Short Stack III: fake jobs aren't really a thing

The pool Tiktok girl has a valuable job, whether you want to hear it or not.

Photo by MariaPolna from PxHere

Good morning, and welcome to our third “short stack,” a collection of smaller news items we thought were interesting throughout the week. This post, like many short stack posts, will be for paid subscribers. I have plenty of interesting things to discuss with you today, so let’s get to it.

Time for some game theory

Contestants on The Price is Right have been getting the game theory wrong for decades, according to four economists. Bouke Klein Teeselink, Dennie van Dolder, Martijn J. van den Assem, and Jason Dana examine how contestants play the Showcase Showdown, and come to the conclusion that contestants are “myopic.” They don’t see far enough ahead.

This game is a bit like blackjack. You want to get as close as you can to the target without going over. The target number is 100, and you spin a wheel with each multiple of 5 up to 100. After your first spin, you are allowed to stick with the number you have, or spin the wheel a second time and sum your two results⁠—but this puts you in danger of exceeding 100 and losing.

Three contestants play this game in order, and the one with the highest score (so long as it does not exceed 100) wins. There are some bonus points and other details, but these aren’t too relevant to understand the findings, so I’ll skip them. The basic question for players comes after their first spin: given the spins you’ve seen so far, do you elect to do a second spin?

Game theorists are very good at “solving” sequential games like this. (That is, describing how you should play, in theory.) You work backward. You start from the final player’s turn: his decision is really easy. If he’s behind the best score so far, he should spin a second time, and if he’s ahead, he should take his win.

Once you know how the third player will play, you can look back to the second player’s turn, and realize that all of the probabilities are knowable. Now that you’ve modeled the third player’s behavior, he is effectively part of the machinery of the game. You could replace him with computer code.

So the second player has a more complicated task than the third, but in the end, they can multiply out the expected values and probabilities, and determine whether they’re likely to win more with a second spin or without one.

Finally, you can iteratively repeat that process for the first player, treating the second player and third player decisions as solved, multiplying out values and probabilities, and figuring out the right strategy.

This is right in theory. But do contestants do this in practice? Depends on which contestant. Contestant 3 has a pretty simple game (spin to beat the best score available, unless you already have it) so they unsurprisingly get things right.

A fair number of Contestant 2 decisions are easy, as well. If you don’t have the high score, you obviously spin. But if your first spin puts you into the lead, you need to consider whether or not to spin again to shore up that lead. And there, Contestant 2 gets some decisions wrong. Sometimes they spin a second time when they shouldn’t, and sometimes they abstain from a second spin when they should take it. But there’s no obvious bias⁠—they underspin and overspin about equally. This suggests they have the right instincts, but they’re not perfectly precise.

The real trouble comes with Contestant 1, who almost always spins too little. As contestant 1, you want to play aggressively; people will know exactly what your score is, and they will attempt to beat it, and unless you get an exceptional score there’s a good chance at least one of your rivals will beat you later. The economists believe you should generally take a second spin even if you score 60 or 65, but contestants often stood pat with those scores.

The authors venture explanations for why Contestant 1 might be too timid. One theory they consider is the idea that taking a second spin counts as “action,” while staying put counts as “inaction,” and people fear incorrect actions more than incorrect inactions.

This is wrong strategically, but it could explain Contestant 1’s underspinning. The problem is that such a theory should predict Contestant 2 would underspin as well.

Instead the authors attribute Contestant 1’s overly-cautious play to “limited foresight.” The idea is that both Contestant 1 and Contestant 2 play as if there’s only one player left, and imagine what kind of score they’d need to beat that player. But Contestant 1 should actually be trying to record a score that withstands two challenges, not just one.

One of the more amusing observations from the paper is that contestants have grown smarter at spinning over time⁠—but only at a glacial pace. Even after 40 years, many contestants (about 24 percent) suffer from the limited foresight bias.

Not so fast, Trump CEA

Two former Trump administration advisors, Kevin Hassett and Tyler Goodspeed, took a victory lap in the Wall Street Journal over what they claimed were good results from the 2017 tax reform. While there are some substantial accomplishments in the tax bill, these advisers overstate their case in one key respect: they claim credit for high corporate tax revenues, which were much higher than expected in our recent recovery.

They are right that corporate revenues have come in above expectations. But they are wrong to suggest the tax reform has very much to do with it. A multi-think-tank collaboration between tax policy experts Kyle Pomerleau, William Gale, and Steve Rosenthal throws cold water on the claim.

Instead, high corporate revenues are an after-effect of the COVID-19 relief bills, which made money and nominal incomes (not adjusted for inflation) much higher. Every money-related quantity in the economy is up, including tax receipts. Structural reforms have little to do with it.

The authors also add a more subtle point. Prices tend to reflect new macroeconomic conditions (for example, high aggregate demand) faster than wages do. So if you have money rushing into the economy, it will probably translate into high corporate profits first, and only later will workers begin to lock in their gains. This means that the corporate income tax is a bit of a leading indicator, and its numbers are more responsive in the short run to cyclical effects than personal income taxes are.

Fake jobs aren’t really a thing

This week there was impassioned discussion on Twitter about the value of management jobs generally, and that of the product manager position specifically. It was prompted by a video from TikTok user durbinmalonster, who explained her job as a product manager at a financial technology startup while working from a pool.

The video became highly-trafficked for a variety of reasons (women in bathing suits, arguments about working from home) that I won’t address here. But it also sparked debate about a style of job⁠—you might call it a “laptop job” or “email job”—in which the employee’s productive output isn’t necessarily tangible, measured in widgets or lines of code. Instead, it’s a bit more nebulous or organizational.

A fair number of commenters thought such jobs were worthless. Some cracked (admittedly funny) jokes:

But others were relatively serious about it, tying it into some larger theory of jobs without a purpose. The late David Graeber wrote a book on this, “Bullshit Jobs,” contemplating the existence of jobs he perceived to be meaningless. And many replies to darbinmalonster’s video referenced that book.

That goes a little far for my tastes. Product Manager is a real job. And Bullshit Jobs was not a very good book.

The video in question describes product management for credit cards: while the programmers write the code, the product manager figures out what users need and helps prioritize the matters that are most important. It’s not too much more complicated than that. Would you want a product manager for a tiny independent videogame company? Probably not. But for a large financial services company? Absolutely.

Several years ago, I worked on coding for tax policy models that were used to help advise Members of Congress on the likely consequences of their tax ideas. For example, if you expanded the standard deduction by $1,000, how much revenue would that cost the treasury per year?

Though I was at a nonprofit, and the coding was not nearly as difficult as that of corporate professionals, this was a software “product” of sorts. It became popular with politicians. As the team grew, a natural division of labor emerged between those who took the model on the road to advise politicians, and those who wrote the underlying code. Those in the former category had a better idea of what kinds of tax policies politicians were thinking of proposing⁠—and therefore, what kinds of features needed to be added to the model. But those in the latter category had a better understanding of the workings of the code.

I did a little bit of both of these roles, and I don’t think of one as inherently better than another. But they were both necessary. Without knowing the code, you can’t create features, but without knowing the users, you can’t know how well the features are working for them.

It’s hard to keep doing both roles for too long. Eventually there’s too much work to do, and specialization is better. In a followup TikTok, the much-maligned product manager offers a succinct comparative advantage argument: "I in no way think that engineers are not capable of doing what a product manager does. However, engineers are expensive!" Even if a highly-competent engineer could do her job as well as her or better, that engineer is likely even more valuable using their primary skills.

I also want to make a more general point about jobs that “don’t do anything.” Unless it’s required by a law you disagree with, the job is probably useful. People don’t file W2s and sign paychecks and enroll you in their health insurance plans for fun. If you read Graeber's original “bullshit jobs” essay, he has an elaborate theory that these jobs are given to people to keep us from working 3-4 hour workdays or organizing against the ruling class. But, well, the evidence of this is scant, to say the least.

Moreover, the bullshit jobs that Graeber names, offhand, aren’t particularly compelling. One that jumps out at me is “actuary.” There are a few tens of thousands of actuaries in the U.S., maybe 50,000 at most, and they run the math to make sure that pension funds, health plans, and insurers don’t run out of money.

Actuaries are about 0.03% of the U.S. workforce. Just 50,000 in a sea of 150 million people. And their job seems perfectly reasonable to me. Perhaps you’re braver than me, and you want to “chance it,” just let pension and health plans and insurers take money in and pay it out without ever considering the math, and hope that none of them blows up in your face. But I’m happy to spend 0.03% of our labor force for that peace of mind.

What do people mean when they call something a “bullshit job?” I think it means “somebody who works on a computer in a way I don’t understand.”

The fate of commercial real estate

Our friend Arpit Gupta has a new paper, and an accompanying Substack post, on the value of commercial real estate in New York City post-pandemic. He and his co-authors, Vrinda Mittal and Stijn Van Niewerburgh, estimate a long-run decline in value of about 28 percent. But there should be pretty large error bars on this estimate. If you read the Substack (or even more so, the paper) you get a sense of what a herculean task Gupta and his co-authors have taken on. Property value is a complicated net figure, the excess of rent and price appreciation over physical depreciation. Net figures are inherently quite wobbly, especially for low-value properties where rent barely exceeds depreciation.

Moreover, most commercial real estate is either privately owned, or part of a large conglomerate, making it hard to identify market value. The authors use publicly-traded REITs to help act as proxies for these properties without market values.

The authors also realize that the value is heavily sensitive to what they call “work from home risk.” So they build out a factor measurable in public markets to get a sense of how big that “work from home risk” is. Essentially, it is the performance of virtual-world firms like Zoom or Docusign, against real-world firms like Las Vegas Sands or Norwegian Cruise Lines. This lets them show how much value is lost in heavy work-from-home scenarios versus light work-from-home scenarios.

There are even more issues to consider: many properties had ongoing leases during 2020, protecting them from the pandemic shock for a time, but would show those rents to be too high if renegotiated under a sustained work-from-home scenario.

And in the longer run, commercial real estate may have some substitutability into residential real estate, creating a relief valve for the troubled asset class. All in all, there’s a lot to consider. I’d describe these numbers as just the beginning of the solution to a highly complex valuation problem, but they deserve a lot of credit for putting it down on paper.

The most compelling result I expect to have implications later: the authors find a “flight to quality.” The best office spaces hold their value relatively well, while the more marginal spaces lose out disproportionately. This makes a great deal of sense. If you’re trying to entice people to come into work, you aren’t going to do it with marginal office space.

Odds and ends

Matt Yglesias: Progressives should admit Joe Manchin was right. As I’ve argued before, his points about the frontload gambit were correct⁠—and they were ignored.

Washington Post: Rents are rising fast even in manufactured homes. As we’ve argued, more could be done to fix this at a policy level. The big picture is that it’s not actually expensive to build trailers for people to live in. What is expensive, though, is the right to place your home somewhere.

Joey Politano: Extra savings from the pandemic are beginning to run down. I wrote earlier that some savings are being sent abroad to pay for a big trade deficit. And some are being returned to the government⁠—for example, in the form of the record corporate tax revenues analyzed by Pomerleau, Gale, and Rosenthal above. Politano notes that most of the pandemic’s extra private savings are a sort of mechanical consequence of government dissaving during the pandemic. As the government tightens its belt to appease fiscal hawk Joe Manchin, that trend is reversing.

Josh Barro: Republicans Found the Weak Point of 'Woke Capital' and Are Beating It. Josh responds to yesterday's post.