Using compensating efficiencies to assess EU merger policy

Pauline Affeldt, Tomaso Duso, Klaus Gugler, Joanna Piechucka 10 January 2022



There appears to be increasing agreement among economists on the general trends of increasing concentration and markups since the 1990s.1 Yet, disagreement persists on its causes. Many economists believe that the increasing globalisation and digitisation of the economy are key potential drivers for these dynamics.2 Others instead contend that the increased and unvetted merger activity, especially by large global corporations, coupled with an insufficient enforcement of antitrust laws is a leading force for the observed trends (Philippon 2018). While empirical evidence in support of one or other of these causes has been found, there is still ample need for more careful assessments. This is especially true concerning the role of competition policy enforcement, for which a large-scale evaluation is still missing.

Among the different areas covered by this broad policy, merger control plays a particular role. By their very nature, horizontal mergers increase market concentration and create unilateral incentives for price increases. Moreover, by enhancing market power, mergers may not only increase the potential for abuses and the monopolisation of markets, but they can also increase the likelihood of collusion.  Merger control is the only instrument via which antitrust authorities can engage in ex-ante prevention of such anti-competitive situations. This is particularly relevant since it is much harder for competition authorities to intervene ex post, which, as suggested by several researchers, is only mildly successful (Baker 2019). Thus, the competitive assessment of mergers and the retrospective evaluation of merger control decisions are key elements to assess the overall effectiveness of competition policy.

Horizontal mergers and their effects on markets

The common wisdom on horizontal mergers is that their static effects are the balance between two countervailing forces: the market power effect and the efficiency effect. The former captures the impact of the reduction in competition through the combination of two competing firms; the latter relies on the assumption of merger-specific synergies. In most mergers, both effects co-exist and what matters for consumer welfare is their net effect. The level of merger-specific synergies necessary to compensate the market power effect so that consumers are not hurt by the merger is called ‘compensating efficiencies’.

While there is a by now sizeable and growing empirical literature assessing the competitive effects of specific mergers, much less is known about their efficiency effects. The relatively few studies trying to assess actual efficiencies from mergers document that they are either nonexistent or, in the best case, quite small. Yet, antitrust authorities worldwide clear most horizontal mergers unconditionally.3 Hence, the presumption of antitrust authorities at the time they make their merger decisions seems to be that compensating efficiencies should be sizeable.

In a recent study (Affeldt et al. 2021), we assess this presumption and, consequently, the effectiveness of EU merger policy by analysing the level of compensating efficiencies that would have been necessary to make mergers notified to the European Commission not hurt consumers.

The EU merger control database

We use a novel database constructed by analysing the merger control decisions of the Directorate-General for Competition (DG Comp) of the European Commission. Reports concerning all of its decisions are published online. We read and categorised almost all of DG Comp’s 6,429 merger decisions from 1990 to 2018. For each of these mergers, the European Commission performs a careful analysis: first defining the relevant antitrust geographic and product market based on the concept of substitutability, then performing a competitive assessment. 

In the assessment of the relevant antitrust market, the Commission identifies the effective competitors acting in the same (product and geographic) market for each affected product of the merging parties. Because market shares are not always or fully reported, we base our analysis on a subsample of 12,325 product/geographic antitrust markets affected by over 1,000 mergers for which we observe separate values for the market shares of the merging firms.

Looking at the raw data, it seems that the majority of the Commission’s interventions happened against mergers with high HHI and/or large changes in HHI (∆HHI) (Figure 1).4 However, it appears that the Commission also conditionally cleared many mergers in such problematic areas, while it intervened against mergers that did not affect highly concentrated markets or increased concentration. 

Figure 1 Concentration thresholds from the 2014 EU merger guidelines and the Commission’s merger decisions 

Source: Affeldt et al. (2021)

Measuring compensating efficiencies

To assess these patterns, we rely on the theoretical framework recently developed by Nocke and Whinston (2021) (see also Nocke and Whinston 2020). They show that, for different models of oligopolistic competition satisfying some aggregative games properties, compensating efficiencies can be derived as a (non-linear) function of the market shares of the merging parties and measures of the elasticity of demand as well as the degree of substitution across products.

We apply this framework to our dataset of EU mergers and compute the compensating efficiencies for three different models: (1) the Cournot model with homogenous goods, (2) the multiproduct price competition model with a constant elasticity of substitution demand system, and (3) the multiproduct price competition model with a multinomial logit demand system. Since we do not have estimates for the primitives of the demand in these models, we make several assumptions. 

Figure 2 reports the distribution of compensating efficiencies calculated for the Cournot model under different assumptions of elasticity of demand. Compensating efficiency estimates are heterogenous across markets and they are sensitive to variations in the assumed price elasticity of market demand. Yet, they appear to be sizeable. Indeed, independent of the assumed mode of competition and primitives of such models, there is a large proportion of mergers for which the merger-specific efficiencies needed to compensate for the merger’s unilateral price increases are substantial. Just as an example, in the Cournot model, with a conservative demand elasticity of two, the average (median) compensating efficiencies that are just enough to prevent the merger from harming consumers are 6.19% (4.54%) – i.e. the marginal costs of the newly formed entity should decrease by over 6 percentage points. Importantly, 41% of mergers require more than 5% marginal cost reductions, 26% of mergers require more than 7.5% marginal cost reductions, and 18% of mergers require more than 10% marginal cost reductions to not harm consumers. These results are also reflected by the findings for the other models we consider. They indicate that, in many mergers, compensating efficiencies appear to be too large to be realistically achievable. 

Figure 2 Distribution of compensating efficiencies: Cournot Model with different elasticities of demand 

Source: Affeldt et al. (2021)

We then look at what factors might explain the heterogeneity of computed compensating efficiencies across mergers within a regression framework. We find that the main determinants of their size are proxies of actual competition – as measured by the number of firms competing in the relevant market – as well as potential competition – as substantiated by the existence of substantial entry barriers. 

Again, as an example, using the Cournot model and assuming a demand elasticity of two, we show that a three-to-two merger increases compensating efficiencies by 1.2 percentage points compared to a merger that leaves five or more firms in the market. Moreover, if the Commission identified barriers to entry in an affected antitrust market, compensating efficiencies increase by 4.1 percentage points relative to markets without barriers to entry.

What are the policy implications?

These findings might be unproblematic if the Commission were to remedy or block all those mergers where compensating efficiencies are too large. To assess this point, we then relate the computed compensating efficiencies to the Commission’s decision. First, and independently on the chosen model and parametrisation, the average compensating efficiencies are the largest for prohibited cases, followed by phase 2 cases with remedies, while they are the lowest for phase I cases that are unconditionally cleared. This suggests that the Commission rightly uses its enforcement tools against the most problematic mergers, on average.

Second, we also try to identify whether the commission potentially committed enforcement errors. We define a potential Type I error to be a merger where the average compensating efficiencies across different markets affected by the merger are “low” and the Commission intervened (either remedied the merger or blocked it).5 A potential Type II error is defined as a merger for which the average compensating efficiencies are “high” and that was unconditionally cleared by the Commission. We use different thresholds for “high” and “low” compensating efficiencies: 3%, 5%, and 7%.6

Figure 3 reports our results for the Cournot model. For every parametrisation of the model, Type II errors appear to be much more frequent than Type I errors. For example, assuming a demand elasticity of two, we find that the Commission unconditionally cleared almost half (49%) of those mergers that would require compensating efficiencies above 5% and thus could be considered to be potentially anti-competitive, while it only remedied one out of ten of those mergers where average compensating efficiency are below 5% and thus could be considered to be pro-competitive. Again, we find similar results for other models of oligopolistic competition.

Figure 3 Type I and Type II errors for different thresholds of critical compensating efficiencies: Cournot model with different elasticities (ε)

Source: Affeldt et al. (2021)


While the exact numbers for the computed compensating efficiencies depend on the assumed modes of competition as well as the parametrisation of these models, our research highlights some patterns. Independent of our assumptions, compensating efficiencies for many mergers seem to be (too) large to be achievable in the real world. Moreover, although the Commission seems to rightly block and remedy most of the mergers for which compensating efficiencies are particularly high, it also appears to commit potential enforcement errors. Importantly, the rate of Type II errors (the Commission unconditionally clearing potentially anticompetitive mergers) is larger than the rate of Type I errors (the Commission intervening in potentially pro-competitive mergers). We read this finding as indicating that merger policy in Europe might have been underenforced. There might be good explanations for this finding such as high standard of proof that the Commission faces, the role of precedents, as well as the Commission’s limited resources. Yet, these results are particularly worrisome, especially because the European Commission can be seen as a relatively tough enforcer.7

The surge in literature on the mostly detrimental effects of increasing concentration for macroeconomic outcomes (e.g. Eeckhout 2020) witnesses a growing awareness on the central role of competitive markets. We add to this literature by showing that merger control – given unachievably high estimates of compensating efficiencies – appears to be too lenient and, thus, could be seen as one of the potential drivers to the observed trends.

Authors’ note: The information and views set out in this column are those of the authors and do not necessarily reflect the official opinion of the European Commission.


Autor, D, D Dorn, L F Katz, C Patterson, and J Van Reenen (2020), “The fall of the labor share and the rise of superstar firms”, The Quarterly Journal of Economics 135(2): 645-709.

Affeldt, P, T Duso and F Szücs (2018), “The EU Merger Control Database: 1990-2014”, DIW Data Documentation 95.

Affeldt, P, T Duso, K Gugler and J Piechucka (2020), “Market Concentration in Europe: Evidence based on Antitrust Markets”, CEPR Discussion Paper 15699.

Affeldt, P, T Duso, K Gugler and J Piechucka (2021), “Assessing EU Merger Control through Compensating Efficiencies”, CEPR Discussion Paper 16705.

Bajgar, M, G Berlingieri, S Calligaris, C Criscuolo, and J Timmis (2019), “Industry Concentration in Europe and North America”, OECD Productivity Working Papers, No. 18.

Bajgar, M, G Berlingieri, S Calligaris, C Criscuolo and J Timmis (2021), “Intangibles and industry concentration: Supersize me”, OECD Science, Technology and Industry Working Papers, No. 2021/12.

Baker, J B (2019), The antitrust paradigm, Harvard University Press.

Bighelli, T, F di Mauro, M Melitz and M Mertens (2020), “Increasing market concentration in Europe is more likely to be a sign of strength than a cause for concern”,, 13 October. ern

Cavalleri, M C, A Eliet, P McAdam, F Petroulakis, A Soares and I Vansteenkiste (2019), “Concentration, market power, and dynamism in the euro area”,, 24 August.

De Loecker, J and J Eeckhout (2018), “Global market power”, NBER Working Paper No. 24768.

De Loecker, J, J Eeckhout and G Unger (2020), “The rise of market power and the macroeconomic implications”, The Quarterly Journal of Economics 135(2): 561-644.

Diez, F, D Leigh and S Tambunlertchai (2018), “Global market power and its macroeconomic implications”,, 28 June.

Eeckhout, J (2021), The Profit Paradox: How Thriving Firms Threaten the Future of Work, Princeton University Press.

Gutiérrez, G and T Philippon (2018), “Ownership, concentration, and investment”, AEA Papers and Proceedings 108: 432-37.

Grullon, G, Y Larkin and R Michaely (2019), “Are US industries becoming more concentrated?”, Review of Finance 23(4): 697-743.

Kwocka, J (2014), Mergers, Merger Control, and Remedies. A Retrospective Analysis of U.S. Policy, MIT University Press.

Nocke, V and M D Whinston (2020), “On concentration screens in horizontal merger review”,, 26 August.  

Nocke V and M Whinston (2021), “Concentration Thresholds for Horizontal Mergers”, American Economic Review, forthcoming.

Philippon, T (2019), The great reversal, Harvard University Press.


1 Starting in the mid-2010s, mounting evidence shows that industry concentration increased for several decades in the US (e.g. Gutiérrez and Philippon 2018, Grullon et al. 2019, Philippon 2019) and, to a lesser extent, in the EU (e.g. Bajgar et al. 2019). These trends are observed across industries, sectors, and countries. Looking at better defined antitrust markets, the picture is less clear cut, but concentration appears to have increased on average, even when looking at this level of aggregation (e.g. Affeldt et al. 2020).To complement these findings, several other studies using firm-level data show that, over the same period, markups are also increasing (DeLoecker et al. 2021, DeLoecker and Eeckhout 2018, Bajgar et al. 2021) while the labour share is decreasing (Autor et al., 2021).

2 More specifically, some believe that increased concentration and markups are determined by the expansion of ‘superstar firms,’ which gained market shares due to their global scale, unmatched innovations, increased use of digital technologies, and the resulting increased efficiency (e.g. Diez et al. 2018, Bighelli et al. 2020). 

3 In the US, among the large horizontal mergers that needed to be reported to antitrust authorities between 2003 and 2012, second requests were issued in only 3.1% of the cases and fewer mergers have been blocked (Kwoka 2014). In the EU, around 94% of mergers are cleared without commitments, whereas only around 6% of mergers are allowed with remedies, and less than 0.5% of mergers are blocked or withdrawn by the parties (Affeldt et al. 2018).

4 In its horizontal merger guidelines, the Commission defines thresholds that should help identify problematic mergers. Accordingly, there is a safe harbor level of the post-merger Herfindahl–Hirschman Index (HHI) of less than 1,000 (“these markets normally do not require extensive analysis”). Moreover, the Commission is “unlikely to identify horizontal competition concerns” in a merger with a post-merger HHI between 1,000 and 2,000 and a change in HHI (also called DHHI) below 250 or a merger with a post-merger HHI above 2,000 and a ∆HHI below 150. See Guidelines on the assessment of horizontal mergers under the Council Regulation on the control of concentrations between undertakings (2004/C 31/03), p.7.

5 Notice that the use of remedies in mergers that are, on average, pro-competitive must not necessarily be an error. Indeed, even mergers that are on average pro-competitive might entail anti-competitive effects that might be rightly targeted by the remedy.

6 Nocke and Whinston (2021) refer to a presumption of 5% efficiencies as "rather" and "unduly" optimistic.

7 Indeed, several studies show that competition policy seems to be more strongly enforced in the EU than in the US and, consequently, industry concentration and markups are increasing in Europe less than in the US. This is the key argument in a book by Thomas Philippon (Philippon 2019) and is discussed by IMF experts (Cavalleri et al. 2019).



Topics:  Competition policy EU policies

Tags:  EU merger policy, European Commission, horizontal mergers, compensating efficiencies, merger policy

Research Associate, DIW Berlin

Prof. of economics, TU Berlin & Head of the department Firms and Markets, DIW Berlin

Professor of Economics, WU Vienna University of Economics and Business

Research Associate, DIW Berlin


CEPR Policy Research