Subscribe
Share
Share
Embed
The Diff is a newsletter exploring the technologies, companies, and trends that are making the future high-variance. Posts range from in-depth company profiles, applied financial theory, strategy breakdowns and macroeconomics.
The Depressing Bull Thesis for Rocket Mortgage
Rocket Companies, the parent company of Rocket Mortgage and Quicken Loans, filed their S-1 yesterday. Rocket is the country's largest mortgage originator, with 9.2% market share, up from 1.3% in 2009. Financial services companies' economics are defined by transaction costs. You can tell because the most profitable such companies have something special going on in customer acquisition:
For exchanges (ICE, CME), it's a dense network: traders go where spreads are low, and spreads are low where traders are.
For card networks (V, MA), it's a two-sided network: merchants want spenders, and vice-versa.
Paypal had a network effect within eBay, which itself was growing thanks to its own network effect.
For lenders, there are basically two ways to control customer acquisition cost: find novel channels for closing new deals at a low cost, or find an excuse to make loans that other lenders reject. The short-term incentive is to do the latter: every new loan raises the denominator of the "cost per acquisition" number, but defaults take a while—and the faster a company grows, the lower defaults look.
Rocket is an originator that doesn't mortgages on its books for long, so it doesn't have a ton of exposure to mortgage performance as such. They originate and quickly sell mortgages. The model is: originate at 100 cents on the dollar, sell at 103.25 cents on the dollar, use the difference to pay for ads, commissions, and overhead, repeat. They can turn these assets over quickly; they originated $51.7 billion in Q1 2020, and had $12.8 billion in assets on the books, so their average time from origination to sale is about 23 days.
Since Rocket has grown so quickly, with origination volume up 75% Y/Y last year and 131% Y/Y last quarter, it's an especially important question: are they doing a great job, or doing a bad job at scale?
Fortunately, Rocket's business lends itself to quantitative rather than qualitative analysis, at least in the short run. Rocket sells almost all of its mortgages to government entities, and their mandate is to provide capital to homeowners by making every loan that fits certain rules. So when Rocket says its average loan-to-value ratio is 73% (slowly trending down) and that the average customer credit score is 747 (slowly trending up), they're not really talking about how likely the loans are to get repaid. They're mostly talking about how likely it is that repayment will be someone else's problem, and only making the secondary point that it won't be an especially big problem. Lending Tree's 2019 statistics show a slightly higher average credit score (765) and a more aggressive loan-to-value (87%).
And they have more direct evidence that they're growing through more efficient spend, not lax underwriting: "In 2019, we closed 6.7 loans per month per average production team member, compared to the industry average of 2.3 according to the Mortgage Bankers Association. In 2020, our year to date average has grown to 8.3 loans per month."
How?
For mortgage originators, there's a growth lever to pull: refinancing. The US mortgage system has, for vague historical reasons, standardized on a mortgage structure that gives borrowers an option to refinance. This has some bad macro side effects, beyond the scope of this piece.[1] But one effect is that it gives mortgage originators and servicers a chance to benefit their customers by helping them decide when to exercise their free option. Everyone who borrows with a fixed-rate prepayable mortgage is long rate volatility, whether they know it or not, and a company like Rocket can profitably collapse this knowledge gap.
The question is: how do you get borrowers to refinance? That's where Rocket's model gets very interesting. In addition to their lending operation, they have a mortgage-servicing business. For a portfolio of 1.8m loans with an outstanding balance of $344 billion, Rocket gets paid a small fee (0.310% of the balance outstanding each year) to collect payments. And these are the mortgages they've originated; 94% of the time, when Rocket originates a mortgage, it services that mortgage, too.
This gives Rocket a useful marketing channel: once a month, they send their customers direct mail—which the customer has to open, if they don't want to get foreclosed. And that direct mail lets Rocket retain their customer relationship.
It's unclear how profitable this business is, because the accounting is brutally honest: the right to service a mortgage is a financial asset, but it's taking the opposite side of the mortgage-holder's refinancing option. When rates drop, mortgages get refinanced, and Rocket's servicing revenue from that particular mortgage drops to zero. In Q1, for example, they earned $257 million from servicing loans, but since rates declined the fair value of their servicing rights dropped by $991 million.
Over a longer period, you might be able to calculate the underlying profitability of the servicing business, but a) the trouble with any statistical analysis of a volatile growth business is that by the time n=20, your 20th most-recent sample is from a very different business than your most-recent sample, and b) mortgage servicing looks more valuable as a marketing channel and partial interest rate hedge than as a as a profit center per se. What Rocket says is that, of the customers who refinance, 76% refinance with Rocket, compared to an industry average of 22%. When your business model is cost-plus-markup, and your end buyer has unlimited appetite, competing with yourself on price is just good business.
Interestingly enough, a study confirms this for the broader Fintech industry: they're faster, more flexible, better able to handle spikes in applications (such as, for example, a rate cut leading to a refinancing surge), and they tend to increase refinancing among customers who would actually save the most money.
The interest rate bit brings up the depressing bull thesis. Rocket originates so many loans because it can convince borrowers to refinance. And that means its growth is partially a bet on a secular decline in interest rates. Rocket is basically designed around this: they accept volatile servicing revenue in exchange for access to one-time upside from refinancing, and the bigger their pool of servicing customers, the more disproportionate their share of refinancing is. This also lets them reduce ad spending as a share of revenue over time. More of their business comes from repeat customers instead of new ones.
If every part of this model keeps working, it's great news for Rocket (and their shareholders), but bad for the economy. Interest rates ultimately represent expectations for economic growth, and naturally move inversely with the state of the economy. So a bet on Rocket is a bet on economic stasis: that we won't face such a catastrophic recession that we have to rethink mortgage subsidies, but also that we won't see so much growth that rates begin to rise again. Rocket is an exciting way to bet on a boring future.
[1] For more, see here. The short version: the duration of a portfolio of mortgages drops when rates fall, so anyone with a long position in mortgages must hedge their rate exposure by trading long-duration assets. Since they're trading in the direction prices are already moving, this exacerbates swings in long rates. And since the assets they trade to hedge are US treasury bonds, they're exacerbating swings in the world's benchmark long-term interest rate. Normally, if there's a trade that gives both parties exposures they don't want, they should both hedge and the effect on markets should be nil. But in this case, it's a nearly-mandatory trade that only one side hedges; mortgage borrowers don't mark their rates options to market.
Elsewhere
China Equities: Not 2015 Yet
China's equity rally has been partly fueled by a rise in margin lending ($), but the amounts are modest relative to 2015: 1.3tr RMB now compared to 2.2tr at the 2015 peak. Or, in market cap terms, 4% of the total value of equities compared to 10% last round.
One of the long-term concerns about China's economy is that its financial system is relatively underdeveloped. "Underdeveloped" is a term of art that doesn't refer to how much you can borrow to do irresponsible trades, but how sophisticated you have to be to over-lever an irresponsible trade. US institutional investors have plenty of opportunities to make bets that will completely wipe them out (for example, Malachite lost 2x its assets under management by shorting volatility, in an act that Institutional Investor described as "YOLOing risk management"). Robin Hood certainly gives retail investors access to vast amounts of indirect leverage through options trades, but a) has finally reached the point that it can fine-tune its risk management faster than Redditors can find loopholes, and b) mostly offers that leverage through options, which don't lead to the same cascading liquidations pure margin bets do.
In other China finance news, Alibaba's fintech arm, Ant, is planning a Hong Kong listing that could value it at over $200 billion. Ant has a somewhat tortured history—it was spun out of Alibaba proper, in response to regulations that did not ultimately seem to materialize. The Hong Kong IPO is an interesting move; Hong Kong's security law makes it challenging for financial companies to do business there, because it restricts their ability to transfer funds (lest they transfer them to someone deemed a threat to the state) and makes it challenging to issue accurate ratings on stocks (a "sell" could theoretically fall on the wrong side of the law). This is a strong vote in favor of Hong Kong remaining a financial center, at least for a while.
Twitter by Subscription
Twitter is hiring an engineer for a vague subscription product. There was speculation about launching a paid product a few years ago, but it's never made much sense: the services Twitter could offer to power users are more valuable as free services for advertisers or as value-added products sold by data licensers. Twitter might be able to create an ecosystem of subscription-only Twitter feeds, but that's a challenge, too; retweets are a core part of the Twitter product, and screenshot-based piracy would be hard to prevent.
Oil as a Currency Area
A natural economic phenomenon is when a group of countries all tend to price trade and borrowings with each other in the same currency, or a convertible currency. This can happen through formal agreements (like Bretton Woods or the Euro), or in a natural, informal way (most trade, even between countries that don't use the dollar, is priced in dollars).
Russia is building a unique currency zone, priced in oil rather than dollars. Russia has slowly taken more control over Venezuelan oil, after using similar tactics to get access to Syria.
It makes sense to think of this as a currency area: the dollar benefits from network effects, but the cost of using dollars is that the dollar financial system is under the increasingly tight control of the US. Sanctions and the dollar give the US broad authority to punish disfavored behavior outside its borders. Oil is readily convertible into currency, and easily convertible into any currency, so controlling more of it is a way for Russia to bet that either a) the US will eventually stop sanctioning them, b) Europe will stop cooperating as much with the US, or c) if neither of those happens, China will be happy to do business with them.
Decoupling
A former CCP official, Zhou Li (once deputy head of the Chinese Communist Party's International Liaison Department) has written a widely-discussed article on future US/China decoupling ($). It's bleak. He believes that the US and China will trade less, that they'll have separate currency zones, that they'll face food shortages(!) and terrorism(!).
This is, broadly speaking, a view that's pessimistic on China and relatively optimistic on the US. It's very hard to bootstrap a currency into global usage, especially in a country that keeps tight control over its financial system. China is a net food importer, the US is an exporter. It illustrates the paradox of China's rise: a major exporter is dependent on a stable world order, but a rising power causes instability.
Recoupling?
Brad Setser has a line-by-line assessment of the Phase One trade deal with China. Depending on the reader's gullibility, the deal either assumes more Chinese imports in the back half of 2020 or will not be adhered to after all. There are different factors for different line-items:
Aircraft exports were hit by the Boeing 737 issues and then the pandemic.
Agricultural exports to China have been weak, despite China's other flu problem (African Swine Fever reduced China's pork output by 30% in late 2019—China is roughly half of worldwide pork consumption, and pork consumption has been a symbol of the country's rising prosperity). The culprit in this case: China has chosen to import soybeans from Brazil rather than the US.
Energy exports are slow due to demand issues and price.
It's unclear whether China planned to adhere to the deal from the beginning, or always expected to underperform. But given the US electoral calendar, deliberately undershooting sounds like the optimal choice: by the time it's obvious that China can't hit its targets, the election will nearly be here, forcing Trump to either a) launch another phase of a not-especially-popular trade war at a politically inopportune time, or b) tacitly concede that, whether or not trade wars are good, they're not so easy to win.