Relevant labour market: missing in the new Market Definition Notice

Abstract

The European Commission has been signalling an ambition to start more vigorously enforcing European Union competition law against business actions concerning labour. Such enforcement may be necessary because empirical research shows that competition among European employers is often limited. However, the Commission’s new Market Definition Notice entirely ignores the markets for labour, where such competition takes place, even though their definition may be necessary for enforcement. Against this backdrop, the present article examines how to define the relevant labour market. It discusses the occupation and geographic dimensions of the market, the role of the hypothetical monopsonist framework, and the relationship of the market to downstream product markets, and it considers which evidence can be used to define the market. The discussion reveals that the relevant labour market consists of jobs for which the respective workers qualify and are also willing to consider. This so-called matching nature of labour markets is a major reason why they tend to be quite narrow. The article calls on the Commission to provide authoritative guidance to market actors and enforcers on how to define relevant labour markets.

1. INTRODUCTION

The European Commission has for some time been signalling its ambition to enforce EU competition law against actions taken by businesses in their capacity as employers, that is, actions concerning labour. In 2021, the Commissioner for Competition remarked that agreements to fix wages or not to poach each other’s staff constitute illegal cartels.¹ In 2024, the Commission for the first time ever, opened a formal investigation into a possible case of such no-poach agreements.² Also in 2024, the Commission published a competition policy brief called Antitrust in Labour Markets, claiming that the Commission is devoting close attention to competition between employers.³ In this respect, the Commission finally seems to be following the example of US competition authorities, which up to now ‘have acted more aggressively to tackle the causes of employer power’⁴ than their European counterparts.

Intervening against violations of competition law by employers is undoubtedly to be desired. For one, the exercise of employers’ market power—that is, monopsony power—against workers will often ultimately harm also downstream consumers. This will be the case when businesses suppress wages by procuring less labour and when that translates into lower output.⁵ Interventions against restricted competition between employers are then justified even under the traditional consumer-focused perspective on competition law. Nevertheless, there are good reasons to intervene even when such restricted employer competition actually does not impact consumers negatively or even benefits them, that is, to protect workers for their own sake. Work looms larger in people’s lives than most consumer products and services because it ‘encompasses the greater portion of most people’s lives and is the source of self-worth and dignity, as well as lasting social bonds, skills, and financial well-being’.⁶ Enforcement against employer actions that harm workers for the latter’s own sake is, moreover, reconcilable with the existing doctrines of European Union (EU) competition law.⁷

Interventions against the market power of employers are also necessitated by the reality of labour markets. Until recently, competition between employers had been believed to be generally fierce.⁸ More and more empirical studies are nevertheless showing that this belief is mistaken and that workers are negatively impacted by employer market power. Many of these studies concern labour markets outside the EU. For instance, Azar and others found that US markets for labour are often highly concentrated.⁹ A study by the Competition and Markets Authority concluded that workers finding themselves in the most concentrated decile of UK labour markets earn about 10 per cent less than comparable workers working in comparable businesses in the least concentrated decile.¹⁰ There is nevertheless also a growing number of studies identifying competition issues in labour markets within the EU. For example, Araki and others show that a sizeable share of labour markets in selected EU countries is moderately to highly concentrated.¹¹ Bassanini and others then conclude that labour market concentration in a sample of EU countries negatively impacts workers not only through lower wages but also through a lower probability of having a permanent contract.¹² While competition law alone cannot solve all the problems associated with the market power of European employers, it can play an essential role in addressing some of these problems.¹³

Given all this, it is a pity that the Commission may actually not be all that serious as regards its ambition to intervene against employers. Namely, the Commission’s new Market Definition Notice¹⁴ (hereafter only the ‘Notice’) does not say a word about markets on which employers compete for labour. To be sure, the Notice does recognize that its guidance applies also ‘when defining relevant markets for the purchase of particular products in a particular area (“purchasing markets”)’.¹⁵ Markets on which employers compete for labour are indeed a kind of such purchasing markets. However, the Notice fails to provide any further specific guidance on how to define purchasing markets in general or labour markets in particular.

The lack of guidance on how to define labour markets in the Notice is a lost opportunity.¹⁶ There are certain ‘characteristics inherent to the worker-employer relationship [that] can create challenges for the definition of relevant antitrust labor markets’.¹⁷ Among other things, the unusual features of labour markets lead to these markets often being narrower than one might expect and in any case than product markets.¹⁸ The Notice itself recognizes that guidance on how the Commission applies the concept of relevant market in its enforcement of EU competition law is helpful to businesses¹⁹ as well as ‘external stakeholders’.²⁰ The latter arguably include other enforcers of EU competition law, that is, national competition authorities and national courts.²¹ By providing guidance on how to define the relevant labour markets, the Commission thus could assist all these actors in applying EU competition law to the conduct of employers. Given that the insufficient understanding of labour markets has been put forward as a major reason for a lack of competition law intervention in these markets,²² such assistance could also lead to a more vigorous enforcement against anti-competitive employer actions.

Against this background, the present article discusses how to define the relevant labour markets. Section 2 offers some preliminary points about defining these markets under the Notice. Section 3 then discusses the actual market definition as identification of alternative jobs available to workers regarding both the substantive and geographic market dimension, and it also addresses evidence used in the definition. Section 4 concludes by calling on the Commission to provide authoritative guidance on labour market definition as soon as possible.

2. THE NOTICE AND LABOUR MARKETS: PRELIMINARY POINTS

The Market Definition Notice provides guidance on how the Commission applies the concept of relevant market in its enforcement of EU competition law.²³ The original 1997 notice has been replaced by a revised document in February 2024 in order to provide a more comprehensive and clear guidance.²⁴

Remarkably, the new Notice has virtually nothing to say about competition between employers. The only related provision appears to be paragraph 7, which says that the guidance set out in the Notice applies also to the definition of ‘purchasing markets’, that is ‘when defining relevant markets for the purchase of particular products in a particular area’. With employers being purchasers of labour, the Notice thereby arguably recognizes also the definition of labour markets.²⁵ Paragraph 7 further advises to define purchasing markets ‘taking into account their specificities’. Such advice, however, seems to have little value when these specificities and their role in the definition are not discussed. The only specificity of purchasing markets mentioned by the Notice is the obvious point that actual competitors of a buyer are given by whom the sellers consider as alternative buyers.²⁶ In any case, the Notice does not expressly mention labour markets.

It will be helpful to make a terminological remark here at the outset. As just shown, the Notice distinguishes between purchasing markets and selling markets.²⁷ Admittedly, one might perhaps find it more natural to talk about the purchasing and selling side of a certain market. Such a framing may, after all, be often found in competition law literature.²⁸ However, it is true that the purpose of defining the relevant market in the context of law enforcement is to identify effective and immediate competitive constraints faced by the business(es) involved in the investigated conduct or concentration.²⁹ After all, the market is relevant to that conduct or concentration. The most effective and immediate constraints then tend to come from competitors, that is, from businesses finding themselves on the same side of the market as the business(es) under investigation.

What is more, the two sides of the market may not even be symmetric: The Notice correctly observes that the fact that seller B is within the relevant market defined for the investigated seller A does not necessarily mean that seller A will be within the market defined when seller B gets investigated.³⁰ In the same way, the fact that the relevant market defined for an investigated seller includes buyer C and buyer D as ‘relevant customers’³¹ does not necessarily mean that both of these buyers belong to the relevant purchasing market defined when one of them gets investigated. That is all to say that it makes perfect sense to distinguish the definition of a purchasing market from the definition of a selling market.³²

The present article focuses on the definition of markets on which businesses purchase labour. In keeping with the majority practice, these markets are referred to here as ‘labour markets’.³³ One should nevertheless keep in mind that it is in principle possible to define also markets on which workers sell their labour to businesses. A perhaps less ambiguous label for the markets considered by this article could thus be employing markets or hiring markets.³⁴

The Notice fails to address labour markets even though the Commission had been repeatedly asked to include them. As recognized by the Commission itself, already in the 2020 stakeholder consultation on the previous Notice, ‘[a] trade union criticised that the Notice fails to take account of the labour markets.’³⁵ A look at the actual submissions to that consultation reveals that the criticism was voiced by the European Trade Union Confederation (ETUC),³⁶ a major trade union organization representing workers at the European level. After the draft new Notice published for comments in 2022 nevertheless remained silent on the topic, the omission was criticized again, this time by the Catalan Competition Authority,³⁷ but still to no avail.

The absence of labour markets in the Commission’s soft law contrasts strikingly with the situation on the other side of the Atlantic. Namely, the Merger Guidelines published jointly by the US Department of Justice and the Federal Trade Commission at the end of 2023 do talk about markets on which businesses procure labour.³⁸ As a matter of fact, the guidelines outright observe that ‘labor markets are important buyer markets.’³⁹ Subsequently they—admittedly rather briefly—discuss special characteristics of labour markets relevant to their definition, concluding that these characteristics possibly make these markets ‘relatively narrow’.⁴⁰

While it is not the objective of this article to diagnose why labour markets do not feature in the Notice, it will be useful to consider two potential reasons. First, some might think that a labour market does not satisfy the criteria of a relevant market laid down by the Notice in the first place. Namely, the Notice observes that a relevant market ‘typically comprises a product … dimension’,⁴¹ whereas a labour market is a market for labour, not a market for a product. This might hence bring one to the conclusion that a labour market cannot in fact constitute a relevant market. However, such an interpretation does not seem warranted. The Notice itself recognizes that the term ‘product’ is used as a shorthand that may cover various objects of trade including also for instance technologies to be licensed.⁴² Labour may constitute such an object of trade, too. Still, it would be advisable to explicitly recognize this in the Notice. The present article will further refer to this ‘substantive’ dimension of a labour market as to its ‘occupation’ dimension.⁴³

Second, one might argue that the Notice does not need to address labour markets because their determination is actually not necessary in the enforcement of EU competition law against employers. More specifically, the Commission has so far been focusing only on interventions against wage-fixing and no-poach agreements between employers. Such agreements have been at the centre of the Commissioner’s remarks, the opened investigation as well as the competition policy brief that were all mentioned in the introduction of this article.⁴⁴ At the same time, the Commission’s 2023 Guidelines on Horizontal Co-Operation Agreements qualify these two types of agreements as restricting competition by object.⁴⁵ Finally, the Notice states that the Commission usually does not define the relevant market when assessing agreements restricting competition by object.⁴⁶ However, this explanation does not seem convincing, given that there are also other types of potential EU competition law cases concerning employer activities, such as agreements restricting competition by effect, abuse of dominance and concentrations, in which the relevant market is actually supposed to be defined.⁴⁷

3. THE RELEVANT LABOUR MARKET

Let us now proceed to the actual definition of labour markets. The Commission maintains that a market consists of actual competitors as businesses that exert the most effective and immediate constraints upon each other. These constraints follow from the threat that market counterparties will switch to a competitor.⁴⁸ Following this logic, a labour market is ‘a group of jobs, between which workers can switch with relative ease’.⁴⁹ The threat of switching is usually discussed in terms of substitutability. The assessment of substitutability of jobs by workers mirrors how substitution of products by consumers is analysed. As regards selling product markets, one looks at whether a business’s customers would switch towards rival businesses’ products when faced with a deterioration in the conditions of supply.⁵⁰ For labour markets, the question is whether workers would switch to rival employers following a decrease in wages or some other worsening of their working conditions.⁵¹

The rest of this section elaborates on the definition of the relevant labour market in greater depth. First, it interrogates how the purpose of the exercise shapes also its content. Then the section, in turn, looks at how to define the occupation dimension and the geographic dimension of a labour market. Subsequently, it examines the role of the hypothetical monopsonist framework. Then it discusses the relation between labour markets and downstream product markets. Finally, it scrutinizes the evidence used in the defining of labour markets.

The purpose of defining a labour market

When defining the relevant market, it is important to keep in mind the practical purpose of the exercise. Namely, the relevant market is defined in the context of competition law enforcement in order to identify the effective and immediate competitive constraints faced by the businesses involved.⁵² Such businesses are typically the businesses investigated under Article 101 or 102 TFEU or the businesses whose concentration is being investigated under the Merger Regulation.⁵³

The enforcement case in question may concern one or more labour markets. For instance, a merger of two businesses may involve multiple markets in which employers compete over labour, which may differ in their geographic⁵⁴ as well as occupation⁵⁵ dimension. Such multiplicity of relevant markets may arise also with regard to product markets: for instance, a merger of large chains that operate drug stores may impact various markets.⁵⁶ The issue nevertheless appears more pervasive when it comes to labour.⁵⁷

In any case, given that the purpose of defining the relevant market is to identify the effective and immediate competitive constraints faced by the businesses involved, the market definition process starts from the activities of these businesses. Namely, as regards product markets, the relevant market gets defined for the particular products traded by the businesses involved and for the particular area in which the businesses trade the products.⁵⁸ That is to say, the starting point of market definition is the businesses’ products covered by the investigation. The question then is which other products constituting the same market should be added.⁵⁹

For labour markets, the starting point is thus the labour concerned by the agreement between employers, the potentially abusive conduct of a potentially dominant employer, or the concentration of employers. The question is then which other businesses these employers are competing over labour with. That is to say, the process of defining the relevant labour market starts with the respective jobs at the businesses involved, to which other jobs constituting the same market are then added.

Occupation dimension

The first dimension of labour markets to be discussed is their occupation dimension. As observed by the Notice, ‘in purchasing markets the substitution assessment focuses on alternatives available to suppliers, rather than alternatives available to customers’.⁶⁰ This sub-section will make clear that the alternatives available to workers as suppliers of labour depend on the preferences of employers as well as of workers themselves.

Preferences of employers

Let us start with how the occupation dimension of a labour market is shaped by the needs of employers. The alternatives available to suppliers always depend on the customers: suppliers can switch only to those customers who want to buy from them. The same holds also for labour markets: the alternatives available to workers are limited by whom employers want to hire.⁶¹

Labour markets in this regard bear a resemblance to markets for differentiated products rather than to markets for homogeneous commodities. In the latter case, a buyer does not distinguish between different suppliers because their products are the same. Take for instance the market for a certain type of oil. When thinking about alternatives available to suppliers of this oil type, they will include anyone who wants to buy this commodity. When it comes to differentiated goods, however, buyers do distinguish between the goods. Consider for example the market for cars: ‘the buyer cares about the identity, nature, and features of the product in question—the car.’⁶² The alternatives available to car suppliers are then very much contingent on the preferences of buyers.

Employers have preferences over various worker characteristics. Most obviously, they require certain skills. To make sure that newly recruited workers have the demanded skills, employers may sometimes impose strict limitations on who qualifies for the given position, concerning for instance certain completed education. Some positions require a more general skill set that also fits other types of positions. For instance, Kucko and others observe that a former quick-service restaurant worker may easily find employment in retail or at a warehouse.⁶³ The literature posits that workers in such generalist positions will thus have relatively many alternatives available to them and, hence, that their employers will belong to relatively broader markets.⁶⁴ Then there are specialist positions that require special skills and knowledge, with limited use in other types of positions. Consider, for instance, doctors, as well as miners⁶⁵ or certain maintenance workers.⁶⁶ The literature claims that workers in such positions are employable only in very similar positions,⁶⁷ which leads to a narrower labour market.⁶⁸ This point is sometimes recast in terms of the distinction between high-skilled and low-skilled workers: ‘Many different employers hire low-skill workers such as customer service representatives or secretaries and administrative assistants, while a high-skill worker invests in skills that may be suitable for only a small number of employers.’⁶⁹

However, one has to wonder whether such a portrayal of competition in labour markets makes sense. Would Amazon really not want to hire someone as a warehouse worker only because that person previously studied medicine and worked as a doctor? Or is such a person rather unlikely to apply for a warehouse job in the first place? That is to say, the eventual exclusion of generalist jobs from the labour market for specialized workers does not seem to follow from the preferences of employers but of the workers themselves. The effect of workers’ preferences on the market definition will be discussed in the following sub-section. As regards the actual role of employers’ preferences over skills in the definition of labour markets, the following conceptualization seems more accurate: Workers cannot switch to a job when it requires special skills that they do not have. A hospital will thus not be competing over doctors with businesses offering software developer jobs because doctors usually do not have the skills necessary to develop software.

Many employers nevertheless have additional requirements beyond skills and knowledge.⁷⁰ These requirements may concern, for instance, time availability: a business may need its employees to work night shifts.⁷¹ They will also often have preferences concerning the match of workers’ personality traits to their culture.⁷² All such requirements imposed by alternative employers⁷³ do limit the set of alternatives available to workers and, thus, narrow the scope of labour markets.⁷⁴

Preferences of workers

What makes labour markets special is that the preferences that shape their scope are not only those of buyers but also those of sellers. Let us return to the example of a market for cars. There, ‘[t]he seller cares nothing about the buyer or (in most cases) what the buyer plans do with the car.’⁷⁵ The alternatives available to suppliers are thus driven only by buyer preferences: whether a buyer is competing with another depends on whether the latter prefers the same car suppliers as the former. When it comes to labour markets, however, workers do care about the identity and characteristics of the buyer, that is, the employer.⁷⁶

Workers may have preferences over various aspects of jobs.⁷⁷ This is hardly surprising: ‘because work is such an important part of people’s lives, people are naturally concerned even about minor aspects of it’.⁷⁸ One aspect concerns the actual content of the job. Naidu and Posner give the example of lawyers who want to work as litigators. Some of them may like better antitrust cases while others may prefer employment litigation.⁷⁹ Sillman similarly observes that ‘[t]o a lawyer who has many years of experience working in immigration law, a position working on corporate transactions is probably not a meaningful alternative.’⁸⁰

Workers’ preferences may concern many other aspects of jobs. These may have to do with practical issues such as shift flexibility, vacation, and sick time.⁸¹ Workers may nevertheless also favour a certain workplace culture: continuing their litigator example, Naidu and Posner observe that some lawyers prefer law firms with highly intense and competitive cultures while others do not.⁸² A similar point is that people care about ‘the identities of the other workers at a workplace—whether they are driven and intense, friendly and laid back, or young or old’.⁸³ Other relevant features of a job may include job security, job autonomy, job prestige, work intensity, participation in training, skill development, relationship with the employer, and potential adverse effects of work on private life.⁸⁴

An important consideration is, of course, also financial compensation. This may be construed as consisting of not only wages but also other monetary benefits such as health insurance and pension plans.

It was discussed above that for specialist workers a job may be a substitute only if it makes use of their special skills. This is, in fact, supported by empirical research, which shows that ‘[w]orkers are more likely to switch between occupations that are similar in the kinds of tasks that are performed’.⁸⁵ The underlying mechanism proposed by the literature was then that employers are not willing to hire such workers holding specialist jobs for other non-specialized jobs.⁸⁶ It, however, seems more convincing that employers would actually happily hire these workers for such positions but that the workers themselves are not interested.⁸⁷ This may be because the wage is lower—specialist jobs often pay significantly more than generalist ones. The workers nevertheless could also enjoy the specialist jobs for other reasons such as putting their training and experience into work. That is to say, specialist jobs may indeed belong to narrow labour markets but for other reasons than suggested by the literature.

Matching markets

The fact that the preferences of both sides of the market affect which alternatives are available to workers makes labour markets so-called matching markets. A possible way of looking at this issue is that the market is doubly differentiated.⁸⁸ First, it is differentiated as regards differences between workers as those differences are relevant to employers. For example, a business employing workers with a specialized skillset will not face competition over such workers from employers requiring a different specialized skillset. Secondly, a labour market is differentiated as regards differences between employers as those differences are relevant to workers. For instance, a business employing workers who have a strong preference to work part time will face competition over such workers only from employers who also offer part-time positions.⁸⁹

As a result of the matching nature of labour markets, their substantive dimension consists of jobs that workers are both qualified for and willing to consider.⁹⁰ This dual limitation then means that there may be few actual job alternatives for a particular worker.⁹¹ The set of firms that prefer a worker may be small to start with, and the sub-set of available firms that are preferred by that worker is even smaller.⁹² This is why labour markets can be quite narrow.⁹³

Geographic dimension

Labour markets have also a geographic dimension. The Notice says that ‘[i]n some markets, the competitive positions of suppliers may depend on the distance between each supplier and the customer’.⁹⁴ This holds analogously also for labour markets: the competitive positions of employers depend on the distance between each employer and the worker. This is because the alternatives available to workers depend on whether they are willing to commute or even relocate to a job.⁹⁵ How willing they are in turn depends on the ‘costs’ of the commute. These costs, however, do not include only actual out-of-pocket expenditures such as the price of public transport tickets or of moving one’s home but also—and perhaps even mainly—convenience and time.⁹⁶ Different workers will perceive these costs differently, depending for instance on their age, family status, or health situation.⁹⁷ ‘For example, younger workers may be willing to commute farther or even move from one location to another, while older married workers or those with children may be less mobile.’⁹⁸ Two-income families may be constrained even further because both parents need to find a job in the same area.⁹⁹

The geographic scope of the market is expanded by the eventual possibility to work remotely. Such a possibility gives workers more alternatives to choose from.¹⁰⁰ The increase in remote work following the COVID pandemic thus may have impacted the market definition for some occupations.¹⁰¹ At the same time, however, ‘[m]ost jobs still require physical proximity to the employer’.¹⁰²

The geographic dimension of a labour market may interact with its substantive dimension. Namely, how far workers are willing to commute or relocate may depend on the content of the job. In particular, high-skilled workers are generally more mobile than low-skilled ones.¹⁰³ ‘Low-skill workers may have less access to transportation, well-situated housing markets, child care options, and job information, and be more dependent on local, informal networks.’¹⁰⁴ Jobs for high-skilled workers are also more likely to allow remote work, which expands the geographic scope of the market.

Many labour markets will be quite narrow in geographic terms. Naidu and Posner, in this context, even maintain that ‘labor market areas are typically (though not always) smaller than product market areas’ because people are less mobile than goods.¹⁰⁵ They give the example of farm-equipment manufacturers.¹⁰⁶ While the market for farm equipment may be even international in scope, the labour markets are only local, situated around the plants.

Hypothetical monopsonist framework

When identifying the alternatives available to workers and, hence, defining the relevant labour market, one may rely on the hypothetical monopsonist approach. This is an adaptation for a purchasing market of the hypothetical monopolist approach, which is intended for selling markets.¹⁰⁷ In the latter context, the relevant market is the smallest set of products over which a hypothetical monopolist would find it profitable to implement a small but significant non-transitory increase in price (SSNIP).¹⁰⁸ This increase is usually understood to be somewhere between 5 per cent and 10 per cent.¹⁰⁹ If such a SSNIP is not profitable, the logic goes, there must be some substitute products to which customers are switching, and these must therefore be added to the market definition. The Notice explains that the Commission may rely on the principles of the SSNIP test in its assessment of the relevant market but is not obliged to apply the test empirically.¹¹⁰

When it comes to the definition of labour markets, the analogous question is whether a hypothetical monopsonist would be able to implement a small but significant and non-transitory decrease in wages (SSNDW),¹¹¹ also referred to as a small but significant and non-transitory reduction in wages (SSNRW),¹¹² without losing so many workers that it would be unprofitable.¹¹³ One starts from the jobs at the businesses involved and then adds readily available substitute jobs until the SSNDW becomes profitable.¹¹⁴ At that moment, the workers qualified for and willing to consider these jobs clearly do not have alternatives to which they could switch and, hence, the jobs constitute the relevant labour market. Nevertheless, also as regards labour markets, this approach serves mainly as a ‘conceptual framework for the interpretation of available evidence’¹¹⁵ rather than an actual empirical test.¹¹⁶

Just like the hypothetical monopolist framework, also the hypothetical monopsonist framework needs to address the so-called ‘cellophane fallacy’.¹¹⁷ The name refers to a US Supreme Court case that was criticized for how the court defined the market in question.¹¹⁸ Namely, the court observed that a SSNIP of the product in question would not be profitable because many customers would stop buying the product. The court believed that there thus had to be many substitute products to which customers could switch. Critics, nevertheless, challenged this interpretation of facts, arguing that there were no such substitute products and that the reason why a SSNIP would lead to fewer purchases was that the actual prevailing price had already been inflated in the first place. As a consequence, rather than switching to substitutes, customers would actually stop buying this type of product altogether because the price would be above their willingness to pay. The same might happen also when defining a labour market: if an employer is already paying infra-competitive wages close to the minimum that workers are willing to accept, a SSNDW may not be profitable despite there in fact being no jobs to add to the market definition. This needs to be taken into account when defining a labour market in the context of a case that requires an assessment of the degree of existing employer market power.¹¹⁹ The SSNDW test may then need to be applied starting from a counterfactual, more competitive wage.¹²⁰

As it was discussed above, workers choose between—and thus derive value from—jobs not only on the pecuniary wage but also on many other aspects. Hence, to actually be significant, the wage cut may need to be greater than it appears at first sight because the decrease of 5 to 10 per cent should be calculated from all the benefits enjoyed by the worker.¹²¹ As noted by Dechamps and others, ‘a 5% change in monetary remuneration may correspond to a 5% change in the overall benefit package in some contexts, and be only a small part of the benefit package in other cases.’¹²² That is why Kucko and others propose to talk about a small but significant and non-transitory reduction in compensation (SSNRC), where compensation is understood to also include valuable non-wage components.¹²³

Labour markets and downstream product markets

Let us now consider two questions about how the scope of a labour market relates to that of a downstream product market. The first question concerns the relationship between the geographic scope of the relevant labour market and of downstream product markets on which the given business is active. It should be clear from the previous discussion that these scopes are independent of each other.¹²⁴ Posner gives the example of a hospital: the downstream market on which it provides health care to patients may be geographically larger than the labour market on which it hires nurses because the former, who visit a hospital occasionally, are likely willing to travel farther than the latter, who need to make the trip daily.¹²⁵ Marinescu and Posner note that mines hire miners locally but sell their products even into global markets.¹²⁶ Hovenkamp observes the following: ‘[A] large Amazon warehouse might recruit most of its employees in a local market, even though the products are destined for national shipment.’¹²⁷ Such a disproportion between the geographic scope of the labour market and the selling market is actually very common.¹²⁸

The second question is whether two businesses operating on the same labour market—or, for that matter, purchasing market more generally—are necessarily competing also on a downstream selling market and vice versa. Regarding the former point, two businesses may operate on the same labour market without being present in the same downstream selling market.¹²⁹ That will be the case when the businesses need the same type of labour but use it to produce products that are not interchangeable in the eyes of downstream customers.¹³⁰ ‘For example, airplane manufacturers and their part suppliers may both hire from the same market for engineers.’¹³¹ Or consider the US case concerning competition between eBay and Intuit for computer engineers even though the former is an e-commerce platform while the latter specializes in business software.¹³²

Regarding the latter point, producers of substitutable, competing products do not necessarily compete for labour. This may be because they each use different production technology and, thus, need a different type of personnel.¹³³ Alternatively, they may rely on the same type of workers but be located far apart and thus not compete for the same actual people,¹³⁴ while still delivering competing products to buyers in a greater, overlapping geographic area.¹³⁵

As a result of both points, the relevant labour market may not correlate much with the industry to which a business belongs.¹³⁶ This is because the industry affiliation of a business depends on what it sells downstream. However, as just established, the downstream activities are irrelevant to the definition of the labour market. For instance, ‘janitors, maintenance staff, clerical employees, and other support employees’ working in hospitals are very likely to be competed over by a much broader range of employers.¹³⁷ As discussed above, workers may move between different—especially generalist—occupations.¹³⁸ But, in addition, the same occupation may also be offered across various industries. ‘For example, a cleaner could find employment cleaning the factory of a manufacturing firm, a hospital in the health industry, or a bank in the financial sector.’¹³⁹ Also, ‘an accountant might leave his accounting job at a manufacturing company to join an accounting firm’.¹⁴⁰ All this only confirms the necessity to define labour markets in their own right, without paying much attention to the businesses’ operations down the stream.

Evidence

We have established that the relevant labour market consists of alternative jobs available to the workers of the businesses involved. It is, however, a different question determining which jobs those actually are. The Notice discusses evidence used to define product markets and its gathering in several of its sections. The present section of this article does the same with regard to evidence used to define labour markets. It first addresses the usefulness of pre-existing administrative classifications and subsequently examines which specific evidence may be considered in defining a labour market.

Administrative classifications

One possible input into the definition of labour markets is pre-existing administrative classifications. As regards the substantive dimension of labour markets, American antitrust commentators have been pointing at the ‘Standard Occupational Classification’ maintained by the US Bureau of Labor Statistics.¹⁴¹ This system classifies jobs into occupational categories based on the tasks that they entail.¹⁴² The EU relies on the analogous international standard classification of occupations developed by the International Labor Organization, which it further extends into more granular sub-categories through its own classification system.¹⁴³ Such occupational classifications are in particular helpful insofar as they cut across different industries because, as discussed above, that aligns them better with the logic of defining relevant labour markets.¹⁴⁴

Pre-existing administrative classifications may also pertain to the geographic dimension of labour markets. For instance, the US Department of Agriculture developed commuting zones, which are clusters of counties, as ‘a spatial measure of the local labor market’.¹⁴⁵ The UK Office for National Statistics maintains ‘travel to work areas’,¹⁴⁶ which ‘are constructed based on observed commuting flows from census data, and hence provide a reasonable approximation to the geographic scope of labour mobility.’¹⁴⁷ Eurostat works with ‘functional urban areas’, which consist of a city and a commuting zone ‘whose labour market is highly integrated with the city’.¹⁴⁸ Such pre-existing administrative classifications are regularly used by researchers studying competition in labour markets.¹⁴⁹

The use of administrative classifications in the defining of labour markets is, however, not without its problems. This is because ‘the correlation between government classification data and relevant antitrust market can be poor.’¹⁵⁰ On the one hand, even the most granular categorizations may still be overly broad, that is overinclusive. As observed by Sillman, the US classification of occupations ‘would group all lawyers … into one market, which is probably too broad.’¹⁵¹ Marinescu and Hovenkamp similarly note that while the classification has no further sub-categories for accountants and auditors, empirical evidence suggests that even the positions of senior accountants and junior accountants are actually not close substitutes.¹⁵² Kucko and others say that the occupational category ‘computer programmers’ may be too broad because it fails to appreciate the role of different programming languages in employment decisions.¹⁵³

On the other hand, the categories may also be too narrow, that is underinclusive. Sillman notes that executive secretaries, legal secretaries, medical secretaries, and other secretaries are classified as different occupations at the highest level of granularity even though they may often be qualified for and considered by the same workers.¹⁵⁴ Marinescu and Hovenkamp make a similar observation about the various categories of cooks: fast food, institutions and cafeteria, private household, restaurant, short order, and others.¹⁵⁵ Kucko and others go even further, stating the following about generalist occupations: ‘[A] retail worker may well leave their job to drive for a ride-hailing app, or work in food services. Pre-defined industry or occupation codes likely would not group retail, ride-hailing app, and food service, but workers may consider all of these opportunities when making employment decisions.’¹⁵⁶ In other words, these classifications do not capture the substitutability of generalist occupations.

In the same way, administrative geographic classifications may be a poor proxy for the geographic dimension of labour markets. Consider for example the observation made above that people in some occupations tend to be willing to commute farther than others.¹⁵⁷ This alone suggests that a commuting zone or travel to work area may again be over- or underinclusive with regard to a specific labour market.

Administrative classifications thus need to be approached with care. What is acceptable in academic research that studies labour markets in an aggregate way from a bird’s eye perspective may not suffice in competition law enforcement that defines particular labour markets for the purposes of adopting consequential individualized decisions. To be sure, OECD argues that an approach based on administrative classifications ‘might be acceptable for [enforcement] agencies that have to make quick decisions.’¹⁵⁸ Usually, however, it will likely be necessary to aim for greater accuracy, using the evidence discussed below. But even then, administrative classifications may provide a helpful ‘provisional definition’¹⁵⁹ of the labour market at stake. Such a provisional definition may then serve as a ‘good starting point that can be revised upward or downward’ on the basis of additional specific evidence.¹⁶⁰ This is in line with the Commission’s approach to start from a ‘working hypothesis’ about the relevant market and to adjust this evidence in the light of the evidence gathered during its investigation.¹⁶¹

Specific evidence

Let us now consider possible specific evidence of substitutability between jobs. Perhaps the most direct kind of such evidence is data on the actual past flows of workers between businesses, that is, on past substitution.¹⁶² Such data may be included, for instance, in existing longitudinal surveys conducted by public bodies.¹⁶³ Information about substitution following eventual past structural changes, events, or shocks in the labour market, such as a closing or opening of a production plant, tend to be particularly informative.¹⁶⁴ It is also possible to consult the employment histories of individual workers found in their CVs and social media profiles or in the human resources records of their employers.¹⁶⁵

One may infer which jobs are substitutable also from the activities of workers and employers in connection to finding a suitable counterpart. As regards workers, one may for instance collect online information about workers’ job searches and interactions with job-posting websites or recruiters more generally.¹⁶⁶ Concerning employers, one can think about data on vacancies that include job responsibilities or documents ‘showing their recruiting strategies, including job fairs, campus visits, specific skills/profiles targeted’.¹⁶⁷

Further evidence may concern hypothetical conduct and views. One possibility is to carry out a worker survey, ‘ask[ing] about the sets of alternative job options and/or about propensity to switch jobs as a result of a change in the wage or working conditions’.¹⁶⁸ Views on job substitutability may be requested also from the workers’ representatives.¹⁶⁹ In fact, in its contribution to the public consultation on the Notice, the ETUC did propose to involve trade unions in the process of defining labour markets.¹⁷⁰ For employers, it is also possible to survey them or their associations. In addition, they may possess internal ‘documents that reveal which employers the company considers as potentially attractive for its employees, and hence as competitors in the labour market’.¹⁷¹ As recognized by the Commission, such documents may be particularly relevant where they have been prepared in the ordinary course of business as opposed to documents prepared in connection with an investigation.¹⁷² For instance, employers may ‘keep records of alternatives revealed by current employees during exit interviews or of other jobs considered by prospective employees when discussed during interviews’.¹⁷³

Last but not least, a strong indication of the scope of the market is business efforts to limit competition over labour.¹⁷⁴ In particular, if employers agree to limit competition over some workers, the workers are likely willing to switch between jobs offered by the employers. Why else would the employers limit their competition in the first place? A wage-fixing agreement or a no-poach agreement thus indicates that the jobs covered by them likely belong to the same labour market.¹⁷⁵ The same applies to non-compete clauses in employment contracts: ‘If one employer feels the need to include another in a non-compete agreement, it is likely because the other is a competitor in the same labor market.’¹⁷⁶ As discussed above, it is then irrelevant whether the businesses operate also on the same downstream selling market.¹⁷⁷

4. CONCLUSION

In principle, EU competition law does not itself preclude the competent bodies from enforcing its provisions also against employers¹⁷⁸ and from defining the relevant labour markets. In fact, some enforcers are already doing so. For example, when applying Article 101 TFEU, the Portuguese competition authority has defined the relevant market as the market for the recruitment of professional male soccer players.¹⁷⁹ The level of enforcement is nevertheless still very low, in part arguably because of the lack of guidance from the main institutional actors, including most of all the European Commission, on how to actually enforce the law in this type of cases. As a point in case discussed by the present article, the Commission’s new Market Definition Notice has not provided any guidance on the definition of labour—and, more generally, purchasing—markets. This is despite the Commission’s own claims of being ready to intervene against anticompetitive actions taken by employers.

Defining the labour market is an important part of enforcing EU competition law against employers. To be sure, certain types of cases do not require the relevant market to be defined¹⁸⁰ and there are even voices to get rid of this enforcement step altogether.¹⁸¹ To the extent that market definition remains a common feature of competition law enforcement, however, being able to carry out such a definition for a labour market is essential in intervening against employers.

It needs to be recognized that competition law is no panacea against employer market power.¹⁸² While such power is pervasive, that is frequently ‘not due to artificial limits on competition nor excessive concentration’¹⁸³ but due to other frictions. What gives employers the ability to offer worse working conditions than they would have in a competitive market are often natural features of labour markets such as the high costs of searching for a new job or of switching to it.¹⁸⁴ Surely, in the EU, even the exercise of this ability may perhaps sometimes qualify as an exploitative abuse of dominance under Article 102 TFEU. Nevertheless, given the very limited actual enforcement of EU competition law in this type of cases, it may be more pertinent to address the exercise of employer power against workers through other tools such as wage regulation, job protection or union support.

Still, this is not to say that the application of competition law to the conduct of employers should be abandoned as futile.¹⁸⁵ Labour markets can be quite concentrated, not least due to their frequent narrowness. Agreements between employers that restrict competition, exclusion of rival businesses through abusive conduct of dominant undertakings, and mergers of employers that significantly impede effective competition should be enforced against as a complement to other interventions in labour markets. Who knows, perhaps even the exploitative abuses of employer dominance will start receiving more attention. Be that as it may, enforcing EU competition law in such cases will often require defining the relevant labour market. It would thus be helpful to have some authoritative guidance on how to carry out the definition. The Commission should remedy the lack of such guidance as soon as possible.

Footnotes

Acknowledgements

I am grateful to Magali Eben, Jan Kupčík, and Viktoria Robertson for their comments.