The US has fourteen restaurants that have earned three Michelin stars. Of these restaurants, nine are located in either New York City or the San Francisco Bay Area. There are all sorts of historical contingencies that make this so, but you can also explain it with one of the most delightful paradoxes of microeconomics: the high cost of literally everything in these cities makes over-the-top luxury relatively affordable. This effect (known as “The Alchian-Allen Effect”) was observed by economists Armen Alchian and William Allen in the early 1980s, and is best explained with a stylized example: suppose that in Florida, a regular orange costs $0.50 and a premium orange costs $1.00. That makes the "exchange rate" between the oranges 2:1—that is, a Floridian forgoes 2 affordable oranges for every expensive one.

Now, imagine that it costs fifty cents to transport any type of orange to New York City, which makes premium oranges $1.50 and regular ones $1.00. Therefore a New Yorker forgoes just 1.5 affordable oranges (instead of 2!) for every expensive one. And as a result, relatively more premium consumption happens in New York, and the median quality of oranges consumed in Florida is lower.

But like every good microeconomic theorem, this makes all sorts of objectionable assumptions. Aren't fresh oranges tastier? And if you're breeding oranges for easy transportation, are you selecting against flavor and juiciness? Regardless, the model holds true: the higher the fixed cost of a given category of good is, the lower the relative cost of the more expensive variant.

Okay—now turning back to restaurants: yes, rent and waitstaff would be a lot cheaper if 11 Madison (a Michelin starred restaurant) were located in Madison, Florida (median household income: $18k) instead of in Manhattan (median household income: $73k). But the folks in Madison, Florida would be sacrificing many, many nights of regular restaurant meals for every 11 Madison dining experience, while the folks in Manhattan would only be giving up a few regular restaurant meals.

Zooming out to the level of cities, there's a powerful self-reinforcing effect that makes superstar locations outperform: if the cost of office space is constant regardless of who is in that particular office, a hedge fund is likely to keep quants and analysts in the city, and move its back-office processes somewhere cheaper. Similarly, a tech company in the Bay is likely to have its AI researchers at the headquarters and its customer support team somewhere more affordable.

As lower-wage jobs get priced out of the city, the median wage rises, which further ratchets up the cost of real estate. And since so much economic activity requires physical space, that increases the cost of everything else—exacerbating The Alchian-Allen Effect that helped drive the cost increase in the first place.

Alchian-Allen is also useful for predicting something else: the price sensitivity for buying critical products. Headlines to the contrary notwithstanding, big companies would generally prefer not to be hacked. The cost of a hack is high, both in terms of direct economic impact and in terms of laboriously training everyone at the firm not to give their passwords out, not to click on suspicious links, etc. Since adopting any kind of cybersecurity product and associated procedures already has a fixed cost, there's less price sensitivity around the marginal cost of the best versus second- or third-best option. So the best cybersecurity companies have a fair amount of pricing power (at least conditional on solving the "be perceived as by far the best" problem, which is of course a nontrivial task).

This model also helps explain salaries. With an experienced team, one of the biggest costs of adding a new entry-level employee is that they'll either ask distracting questions or make expensive mistakes. And one way to get around this is to be very choosy, and then to win a bidding war among other similarly-choosy employers. It's not that the engineers hired at elite firms are necessarily that much more productive than less fortunate 22-year-olds; it's that the implicit fixed cost of adding one new employee of uncertain ability is probably comparable to that employee's expected salary, so a 50% pay bump to get the ideal candidate is really more like a 25% premium on the all-in price.

You can use this model to your advantage. As in many parts of life, a big chunk of success is choosing a domain in which success is relatively easy, either because other people aren't trying very hard or because you're doing something that plays to your strengths. And the easiest way to do this is to get good at a set of complementary skills, such that skills A or B provide a decent backup plan, but companies hiring at the intersection of A plus B have a limited set of options. (A nice lemma here is that the more specific a company is about what it needs, the more desperate it is for talent—plenty of companies need CFOs, for example, but there's probably only one in the market for a CFO who is familiar with cryptocurrency derivatives, fluent in bankruptcy procedures in the Bahamas, and has a thick skin.)

And the other way to use this rule is to illuminate hidden factors behind what seem like irrational decisions. Any time there's a job, or a product category, where pricing starts going nonlinear and it's hard to see why it could be that valuable, it's worth knowing that there's probably a big fixed-cost iceberg underneath the visible bit that's poking out.

Share Capital Gains

Subscribed readers can participate in our referral program! If you're not already subscribed, click that Get Started button below and we'll email you your link; if you are subscribed, be sure to check out the email version of this edition.

Help us out!

We'd like your feedback so we know what to cover, so if you haven't already, please fill out this one-minute form.