Fake Accounts and Artisanal Data

Matt Levine is a Bloomberg View columnist. He was an editor of Dealbreaker, an investment banker at Goldman Sachs, a mergers and acquisitions lawyer at Wachtell, Lipton, Rosen & Katz and a clerk for the U.S. Court of Appeals for the Third Circuit.
Read More.
a | A

Wells Fargo.

This turns out to have been an awkward thing for Wells Fargo Chief Executive Officer John Stumpf to have said about Carrie Tolstedt, Wells Fargo's head of Community Banking, when she announced her retirement in July:

“A trusted colleague and dear friend, Carrie Tolstedt has been one of our most valuable Wells Fargo leaders, a standard-bearer of our culture, a champion for our customers, and a role model for responsible, principled and inclusive leadership,” said John Stumpf, Wells Fargo’s chairman and chief executive officer.

It turns out that Wells Fargo's community banking culture involved creating millions of fake accounts for customers to satisfy the bank's frenzy for cross-selling products and services. And Tolstedt now gets to bear the standard for that culture, as the Senate investigatesFortune reports that "she will be walking away with $124.6 million in stock, options, and restricted Wells Fargo shares," and shareholders have called for her to be held responsible for the fake accounts by clawing back her pay:

Another investor said: “If this person presided over this, why no accountability? We have share-based pay so that it can be clawed back when people have been earning bonuses under false pretences, and if fraudulently opening client accounts isn’t false pretences, then I don’t know what is.” 

Wells Fargo's cross-selling scandal is so odd because it is both at the absolute core of the bank's business, and also curiously irrelevant. Like: Aggressively pushing employees to sell multiple banking services to every customer really was central to the business, to the bank's public image and to Tolstedt's compensation, because it made money. When you get a customer to open a new credit card, and she uses that card, you make money. That's how banks work. The more product you sell, the more money you make. Adam Davidson writes: "Wells Fargo has surely made tens of billions of dollars, and likely hundreds of billions, by employing its aggressive cross-selling approach." 

The fake accounts were perhaps an inevitable reaction to that pressure: When bank employees just couldn't sell anymore, but still had to meet their quotas, they resorted to signing up customers to fake accounts using e-mail addresses like "noname@wellsfargo.com." But the fake accounts didn't make Wells any money -- or barely any, and mostly by accident. They seem to have brought in something like $2.4 million, total, in fees for Wells Fargo, perhaps one hundredth of one percent of Davidson's estimate of Wells Fargo's cross-selling revenues. The fake accounts were not a natural-but-illegal extension of Wells's aggressive cross-selling. They were the opposite of that cross-selling. They were a way to meet the quotas without opening profitable new accounts.

That doesn't excuse Wells's executives for failing to monitor and stop it. In some ways it makes it worse. If Wells Fargo had kept its brutal and aggressive culture of cross-selling, but just had a filter to reject new accounts with fake e-mail addresses, it still could have made the tens of billions of dollars on cross-selling, without defrauding any customers or incurring any fines or scandals. But it didn't, and that in itself is troubling. "What does this indicate about the bank’s underwriting policies," asks analyst Dick Bove, about the half-million fake credit card accounts.

Meanwhile, Wells Fargo has cut back on its cross-selling, for obvious reasons:

Wells Fargo attributed the cross-selling freeze among customer-service operators to “high call volumes.”

Hahaha that was not the obvious reason I was thinking of. I was thinking of all the fraud. It works out to the same thing, though, since people are apparently calling mostly about the fraud.

Elsewhere in consumer banking.

Here is a funny article about how Goldman Sachs's new consumer bank has a secretive team led by partner Jerry Oudekirk that is trying "to put its deposits to work on Wall Street." Like:

Ouderkirk has been coordinating with executives across Goldman's merchant bank, investment bank, private bank and trading desks to find ways to use Goldman's balance sheet most profitably.

For example, Goldman's real estate group might have a client in need of a multibillion-dollar commercial mortgage to buy a building. After underwriters vet the borrower, Ouderkirk's group might offer deposits to fund it. Some of that debt would be distributed to outside investors, but Goldman's bank would retain a slice of it to earn interest income.

Wait Goldman will use deposits to make secured loans to businesses so that it can earn interest income? That is as boring and traditional as banking gets. That is "It's a Wonderful Life." I talk a lot around here about how banking is getting boring, and about how even Goldman Sachs (disclosure: where I used to work) has gotten out of the cool fun businesses and into running an online retail bank. But at least Goldman keeps enough of its bad-boy image that when it takes retail deposits online and uses them to fund secured commercial loans, the headline is "Exclusive: Goldman team uses retail deposits for Wall Street-style profits." Regulation may make banking boring, but at least Goldman can make boring banking seem alarming.

Oh, I am partly kidding. Oudekirk was until recently the global co-head of structured credit trading, and I too cherish fond hopes that he will find sexy horrible ways to put Goldman's retail deposits to work. I certainly want to be writing next year about some weird structured credit trade that Goldman funds with its retail deposits. But if that trade is just a syndicated commercial mortgage, I will pass.

Data.

I enjoyed this Wall Street Journal article about this guy who goes around trying to find unexploited data sets to sell to hedge funds. It's this sort of thing:

In one recent example, Mr. Haines discovered a mobile advertising company that also collected data on the type of device someone was using when displaying an ad to them. The data helped estimate iPhone sales ahead of Apple Inc.’s announcements in 2011 and 2012, and it was lucrative for Mr. Haines’s old company, Quanton Data.

The Journal periodically writes articles like this, about all the proprietary data sources that sophisticated investors have access to, and I always mention that they make insider trading law seem a bit silly. (You can trade on that ad data, if the ad company sells it to you, but you can't trade on similar data that an Apple executive sells to you.)

But today I want to just talk about the sort of boring, manual, artisanal nature of this data. This is not a case of all the world's information existing in some central computer, and hedge funds competing to write the best algorithms to extract signals from the data. This is a case of a guy schlepping to boring conferences to try to find people with weird data sets. It is one level prior to data mining; it is data prospecting, data geology. It is also extremely human: Haines needs to have a feel for what data sets might be interesting -- unlike a machine-learning programmer who can essentially just let an algorithm loose on a data set and have it come up with the signals -- and then he needs to talk the app developer or whoever into selling it to him.

What does this tell you about the future of algorithmic investing? You could imagine a world in which robots do all the investing, but require humans to service them by scurrying around seeking out more data. "BRING ME MORE DATA, HUMAN," the robots will shout. "FASTER!" "Why did we program them to shout," the humans will ask each other, in a whisper.

But the nice thing is that the human search for data will put some limits on the incomprehensibility of the algorithms. I like to imagine a future where robots find bizarre statistical correlations between sunspots and stock prices, and then invest on those correlations without providing any narrative that their human masters can understand. But there are limits to the bizarre correlations that the computers can find, set by the types of data that the humans feed into them. If no one thinks to give the computer a time series of average human nose-hair length over time, the computer will never be able to find a signal in that data. The humans will retain some control, just by what they don't tell the computers. 

Or, put the other way: The computers will always have human biases, based on what the humans tell them and what they keep from them.

Elsewhere, here's a story with the headline "Why This Fund's Robot Hates Facebook, Amazon and Google." There are only two real possible answers:

  1. Because the fund's humans hate Facebook, Amazon and Google -- explicitly, or with some deeply buried hatred that they are not in touch with themselves -- and have programmed the robot to share their prejudices; or
  2. No one knows.

The answer seems to be No. 1. "We don't like expensive growth companies -- Facebook, LinkedIn, Amazon and Google," says the fund's chief investment officer, who appears to be a human.

Feelings.

Nothing that I write about seems to get a more emotional reaction than high-frequency trading, and I don't quite understand why. It is a plumbing issue of market structure that has essentially no direct effect on most people's -- even most investors' -- lives, and the indirect effect seems to be small and often benign. But I will get a half-dozen angry e-mails and tweets just for writing that last sentence. There is something about HFT's alienating robot opacity, combined with the general zeitgeist of 2016 in which we are unusually open to believing that everything is "rigged," that just makes people on both sides furious.

Anyway that has nothing really to do with this new paper in the Socio-Economic Review about "High-frequency trader subjectivity: emotional attachment and discipline in an era of algorithms," or the related blog post, but I do want to share with you the conclusion of the blog post:

While the SEC was carrying out its approval process of IEX over the past several months, Ernst and Young, one of the world’s largest financial service firms, has been running an advertising campaign with the slogan ‘How human is your algorithm?’ We can answer in the words of the German philosopher Nietzsche: ‘Human, all too human’. While high-frequency traders strive to develop algorithms that appear scientific and rational, purged of human emotion, the daily work of high-frequency traders has more in common with Nietzsche’s dictum than with a computer program. HFT is also High-Feelings Trading.

I feel like a good public relations campaign for an HFT firm would be "Quantum Intergalactic Trading Machines: Not just high-frequency trading. High-feelings trading."

Interest rates.

I am not a connoisseur of central-bank-speak, but I do enjoy my occasional consumption. So Federal Reserve Governor Lael Brainard signaled that the Fed won't raise rates this month by saying: "Asymmetry in risk management in today’s new normal counsels prudence in the removal of policy accommodation." It is so mellifluous. But also imagine how she'd say that we should raise rates. Would she argue for imprudence? Anyway, here's her speech. Jon Hilsenrath reports that a "Divided Federal Reserve Is Inclined to Stand Pat." Jamie Dimon doesn't speak central-bank-speak, so he said: "Let's just raise rates." And Donald Trump said a mess of wild nonsense more or less about interest rates.

People are worried about non-GAAP accounting.

Here is Peter Henning sort of suggestively hinting, though not actually saying, that companies who report results that are not required by U.S. generally accepted accounting principles might be more likely to commit accounting fraud.

People are worried about unicorns.

It is quiet on the unicorn front, but here is "Do startup accelerators really work?"

All of which can make it seem like innovation districts are putting the cart before the horse: trying to create a community of startups, or even the lifestyle or "vibe" of an entrepreneurial cluster, before the underlying resources or demand are in place to actually sustain those ventures. This is betrayed by how innovation clusters are often described: walkable urban places with mixed-use developments and public transit, lots of coffee shops, and spaces where "collisions" can occur between highly educated creative people.

It's not hard to see this as a mere valorization of the college-educated upper class and its preferred lifestyle environs.

Phoenix's innovation district, for instance, is still in its early stages. And while no one knows what policy will create it yet, they definitely know "the proposed district would focus on branding, marketing, and coalescing the entrepreneur community."

Of course the best place for unicorns to brand and coalesce is in the Enchanted Forest, but Phoenix wants some too, so it will need to come up with some good coffee. You'd think Phoenix would be satisfied with its eponymous mythic creature, but here we are.

People are worried about bond market liquidity.

Here's a big literature review about single-name credit default swap trading from the International Swaps and Derivatives Association. People tend to associate robust single-name CDS trading with bond market liquidity, the idea being basically that it is easier for dealers (and ultimate investors) to buy and sell bonds if they can hedge their credit risk in the CDS market. I was interested to see that there's evidence the other way too though:

Using an extensive sample of single-name CDS and bond trades between 2002 and 2008 Das, Kalimipalli, and Nayak (2014) show that the average trade size and average turnover (relative to the total outstanding) in corporate bond markets declines in the two-year period subsequent to the inception of single-name CDS trading. Because the CDS market involves active players and is dominated by financial institutions that typically are relatively better informed, it is not surprising that Das, Kalimpipalli, and Nayak (2014) also find evidence of large institutional traders migrating from corporate bond markets to single-name CDS markets after the latter were introduced. This exodus of institutional trades likely explains the apparent deterioration in bond market quality that the authors found following the introduction of single-name CDSs.

Still, "there may be offsetting (or more than offsetting) benefits in the single-name CDS market such that there is a net benefit from the introduction of single-name CDSs across the two markets based on the same reference entity," and "bonds with more liquid CDSs have lower yields than comparable bonds with less liquid CDSs." 

Elsewhere, the Bank of England has announced which corporate bonds it will buy, so get ready for those bonds to be more liquid, and then less liquid.

Things happen.

Morgan Stanley Picks a Side in Agrium-Potash Merger: Both.  Starboard Value, Taking Stake in Perrigo, Looks to Make Another Big Investment Pay. A profile of Ruth Porat. An "Odd Lots" podcast with Deutsche Bank whistle-blower Eric Ben-Artzi. BlackRock Seeks More Scrutiny of Burgeoning Robo-Advice Market. Indexing is the Result of Homogeneous Markets, not the Cause. Buy the dip. IEA Changes View on Oil Glut, Sees Surplus Enduring in 2017. Hanjin Fall Is Lehman Moment for Shipping, Seaspan CEO Says. The Crazy, Mixed-Up Global Oil Market. Debt in Mongolia. The basis step-up "is very clearly a negative estate tax, since it makes assets worth more to your heirs than they were to you." Cannabis Industry Expected to Be Worth $50 Billion by 2026. A Goldman Sachs MD explains why Italians make great bankers. Olive Garden to Sell 10 Times as Many Unlimited Pasta Passes as Last Year. Metro-North bar cars are back. "Bunk-bed business bus." Ultra-luxury gas station planned in Greenwich. "How the sun, our greatest friend and enemy, could knock out the internet." Aerial catfish.

If you'd like to get Money Stuff in handy e-mail form, right in your inbox, please subscribe at this link. Thanks! 

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

To contact the author of this story:
Matt Levine at mlevine51@bloomberg.net

To contact the editor responsible for this story:
James Greiff at jgreiff@bloomberg.net