January, 2017
Learning from past decisions in South Africa – By Phil Alves and Fatima Fiandeiro (Genesis Analytics)[1]
Competition authorities the world over regularly set out to review and assess their own performance. In South Africa, this work has focused heavily on the consumer welfare benefits of cartel prosecution. The South African competition authorities have dedicated relatively few resources to learning from previous investigations in an effort to improve decision-making, whereas such review activity is common in jurisdictions across Europe, North America, and Australasia.
Why spend precious time and (public) money trying to learn from the past? The simple answer is that it saves time and resources in the future by facilitating better decisions. Most of the decisions that competition authorities are required to make are subject to uncertainty, and thus rely on hypotheses or theories about future market dynamics and firm behaviour. Learning from the past can refine these hypotheses by producing better insights into the competitive dynamics of different industries.
This learning process should improve the efficiency and quality of decision-making over time, be it in merger investigations, identifying appropriate abuse of dominance remedies, or merely identifying where and how competition law intervention can generate the biggest impacts. This would seem to be an important priority given what is expected of competition policy in developing countries—development practitioners, donors and governments increasingly see its effective implementation as a means of generating inclusive growth, among other objectives.
There are a few ways in which competition authorities (and their critics) go about assessing their own decisions, some being more complex than others.
Econometric models can be used to try and determine whether a decision to approve a merger, for example, was ‘right’, by estimating the impact of the approval on prices. These modelling processes generate a great deal of additional insight into market dynamics through the two basic requirements of this approach: (i) the need to isolate the impact of the decision from all of the other forces acting on prices, and (ii) the need to identify what would have happened to prices had the authorities prohibited the merger.
These activities demand considerable resources, including the all-important availability of good-quality data. They also present considerable challenges, particularly when it comes to estimating what is likely to have happened but-for the merger. It is thus quite widely-acknowledged that these studies can generate results that are hard to interpret, and which can obfuscate the lessons the study set out to generate.[2] Greg Werden has gone as far as saying that, “[a]n inconvenient truth is that precisely estimating the tangible effects of a consummated merger might not be possible.”[3]
Some authorities have opted for simpler approaches. These don’t focus on trying to determine whether a decision was ‘right’. They focus instead on trying to identify, for example, the deficiencies in an authority’s understanding of how a market works, or whether an authority tends to ascribe too much or too little importance to a particular prediction or expectation, say in respect of the likelihood of entry post-merger.
Essentially, these approaches identify the handful of key predictions that drove a particular decision, and then test those predictions against the facts of what actually transpired after the decision was taken. Authorities can typically gather these facts through desk research and interviews—the resource demands need not be significant—but research instruments such as consumer surveys can be equally-well employed in this context.
For example, where an authority expected entry or the threat of entry to constrain a merged firm, it can revisit the affected market and assess the veracity of its prediction. In so doing it would likely learn more about the barriers to entry in that market, putting it in a position to make better decisions in that market in future. Or an authority could test its predictions concerning the strength and effectiveness of countervailing buyer power, or virtually any other component of standard merger analysis.
Hospital markets would provide a good testing ground in South Africa for this relatively simple approach. The Competition Commission (Commission) has opposed many mergers over the years in South Africa’s private hospital markets, with limited success. This was often because each merger individually was set to cause only modest increases in market concentration. Yet, this sector now exhibits high levels of concentration, and ongoing concerns over price increases and the presence of market power constitute some of the reasons for the Commission’s current Market Inquiry into Private Healthcare.
This leads to one obvious question—did the authorities define markets correctly when these mergers were assessed? It would not be the job of relatively simple ex-post assessment of previous decisions in private healthcare to answer such a question. But one can interrogate specific components of the market definition analysis. For example, did the authorities accurately anticipate the willingness of patients to travel to more distant hospitals in response to price increases, and did medical insurers require or encourage their customers to adopt such behaviour? The answers to these kinds of questions may inform a different approach to geographic market definition in the next private hospital merger.
United States and Dutch authorities have previously investigated these kinds of questions, and perhaps others too. While those authorities chose to investigate them using complex econometric studies aimed at trying to determine whether previous merger approvals were ‘right’ (the FTC was convinced they were ‘wrong’), authorities can assess them using the simpler approach too, for example through a consumer survey designed to reveal the actual responses of patients to small but significant price increases.
Mergers in private healthcare markets represent only one area where authorities can apply this approach. Wherever and whenever an authority has relied upon a specific or prediction, it can test that prediction, and refine it if needs be. As the New Zealand Commerce Commission has noted, “[t]he benefit [of this approach] is not in an attempt to determine whether the Commission got a decision right or wrong, but in ascertaining whether aspects of the market relied upon in that decision proved correct. This allows for the identification of deficient techniques that can be improved and good techniques that can be again relied upon.”[4]
Click here for a pdf version of this article.
[1] Genesis Analytics is a Johannesburg-based economics consulting firm with a leading competition and regulatory economics practice.
[2] OECD (2016), “Reference guide on ex-post evaluation of competition agencies’ enforcement decisions,” pg. 12.
[3] Werden, G. J. (2013), “Inconvenient Truths and Constructive Suggestions on Merger Retrospective Studies,” Retrospective Analysis of Agency Determinations in Merger Transactions Symposium, June 29, 2013, pg. 2.
[4] Csorgo, L., and H. Chitale (2015), “Targeted Ex Post Evaluations in a Data Poor World”. Organisation for Economic Cooperation and Development.
