Post-COVID-19: What Will Be Normal Five Years From Now?

Plus! Podcast, Crude Contango, Supply Chains

Note: This is the once-a-week free edition of The Diff. Enjoy. Paywalled pieces in the last week included:

  • How the Fed can use targeted quantitative easing to let states and cities run their own emergency bailout programs.

  • A natural experiment showing that the market is still mostly run by people, not robots.

  • Slow feedback as a hedge against crises.

  • A Bill Gates retrospective

Post-COVID-19: What Will Be Normal Five Years From Now?

I first visited New York in April of 2006, and moved there that May. I had never experienced pre-9/11 NYC, so to me it’s always been normal that you can’t get into an office building without showing ID, and that downtown is full of giant concrete barriers. Investment banks have always been in midtown, not downtown.

Crises always have an obvious immediate impact (death, destruction), first-order effects that revert soon after (post-9/11: a near complete collapse in the commercial aviation industry, various foreign policy adventures that are slowly winding down, a housing bubble), and nth-order effects that persist for a strangely long time (the aforementioned New York quirks, the TSA).

Some effects of COVID-19 will doubtless mean-revert: even though restaurants, hotels, airlines, and theme parks can’t survive in a locked-down and virus-aware economy, they are viable businesses. Six Flags might need to go bankrupt again, but between seasonality and herd immunity, people will stop getting sick and come back to roller coasters some day.

Cruise lines are likely toast, though. Not only are they a common source of mini-epidemics, but they’re a politically un-sympathetic group. Their customers and investors are disproportionately American, but their crews don’t get US minimum wage or occupational safety standards, and cruise operators don’t really pay taxes. All this was true before; what’s different now is that more investors are aware of just how bad an epidemic can get. Cruise ships are long-lived assets, and it takes years to build them (roughly 3-4 years from order to construction, with a useful life of about two decades).

Historically, the industry has been brutally cyclical: capacity rises based on demand, but it rises on a lag, so the biggest capacity increases happen during recessions. Like any other company that owns assets that last a long time, cruise line operators lever up to buy ships, which means they face high fixed costs. This epidemic likely represents a permanent impairment of cruise companies’ perceived credit quality, which may make it simply unaffordable for them to operate more ships.

This doesn’t mean the industry disappears overnight, just that most of today’s operators restructure (i.e. go bankrupt), and the industry enters a long-term runoff; existing ships continue to sail, but the fleet slowly shrinks to nothing.

More of the economy is dominated by Amazon. Next week, initial claims will likely top a million next week, up from 280k yesterday. But Amazon is hiring 100,000 people and raising wages as their delivery network is swamped. They’ve had to suspend shipments of goods other than medical supplies and staples for the next few weeks. That has the obvious implication that they’re getting enormous order volume for those products, and the non-obvious one that they’re getting plenty of orders for other things, too. (Expect a run on coloring books, toys, and—sigh—tablets, as more parents learn what it means to work from home when their kids are home, too.)

And that’s just Amazon’s retail operation, the loss-leader they used to bootstrap their real business, cloud computing. AWS earned a 26% operating margin last year, compared to 5% for Amazon overall. Amazon’s margin mix has always consisted of a few ultra high-margin products and lots of breakeven ones that fund them (they can lose money on Kindles if it means selling more ebooks, and like any grocer they know they can afford losses on healthy food since high-margin snacks have a way of sneaking into shopping carts).

If Amazon Retail didn’t exist, Amazon would still be holding the economy together by keeping the Internet going. Internet infrastructure is already under immense strain—the EU literally asked Netflix to reduce quality. Since AWS’s margins rise with usage, it’s also a far more profitable asset than it used to be. (There’s some chance the incremental operating expense from adding capacity will hide this in Q1, especially since setup costs are slightly frontloaded compared to revenue. But Q2 will be incredible.)

Even better for Amazon, this crisis has led to a weird rise in social cohesion: even though we’re all very emotional, I can’t think of anyone who has been career-assassinated over a verbal gaffe. And we’ve produced lots of gaffes! High levels of social cohesion are good for the current elites and bad for underdogs; even if what’s literally happening is that small businesses are dying and Amazon is thriving, nobody’s going to suggest that we break up Amazon over this in the next few months.

Fewer flu deaths. So far in Japan, COVID-19 has produced a negative body count. Japan has had 19 deaths from COVID itself, but flu cases are down 30-70% due to more social distancing, more masks, and more hand-washing. That won’t necessarily hold; COVID could worsen significantly for Japan, and the flu season is almost over. While COVID is a bigger deal than the seasonal flu, because of its higher mortality rate, rapid doubling time, and absence of herd immunity, a long-term change in attitudes towards infection will push flu deaths down. Masks got popular in Japan after the Spanish Flu, and there’s a chance that this trend, too, will catch on in the US—at least if people are still dying 90 days from now, when a new mask production facility can get approved.

Higher status for technologists, lower status for journalists; ambiguous changes for teachers. Vox will never live this down. The people most freaked out about COVID-19 were people who pay attention to exponential growth for a living; if you design viral mechanics for apps, you probably have a good idea of how to extrapolate 33% daily growth. The journalistic establishment has largely caught up at this point, and in fact they’re doing quite good work covering developments in social distancing, hospital capacity, and government responses. But every day of delay costs a lot of lives. The people who got this right will get, and take, credit; the people who got it wrong won’t be trusted on big issues again.

Teachers’ social status is more complicated. Lots of parents are learning how hard it is to keep kids focused and entertained for an entire schoolday—but they’re also starting to suspect that the role of school is not so much to teach as to keep kids out of earshot and make sure they don’t maim themselves. We might end up saying that teaching is a more important job than we thought, because it’s a different job than we suspected: as it turns out, it takes patience and conscientiousness, but we don’t need to pay up for an advanced degree. You don’t need to have the literary and analytical skill to teach Lord of the Flies, just the sang-froid to prevent it.


I’ve been ramping up my consulting business lately. Oddly enough, straddling two different career paths makes it possible that I’m the right person for a variety of different projects—recent work includes qualitative and quantitative investment deep-dives, email marketing, and full-stack growth hacking.

If you have a project somewhere in the venn diagram of online marketing, investment research, and writing, hit reply and let’s talk.



I had my podcast debut a few weeks ago, and spoke to Erik Torenberg of Village global for four and a half hours about basically everything, from geopolitics to financial bubbles to the game theory of marriage to demographics. Listen to part one here and part two here. This was fun; not my last podcast.

Crude Contango

One financial indicator I like is the free-money indicator: is there a way, right now, to make a risk-free profit? Normally, the answer is no. Today, the answer is yes: buy a tanker’s worth of oil today. Put it in a tanker. Sell a tanker’s worth of oil futures. Wait. In 2008, the reason this worked was that nobody could fund the trade: to execute it, you had to pay upfront for oil and pay upfront for the tanker, and essentially no one had capital. Today, it sounds like the profits mostly accrue to tanker companies. So as a straightforward read on whether or not capital can be deployed in oddball trades: the market is not experiencing 2008 levels of stress, despite the 2008-sized drawdown.

The Other End of the Supply Chain

China is largely back to work, as evidenced by better availability for most Apple products, but places like Malaysia are now experiencing delays. A few weeks ago, it looked like COVID-19 was largely a China story, and that the disruption could be contained in other countries. Now, of course, everyone else is going through a crisis similar to China’s, on different time scales—South Korea and Singapore acted early, and have had little interruption in their manufacturing. Iran and Italy acted too late. The US is an unclear case, due to fractal incompetence: until we get more tests, we won’t know how far the contagion has progressed, so we won’t know he far behind we were.

Either way, this forces me to reconsider a claim I made a few weeks ago: back then, I argued that the epidemic shows how dangerous it is to have so many supply chains running through Shenzhen. Now, it looks like the real danger is having so many supply chains run on a just-in-time basis. It’s a question of time rather than space; for the right kind of risk, nowhere is safe.