1) (2) (3)All A�ected (A=1) Establishments

establishments establishments with anticipatorywage adjustments

Absolute numbers:Establishments 9,321 1,582 1,019

Means of covariates:Works councils 0.387 0.219 0.476Collective bargaining 0.321 0.348 0.362Orientation at 0.233 0.116 0.275collective bargainingOvertime hours 0.505 0.470 0.578Working time accounts 0.674 0.650 0.790Further training 0.684 0.599 0.813

Observations 9,321Sample year 2014Source: IAB Establishment Panel, analysis sample.

25

Page 26: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Table 2: Partial e�ects of industrial relations on the establishment-levela�ectedness

(1) (2) (3) (4)A�ected establishments A�ected establishments and

anticipating establishmentsOLS Probit OLS Probit

Collective bargaining -0.077*** -0.079*** -0.072*** -0.073***(0.010) (0.010) (0.011) (0.011)

Orientation at bargaining -0.045*** -0.036*** -0.027** -0.023**(0.010) (0.008) (0.010) (0.009)

Works council -0.089*** -0.081*** -0.097*** -0.084***(0.012) (0.012) (0.014) (0.013)

Controls yes yes yes yes

Observations 9,321Sample 2014

Note: Partial e�ects and iid-standard errors in parentheses. Asterisks indicate signi�cance levels: *** 1%, **

5%, and * 10%. Control variables include the shares of females, quali�ed, high quali�ed and part-time employees

as of all employees; dummies for product and process innovations; the self-assessed competition (4 categories);

dummies for skilled labor demand, single establishment-organizations, foundation before 1990, the legal form

(4 categories), association membership, industries (17 categories), states (16 categories), and establishment size

(10 categories).

Source: IAB Establishment Panel, analysis sample.

26

Page 27: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Table 3: Partial e�ects of industrial relations on the fraction of a�ectedemployees

(1) (2)Fraction of a�ected employeesOLS Fractional Probit

Collective bargaining -0.022*** -0.022***(0.005) (0.005)

Orientation at bargaining -0.020*** -0.014***(0.005) (0.004)

Works council -0.032*** -0.043***(0.005) (0.006)

Controls yes yes

Observations 9,321Sample 2014Note: Dependent variable is the fraction of employees with a wage below

e8.50 per hour of work. Robust standard errors in parentheses. Asterisks

indicate signi�cance levels: *** 1%, ** 5%, and * 10%. Control variables as

in Table 2.

Source: IAB Establishment Panel, analysis sample.

27

Page 28: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Table 4: Partial e�ects of industrial relations on anticipatory wage setting

(1) (2)OLS Fractional Probit

Collective bargaining 0.111*** 0.110***(0.031) (0.030)

Orientation at bargaining 0.096*** 0.094***(0.026) (0.025)

Works council -0.018 -0.017(0.038) (0.037)

Controls yes yes

Observations 2,098Sample 2014

Note: Dependent variable indicates establishments which set wages in antic-

ipation of the minimum wage. Robust standard errors in parentheses. Aster-

isks indicate signi�cance levels: *** 1%, ** 5%, and * 10%. Control variables

as in Table 2.

Source: IAB Establishment Panel, analysis sample.

28

Page 29: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Table 5: Anticipatory adjustments in tari� commitment

(1) (2) (3) (4) (5)Baseline Baseline Placebo With Withoutwithout with regression works workscontrols controls council council

A�ected -0.010 -0.011* 0.005 -0.009 -0.010(0.006) (0.006) (0.006) (0.019) (0.007)

Controls no yes yes yes yes

Observations 9,321 9,321 7,721 1,985 7,336Sample 2013-2014 2013-2014 2012-2013 2013-2014 2013-2014

Note: Robust standard errors in parentheses. Asterisks indicate signi�cance levels: *** 1%, ** 5%,

and * 10%. Control variables as in Table 2, but in di�erences (∆x).

Source: IAB Establishment Panel, analysis sample.

29

Page 30: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Table 6: Anticipatory adjustment in overtime hours

(1) (2) (3) (4) (5)Baseline Baseline Placebo With Withoutwithout with regression works workscontrols controls council council

A�ected 0.022** 0.023** -0.009 -0.006 0.025*(0.011) (0.011) (0.012) (0.020) (0.013)

Controls no yes yes yes yes

Observations 9,321 9,321 7,721 1,985 7,336Sample 2013-2014 2013-2014 2012-2013 2013-2014 2013-2014

Note: Robust standard errors in parentheses. Asterisks indicate signi�cance levels: *** 1%, ** 5%,

and * 10%. Control variables as in Table 2, but in di�erences (∆x).

Source: IAB Establishment Panel, analysis sample.

30

Page 31: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Table 7: Anticipatory adjustments in the use of working time accounts

(1) (2) (3) (4) (5)Baseline Baseline Placebo With Withoutwithout with regression works workscontrols controls council council

A�ected 0.018* 0.018* -0.004 0.011 0.025**(0.011) (0.011) (0.012) (0.023) (0.012)

Controls no yes yes yes yes

Observations 9,321 9,321 7,721 1,985 7,336Sample 2013-2014 2013-2014 2012-2013 2013-2014 2013-2014

Note: Robust standard errors in parentheses. Asterisks indicate signi�cance levels: *** 1%, ** 5%,

and * 10%. Control variables as in Table 2, but in di�erences (∆x).

Source: IAB Establishment Panel, analysis sample.

31

Page 32: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Table 8: Anticipatory use of further training

(1) (2) (3) (4) (5)Baseline Baseline Placebo With Withoutwithout with regression works workscontrols controls council council

A�ected 0.019* 0.018* -0.008 -0.031 0.027**(0.011) (0.011) (0.011) (0.020) (0.012)

Controls no yes yes yes yes

Observations 9,321 9,321 7,721 1,985 7,336Sample 2013-2014 2013-2014 2012-2013 2013-2014 2013-2014

Note: Robust standard errors in parentheses. Asterisks indicate signi�cance levels: *** 1%, ** 5%,

and * 10%. Control variables as in Table 2, but in di�erences (∆x).

Source: IAB Establishment Panel, analysis sample.

32

Page 33: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Table 9: Anticipation e�ects using alternative de�nitions pf treated plants

Panel A: Panel B:Tari� commitment Overtime hours

a - A�ected without adding -0.008 0.020*anticipating now una�ected (0.007) (0.012)b - A�ected excluding all -0.010 0.005anticipating establishments (0.008) (0.014)

Panel C: Panel D:Working time accounts Further Training

a - A�ected without adding 0.024** 0.021*anticipating now una�ected (0.012) (0.012)b - A�ected excluding all 0.030** 0.010anticipating establishments (0.013) (0.014)

Observations 8,787 8,302 8,787 8,302Sample 2013-2014

Note: Robust standard errors in parentheses. Asterisks indicate signi�cance levels: *** 1%, ** 5%, and

* 10%. Control variables include the shares of females, quali�ed, high quali�ed and part-time employees

as of all employees; dummies for product and process innovations; the self-assessed competition (4 cat-

egories); dummies for skilled labor demand, single establishment-organizations, foundation before 1990,

the legal form (4 categories), association membership, industries (17 categories), states (16 categories),

and establishment size (10 categories).

Source: IAB Establishment Panel, analysis sample.

33

Page 34: A ectedness and anticipatory behavior in response to the ...Soure:c IAB Establishment Panel, 1996-2014. olloFwing the political discussion on the decreasing collective bargaining coverage

Table 10: Non-linear interaction e�ects, baseline speci�cation with controls

(1) (2) (3) (4)Tari� Overtime Working time Further

Commitment hours accounts training

A�ected*Y2014 -0.011 0.024** 0.020* 0.018*(0.007) (0.011) (0.011) (0.011)

Observations 18,642 18,642 18,642 18,642Sample 2013-2014 2013-2014 2013-2014 2013-2014

Note: Cluster robust standard errors from a block clustered bootstrap (500 replications) in

parentheses. Asterisks indicate signi�cance levels: *** 1%, ** 5%, and * 10%. Control variables

as in Table 2, but in di�erences (∆x).

Source: IAB Establishment Panel, analysis sample.

34