The identification and quantification of horizontal merger effects is one of the main concerns of antitrust authorities. The reason is that horizontal mergers may enhance and/or strengthen market power which in turn could facilitate firms to increase prices, reduce output choice or quality and deter innovation at the expense of consumers.† Since competition authorities generally delineate welfare as consumer surplus, the previous merger effects are perceived as welfare detrimental. A major concern to policy makers is then the availability of effective analytical and quantitative tools that make possible to identify whether and to what extent a merger will induce market power.
Responding to these concerns, a growing body of models of competition that investigate for horizontal merger effects have been proposed in the literature of industrial economics. On its side, the literature of applied industrial economics has also considerably advanced in providing quantitative methods that answer the analytical complexity of the theoretical models. In turn, competition agencies have increased reliance on such academic contributions to merger analysis and updated their modes of investigation and implementation. For instance, in 2004, the European Community Merger Regulation (ECMR), has undergone into a reform that modifies the (previously more restraint) scope of investigation for merger cases. In particular, the criterion of market dominance has been enlarged to include any form of dominance (not only collective dominance) as well as cases that do not involve dominance but still entail anticompetitive concerns. This restructuring of merger regulation takes concrete form in the accompanying Horizontal Merger Guidelines (HMG) which state a list of factors that have to be analyzed in merger cases. Among them, two major factors are: the degree of market concentration and the possible anticompetitive effects of the merger.
For the analysis of market concentration the HMG stipulate applying the Herfindahl Hirschman Index (HHI) and its post-merger change besides the combined market share levels of the undertaking firms. For the assessment of a possible anticompetitive harm, the HMG do not stipulate specific tools but state that a “but for” analysis is required. Such analysis is actually interpreted as a quantitative projection of the (short-term) price change as a result of the merger. Thus, merger simulation models that perform such a price projection are implicitly call and indeed put into practice more and more frequently. More specifically, whereas the HHI aims at determining market power in terms of concentration, merger simulation aims at predicting market power through the so-called merger unilateral effects, i.e., the increase in prices due to the merger.
In practice, thresholds for market shares and the HHI are systematically applied to assess horizontal merger cases whereas the merger simulation model is mostly implicitly called during the investigation process. The reason for this preference towards the HHI is that, in terms of data, the HHI only requires the firms’ market shares of the market under analysis. The simulation model, in turn, requires further information about the market’s demand, and not necessarily but preferable, about the undertakings’ cost structure. In terms of implementation, the HHI entails a simple calculation whereas the simulation analysis requires structural econometric estimation and therefore further time and technical abilities. As a result, the European competition authorities still hesitate in going forward with the more structural economic approach of the simulation model and mostly rely on the HHI criteria to close horizontal merger cases.
Nevertheless, the simplicity of the HHI does not come without a cost. It can accurately predict the competitive effects of a horizontal merger only if it takes place in a market of homogeneous products. In fact, within such a market structure the predictions of the two tests coincide because the measure of market power, the mark-up, resulted from the structural analysis is proportional to the HHI. Yet, how much the two methods converge or diverge according to changes in different factors of more general merger models remains to be systematically investigated. In particular, in markets of differentiated products, the focus on market shares and concentration is problematic and merger simulation is predominantly useful. When the analysis deals with such markets, there is a tradeoff between the simplicity and accuracy of these two market power tests. Indeed, a still open question to analysts and practitioners is whether the two types of measures should be substitutes or complements.
In this study, we intend to answer under which circumstances one test should be preferred to the other. That is, we aim at determining under which conditions the various strands of the merger simulation approach, may bring value added, in light of the market power test in markets of differentiated products, with respect to the traditional HHI test. In particular, we want to answer to questions like: How accurately can an impediment to competition be estimated by the HHI and the simulation quantitative tools? Is simulation analysis a more appropriate tool for accurately predicting anticompetitive merger effects? Does the new substantive test effectively reduces the probability of incurring on errors type I and II (prohibiting a pro-competitive and clearing an anti-competitive merger, respectively) compared to the previous dominance test?
To address these questions, we propose a generic methodology that evaluates the performance of both market power tests and allows to critically asses the advantages and drawbacks of the simulation analysis relative to the concentration test. Our methodology consists in constructing and econometric workbench in which we implement the two tests and subsequently compare their results. We do so, by generating market-level data of a differentiated products’ industry that supplies to heterogeneous consumers. In this true economy a merger takes place and its effects in terms of HHI and price increases (unilateral effects) are measured. Then, with the generated data of the true economy, an approximate economy is estimated to recover the actual equilibrium of the market (as it is usually done in the case-by-case analysis). In this approximate economy, we estimate the unilateral effects with the merger simulation model. Then, by correlation analysis we compare the relationship between the resulted predictions of the two market power tests. Our results confirm that the HHI test tends to be upwards biased compared to the test of unilateral effects. These results are sensitive to the choice of the size of the market, or more precisely to the size of the market share of the outside good (composed of the goods that do not belong to the actual market under scrutiny). In particular, when the size of the outside good is small, a prediction based on the HHI test is roughly in accordance with a prediction based on the unilateral effects test. On the other side, when the size of the outside good is large, a prediction based on the HHI test is different from that based on the unilateral effects test. That is, in the latter case, the increase in prices and in HHI, induced by the merger, are not related to each other. Consequently, a decision based on the dominance test, HHI would have nothing in common with a decision based on the substantive test of unilateral effects.