This browser is not actively supported anymore. For the best passle experience, we strongly recommend you upgrade your browser.
| 6 minutes read

Big data in Merger Control Part 3 - Lessons from the coal face

In this series of posts, we are focusing on the increasingly important topic of Big Data and its role in merger control assessments. Our previous posts set out what we mean when we refer to Big Data and what the competition authorities are looking to understand from Big Data. We defined structured versus unstructured data and considered the importance of layering facts and context when interpreting the information in front of you.

We now pick up on the challenges of responding to expansive data requests from competition authorities, including when best to engage with Big Data and the importance of providing correct and timely responses to information requests. 

Authorities have statutory powers (e.g. s109 of the Enterprise Act 2002 in the UK) to ask for documents and structured data when a relevant merger situation is created. Submitted information plays an important role in the assessment process. Timelines can vary but, in the case of the CMA, these are often short (typically a few weeks) and failure to comply where there is no reasonable excuse can lead to financial penalties of up to £30,000 and/or a daily penalty of up to £15,000. Compared to the European Commission’s fine of €110m levied on Facebook for providing misleading information during its acquisition of WhatsApp, the levels remain low. However, the CMA can also “stop the clock” and can even reject the parties’ merger notice, both of which can lead to delays.

As we covered in our previous post, competition authorities are increasingly focusing on internal documents and structured data to understand the merging parties’ motives for the transaction, rivalry between the parties and their competitors, broader competitive dynamics, and likely changes in the market.

So, when does Big Data come in? The CMA’s Guidelines on internal documents suggests that at Phase 1, merging parties are unlikely to have to provide ‘material volumes’ of additional internal documents beyond those specified in questions 9 & 10 of the merger notice. These constitute documents received by board members or senior management covering merger rationale and analysis of competitive conditions – including reports, presentations, studies, internal analyses, industry/market reports or analysis, customer research and pricing studies. With some assistance from advisors, it is therefore usually straightforward for the parties to locate these documents themselves.

However, in our experience, requests at Phase 1 are increasingly becoming too burdensome for the merging parties and data experts are often required. This is particularly the case for local retail mergers where catchment area / drive time analysis is required to understand overlaps and local market shares. In some cases, a focused data analysis exercise may also help advisors get to grips with the business and market dynamics and advance some early arguments on the lack of competitive effects (e.g. detailed analysis of pricing or product positioning).

If a Phase 2 investigation is likely, then it is best to engage data experts as early as possible. Our experience of Phase 2 cases is that the competition authorities expect data requests to be responded to very quickly, typically within 2-3 weeks. This can prove very challenging, even for larger businesses with very large IT and business intelligence teams. After all, most businesses do not have teams whose job it is to respond solely to ad hoc regulatory requests. Our experience is that the IT system discovery phase, i.e. working out what systems are (and were) in use, what data they contain and how this can be obtained, should start as soon as possible, ideally before the information request has been received. External advisors will be able to predict the majority of what the competition authority will request, so the IT system discovery can remain focused, and not be overly burdensome or expensive for the business, whilst also allowing the external advisors to prepare for the receipt of the information request.

Internal knowledge is critical

For the collection of data, client IT staff are your best and most precious asset. A key challenge in collecting, interpreting and analysing data is the passage of time, as often both current and legacy systems have to be understood before key information can be identified and extracted. External advisors can leverage their experience to work alongside already busy IT teams who may not be familiar with responding to such time-critical regulatory requests. Documenting key decisions and validating assumptions as you go, including the harmonising of definitions from one system to another, is essential to both ensuring nothing is missed and to provide transparency to the competition authority.

In the UK, the CMA’s document requests tend to be very open-ended, so identifying the starting point for analysis is a challenge in itself. For example, custodians and search terms are not specified, which can obviously leave room for interpretation and potential disagreement. The CMA might provide comments (and in our experience be willing to engage on the detail) but the burden remains with the parties to provide an appropriate response. Unlike in the EU, where hundreds of thousands of internal documents can be routinely handed over without human review (subject to any privilege review, of course, where they are “responsive” to a prescribed list of search terms), the CMA prefers to receive a very targeted set for each request category. Reducing a potentially large population to only the key documents requires effort and flexibility, especially as legal advisors can often still be in negotiation with the CMA as to what will be handed over, right up until the last few days of the deadline. Careful consideration needs to be given to search terms, striking the balance between making these comprehensive enough to capture potentially relevant documents, versus the risk of producing huge swathes of documents containing a large proportion of false positives.

The use of analytics under its many guises is an acceptable means of identifying categories of documents which can be excluded from the review population, but each of these needs to be carefully considered and clearly explained in the methodology. Common analytical techniques include: email threading (removing messages within an email conversation that do not contain unique content), near-duplicate analysis (comparing similar documents based on a percentage similarity), and predictive coding (using machine learning to predict relevance within a document population, based on reviewer input). 

Non-English language documents, encrypted files and the presence of personal information are among the challenges that also need addressing. Machine translation, password harvesting and pattern recognition/auto-redaction can all help with these respective challenges. Advisors should also, as soon as possible, verify the completeness of the collection via some form of ’gap analysis’ across the custodians’ data (e.g. to help identify the instance where an email system failed to back up or where a potential data source was missed from the collection). It is far better for the parties to proactively identify that a gap exists than for the competition authorities to discover it themselves.

On the structured data side, the requests from the CMA tend to be more closed and specific, albeit they will likely require some interpretation as they may have implicit terms within them that the business does not define (e.g. “repeat customer”, or “product hierarchies”). Two trends we have observed recently are: the CMA being very focused on the internal consistency of responses, where their questions overlap or repetition exists; and the CMA expecting that responses across the two (or more) parties are answered on a consistent basis – i.e. wherever possible using common definitions of, for example, margins, revenues or even reporting periods.

Keep communication flowing

Finally, a key element of the entire process is to maintain an open dialogue with the competition authority. If a data methodology is required up front, any subsequent deviations in search strategy or assumptions must be communicated and the end-product must reflect what actually happened. We are already seeing a rise in parallel investigations between the UK and other jurisdictions, each with distinct requirements, and so the challenges repeat themselves.

In summary:

  • Requests come thick and fast during the very short statutory window that competition authorities have to investigate mergers.
  • Choose a search methodology that is responsive enough to capture documents, while also practical in light of tight deadlines (and agree it with the authority).
  • Have quality control processes in place to ensure consistency across responses.
  • Maintain an open and transparent dialogue with the authority about responding to their requests (and what is realistic).
  • Do not underestimate how much effort can be involved in responding to requests for information (and clients need to be warned and line up internal and external resources - it is never too early to start engaging with your trusted economist and data forensic specialists).

In the final part of this series, we will explore the techniques advisors can use to turn the mountain of data and documents into targeted proactive analysis. It is becoming increasingly important to understand helpful and unhelpful messages early so that advisors can focus the competition authority’s attention or be front-footed with regard to the further work required.

The 'Big Data in Merger Control' series: 

Tags

regulation, legal, merger control, big data

Related Insights