Before we get into it this week, a quick note about our coverage. On the morning of November 30, it was formally announced that S&P Global was acquiring IHS Markit. We didn’t write anything about the deal that day. Or the next day. Or the day after that. Or the day after that. No, our story went live on Friday afternoon, New York time.

If you’re able to read what’s below this, that means you’re a subscriber (Or a subscriber sent this to you…you lousy cheapskate, you! But hopefully, you enjoy this, nonetheless). As you well know, the price of a WatersTechnology subscription is, uh, not cheap, to say the least. As editor, I believe that our job is to not provide you, the reader, with a knee-jerk reaction story, or a story that does little to expand on coverage that already exists at other outlets. Rather, I’d prefer to take a couple of days to write something that is more definitive and that you can’t find anywhere else.

We won’t always be successful in that aim, but that’s the goal. Also, Max Bowie will be providing some of his thoughts on the acquisition on Monday or Tuesday, depending on when the ol’ boy files. You can see some of my initial thoughts below, and as we learn more about the purchase, we’ll write more.

This is all to say that when there’s a major acquisition or some other form of major breaking news, we won’t always be first on the scene, but hopefully, ours ends up being the definitive piece. And if our story isn’t the definitive piece, please feel free to let me know what we’re missing: anthony.malakian@infopro-digital.com.

Land of Giants

Let’s first address the elephant in the room. I’m not going to go deep on the S&P/IHS acquisition, as Max is writing an opinion piece on the subject and I don’t want to cause him the anxiety of having to go one-on-one against my genius. So I’ll just go with some bullet points, but as I’ve written previously, the name of the game when it comes to M&A and tech development is creating a sticky ecosystem, from front-office trading to back-office operations. It’s a land of giants, and those giants are only getting bigger.

• As was mentioned in our original coverage of the acquisition, this deal—should it clear regulatory hurdles—might set the stage for another S&P acquisition. Because this was an all-stock purchase, it would seem that the data and index giant is purposefully keeping cash on hand. S&P Global CEO Douglas Peterson noted that S&P was able to maintain a strong balance sheet and credit profile to allow for “future capital deployment,” while CFO Ewout Steenbergen said the new company will generate annual free cash flow in excess of $5 billion by 2023 to use to accelerate organic growth or to pursue strategic M&A.
  
  If there is another acquisition in the near future, some people believe it will be a trading platform provider.
  
  “Looking at the combined business and who they’re going up against—LSE/Refinitiv, Bloomberg, Intercontinental Exchange—one of the things that’s different is that they don’t have any sort of trading platform. So maybe the next step would be to close that gap … if they wanted a more complete value chain,” says Tobias Sproehnle, CEO of index startup Moorgate Benchmarks. Sproehnle spent almost eight years at Markit between 2006 and 2014 as head of credit indexes and head of cash bond indexes, and his business partners Gareth Parker and Mark Pralle held senior index roles at S&P and Markit.
  
  As I’ve noted previously, I hate speculating about “who will be acquired next”—too often, you are proved wrong. I remember after Refinitiv was bought by Blackstone, there were several articles in the media that said that FactSet was surely the next to be dealt. Um, not so much.
  
  But, here’s the thing: the addition of a trading platform might make sense. S&P now has the pre-trade data and post-trade data; what they don’t have is where those meet in the middle. Yes, IHS Markit has the thinkFolio order and portfolio management system, which specializes in fixed income, cash management, foreign exchange, and loans, but to compete with the London Stock Exchange (assuming the Refinitiv deal goes through, which is looking likely at this point), the Intercontinental Exchange, and Bloomberg, S&P might need a more robust offering.
  
  For fixed income and credit derivatives trading, that likely leaves MarketAxess, Tradeweb, and MTS, maybe Trumid, BGC Group, and Trading Technologies (rumors about a Goldman Sachs acquisition have been in the news), and, of course, Bloomberg (sorry, I’m not seeing that one). Liquidnet also falls into this crowd, but they’ve already been swallowed up.
  
  Again, this is all speculation, but the whole point of the acquisition is to create a full front-to-back offering, and that might mean that there’s still some Christmas shopping (or mid-2021 shopping) ahead.

  

• “I think Mike Bloomberg wants to be Bill Gates—I’m not just talking about curing the coronavirus, I’m throwing a billion dollars at it. And in order to that, he needs to be liquid. It’s crazy to think that Mike Bloomberg is not liquid, but in the Bill Gates walk of life, he’s not. And he’s not young. It has to happen.”
  
  This comes from a source whom I trust a lot and knows the companies involved very well. Again, I do not see it, but it is being talked about by very senior executives in the industry.
   
• I should note here that sources have indicated that this is a good deal for both IHS Markit and for S&P Global. First, IHS Markit has in its DNA a history of major mergers, so overcoming those awkward culture issues might be easier. Additionally, from a product standpoint, there isn’t a ton of competition. For example, as one former employee of IHS Markit puts it, S&P’s loans business will complement what Markit already does; when it comes to indices, S&P is more qualitative and Markit is more quantitative; Markit is big in the software business, S&P is not; in the primary market, S&P focuses on research and ratings, while Markit is bigger in the IPO-issuance world.
   
• As we noted in our article, one of the more interesting components of this deal might be how Kensho utilizes and enhances Markit’s Data Lake.
  
  S&P bought the highly-respected artificial intelligence startup in 2018, and they are already producing sophisticated new research products for S&P users. IHS Markit has a very diverse business model, which means that it has a lot of unique datasets flowing into the company. You stick the data scientists at Kensho on that repository of information and they are likely to create some impressive analytics tools.
   
• IHS Markit owns MarkitServ, which has been described as the nerve center of the post-trade market in derivatives. In September 2018, MarkitServ launched TradeServ, a cloud-based platform that will eventually house its other platforms, including MarkitWire, Markit Trade Manager, and DSMatch. It was an impressive transformation overhaul, even as there have been delays in the rollout.
  
  Now, even before that 2018 launch, IHS Markit had announced it was going to sell MarkitServ, but revered those plans later in the year. The company had troubles selling the unit, hence why it backed off its initial plans. Does S&P view the TradeServ platform as a valuable piece of data-creating machinery that’s already done the hard work of transitioning to the cloud? Or is a sale back on?
   
• Last month we broke the news that Fidelity Investments is in the process of shutting down its Fidelity Corporate Actions Solutions (formerly known as Fidelity ActionsXchange) business line and its corporate actions data validation service, prompting an exodus of clients to alternative solutions.
  
  The two leading candidates to win those Fidelity customers, according to sources, are IHS Markit (IMActions) and FIS (XSP). The reason Fidelity got out of that business—again, according to sources—is because corporate actions processing is a low-margin business. The costs of operating in this space can be high—particularly the cost of acquiring corporate actions datafeeds from multiple vendors before even applying any proprietary value—and as the oldest provider in the space, Fidelity may have needed to make technology upgrades that simply weren’t worth the expense.
  
  This begs the question: Does S&P have an appetite for this low-margin business, or, being that IHS Markit is arguably the biggest player in this space, can S&P use that data to grow its own corporate actions business?

There are likely other questions that I’m missing, but these are my initial ponderings. I’d truly love to hear your thoughts: anthony.malakian@infopro-digital.com.

AML Mea Culpa

A couple weeks ago, my column was titled “Is AML Tech Worth the Cost?” The thesis was this: Perhaps anti-money-laundering compliance is a Sisyphean task, but one can’t help but wonder just how effective these platforms are at preventing money laundering. When vendors walk into a bank to sell these systems, they come equipped with headlines of massive AML-related fines, and say, “We can help you avoid these charges.” The small print that looms large, though, is that when the regulators come knocking, it’s going to be the bank and not the vendor on the hook for paying up.

It’s clear that there are flaws when it comes to AML tech, but perhaps I went too far in my assessment of these tools. A former chief information officer at a tier-1 bank wrote to me after reading my column and said that they felt my article was lacking due to “observational bias”. First, they say that banks are very sensitive about talking about successful examples of AML prevention for several reasons, not least of which are jurisdictional laws.

More importantly, though, was this:

“Huge fines for money laundering have very little to do with technology. Generally, huge fines are levied because management didn’t take AML seriously. Either they underinvested (people, process, technology) or they deliberately obscured the purpose for transactions. At one of the top-5 banks the fine was because the bank actively promoted the use of offshore accounts and ways to hide the source of funds. … Money launderers look for weak bankers—people that they can co-opt within banks to facilitate the process. Often when a big AML case is found, there is a common centralized theme—a specific branch or set of branches, a small group of bankers, a network of inter-linked accounts. They co-opt the people rather than the technology. Very rarely is the failure a technology failure.” 

They noted that it would be interesting to see how many of the 23 million transactions—in the case against Westpac in Australia—were flagged or would have been flagged if the AML technology was properly implemented and personnel were trained properly. They also wondered how many investigators were assigned to the transactions that were flagged. “It’s not unheard of for investigators to be so overwhelmed that ‘aged’ alerts get purged rather than investigated. That is, an investigator gets 1,000 alerts a day and has the ability to investigate 200. After 30 days, the backlog is so great that anything older than 30 days is purged. This is a management decision, not a technical problem.”

And, as always, there’s a training data problem.

“Let’s face it, if my training set of data contains 23 million cases of money laundering that are misclassified as legitimate, why would we expect any machine learning algorithm to work? One of the weaknesses of algorithms that are based on data—whether ML or statistical—is that bad data will lead to bad parameters. And many banks don’t want to clean the data, because it can only open up more investigations. That is, cleaning the historical data will uncover more false negatives; presumably they’ve already dealt with the false positives via investigating and closing the investigation as spurious.”

Further to that point, ML needs high-quality data—bad data in, bad data out. Because banks have grown through acquisitions and mergers, there’s always a fight against dirty and fragmented data. You start running ML algos on that data to prevent money laundering, and it’s likely going to produce mixed results of false positives and negatives.

And here’s maybe the most important point of all:

“Here’s the problem for banks: They are required to fight AML; they aren’t compensated for it. Legitimate customers get offended when they are treated like potential criminals—either during the account-opening process (KYC) or during an investigation. Banks have lost legitimate customers because of how an investigation is handled; and remember, these are the same overworked, under-appreciated staff who get yelled at by revenue generators within the bank. The volume of money laundering transactions in a bank versus the legitimate ones is well less than 1%—probably less than 0.01%. The tech solutions help. They do find legitimate cases but given the ratios (0.01% actually illegitimate) they find many more false positives. Without the tech it would be impossible to scan the transactions at all.”

The source did note, though, that they do wonder if some banks spend on AML tech so that the bank can show that they are trying to fight instances of money laundering so that the next fine is lower. “It helps with negotiation.”

There’s a problem to be solved, for sure, but perhaps it rests more at the bank executive level and among the regulators. As always, I value your feedback, so keep it coming.

The image at the top of the page is Paul Gavarni’s “Gulliver Awed by Three Giant Beggars in the Land of Brobdingnag” courtesy of the National Gallery of Art.