The Business of Benchmarking

Plus: Canadian Banks, Visa/Plaid II, Lufax

Welcome to another issue of Net Interest, my newsletter on financial sector themes. If you’re new here, thanks for signing up. Every Friday I go deep on a topic of interest in the sector and highlight a few other trending themes below. If you have any feedback, reply to the email or add to the comments. And if you like what you’re reading, please spread the word.

The Business of Benchmarking

“Yes, we think of ourselves as a benchmark company. I mean, data is in vogue now, and people are really kind of a bit obsessed with data, data companies… I think data is nice, it’s interesting. But if you could turn something to a benchmark, it really transcends data.” —Senior Vice President of Investor Relations, S&P Global, November 2020. 

It’s easy to miss, but under the arches by the Senate bus stop in Paris, across the street from the Palais du Luxembourg, you can find this:

Credit: Allison Meier/Atlas Obscura

The “mètre étalon” is one of sixteen metre rules that were installed throughout Paris shortly after the metre was first established as a unit of measure in 1791. Before then, measurement varied hugely from place to place. People had moved beyond subjective measures such as “a stone’s throw” but not by much. The aune was used as a measure of length for cloth, yet it varied according to the fabric (a silk aune was less than a linen aune) and across France there were over 17 different aunes.

The lack of standardisation was more than a quaint curiosity; measurement was used as a power play. As James Scott describes in his book, Seeing Like a State, a common measure of length, the ell, was used to mark off areas of land to be plowed as part of feudal dues. It was hard for collectors to increase the rate of feudal dues directly but what they could do was try and nudge up the length of the ell—move the goalposts as it were. Attempts to simplify or standardise measures across France were regularly corrupted. 

Three factors ultimately coalesced to allow a “metrical revolution”. One was the growth of market exchange which encouraged uniformity in measures. A second was popular sentiment and Enlightenment philosophy, which favoured a single standard through France. And third, the Revolution, and especially Napoleonic state building, enforced it.

According to Scott:

The metric system was at once a means of administrative centralisation, commercial reform and cultural progress. The academicians of the revolutionary republic, like the royal academicians before them, saw the meter as one of the intellectual instruments that would make France “revenue-rich, militarily potent, and easily administered.”

The job of establishing national measurement standards fell to the Académie des Sciences. They initially defined the metre as one ten-millionth of the distance from the equator to the North Pole. A nice round number, steeped in earthly significance. But of little use to the tailor who wants to measure off a roll of cloth. Once precise measurements of that distance had been surveyed, French authorities smelted a platinum metre bar to serve as an official benchmark. To popularise its use, replicas were installed at various places around Paris. The one in the picture above, by the Palais du Luxembourg, is the last remaining replica in situ. 

The platinum reference bar was maintained for ninety years until a new one was made out of a platinum alloy, its length measured at the melting temperature of ice to avoid heat expansion. In 1960, the metre was refined to reflect precisely 1,650,763.73 wavelengths of a specific frequency of light emitted by an atom of krypton-86. And then, in 1983, it was refined again, to become the metre we know and love today: “the length of the path travelled by light in vacuum during a time interval of 1/299 792 458 of a second.” [1]

Uniform measures became a lynchpin of economic development. While parochial measures were fine for trading locally, large scale commercial exchange was only possible with standardisation. 

Indeed, throughout history, growth has been fuelled by standardisation. Measures of distance, weight and money provide the foundation but it goes beyond that. In 1946, a conference was held in London to discuss the formation of a global organisation to agree on a range of standards. Out of this conference the International Organisation of Standards, or ISO, was born. It sets standards for all sorts of things. For example, international standard ISO 68-1 lays down design principles for screw threads so that they are consistent around the world. 

The best example of standardisation in recent history is the shipping container, invented in the 1950s by Malcolm McLean. Like the standard bearers of pre-revolutionary France, McLean was confronted with resistance by parties with competing interests—unions, regulators, port authorities, railroads. But he persevered, building a business – SeaLand Industries – around the innovation. Even though his inventions had patent protection, McLean chose to make them available by issuing a royalty-free lease to the ISO. This, and sponsorship from the US military during the Vietnam war, helped standardised shipping container technology gain widespread adoption. By the end of the century, 90% of the world’s trade cargo was being transported using container shipping.

In 1969, McLean sold SeaLand Industries for $160 million. Yet the value created by his invention is an order of magnitude greater than that. Having cost around $420 to ship a tonne of goods across the Atlantic before his time, that cost is now below $50. These efficiency benefits have been widely distributed. In the trade-off between his slice of the pie and the size of the pie, it is interesting to consider how much money McLean may have made had he chosen to retain royalties. 

In all of these cases, it doesn’t really matter which standards ultimately prevail, so long as they are widely accepted. Shipping containers are measured in twenty-foot equivalent units (TEUs) but the number is fairly arbitrary. The metre works well, but so does the foot. In other cases, convergence may not even alight on the best alternative. The videotape format wars between VHS and Betamax are a case in point. Betamax was largely considered the better standard: it had higher resolution, slightly superior sound, and a more stable image. Yet JVC, which designed the VHS technology, licensed it to any manufacturer that was interested and it won out to become the de facto standard. 

Standardising Financial Services

Standardisation is no less relevant in financial services. Consider financial derivatives. Derivatives have been around for a long time. The first evidence of one is in a document dating from Hammurabi’s reign in Assyria, authorising the bearer to receive in 15 days in the City of Eshama on the Tigris 8½ minae of lead deposited with the Priestess of the Temple. The terms of this contract didn’t really catch on. However, over time contract terms became more standardised and when that happened, derivatives trading took off. 

All derivatives are tied to an underlying commodity and in the case of equity derivatives, that means an index. Stock indices emerged as a way to measure the market. Just as metres are used to measure length and TEUs are used to measure shipping container capacity, index points are used to measure stock markets. 

Originally, newspapers took on the role of defining indices, as the Académie des Sciences had the metre. The first index was the Dow Jones Transportation Average, published by Charles Dow in 1884. It was followed by the Dow Jones Industrial Average (DJIA) a few years later. Like the VHS to come, Dow Jones didn’t produce the best standards. The DJIA includes only 30 companies, not the whole market, and simply adds together the stock prices of its constituents rather than weighting them. But it had a broad enough distribution that the index became entrenched and, over a hundred years on, people still talk about Dow points as a measure of the market. 

The model the newspapers employed was somewhat different to McLean’s or the Académie’s. They gave the measure away for free to their readership to secure its position as a standard, and then they made money licensing their indices to exchanges as the basis for derivatives contracts. It was consistent with their core operating model: to sell attention.

Unfortunately for the newspapers, they underestimated just how much attention indices would attract. Over 2009 and 2010, Dow Jones sold most of its indices to CME Group for less than $1 billion. The Financial Times sold its 50% stake in FTSE to the London Stock Exchange for £450 million in 2011. Although these are big numbers compared with their core businesses (what was left of the Financial Times went on to sell for £844 million three and half years later), their buyers got excellent deals.

That’s because of the unwavering growth in passive asset management, whose business model feeds off an index. In the US, the passive share of assets under management is now at 50%; in Japan it’s over 70%. Exchange traded funds in particular have been a boon for index providers. They pay a fee to the index provider that is typically linked to the volume of assets benchmarked. When the London Stock Exchange bought control of FTSE in 2011, $5 trillion of assets globally were benchmarked to it. The exchange has since also bought the Russell group of indices, but in combination $16 trillion of assets are benchmarked today. 

Enter S&P Global

One publisher that went in the other direction was McGraw Hill, the owner of the S&P family of indices. In 2011, it spun off its publishing arm to focus on financial information. In 2012, it merged its indices with the Dow Jones ones bought by CME Group, but it retained control. Then, in 2016, it changed its name to S&P Global. This week it announced a $120 billion merger with IHS Markit. [2]

The merger brings together the #2 index player in the world by revenue with the #8 player, allowing them to leapfrog MSCI into #1. Combined, they will have over $1 billion in index revenues, representing around a quarter of the market. S&P’s strength is in its equity indices, while IHS Markit’s is in fixed income. Driven by exchange traded funds, fixed income indices are gaining increasing traction (as discussed in prior Net Interest pieces on Bloomberg and on ICE). 

The index market is already very concentrated. After S&P and MSCI, there’s FTSE Russell, which is owned by the London Stock Exchange. Between them, the three have ~70% of the market. Competition could emerge from the large passive managers themselves, but existing indices are pretty entrenched. Years of reading about them in the Financial Times and the Wall Street Journal leave asset owners very comfortable.

S&P Global’s index business sits alongside its original business which is credit ratings. On the face of it they look like quite different businesses. In ratings, companies and governments pay for an evaluation of the credit risk associated with debt they issue. But across both, S&P is selling a standardised metric. In the case of ratings, it’s a measure of credit risk—across names and over time. [3]

Just as indices have become embedded in the workflow of passive asset management companies, standardised credit ratings have become embedded in the workflow of banks and fixed income investors. In the case of both credit ratings and indices, the value of the benchmark is in its widespread adoption. An S&P credit rating and the S&P 500 index both provide what the metre does—there may be other metrics, but their value is diminished by not being universally acknowledged.

For the same reasons, the market is very consolidated. In ratings there are only two main players, Moody’s being the other; they have ~40% of the market each. The structure of the broader ecosystem is different to the index business, though, where an oligopoly of index providers faces an oligopoly of customers. Here a duopoly of credit rating agencies faces a fragmented customer base. Consequently, they retain considerable pricing power, able to raise prices by around 3-4% a year.

The resilience of their business model was underpinned by their performance following the financial crisis. Despite coming under intense scrutiny for their role in the crisis, their position was not neutered. Their ratings on structured credit were wildly wrong, but as we know from the DJIA and VHS, it’s not about being right, it’s about being the standard. 

Going forward, S&P Global has a huge opportunity in ESG. It’s a big theme in asset management as firms look for exposures to be standardised and benchmarked. S&P Global has bought several new datasets linked to ESG. This has led to the creation of new indices, but it also provides an opportunity in its ratings business as companies seek to have their ESG credentials rated alongside their financial debt. 

Other Financial Industry Measures

The quote at the top of this piece, attributed to S&P Global’s head of investor relations, reflects what S&P does with data. It takes credit information, stock prices and other data sources and turns them into benchmarks—uniform measures that are widely legible.

S&P Global and its peers are not the only financial services companies to have done this. Another is Fair Isaac Corporation, the purveyor of the FICO score. It’s one of the first data science companies, founded in 1956 by Bill Fair, an engineer, and Earl Isaac, a mathematician. The company maintains an algorithm that spits out a credit score. The underlying data is managed by the credit bureaus, who license the algorithm from Fair Isaac.

The FICO credit score has become the industry standard for consumer creditworthiness. Like credit ratings and indices, it is a) ubiquitous, b) embedded in end-users’ workflows and consequently c) commands a very high market share, in its case 90%+ in B2B credit scoring. Fair Isaac recently exerted its market power to push through price increases after years of stability. Given its take represents such a small part of the costs associated with a mortgage application, the price increases went through seamlessly. With little incremental cost attached to generating scores, operating margins have recently pushed up above 85%. 

Like in the index business, the risk is that Fair Isaac’s customers – the credit bureaus – cut it out of the picture and churn out their own model. They did that, launching VantageScore in 2006. But it gained little traction, despite assertions that it had better predictive power. Yet again, it’s not about being right!

A related risk is that end-users develop better algorithms themselves. Many fintechs are showcasing credit models they believe outperform. In its S-1, Affirm presents a chart showing how “for any given consumer sub-segment, our model produces lower risk outcomes when compared to FICO.” Same at Upstart. Although in order to do the comparisons, they still need to be pulling FICO data. And in the meantime, Fannie and Freddie, who underwrite more lending decisions than anyone, are firmly committed to the old standard. 

Universal standards are usually unassailable. The risk for companies that manufacture them is less that their moat is crossed and more that their castle becomes irrelevant.

One threat to the castle could come from regulators. Right now, the credit rating agencies and FICO enjoy regulatory favour. But when it turns, it turns. The French don’t count their wealth in French Francs anymore and LIBOR is on its way out as a benchmark of interest rates on loans, bonds and derivatives (albeit on a longer timetable than anticipated).  

The other threat is that the market renders the castle irrelevant.

Arthur Charles Nielsen released his first ratings for radio in 1947 using “black boxes” installed in people’s homes. Before that, radio audiences were measured using telephone polling. Nielsen’s approach provided a more quantitative measure of attention capture, which allowed broadcasters to place a more accurate value on their airtime. When TV took off, Nielsen deployed his ratings technology there. His measures became the de facto standard for measuring attention. In the late 2000s, over $70 billion of TV advertising spending per year was linked to his benchmarks. [4]

The Nielsen company IPO’d in 2011, after having spent four years in the embrace of private equity. Its S-1 makes a single reference to video-on-demand (and none to streaming):

Traditional methods of television viewing are changing as a result of fragmentation of channels and digital and other new television technologies, such as video-on-demand, digital video recorders and internet viewing. If we are unable to continue to successfully adapt our media measurement systems to new viewing habits, our business, financial position and results of operations could be adversely affected.

Until that day comes, there’s scope for benchmarking businesses to remain very profitable. Goodhart’s Law states that when a measure becomes a target, it ceases to be a good measure. A variation may be that when a measure becomes a target, whoever owns the intellectual property has a license to print money. 

Thanks to David Kim a.k.a. Scuttleblurb for his amazing insights on the market infrastructure industry. If you don’t already subscribe to his work, I would strongly recommend it. 

[1] There’s an interesting story why the metric system never made it to America. In 1793, Thomas Jefferson requested artefacts from France that could be used to adopt the metric system in the United States. Joseph Dombey was dispatched from France, bringing with him a standard kilogram. However, before reaching America, Dombey's ship was blown off course by a storm and captured by pirates. He died in captivity in Montserrat and the kilogram was auctioned off with the contents of his ship.

[2] After Markit was founded in 2003, it attracted investment from banks such as Bank of America, Deutsche Bank, Goldman Sachs and JPMorgan. Banks have similarly helped to seed many other companies which operate at the fringes of tech and finance. These include Visa, MSCI, ICE, MarketAxess, Bloomberg and more. In fact, banks may have created more value outside their legal structures than within.

[3] A number of standardised measures trace their history to the railroads. These include railroad gauges, obviously. Also, time. And credit ratings: to help investors navigate the sector, Henry Varnum Poor published information about the financial and operational state of US railroad companies.

[4] In his book The Attention Merchants, Tim Wu quotes Robert Elder, the inventor of Nielsen’s “black box” as saying that broadcasting “suffers greatly from the misuse of the [ratings], and for that reason I am not too happy about my part in getting it started.”

More Net Interest

Canadian Banks

These days most companies have a December year end. Some, like in India, follow the government’s fiscal year so may have a financial year ending in March. Others choose a date that tracks their seasonality—US investment banks used to close their books in November so they could award a year end bonus that coincided with the calendar year.

Canadian banks are weird. Their financial year ends in October, which doesn’t map with the government’s fiscal year, nor the personal tax year, nor anything. The reason dates back to 1965, when banks agreed to shift their reporting period as a favour to overworked accountants, who were typically swamped between January and April. 

What it means though is that when Canadian banks report, we get a little peek into the future. They have just reported their full year 2020 numbers, including results for October, which we will have to wait nearly two more months to get from the US banks. And the news is good!

Credit was especially favourable. TD Bank reported loan loss provisions of C$917 million for the fourth quarter, versus consensus expectations of C$1.5 billion. Royal Bank posted provisions of C$427 million against expectations of C$740 million. In most cases, deferred loan balances came in lower than prior quarters. As the Chief Risk Officer of Royal Bank said on his earnings call: “Overall, the macroeconomic environment has proven more supportive than originally forecasted at the onset of the pandemic, due in part to the extent of government support programs, which resulted in better-than-anticipated credit performance this year.”

However, they don’t reckon they’re out of the woods yet (although they are a traditionally conservative bunch). Royal Bank upped the weight of its downside scenario in its planning and TD’s CEO said that “we’re really going to see impairments start in the second half of the year… until the stimulus ends and we start seeing the cash balances come down, we’re not going to see the impairments… We could see a situation where PCLs [provision for credit losses] remain low for a couple of quarters and then start rising.”

Visa/Plaid II

The Department of Justice case against Visa in its proposed acquisition of Plaid was the subject of More Net Interest a few weeks ago. Visa filed its defence this week. The DOJ’s case rests on the notion that Visa is a monopolist in debit cards and wants to extinguish Plaid as a potential threat. Visa comes back firing:

“Plaintiff's narrative… is nothing more than a patchwork of excerpted party documents and testimony taken out of context, stitched together with conclusory allegations where facts do not exist, and embellished with irrelevant and stale customer complaints unrelated to Plaid and this acquisition.” 

Visa doesn’t dispute its market position in debit, but argues that the market it competes in is much larger than that (chapter three of Zero to One). It also cites Mastercard’s acquisition of Finicity as an analogous deal which didn’t get the same degree of scrutiny (the “it’s not fair” defence). And it argues that Plaid is nowhere near being ready to compete with Visa if left to its own devices: “Plaid’s pipeline products having no consumer awareness or merchant adoption… no experience in the payments space, being nowhere close to having the requisite feature set to operate a network… and having no reasonable path to developing all of the necessary features of relationships.”

It’s an interesting case and, for an investor, cases like this are valuable for the unique access to internal management strategic thinking they provide. One of my favourite resources when researching UK names is the Competition and Markets Authority website. Case management conference is set for 18 December. Bring it on!

Lufax

So Ant Group didn’t make it into the public markets. But Lufax did. Although it’s listed in New York rather than Shanghai/Hong Kong, it has nevertheless attracted the attention of Chinese retail investors. Since listing at the end of November, it has been a highly volatile stock. In particular, activity seems to ramp at around 4am ET each day, signalling Chinese retail involvement. The volatility is perhaps unsurprising given the regulatory uncertainty swirling around non-banks in China. On their call, management reflected, “post-IPO, our biggest challenge is to get a clear view of what regulators are thinking about.” But volatility creates trading opportunities and in the absence of Ant, this is the next best thing. (h/t Josh)