Marks to Market

Zillow, Car Loans, Wirecard, TMTG, and Crypto


Back in August, people were saying there was an “arms race” to buy up as many US homes as possible because the real estate market was overheated and house flipping was going to be very profitable:

High-tech middlemen like Opendoor and Zillow Offers, Zillow’s home-buying platform, first inserted themselves into the housing scene a few years ago, armed with cheap money and hoping to profit off the bedrock of American middle-class wealth. iBuyers target mid-level homes that are in decent condition, offer to buy the house with cash, and make the selling and moving process quick and convenient. They then make a few repairs and quickly put it back on the market, ideally at a higher rate.

One thing tech companies love to do is talk about solving market problems with algorithms. They might say, hey, we’re the largest online platform for buying and selling houses in the country, we’re a public company with a lot of money, and we’ve already built a tool (called Zestimate) that millions of people are using to determine a fair market value for their homes. What if, we, Zillow, used all our data and knowledge and built an algorithm to buy and sell the homes ourselves? If you were the CEO of Zillow you might say this in a very legal-department-approved way on an earnings call:

When we decided to take a big swing on Zillow [O]ffers 3.5 years ago, our aim was to become a market maker not a market risk taker. And this was underpinned by the need to forecast the price of homes accurately three to six months in the future.

We used historical data and countless simulations to test this belief. We set unit economics targets that required us to stay within plus or minus 200 basis points in breakeven, holding ourselves accountable to these levels publicly with you all.

200 basis points is 2% for those of you without a finance degree. So Zillow spent years building a home buying-and-selling system based on all its data, and its goal was to stay within 2% profit or loss margins in any given quarter. How did that work out? Well, the reason the CEO of Zillow was saying those things on an earnings call was to preface the announcement they were shutting down Zillow Offers and laying off a quarter of their staff because it lost a lot of money:

In a statement Tuesday, Chief Executive Rich Barton said Zillow had failed to predict the pace of home-price appreciation accurately, marking an end to a venture the company once said could generate $20 billion a year. Instead, the company said it now plans to cut 25% of its workforce.


Zillow, which released earnings Tuesday, said its home-flipping business, Zillow Offers, lost $381 million last quarter, as measured by adjusted earnings before interest, taxes, depreciation and amortization.

Oops! Over the summer, when other “iBuying” (ugh) companies had pulled back, Zillow accelerated their home buying, essentially buying at the top of the market. As it cooled, Zillow was forced to sell hundreds of homes for less than what it paid, which is not great when you are trying to stay within 2% of break even.

Zillow has around 17,000 homes left to sell, and it’s written off another couple hundred million dollars of expected loss. The thing about software companies is that they sometimes mistake having a lot of data and hiring a lot of smart people as a replacement for, I don’t know, being an expert at buying and selling houses? Normally, when a software company writes an algorithm that doesn’t behave as expected, they can just turn it off and say oh, whoops, we won’t use this anymore. However, when that algorithm has purchased a few billion dollars’ worth of houses in Scottsdale or wherever, you need a bunch of non-algorithmic people in the finance department to clean up the mess, and maybe you need to fire all the people who wrote the software.

Car Loans

Consumer Reports has published an exhaustive piece on car loans in America and it is maddening:

Today, Americans with new-car loans make an average monthly payment approaching $600—up roughly 25 percent from a decade ago.

This is not because cars are more expensive - some models are cheaper than they were 20 years ago. It is mostly driven by the cost of financing a car, which has increased because, well, lenders can just do that:

Experts say that CR’s analysis suggests a broad problem with the way car loans are arranged in this country: Dealers and lenders may be setting interest rates based not only on risk—standard loan underwriting practice—but also on what they think they can get away with.

Some of it can be attributed to discrimination - a constant problem in the US, where Black and Latino borrowers are often charged higher interest rates or tricked into taking subprime loans when they have good credit - but a lot of it likely has to do with lack of government oversight:

Many states have confusing and contradictory laws regarding how high rates can be set, according to interviews with regulators in all 50 states and the District of Columbia. At the federal level, the Consumer Financial Protection Bureau has limited oversight of auto lenders.

Very little information about car loans is publicly available, and Consumer Reports was only able to access this limited dataset because the lenders in question had securitized packages of their loans, and had to file SEC disclosures about the underlying borrowers. Here are some of the key findings:

  • A credit score doesn’t necessarily dictate the terms of the loan offered. Borrowers in every credit score category—ranging from super-prime, with scores of 720 and above, to deep subprime, with scores below 580—were given loans with APRs that ranged from 0 percent to more than 25 percent.

  • Many borrowers are put into loans they might not be able to afford. Experts say that consumers should spend no more than 10 percent of their income on an auto loan. But almost 25 percent of the loans in the data CR reviewed exceeded that threshold. Among subprime borrowers, that number is almost 50 percent, about 2.5 times more than prime and super-prime borrowers.

  • Underwriting standards are often lax. Lenders rarely verified income and employment of borrowers to confirm they had sufficient income to repay their loan. Of the loans CR looked at, these verifications happened just 4 percent of the time.

Cool! When CR controlled for other factors, it found that many borrowers with good credit, who should have qualified for low interest loans, were instead put into high APR loans, costing them thousands of dollars’ worth of extra payments:

The average APR for car borrowers with prime and super-prime credit scores in the most recent quarter was 3.48 and 2.34 percent for new cars, respectively, and 5.49 to 3.66 percent for used cars, according to a recent report from Experian, a credit reporting agency.


Yet nearly 21,000 borrowers— about 3 percent of all prime and super-prime borrowers in the data CR reviewed, representing $439.6 million in loans—had loans with APRs of 10 percent or higher. Several thousand had loans with APRs of 15 percent and above.

This isn’t accidental, and is a profit center for lenders:

All told, consumers in the data CR reviewed with high APRs for their credit tier will spend $790 million more on their vehicles than if they had received average interest rates based on their credit score of our dataset.

It is largely due to consumer ignorance, and exacerbated by the way buying a car works in America - dealers push a target monthly payment, which can disguise high APR loans or allow the lender to fiddle with the terms to skim more profits. Also, with current car shortages, consumers don’t often have the luxury of shopping around for the best rate - many dealers will insist they use whichever lender they contract with, and it’s rare for consumers to walk in the door with a pre-approved loan in hand.

Then there’s the issue of lenders writing expensive loans to people who can’t afford them. The rule of thumb is a car payment should be 10% or less of monthly income. In fact, the co-founder of Exeter Finance - a lender who’s had to pay settlements for writing predatory loans - wrote an op-ed in an industry publication saying that payment-to-income ratios of more than 14 percent leads to 50% higher defaults at every credit level. People with lower income are more likely to have a car payment they can’t afford:

About 20 percent of subprime borrowers had a payment-to-income ratio of 14 percent or higher, while just 6 percent of prime and super-prime borrowers did. 

So, why are so many people ending up with car payments they can’t afford? Isn’t it better for auto lenders to issue loans to people who can pay them back? It’s easier to repossess a car than a home, but surely that isn’t their business model. Surely!

Misrepresentations of income and employment have been a sticking point in the industry in recent years. Such misrepresentations led to about $4.4 billion in auto loan losses in 2020, according to a report this year by Point Predictive, a company that sells software to detect auto loan fraud.


In a 2017 case against Santander, the state attorney general alleged that the lender predicted an estimated 42 percent of loans the company made through certain high-risk dealers to Massachusetts residents went into default or were expected to default. Yet the company continued to fund loans originating from dealers who inflated incomes and had issues arise.

Man, I don’t know. Banks like Santander have done big business in recent years issuing risky, subprime auto loans to people who may be unable to pay them back, then packaging those loans and selling them to Wall Street investors. “Average” auto loan default rates are between 4 and 5% according to national statistics, but the subprime loans lenders like Santander originate are much worse:

According to the results of a survey presented at the annual Non-Prime Auto Financing Conference last fall, the average 30-day delinquency rate reported by the 28 nonprime lenders included in the report was about 11.2 percent for the fourth quarter of 2019. The repossession rate was around 13 percent in 2019—or roughly 1 in 8 vehicles financed by those lenders.

Isn’t it cute that they call it “Non-Prime” when lenders attend conferences discussing new ways to fleece low income borrowers? Default rates fell last year, likely due to the stimulus and increased unemployment payments and, perhaps, people being able to sell cars they couldn’t afford into an overheated used car market. Now that workers are returning to offices, pandemic-related stimulus efforts are likely gone for good, and car prices are temporarily inflated due to supply issues, expect to see these problems get much worse before they get better.


We have talked a little bit about Wirecard, the German bank that collapsed amid a fraud scandal last year. Now, one of the Financial Times reporters who helped expose the fraud has shared his version of events and wow, it is a ride:

At first, I thought the note was a hoax but, astoundingly, it was real. It was the latest episode in a skirmish that was to last 18 months, leaving me under attack as German banks and regulators waved away evidence of corporate criminality to place me at the heart of a conspiracy theory. At times, it seemed like the world had gone mad.

Essentially, for the two or so years McCrum and his team were investigating Wirecard, the company used a scorched earth strategy to discredit, and eventually get them criminally charged by German bank regulators. It paid a small army of private investigators and former military intelligence officers to surveil the reporters, and forged documents to discredit them. It used paid snitches to plant stories that the FT was colluding with short sellers to tank Wirecard’s stock. It got the German bank authority - BaFin - to restrict short selling of Wirecard’s stock for two whole months.

Skipping ahead - spoilers, sorry - most of the executives involved in the Wirecard fraud are in jail awaiting trial, and the former COO who was in charge of the smear campaign against the FT is on the run from the authorities. So, in the end, the good guys did win, but it was a harrowing journey.

I think we expect this sort of corrupt corporate conduct in the US, because we’re a crumbling society largely governed by the whims of capital, but it’s important to remember that we aren’t the only country with evil corrupt corporate executives and their government enablers. Add Germany to that list!


I feel obligated to do something I do not do very often and issue a correction about a prior newsletter. Last week, I wrote this:

There is Trump admitting he created TMTG and then immediately merged it with DWAC via a SPAC - I am dying typing this - and suddenly the company had $293 million dollars, which DWAC had sitting around waiting for a company to take public. Normally, you’d start a company and run it for awhile, and once you had a viable product, or a business plan, or a need for investment money, you’d find a SPAC and convince them to give you money, but if you’re Trump you cut the deal in advance, using an imaginary company with no business plan or whatever, and congratulations, you’ve got a multibillion dollar company. Finance!

Matt Levine and I believed - based on Trump’s own press release! - that Trump had created TMTG last month, right before taking it public. But! The New York Times has the full story, and it turns out we were both wrong:

On Feb. 8, Trump Media was incorporated in Delaware.

Damn. The provenance of TMTG is important in the Times story, because the path it took before merging with DWAC and going public is now under scrutiny because they maybe did some securities fraud, oops:

Lawyers and industry officials said that talks between Mr. Orlando and Mr. Trump or their associates consequently could draw scrutiny from the Securities and Exchange Commission.

Another issue is that Digital World’s securities filings repeatedly stated that the company and its executives had not engaged in any “substantive discussions, directly or indirectly,” with a target company — even though Mr. Orlando had been in discussions with Mr. Trump.

Basically, when Trump founded TMTG in February, the plan was to take it public right away with a different SPAC controlled by Patrick Orlando, but that SPAC only had $100 million dollars in the bank, and Trump didn’t think that was enough money. So, eight months later, Orlando created DWAC, which had almost $300 million dollars, and Trump liked that just fine, so they merged. The problem is Orlando wasn’t supposed to be creating a SPAC with a merger in mind, because that’s not how they are supposed to work, legally. So, Orlando probably broke the law a little, and Trump probably made some misrepresentations in his securities filings, but it’s not the sort of thing anyone goes to jail for. It definitely is something that will get them sued for fraud by teams of lawyers whose primary revenue stream is suing companies for securities fraud, and who maybe hate Trump and want to annoy him, which is perhaps not the sort of thing that our federal courts should be spending their time on, but this is the society we are stuck living in.


Here’s a fun round-up of stuff that happened this week in crypto:

Monkey Jizz turned out to be a scam. So did Squid Game (the coin, not the TV series). A CryptoPunk NFT sold for $532 million dollars, but didn’t really sell at all. The whale holding billions of dollars’ worth of Shiba Inu coin might be about to sell. People are using “right-clicker mentality” as an insult, based on a post about Salt Bae. An MLM guy got his Bored Apes stolen and threw a tantrum on Twitter.

Those are all real things, reported in major news outlets, that I typed out on my computer and put into a newsletter, in case you want to read them, but I wouldn’t blame you if you did not.

Short Cons

FTC - “Amazon will pay more than $61.7 million to settle Federal Trade Commission charges that it failed to pay Amazon Flex drivers the full amount of tips they received from Amazon customers over a two and a half year period.

CFPB - “The Consumer Financial Protection Bureau (CFPB) today released research finding that consumers in majority Black and Hispanic neighborhoods, as well as younger consumers and those with low credit scores, are far more likely to have disputes appear on their credit reports.

Wired - “The NRA has not confirmed the attack nor the validity of the purported stolen documents, which researcher say include materials related to grant applications, letters of political endorsement, and apparent minutes from a recent NRA meeting. It appears, they add, that the NRA was hit with ransomware late last week or over the weekend, which lines up with reports that the organization's email systems were down.

Wired - “Despite never reaching the peak trading volumes of its more-famous cousins Silk Road and AlphaBay, White House Market had established itself as one of the most popular—and secure—markets on the dark web. So when WHM unexpectedly closed on October 1, it came as a shock to the platform’s dedicated user base.

Tips, thoughts, or deeply discounted suburban homes to