The Diff - Credentialism and the Money Illusion

In this issue:

• Credentialism and the Money Illusion—Money and credentials are both useful abstractions that allow us to interact with a complex world without dealing with all of that complexity. But they run into a similar set of problems when they're overused.
• Capacity Constraints—A big fund makes the surprising choice to let itself get a little bigger.
• IPO Window—Time for the IPO cycle to turn?
• Strikes and Opportunity Costs—The same features that make ski resorts a good business to own make them a good business to unionize.
• The Upside and Downside of Ads—Running ads means running the risk of associating your brand with something unpleasant, and this risk ultimately helps make the biggest platforms more dominant.
• Backcasting Returns—Quantitative models are powerful, but you never sample from exactly the same distribution you backtested on.

The Diff January 6th 2025 14:01 • Click to play audio

Credentialism and the Money Illusion

In a financial system, you can get by with a sort of writing-free financial system, where tangible wealth is transferred in the form of physical coins that are made of precious metals, and intangible wealth consists of reciprocal obligations (so a medieval peasant probably gets protected by their local lord in exchange for definitely paying that lord a cut of the harvest, and some of that cut gets paid to the local monastery in exchange for which the monks periodically hold a mass for the soul of said lord's ancestors). These systems could be quite sophisticated, since so many obligations were in-kind trades of different varieties, and they implied a pretty elaborate system of exchange rates—the system above is implicitly setting an exchange rate between masses per year and pillagings averted, but it doesn't explicitly do this. A modern dollar-based system does set an implied exchange rate between all available goods and services, but we don't ask those questions ("How many bushels of apples did you say that Tribeca 3-bedroom was?") because the dollar is such a convenient tool for measuring everything. That convenience leads to a “money illusion,” where people fixate on nominal changes and don’t look at real numbers. But in a way, that’s perfectly rational, because the nominal values are right there, and the real ones are estimates.

That shift to a more universal economic abstraction was supported by an edifice of more particular abstractions, namely ideas like:

1. Transactions are values-neutral by default; your obligation to Starbucks ends when you pay them, and their obligation to you ends when you get your order.^[1] Similarly, your job is not some permanent, immutable, and possibly heritable aspect of who you are, but something that you or your employer can change their mind about at any time.^[2]
2. Obligations are transferable, and that compounds: a house can support a mortgage, that mortgage can be bundled into a mortgage-backed security, that security can be referenced by a derivative, and that derivative can be used to neutralize or double-down on risk somewhere else.
3. As a related note, most of the money people interact with is some form of credit, which almost never gets settled physically. You can buy something with a credit card and autopay your bill after receiving your salary in electronic form through direct deposit, turning one purchase into three separate transactions, not a single one of which involves ever touching cash or coins.
4. As a corollary to that, these abstractions flex in size and importance based on confidence in the future, and the future they're confident in is increasingly the output of these complex layers of economic abstractions. The most important of these is credit: a credit crunch is partly a process whereby some market participants discover that things they thought were roughly equivalent to cash were riskier than expected, and when the moneyness of an asset declines after someone has levered up to borrow it, the moneyness of the lenders' assets declines, too. In the financial crisis, 2007 was the year that mortgages started getting impaired, but late 2008 was when that resulted in large swathes of the banking system lacking the liquidity necessary to roll over debts that propped up the value of mortgage-backed assets and derivatives.

Credentialism follows a similar parallel to this moneyness. Most of the time, and for most purposes, you do the rational thing and treat a credential as synonymous with the skill required. Years ago, I needed a lawyer to look at a contract for me, and I put very little effort into evaluating the skill of the lawyer in question—the skill to do that is pretty close to the skill required to just look at the contract myself and figure it out. There's a narrow band where you can hire service providers whose skill you're able to assess—plenty of people dining at restaurants are perfectly capable of cooking a meal at home—but also large areas of activity where the optimal strategy is to assume that credentials are a good guide to what's actually going on. Assuming that someone who went to law school and passed the bar knows how to read a contract is almost precisely analogous to assuming that the numbers on your screen when you check your bank account are equivalent to cash you can spend from that account.

But, as with financial crises, credentials can get overextended, and sometimes they're borrowing against expertise that isn't really there. And that's hard to spot, because credentialing organizations are self-perpetuating. An accountant is anyone who's followed a formal process of convincing other accountants that they're an accountant; a dentist does the same thing with dentistry. It's like borrowing against a loan, which is something that banks do all the time—and that, from time to time, makes their depositors worry that the money's disappearing.

In finance, you can trace the economic logic of some purely financial transaction like buying calls on the pound back to some real-world referent—Britain has an economy and produces goods and services. It has a government that taxes this output, and those taxes create demand for pounds. Pounds get borrowed and lent at some rate that's determined by policy in light of economic reality, and rate differentials tend to push currencies around. So, you can connect this abstraction to real-world behavior and ask if it makes sense. This kind of question is a useful one to ask, because sometimes it doesn't make sense. During DeFi Summer in 2021, there were lots of ways to lock up cryptocurrency and earn interest, but it was harder to understand what economic activity could be best funded with Terra, and which would earn a return higher than the 20% cost of capital. Trace the connections, and you found that these crypto rates were driven by one of three things: pure ponzi schemes, martingale bets against volatility (hello again, Terra!), or, at best, funding margin loans for crypto speculation.

For credentials, the equivalent thought experiment is to replace "I heard it from a lawyer" or "I read it in the science section of the New York Times" with "I heard it from Some Guy." Some Guy might be right, or wrong, but it's good to trace the process whereby Some Guy becomes, say, a science correspondent, and to ask what this process selects for and against. This is also a useful heuristic when faced with Credential Smearing, where someone who has a meaningful title in one domain opines about something that's related but not necessarily adjacent. Doctors interact with health insurance companies all the time, and they're close to the front lines when patients find out that a proposed treatment isn't going to be covered. But they aren't looking at the margins of health insurance companies and comparing them to the financial performance of their own practices. (To the extent that you're worried that a small cohort of highly-paid individuals are responsible for American healthcare costs, it's good to know that the single most common job among the top 1% of earners is "physician," at 20% of the total. And this probably underestimates their prevalence in the top 1%, since their compensation is more stable than that of salespeople, managers, or the best-paid programmers who are paid disproportionately in commissions or stock.)^[3] Of course, someone who does have the relevant context works in insurance, and has some biases about how generous health insurance should be and how defensible their messy paperwork is. On the other hand, those payouts are doctors' revenue, and the lower the rate of claim denial, the more predictable their income is.

This doesn't preclude them from being right! Nobody is stopping them from reading a few 10-Ks, or looking at differences in healthcare consumption per capita and how wages change as countries get richer. But the Some Guy heuristic helps to put this in context; there isn't a specific reason to think that doctors would have a deeper view of healthcare economics, as opposed to a view of what might be causing your symptoms, any more than you'd expect a plumber to be able to derive the Navier-Stokes equations.

Credentials and credit are still enormously useful social technologies. The world is simply too complicated for everyone to be an expert on everything, and that complexity depends on both the existence of experts and some means for the rest of the world to identify them. But those mechanisms will never be perfect, and sometimes fall prey to incentive problems. Credentials and credit always want to expand, and the place where they want to do it the most is where they sacrifice a bit of quality. In light of that, it's something of a miracle that they work as well as they do, but it's never something to count on.

Diff Jobs

Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:

• A premier proprietary trading firm is looking for smart generalists to join their investor relations team, working with external investors, rating agencies, and the internal finance team. Investment banking and/or investor relations experience preferred. Quantitative background and technical aptitude a plus. (NYC)
• A Google Ventures backed startup founded by SpaceX engineers that’s building the data infrastructure and tooling for hardware development is looking for a staff product manager with 5+ years experience, ideally with AI and data intensive products. (LA, Hybrid)
• Ex-Ramp, Truework founders and team are hiring a high energy full-stack engineer to help build the automation layer for the US healthcare payor-provider eco-system. (NYC)
• A well-funded startup that’s building the universal electronic cash system by taking stablecoin adoption from edge cases to the mainstream is looking for a senior full-stack engineer. Experience with information dense front-ends is a strong plus. (NYC, London, Singapore)
• The treasury management arm of an established stablecoin project is looking for a research economist to bring a macro perspective at the intersection of traditional monetary theory and digital currency innovation. Advanced degree and portfolio of high quality research output preferred; no crypto experience necessary. (Remote)

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.

Elsewhere

Capacity Constraints

When you look at the returns of the best hedge funds, one temptation is to think of them as compounding numbers. But the better the fund, the less compounding you get. (At Renaissance, for example, outside investors were unable to add new money after a while, and got bought out entirely in the early 2000s. For the last few years, Citadel has been letting outside investors maintain consistent positions, but not reinvest their profits. This year, they gave them the option to reinvest, and almost all of them took it. This is a surprising change, because if anything the business has gotten less capital-intensive over time, while the opportunities to add new return streams have diminished. If there's something that produces incremental Sharpe or allows a firm to deploy a bit more capital with the same risk/reward profile, they've probably considered them. And the better the firms get at managing risk, the less capital they need for a given set of positions. So it's notable that at a time when peak pod narratives are strong, at least one big pod wants to be bigger.

IPO Window

Might Citadel (and perhaps others) be anticipating a newly-reopened IPO window in 2025 ($, FT)? After the peak in 2021, non-SPAC IPO volume has been quite low—in the last three years, IPO activity was actually slower than even pre-pandemic, because many of the companies that should have waited a few years had their offerings pulled forward in that bull market. But every year, more VC funds get closer to the end of their life, and the companies that didn't burn through all of their capital after the slowdown in 2022 have straightened things out and gotten onto a reasonable growth path. It'll be a while before we see anything like the Covid-era IPO peak, but the pace of offerings is more cyclical than the pace at which IPO-worthy companies are produced.

Strikes and Opportunity Costs

Is it worth it for a fairly levered ski resort operator to risk a good ski season in order to save $2/hour on some of its labor costs? Vail Resorts seems to think so, and is holding out on offering ski patrollers a raise, leading to long lines and irate customers. One feature of the economics of organized labor is that threatening a strike is much more effective in businesses whose economics are dominated by utilization of fixed assets. So you see unions in hotels, airlines, and steel; you don't see as much union activity in asset-light businesses, because there aren't as many other costs the company is paying when it's not paying workers. And you see them in ski resorts.

Vail is happy to tell investors that part of the bull case for their business is that there just won't be many more ski resorts in the US any time soon, and that building new brands is even harder. So funneling a growing number of customers into the same set of lifts, hotel rooms, equipment rentals, etc. is a winning model. Vail has also shifted its economics from selling one-off tickets to season passes, which means that one of the fixed assets they're trying to maximize utilization of is their brand name and customer database. Strikes have plenty of leverage in disrupting this, but it cuts both ways: Vail is negotiating against all future unions, not just the current one. As long as it has a longer time horizon than any given counterparty, it has an incentive to hold out for a slightly better deal.

The Upside and Downside of Ads

News sites have struggled to attract display advertisers, who often blacklist content so widely they refuse to run ads on content like crossword puzzles and recipes ($, WSJ). As with so many online advertising phenomena, the root of the problem is scale: if you whitelist a given kind of content, there's always some risk of a false negative, i.e. content that ticks one box for being innocuous even though the content isn't (suppose an advertiser targets all content on a news site that involves cooking, and finds that they're running ads on stories about cannibalism or about industrial accidents at food preparation facilities). So their natural instinct is to be extremely restrictive. This is a case where a black box model can outperform, because someone who's using a complex machine learning model to determine context, rather than matching to a predetermined list of words, will probably catch those edge cases. Those models need a big sample to work, but the better they get the more they produce a long list of close calls that can be used to manually refine the model. So that approach improves at a compounding rate, and means that both advertisers and publishers a) get better performance, and b) have less information about what actually drives it.

Backcasting Returns

Cliff Asness has a piece imagining an asset allocator in 2035 looking back at the prior ten years. It reads equally well as an inversion of an allocator in 2025 looking back at the last ten years. It's true that over time, trends tend to mean-revert—at some point, US stocks are so expensive relative to the rest of the world that investors are overpaying for higher-quality American companies, eventually value stocks are simply too cheap, at some point the premium people pay for private equity's volatility laundering offsets the premium from PE firms' asset selection and deal structuring, etc. But it's also useful to remember that deviations in these long-term returns can be very long-term themselves, and every year of data slightly shifts the long-term returns we should expect. Some factors might produce durable excess return indefinitely, but in other cases they produce excess return until they're identified and exploited, at which point bets on mean reversion just create opportunities on the other side. For example, consider the bet against US stocks. The further back you go in history, the more national stock markets are primarily bets on the local economy, but as industries globalized, "US" stocks were really global companies that happened to have a primary listing on a US exchange. So the Magnificent 7 don't mean-revert to a valuation cohort dominated by AT&T, GM, and Exxon, but to a different mean entirely.