V-Shaped Recovery for Me, L-Shaped Recovery for Thee: The Aftermath

Plus! Amazon & Twitter Earnings, Bullwhips, Commoditizing the Complement, Remote Work, Apple Demographics, State Bailouts, Adpocalypse

This is the weekly free edition of The Diff, the newsletter that tracks inflections in finance and tech. Last week was a five-part series on the most important trend in the economy today: the V-shaped recovery we see priced into the market and an L-shaped recovery for the rest of the economy. It’s not a call for pitchforks and torches, just an assessment of what’s going on right now. Some of it stems from forces that already existed (the world’s best-run companies are high margin and have more cash than they know what to do with). And some is because of policy. In short: the Fed has done at least an A-minus job, while Congress gets a generous D. But we can grade them on a curve: the Fed is an economic A-team, and even if Congress were full of epidemiologists, economists, and executives, a committee of 535 people can’t accomplish anything. Last week’s posts:

V-Shaped Recovery for Me, L-Shaped Recovery for Thee: The Aftermath

Over the last week, I’ve laid out the most straightforward explanation for the divergence between equities and the broader economy: large companies were unusually well-equipped to survive, and they’re better able to benefit from monetary interventions—which have been far faster and more effective than fiscal ones. Meanwhile, small companies, individuals, and municipalities just don’t have the cash reserves or flexibility to react.

In the last few posts, I’ve focused on what this means in the short term. Now I’d like to take a step back and explore two longer-term possibilities. The pessimistic one is front-end corporatization: small businesses just evaporate, their real estate is taken over by big companies, and (some of) their employees find new jobs at these companies. This isn’t merely bad in the aesthetic sense that more of the country will look like Breezewood, PA, but in the economic sense that it destroys a lot of wealth: it wrecks the balance sheets of small business owners and their employees, dissolves institutional and social capital, and generally represents a lot of waste. (It would be good for productivity—big businesses are, on average, more efficient than small ones. But in macro terms, if we reduce the amount of labor needed for a given level of output, we’re trashing demand unless we find offsetting policies.)

The optimistic view is back-end corporatization: that software companies and lenders launch an all-out sprint to modernize and recapitalize US small businesses, applying the scale advantage of big companies to solving the problems of local ones. In that happier outcome, small companies hang on for dear life, and come back leaner and ready to fight. Unlike big companies, they won’t necessarily respond to efficiency growth with layoffs; it’s much easier to fire 500 people who work a thousand miles away than it is to fire one person who lives in your neighborhood.

Front-End Corporatization

Here’s the baseline: in retail, restaurants, and medicine, thinly-capitalized companies will run out of cash in a few weeks of lockdowns, or a few months of partial-reopenings during a brutal recession. When they do, they’ll close down the business, fire everyone, and leave the landlord the keys. That’s going to lead to a lot of cheap real estate, and anyone who a) expects things to be fairly normal in a few years, and b) has access to capital, will take advantage. Point a) describes almost everyone, and point b) describes any company big enough to sell bonds.

Since landlords are also distressed, that might take the form of rock-bottom prices for long-term leases in locations that won’t be occupied for a while. So Starbucks will grab the spot your neighborhood coffee shop was in; your GP and dentist will sell to a private equity-backed chain; your local gym will become a Planet Fitness. (The independent fashion retailer will probably become a Chase Bank or something; apparel has been hit harder than other sectors and was in secular decline already) More familiar logos, fewer familiar faces.

The math works out fine, as long as the price is low: if a buyer just bakes one year of significantly-impaired revenue and one year of post-recession revenue into their model, and adds it to the acquisition cost, they can still project a healthy rate of return over a long period. That’s especially true when big brands can take advantage of soft network effects: a logo you see on your drive to work might prompt a purchase over the weekend, and every store is a touchpoint for getting you to install the company’s app.

In theory, big companies shouldn’t have much appetite for risk right now; they see the same headlines as the rest of us. But in practice, liquidity finds a way: buybacks are politically challenging right now, but raising money is feasible, so for companies that are hurt but not crippled, the pressing question in the next few quarters will be where to put cash.

Large companies are certainly not an intrinsic problem: they’re more productive than small companies, and they tend to pay better, too. But getting more output per hour from the same footprint of physical assets means lower employment, and long-term unemployment reduces earnings even decades after the fact.

That problem is especially pressing because of demographic issues. Specifically, the Great Baby Boomer Services Hedge. It works like this: as you age, your productivity tends to decline, but your income moves at a different pace.[1] This means that older workers in jobs that aren’t physically demanding can choose to retire later—which means they can save less and make up the difference by working longer, or save in riskier ways and make it up the same way. These older workers are an important issue in two directions: first, if they get fired, they lose a lot of that accumulated social capital, lowering their productivity. But if they keep working, their marginal propensity to save goes up as the market goes down, so they actually make consumption more cyclical, which hurts aggregate demand.

It adds up to a bleak, deflationary picture. A slow recovery, with continued weak unemployment. And a lot more visible homogenization.

Back-End Corporatization

That’s the bad scenario. Here’s the good one. Those same local businesses are running down their cash reserves, but lenders are banging down the door with a crazy offer: borrow enough to meet payroll now, pay nothing—until business starts coming back. Revenue-based loans transfer macro risk from borrowers to lenders. Why would lenders do this? Because they know that the lifetime value of a quality borrower is worth the temporary income hit from delayed payments.

They also know the iron law of growing a lending business: the best problem to have is customer acquisition cost. (If you’re running a growing lender, you’ve tapped out all your organic cross-selling channels, and CAC isn’t driving you up the wall, congratulations! Your underwriting sucks, and your borrowers are going to default. Sorry!)

Q2’s ad CPM decline means that these lenders will be able to get as big an audience as they can handle, although the challenge of sorting the unlucky from the unbankable still looms.

One way they can meet that challenge, though, is by getting more involved in the borrower’s business—get them good bookkeeping software and a modern point-of-sale system. Band together a bunch of borrowers and start negotiating with suppliers and landlords.

In short, use software economics to give small businesses the same economies of scale that large ones already benefit from. That’s an enormous challenge, but it’s exactly the model fintech companies are pursuing. They don’t have the scale to do it on their own, but PPP loans have been a de facto customer acquisition subsidy. As long as the program keeps getting funded, more small businesses will have their financial backend linked to robust APIs, and will be able to start optimizing everything from cash flow to marketing to suppliers.

In that version of the future, the average strip mall and downtown looks a lot like it did pre-Covid. But it runs more efficiently and at higher margins.

Which Way?

These aren’t mutually-exclusive outcomes. The back-end version has a head start in execution, but the front-end story has a head start in funding. And, interestingly enough, both versions are compatible with the market-vs-main street divergence. In the pessimistic outcome, big companies have a wobbly recovery and smaller ones suffer a depression; in the optimistic one, small companies struggle through a recession and then recovery, while a few big tech companies become massively more influential (which will, inevitably, benefit the top lines of most of the existing tech companies. Facebook and Google will not exactly suffer if small business spending recovers; Microsoft will do great if more store owners can export their sales data to Excel; and a big chunk of that incremental computing will end up on AWS).

Monetary intervention is fast; fiscal intervention is thorough. And the one we choose will determine whether American small businesses catch up to bigger ones or get trampled by them.

[1] Typically, incomes peak in the 50s, but measured output at specific tasks tends to peak earlier. This is not ageism; it’s information asymmetry: older workers are a better-known quantity, and that’s actually valuable. Deirdre McCloskey has argued that, if you sum up finance, management, and advertising, you conclude that “persuasion” is about a quarter of the nation’s GDP. So anyone you know and trust can cost up to 25% less because the persuasion overhead is so much lower. Their productivity at individual tasks might be lower, but they’re more likely to work on the right stuff.



Amazon reported earnings last night, and they’re a challenge to sort through. Retail revenue beat, AWS revenue slightly missed. Amazon’s retail segment is basically one big one-time-item this quarter, as they deal with a bizarre range of moving pieces, including—a step down in aggregate retail spending; a surge in one-time spending for things like home furnishings, high-quality webcams, toys, and games; a temporary halt to shipping for non-essential goods; and a step-function increase in food and healthcare purchases. Next quarter, Amazon expects to spend all of their operating profit of ~$4bn on Covid-19 costs. Basically, Amazon retail became America’s temporary quartermaster.

Amazon is part of the “reverse war economy” of the US’s Covid-19 response. When a country faces an existential threat, the government typically does a de facto nationalization of key assets like factories and transportation systems, and enacts wage and price controls. In this crisis, it’s private companies that are taking over public health and setting wages. They’ve also helped states make SNAP benefits redeemable online.

Amazon has also done a little monetary policy, by pausing loan repayments and cutting some warehouse charges. (This sounds fiscal, but in mid-twentieth century recessions, one of the ways monetary policy helped was by reducing the carrying cost of inventory so firms wouldn’t go bankrupt working through it. Since Amazon’s fees are proportionate to the amount of inventory held, the net effect looks like a targeted rate cut.)


Twitter reported earnings, with the now-obligatory note that ad spending was great in January and February, and took a nosedive in March. In their case, total spending dropped 27% YoY from March 11 through quarter-end. Given Twitter’s exposure to event-driven ads, this drop is not representative of ad demand broadly—a decent share of their customers are in the “demand literally went to zero” category. But that’s a category that can snap back fairly quickly; there are still 24 hours in a day, and “what’s happening” is increasingly happening online. The monetization is worse, but the attribution is A+—so while this is a disaster for event-driven advertising generally, it’s a huge victory for Twitter’s share.

(With an activist breathing down their necks, Twitter management spent a lot of time talking up various improvements to ad products.)


I went on the Urbane Cowboys podcast to talk about markets, including further discussion of the V/L recovery question. In Marker, I have a new piece covering fintech in the Covid-19 era, including much more on the nuances of revenue-based lending. And in American Greatness, a piece on how higher education will be devastated by Covid-19.

Supply Chains and Deflationary Bullwhip Effects

A subtle reason for the “long deflation” is that supply chains are more tightly-managed than ever before. But that optimization requires automation, and automation means relying on historical data. Today, those historical models don’t apply; for a broad swathe of products, demand has either been shifted (luxuries to staples), pulled forward (bulk food instead of just-in-time), or has suddenly raced up the S-curve to full adoption (webcams, pull-up bars, bidets). This can easily lead to a classic bullwhip effect, magnifying distortions further back in the supply chain. Since retailers are very reluctant to raise prices, the visible effect will be deflationary: for products that were overproduced, retailers will want to get rid of them; for the ones that were underproduced, they’ll just show up as out-of-stock. Whether you treat that as inflationary or deflationary depends on whether you treat out-of-stock as a null entry or infinity.

Microsoft Commoditizes the Complement Some More

A few weeks ago, Microsoft drastically cut the price of Github (see the Diff writeup here. Yesterday, they dropped the price of Visual Studio and made one of their conferences online-only and free (down from $2,395). All of these moves make sense as part of a secular trend towards monetizing compute time and treating every other developer-facing product as a way to market Azure. The customer lifetime value calculation bakes in the fact that 1) cloud computing time starts out very cheap, but tends to compound over time, and 2) if you sell a usage-based complement to the output of programmers whose fully-loaded cost is $200k+, you have a lot of room to collect high-margin incremental revenue before anyone’s tempted to cut costs.

And elsewhere in cloud news: financial companies are adopting cloud computing to deal with spikes in demand, and to compete with agile challengers who are already using it. It’s good to upgrade your IT, but it’s great to be an arms-dealer in a market where everyone is scaling as fast as they possibly can.

Remote Work: “I think this is the future”

Nationwide Insurance is transferring 4,000 of its 28,000 workers to permanent work-from-home status, closing several regional offices. Interestingly, their larger offices are staying open while smaller ones are closing. It’s likely that the agglomeration effects of having people work in the same place are still worth the murderously high cost of commercial real estate in major cities, so this will exacerbate the long-term trend of more extreme real estate price differences between major business hubs and other locations.

Apple’s Demographic Crunch

The Information profiles an interesting cohort: older Apple employees with extensive manufacturing experience. A few months ago, The Economist referred to Shenzhen’s geoepistemological advantage, a phrase that’s stuck with me ever since: there are social networks tied to expensive capital goods, and they’re extraordinarily hard to dislodge. The supply-chain trend over the next few years will be fault-tolerance rather than pure efficiency, which will slowly erode Shenzhen’s advantage. But American manufacturing experts' demographics might move faster.

State Bailouts

New Jersey’s governor wants a bailout. STUMP has a great overview of the case against bailouts, at least bailouts for pension funds. State bailouts may turn out to be a moot point: since the Fed is purchasing muni debt, and more or less sets its own limits, we could end up in a scenario where states engage in macroprudential policy (and clean up their finances a bit) while the Fed monetizes the debt. This would be a very odd situation, because it basically amounts to the Fed running a large chunk of fiscal policy.

Adpocalypse: Bad News and Green Shoots

Anyone who was trading in 2009 remembers when the rally started: it wasn’t when the news got good, but when the news started getting worse at a slower pace. So this survey is very interesting to read in light of the digital “Adpocalypse” narrative:

While buyers said they were making 33% cuts to their digital media spend one month ago, that number has since changed to just 29%. Meanwhile, their cuts spend in traditional media (linear TV, out-of-home, direct mail, print and radio) are now estimated to decline 44%, compared to just 39% in the previous survey.

If you compare what ad agencies are saying to what Internet companies are saying, there’s a serious disconnect in the ad spending narratives. Non-digital advertising may fill that gap.