Project topics and materials
Get Complete Project Material File(s) Now! »
Large firms’ collusion in the labor market : Evidence from collective bargaining
This chapter is a joint work with Bérangère Patault (CREST). In several countries, including France, industry-level agreements are binding for all firms of the industry, whether they sit at the negotiating table or not. This paper provides a theoretical framework showing that such agreements can be used by dominant firms to reduce competition. In this framework, the higher the over-representation of large firms in employers federations, the larger the bargained wage floors, which entails in turn the eviction of small firms. This prediction is tested using French administrative data. We document the domination of large firms within federations and devise an instrumental strategy to causally show that the larger the bargaining firms relatively to the other firms of the industry – ie the lower the federation’s representativeness, the higher their incentives to raise wage floors.
Introduction
Wage setting can occur at different levels, from the most decentralised level – firm level – to the most centralised one – national level. In their seminal paper, Calmfors and Driffill (1988) show that the intermediate level of centralisation – industry level – leads to the worse macroeconomic performance. The simultaneity of the German decentralisation of wage bargaining and resurgence of the German economy in the 1990s seems to corroborate such findings (Dustmann et al., 2014). We argue in this paper that some common features of industry-level wage bargaining can produce the effect of an anti-competitive tool. Indeed, in several countries, among which France, Italy or Portugal, the bargained wages are extended to all firms of the industry, whether they sit at the negotiating table or not, and firms cannot opt out from these agreements. Because of this extension system, the characteristics of bargaining firms are a crucial component of the bargaining outcome. If bargaining firms have different characteristics, and thus different objectives, as the average firm in the industry – ie are unrepresentative of the industry, the bargained wage may favour affiliated firms. In particular, the domination of employers federations by large firms 1 – that we will denote unrepresentativeness in the following – , tilts the bargaining process in their favour, generating a cartel effect. Therefore, dominant firms can use collective bargaining as a tool to raise the labor cost of competitors, and in doing so, reduce the number of producing firms. The following quote, extracted from an Economic survey of the OECD on Portugal (see OECD (2012)), summarizes this mechanism.
« […] dominant firms impose wage and working conditions on others via the administrative extension of collective agreements, reducing competition and entry, thereby hurting competitiveness. ».
OECD, Economic surveys Portugal, 2012.
In the first two parts of the paper, we compare within a Melitz-type model (Melitz (2003)) two different levels of wage bargaining : firm-level and industry-level bargaining. First, we find that the higher the productivity-level of the firm, the higher the rent to be shared, so the higher the wage negotiated at the firm-level. As a consequence, when there is an industry-level wage floor, it is binding only for small firms, and it raises the wages they pay above their optimal level, thus driving them out of the market . The higher the domination of large firms on the employers federations, the higher the wage floors, which is detrimental to small firms. Equivalently, the more employers federations are dominated by large firms, the higher the negotiated wage floor and, as a result, the lower the product market concentration. We depict the main results of our model in Figure 1.1.
FIGURE 1.1 : Results from our theoretical model
Note : All the mechanisms depicted above are results of our model. One result of our model is that large firms have higher incentives to raise wage floors. The higher the unrepresentativeness of employers federations, the higher the incentives of bargaining firms to raise wage floors.
We then empirically confirm the collusion effect highlighted by the model. We first derive novel stylized facts on the relation between the representativeness of employers federations and the degree of competition of an industry. To measure representativeness we construct a novel proxy using unique data from the Minister of Labour. This dataset enables us to compare for the first time the average size, for each industry agreement, of the bargaining firms as compared to the average size of all firms of the industry – ie bargaining and non-bargaining firms. The index built therefore proxies the domination of employers federations by large firms, ie the federations’ unrepresentativeness. We find a positive correlation between unrepresentativeness and product market concentration, as well as between unrepresentativeness and small firm’s destruction rate.
In our model, the mechanism explaining the positive correlation between federations unrepresentativeness and product market concentration is that bargaining firms have higher incentives to raise wage floors the larger they are compared to the average firm of the industry – ie the more unrepresentative the employers federation. Our model indeed establishes that large firms always have higher incentives than small firms to raise the wage floors because it enables them to evict the small firms from the market. However, for that to translate into higher wage floors, bargaining firms must be the large firms. Therefore, the over-representation of large firms in employers federations – that we call unrepresentativeness of federations – is a crucial component to understand the outcomes of the bargaining system. In other words, bargaining firms have differential incentives to raise wage floors whether they are representative or not of the average firm in the industry.
However, this mechanism cannot be directly tested because bargaining firms incentives to raise the wage floor are unobservable by nature. We solve this problem by using a variable shifting the large firms incentives to raise wage floors : the share of workers employed by small firms. The higher the share of workers employed by small firms, the higher the incentives for large firms to increase bargained wage floors. Indeed, the higher this share, the higher the competition from small firms, and thus the higher the large firms incentives to evict small firms from the market (a similar argument is used in Magruder (2012)). If bargaining firms are the largest firms – ie in unrepresentative industries, then the share of workers employed in small firms should have a positive effect on the bargained wage floors. On the opposite, in representative industries, the share of workers employed by small firms should not have any effect on the bargained wage floors. We use our index of unrepresentativeness of the employers federations and estimate in the same regression the effect of the share of workers employed by small firms for both representative and unrepresentative industries on wage floors evolution.
Two variables are likely to be endogenous in our setting : the share of workers employed by small firms for representative industries and the share of workers employed by small firms for unrepresentative industries 2. To achieve causality, we thus use two instrumental variables : for each industry, we construct the share of workers employed by small firms in both Denmark and Germany. Our instrumental strategy is based on the assumption that there is no unobserved variable affecting both those foreign shares and French domestic labor costs. Naturally Danish and German sectoral shares are correlated with the industry unobserved comparative advantages through international trade. To mitigate this issue, we exploit the existence of several wage floors per industry agreement and add industry year fixed effects, thereby controlling for such comparative advantages. Moreover, we ensure that there is no common shock affecting both wages negotiated in France and the share in Denmark and Germany by using lagged values of our instruments. Another potential threat to identification could be the existence of a technological shock hitting small firms that would be common to France and the country of interest (either Denmark or Germany). Such technological shock would induce a change in the French labor demand by small firms, thus probably changing the wage floors, and would be correlated to both the Danish (for instance) share of workers employed in small firms. We alleviate this concern by controlling for the evolution of the share of workers employed in small firms, which enables to capture labor demand shocks that would affect proportionally more small firms.
Consistently with our model predictions, we find that the share of employees working in small firms has a positive and significant effect on wage floor variations only for unrepresentative industries. This confirms that bargaining firms have differential incentives to raise wage floors whether they are representative or not of the average firm in the industry. The representativeness of employers federations therefore plays a key role in determining collective bargaining outcomes.
This paper speaks to several strands of the literature. First, it relates to the theoretical literature studying the effect of industry-level agreements on competition (see the seminal work of Calmfors and Driffill (1988)). The main effects highlighted by the literature are twofold. The closest paper to ours is Haucap et al. (2001), which established that incumbent firms have an interest to raise the industry wage floor in order to impede the entry of firms. Our contribution is to add firms heterogeneity, which enables to distinguish large from small firms interests in the bargaining process, and thus to assess the effect of the domination of employers federations by large firms. Jimeno and Thomas (2013) show that the industry wage floors are a source of wage rigidity as it impedes less productive firms to settle a lower wage. We generalize the approach used by these papers by integrating industry-level wage bargaining in a standard firm dynamics model (Melitz (2003)) that allows to account for competition both between heterogeneous monopolistic firms and between industries. Our framework allows for firms entry, which is crucial to analyze the barriers to entry induced by sector-level agreements. Therefore, our model is able to disentangle the different effects highlighted in the previous literature, and show that they do not offset each other. The cartel effect that we uncover thanks to our model has been confirmed before by the empirical literature. Indeed, Martins (2014a) and Magruder (2012) demonstrate that the implementation of industry-level wage floors has a strong negative effect on small firm survival.
The main novelty of the paper is to emphasize the role of representativeness on collective wage bargaining. To the best of our knowledge, we present the first model focusing on the impact of the formation of the employers federations’ objective, in presence of firms heterogeneity.
On the empirical side, Martins and Hijzen (2016) and Hijzen et al. (2017) are the only papers underlining the importance of federations lack of representativeness. More precisely, Martins and Hijzen (2016) explain that ’the lack of representativeness of employer associations is a potentially important factor behind the adverse effect of extensions’. However, their degree of representativeness, computed as the share of the workforce in affiliated firms to the total employment of the sector, does not increase significantly the effect of extensions. We argue instead that the primary factor is the difference between the interests of decision-makers among employers federations and those of covered firms. More precisely we look at small firms representation, rather than overall representativeness.
This paper relates naturally to the large literature on the effects of industry-level wage bargaining on unemployment, employment losses and wage rigidities (Díez-Catalán and Villanueva (2015), Dustmann et al. (2014), Guriev et al. (2016), Hartog et al. (2002), Martins (2014a), Murtin et al. (2014), Villanueva (2015)).
This paper also contributes to a broader debate on the wage inequality effect of unions. Numerous papers analyze the presence of a wage surplus associated with union membership (DiNardo and Lee (2004), Hirsch (2004) or Lewis (1986)). While this may increase the dispersion of wages throughout the economy, collective agreements mechanically raise the wage compression in covered firms. Although these countervailing forces result in an ambiguous theoretical effect, empirical studies have tended to set forth a negative effect of unions on wage inequality (see Frandsen (2012) or Farber et al. (2018)).
Finally, this paper relates to the recent literature exploring the rise of both the labor and product market monopsony. These trends, and the major concerns about its harmful effects, received a large attention from economists (see Van Reenen (2018) for a literature review) and by economic institutions (see CEA (2016)). Indeed, the rise of the product market concentration undermines productive efficiency (see Van Reenen (2011)), raises prices (see De Loecker and Eeckhout (2017)) , reduces real wages (see Benmelech et al. (2018)) and increases inequalities (see Hershbein and Macaluso (2018)). Several causes has been presented in order to explain this movement of concentration (see Grullon et al. (2017) who argue that it is generated by a decrease of the anti-trust legislation). Autor et al. (2000) argue that it is generated by superstar firms where more markets become ’winners take all’. In this article we argue that regulation increases the monopsony as it introduce barriers to entry, and allows larger firms to strengthen their dominant position.
The structure of the paper is as follows. Section 1.2 provides some information about the French institutional setting. Section 1.3 lays out the model and compares industry-level wage bargaining with firm-level bargaining. Section 1.4 characterizes the theoretical impact of representativeness of employer federations. Section 1.5 explores the empirical validity of the model’s predictions. Section 1.6 concludes.
Institutional setting
Collective wage bargaining in France
In almost every OECD country, three distinct levels of minimum wage may coexist : national minimum wage, industry-level minimum wage and firm-level minimum wage. Yet, the predominant level of bargaining starkly differs across countries 3, as exhaustively explained in the OECD Employment Outlook 2017. In France, the predominant level of wage bargaining is sectoral bargaining : each year around 70% of French total workforce is covered by an industry-level agreement 4
(see Dares (2015b)).
Employers associations and unions negotiate over several topics, among which wages, working time, training, health, severance pay and bonuses. In order to negotiate over wages, the parties must first agree on qualifications levels, and then on a wage floor for each of them. A single industry-level agreement thus often includes several wage floors levels. Figure 1.1 provides an example of job qualifications and corresponding wage floors for Hairdressing in 2013. In this paper, we ignore the other components of collective bargaining and restrict our attention to wage floor levels as, except for wage floors and qualifications, the negotiation at the firm-level enables to opt out from industry-level agreements 5.
The perimeter of an industry is decided by unions and employers associations and is validated by administrative controls 6. Extensions of collective agreements are quasi-automatic in France : once the agreement is signed, except very rare cases that are not economically significant, the Minister of Labor extends them to the entire industry, usually within two or three months 7. The Minister may, in principle, exclude from the extension certain clauses of the agreement for legal reasons or reasons of general interest (Labor Code L.2261-25). However, refusals to extend the entire agreement are rare and above all founded on the legal validity of the text – never on economic or social arguments. The possibility for the Minister of Labor to refuse an extension on a ground of general interest, in particular the objectives of economic and social policy or the protection of the situation of third parties, exists but it is practically never used. Such quasi-automatic extensions prevail in a large number of European countries, including Italy, Portugal or Spain. The main rationales for such extensions are fairness considerations for workers – ie to ensure that all workers in a given industry are treated the same way – and transaction costs reduction – ie to avoid some firms from engaging in lengthy negotiations. However, it has been argued before that extensions could be a tool for ’insider firms’ to drive competitors out of the market (Haucap et al. (2001), Magruder (2012), Martins (2014a)).
The issue of representativeness
Whether extensions are desirable or detrimental is closely intertwined with the representativeness of bargaining institutions – namely the representativeness of both employee unions and employer federations. In France, for an industry agreement to be signed, employees unions have to be representative enough according to a legal threshold which corresponds to 8% of the votes in the last work council elections. In several countries an employer federations representativeness criterion is also applied : for instance, Portugal requires that workers in signing firms represent at least 50% of workers of the industry. In France, no such criterion applied until the 2014 and 2016 laws. Those laws established that an employer federation should represent either 8% of all firms pertaining to employers federations or 8% of all workers of these corresponding firms.
Employer federations representativeness is often defined as ’the share of the workforce in affiliated firms in the total employment of the relevant sector ’ (Martins and Hijzen (2016)). However, a criterion based on this definition does not ensure that signing firms are representative of all the firms covered by the industry agreement. In other words, the representativeness criterion can be met even if employers federations over-represent large firms, and therefore large firms’ interests. Such concern has been expressed in the OECD Employment Outlook 2017 : ’Extensions may also have a negative impact when the terms set in the agreement do not account for the economic situation of a majority of firms in the sector. For instance, when the employer association is representative only of large and relatively more productive firms (and hence willing to pay higher wages), it may agree on wage floors and other components that are not sustainable for smaller and less productive firms.’.
A cruel lack of statistics on employer organisations’ membership (OECD Employment Outlook 2017) often prevents from providing an adequate picture of employers’ federations. Figures on the population covered by collective agreement are usually available : in OECD countries, 26% of small firms workers are covered by a collective agreement while 34% of large firms workers are covered 8. Yet, information about which firm pertains to which employers federations is largely ignored. Still, it has been documented that large firms are more willing to affiliate than small ones (Traxler (1995), Traxler (2000), Traxler (1995), Barry and Wilkinson (2011), Mortimer et al. (2004)). We present in Figure 1.2 an index of large firms domination in employers federations, constructed from French data 9. An index higher than one means that the average size of bargaining firms is higher than the average size of all firms in the sector. In other words, for an industry to have a high index means that in this industry, large firms dominate the bargaining process. The histogram exhibits a positively skewed distribution, and most industries display an index superior to one : in most industries, the bargaining firms are the largest firms.
Reasons for this lack of small firms participation can be manifold : lack of time, lack of information, membership contributions. Sociological studies (Giraud (2012), Offerlé (2013)) put forward the limited time of small-firms CEOs to fully participate to federations, and therefore negotiations. It is even sometimes argued that some federations refuse small firms as they would not be cost-efficient for the federation : they would contribute low amounts while consuming a lot of the federation’s services (Offerlé (2013)).
Model : the impact of industry-level wage bargaining
We use in this section a model very close to Melitz (2003), in which we introduce firm-level bargaining and industry-level bargaining. Going from firm-level to industry-level bargaining raises the productivity threshold, which drives the least productive firms out of the market and which thus benefits the dominant – or most productive – firms. The general equilibrium effects of such sectoral bargaining are an increase in the unemployment rate and a decrease in the utility of consumers. We first lay-out the basic set-up of the model and then study the impact of the bargaining level on the economy 10.
Setup of the model
Suppose the national market consists of J industries. An industry is composed by a continuum of heterogeneous firms which produce each a single product and operate in situation of monopolistic competition. A representative consumer allocates her consumption between industries on the basis of an aggregate industry price, and then between firms of an industry on the basis of the price they charge. We first focus on the demand side, and then we turn to the supply side.
Table of contents :
1 Large firms’ collusion in the labor market : Evidence from collective bargaining
1.1 Introduction
1.2 Institutional setting
1.2.1 Collective wage bargaining in France
1.2.2 The issue of representativeness
1.3 Model : the impact of industry-level wage bargaining
1.3.1 Setup of the model
1.3.2 Impact of the introduction of industry-level bargaining
1.4 Model : the role of the representativeness of bargaining institutions
1.4.1 Representativeness of employers federations
1.4.2 General equilibrium
1.5 Empirical evidence on the impact of employers federations unrepresentativeness
1.5.1 Testing the model’s predictions
1.5.2 Data
1.5.3 Indices construction and descriptive statistics
1.5.4 Stylized facts
1.5.5 Estimation strategy
1.5.6 Results
1.5.7 Robustness
1.6 Conclusion
1.7 Tables and figures
1.A Collective bargaining in OECD countries
1.B Aggregate variables
1.C Firm-level bargaining
1.D Industry-level bargaining
1.D.1 Equilibrium structure of the industry
1.D.2 Value of the wage floor
1.E General Equilibrium
1.F Impact of the representativeness of bargaining institutions
1.F.1 Micro-foundation for the existence of employers federations dominated by large firms
1.F.2 Proof of Proposition 5.1
1.G Descriptive statistics
1.H OLS results
2 Impact of sectoral agreements on creative destruction
2.1 Introduction
2.2 Demand, supply and Schumpeterian growth
2.2.1 Aggregate Demand
2.2.2 Production
2.2.3 Innovation
2.3 Wage bargaining
2.3.1 Workers objective
2.3.2 Firm-level bargaining
2.3.3 Industry-level bargaining
2.4 Impact of level of bargaining on the equilibrium
2.4.1 Definition of the general equilibrium
2.4.2 Effect on the size distribution
2.5 Quantitative analysis
2.5.1 Calibration
2.5.2 Impact of the level at which the bargaining takes place
2.A Incumbent firm program
2.B Effect of the aggregate Industry revenue
2.C Equilibrium
2.D Alternative scenario
3 Collusion in the Labor Market and International Competition
3.1 Introduction
3.2 The model
3.2.1 Product Market Structure
3.2.2 International Trade
3.2.3 Labour Market
3.2.4 Equilibrium structure
3.3 Impact of trade frictions on wage floors
3.3.1 Equilibrium value of m
3.3.2 Variation of wage floors in response to trade frictions
3.4 Empirical implications
3.4.1 Identification using the China Shock
3.4.2 Data
3.4.3 Descriptive Evidence
3.4.4 The impact of international competition on sector-level wage negotiation
3.4.5 The impact of international competition on revenue of firms
3.A Method used for figure 3.1
3.B Descriptive statistics
3.C Firm-level bargaining
3.D Industry level bargaining
3.E The impact of the wage floor on the trade equilibrium
3.F Impact of trade frictions on the wage floor
