Weekly Wrap-Up: The War, Jobs/Wages, Hot Hands!

The war drags on, as Iran learns the power of the card they hold; low breakevens in the job market; I may never have had the hot hand in basketball, but I've always thought it's a thing!

Happy Easter! I’ll be brief on the first two items, so you’ve got the time to get into the last one, which involves two of my favorite things: whiskers on kittens basketball and probability!

The War

I did a deep dive on war econ this week, and none of that has changed, expect oil and gas prices have gone up since then. I also learned a lot from listening to Catherine Wolfram on a war-econ-update webinar in which we both appeared (here’s my deck).

One point we leaned into is the fact that oil is sold on a global market (see figure below; source: GS). Many people have asked me, “if U.S. oil and nat gas production has gone up so much, why aren’t we insulated from the chokepoint of Hormuz?” The answer, as Catherine put it, is to think of the oil supply not as bunch of countries holding their own buckets of oil, but as a big bathtub, wherein the spigot delivers oil from all the producers around the globe.

Okay, well then why not enforce a move from the communal bathtub (gross!) to individual buckets, i.e., export restrictions? One reason the U.S. natural gas price has stayed lower than Europe’s, e.g., is that we produce a lot of it but we don’t have the same export capacity we do for oil, gas, jet fuel and the other distillates.

Some countries, most notably China, have been curbing exports for awhile, in China’s case, mostly refined product.

And I wouldn’t put it past the Trump admin to go there, though not before the gas price goes a lot higher. Major US oil producers, a group Trump listens to and a group that’s reaping handsome profits right now, would scream their heads off if the admin went there.

But the more important evolving story regards the key variable that determines war econ outcomes above all others: duration. The longer the war goes on, the greater its impact. Yes, at some point other oil pathways could be developed that substitute for the SoH, but, as someone who went through the ‘21-’22 supply-chain snarl, with its accompanying inflationary impacts, you can trust me when I tell that energy supplies are near-term inelastic to price. The adjustment for now, and for weeks/months to come, has to come on the demand side, meaning higher prices.

Yes, shifting to renewable energy sources also makes sense, but of course, Trump 2 has us moving backwards on that one.

What we learned last week, including the shooting down of U.S. aircraft—an event Trump implied couldn’t happen given our destruction of their air defenses—is that, even amidst our clear tactical advantages, Iran recognizes that its ability to shutdown the SoH gives it, if not an upper hand, a very powerful card. They know the duration costs as well as we do, and they know they can keep throwing this Trump card down all day, everyday.

And as we can see—and “we” here includes the Iranians—up to this very morning, that’s making an already unhinged Trump go even crazier (I’ve cleaned this up for my family audience…but jeez—this dude is really out of control):

“Tuesday will be Power Plant Day, and Bridge Day, all wrapped up in one, in Iran. There will be nothing like it!!! Open the F***in' Strait, you crazy bastards, or you'll be living in Hell - JUST WATCH! Praise be to Allah," Trump wrote on his Truth Social account on Sunday morning.

For all such bluster, this is also why we’re starting to hear Trump talk about just getting the hell outta there are letting Europe et al fix what he broke. One worrisome outcome of that approach is the distinct possibility that the SoH becomes a toll booth. Every cargo ship with an EZPass goes through while transferring a $2 million (or whatever) payment to the thoroughly intact, as hardline-as-ever, Iranian regime.

If anything like that scenario is the outcome of this terrible misadventure, it will represent a huge blunder by the Trump admin, and, as I always emphasize, his Congressional enablers, the R majority that sits and watches, earning a fat paycheck for doing less than nothing. If that’s the outcome, Trump will have done nothing but reminded the Iranian regime that they have a powerful weapon to wield against the rest of the world, while costing thousands of people their lives in the process.

A Few More Points Re the Hard-To-Read US Job Market

Is the U.S. labor market terrible, as some (not me!) said in February, when we lost 133,000 jobs (according to the latest revision) or is it great, as some said when payrolls shot up 178,000 last month?

These monthly payroll numbers have been like Trump sentences: you start out with one point and end with a contrary point. But at least in this case, we can back out a coherent meaning, by running a smooth trend through the data:

The trend is currently in the 30K-40K range, which, if you’ve paid attention to these monthly numbers, or you just eyeball the figure, seems pretty low. And, historically, it is. We know hiring rates are historically very low, though thankfully, so are layoffs: low-hire, low-fire.

And so, importantly, is labor supply, largely due to deportations and the chill Trump/Miller have created in terms of labor mobility, leading to both much reduced inflows and heightened outflows. This figure, from a new paper with new data on deportations, says it all. Note the huge upswing and crash of unauthorized immigration as measured by these authors. To be clear, this is an estimate, and may not be right (other data shows fewer exits/self-deportations), but it’s very much in the ballpark.

The president of San Francisco Regional Fed, Mary Daly, featured the breakeven job numbers derived from that cliff-dive above, showing that labor supply is so low, that even if we added no jobs at all, we shouldn’t expect the job market to weaken much or for unemployment to go up.

I think the breakeven job number is a bit higher than that, probably close to about where we are. But I also worry that the job market is weaker than the 4.3% unemployment rate suggests. One reason for that is these nominal wage trends of mid/low wage workers. The 5+% growth rates in the very tight post-pandemic labor market were too high, so some deceleration was expected and necessary, inflation-wise.

But, especially with war-induced inflation going up, there’s a chance the yearly wage growth rates at the end of the figure will lead to near-term flat and even falling real wages. The March service sector (non-managers) wage growth rate is 3.2%, which is likely around where April inflation will end up (the Cleveland Fed’s CPI nowcast for April is 3.4%). This is especially problematic as productivity has strengthened lately, which, in a tight labor market, should boost mid/low real wage growth.

To be clear, I was as happy as the next wonk to see the 178K at 8:30am ET on Friday. But it is one noisy signal in an unusually hard-to-read, and imho, not great, labor market.

Is the Hot Hand a Thing?

I’ve been doing a fair bit of media lately, so when NPR tapped me for an interview on Friday, I was sure it was about jobs day or war economics. But no! It was, in honor of the Final Four, about the existence—or not—of the “hot hand” phenomenon, the idea that players can sometimes get on a streak wherein they hit a few shots in a row and that momentum boosts the likelihood that their next shot will go in. Instead of each shot being an independent event, like a coin flip, whose outcome is determined by your average shot percentage, that percentage goes up due to your hot hand. You’re in the zone!

Given that basketball and probability are two of my favorite things, imagine my excitement about getting to play some chin music at their intersection! Here’s the interview, with the great Scott Tong, who happens to be an old friend.

The history of this debate is highly interesting in itself. For decades, the debate was settled by this GVT paper, who’s subtitle says it all: “On the misperception of random sequences.” We humans are evolutionarily programmed to misinterpret random patterns as non-random. If the bush is agitated, the pattern our ancestors jumped to was “this is consistent with a saber-tooth tiger eating me” versus “the odds suggest that this is just the wind.”

An interesting proof of this is studies that show if you ask people to make up a list of random numbers, they’ll have too few repeats. But true strings of random numbers have streaks. They get the illusory hot hand!

But GVT made a consequential mistake, by not accounting for selection bias. Once future researchers made that correction, the hot hand emerged. As I suggest in the interview, this generated a big “I knew it!” from the rest of us, who had played enough ball to know the hot hand was a real thing.

Understanding GVT’s mistake is both interesting but counterintuitive, which is often the case with deeper probability questions. I’m not sure I can do it justice here, but there are many good articles explaining it. I find that this table gets the message across, but see if it works for you.

The question is should we not, as GVT did, test whether, once you’ve hit a few shots in a row, how the odds of making your next shot compares to your average? If it’s about the same, which is what they found, then no hot hand, right?

Wrong! When we’re looking at streaks in a limited sample (like shot sequences from a b-ball game), because some patterns fall out of the sample, the odds of a streak are biased down. Therefore, if you see shooters with a hot hand maintaining something close to their average, that’s actually proof of the theory.

Let’s try to see this using coin flips wherein the probability of each individual flip is 50%. But—this is the key—we’re not tracking individual flips. We’re tracking sequences, and that changes the math.

In this experiment of flipping a coin three times, evidence of a hot hand occurs when a head follows a head. In the first sequence in the above table, there are two chances for that to happen, so HHH represents a 100% success rate. THH is also a 100% success rate because there’s one chance for a streak to occur in that sequence and it does so. Etc…

But in the last two cases, there’s no chance for a streak to show up, so we must drop them from the sample. Now, if we divide the sum of the success shares by the number of valid opportunities for streaks, we get 2.5/6, or 42%, which ain’t 50%!

Gemini AI explains it like this, which some may find clearer:

“Limited Room” Logic:

Think of it like this: You have a bag with 2 Red balls and 2 Blue balls.

—If you pick a Red ball first, there are now more Blue balls left in the bag than Red ones.

—Therefore, your next pick is more likely to be Blue.

Even though a coin doesn’t have a “memory” like a bag of balls, when you look at a fixed string of 3 or 100 flips, you are essentially looking at a “bag” of results. If you are currently looking at a “Head” in a finished string, you have “used up” one of the Heads available in that specific sequence, making it mathematically more likely that the rest of that specific sequence contains a Tail.

A bit mind-blowing, if you ask me, but that’s probability for you. Whether it makes sense to you or not, the bottom line is that you should pass the ball to the gal or guy with the hot hand. As I point out in the interview, that won’t be me, but I’ll well understand why you did it.