The assumption that companies won't offshore is doing a lot of heavy lifting.

Companies already do a lot of offshoring - you think any rational actor in this space that was hiring H1Bs isn't going to simply relocate them to more friendly jurisdictions for immigration?

On top of this, these are workers who would have otherwise paid tax in the US!

Overall the US economy employs about 800,000 software engineers, with 200,000 or so of them being H1B holders.

Now you can argue you would prefer that those 200,000 jobs go to Americans, but on the scale of the overall economy, it really doesn’t matter. What’s far more important is the massive impact those 800,000 software engineers have on the rest of the economy. Four million IT jobs, the entire finance and healthcare and retail industries that are propped up on technology built by those people; whole technology companies like Uber or doordash that create entirely new labor markets.

Risk 25% of that capacity on the idea that we would rather have those industries built solely on domestically-grown engineering talent? Why would that be a good tradeoff?

You also get the baumol effect increasing wages even for unrelated sectors (sounds helpful at first).

The flipside is that every american industry becomes less competitive globally without the H1b guys.

you'd really need to look at the median pay for specifically companies that hire a lot of H1b SWEs. I'd suspect that would be higher.

Maybe? But what about training and talent pool? Imagine how many companies would not take off because there’s no one to implement the founder’s idea. Imagine you’re a startup and you have hiring difficulties because all the good ones are over at Oracle or Microsoft (doubting the existence of FAANG).

It would definitely be higher.

Lower supply tends to drive the price up.

Whoa whoa whoa, that's (1) not correct[1], but (2) shameless goalpost motion in any case.

The whole premise of your original contention was that we should measure like-profession salaries to see whether or not there is an effect. Then when no effect was shown, you switched it up in favor of an argument that (again, incorrectly) predicts that such an effect can't be shown at all. That's not good faith discussion.

[1] Immigrant labor is arriving, by definition, in a pre-existing market. If immigrants can't be hired more cheaply than existing labor, by definition they can't be pulling wages down.

Underpaid or overpaid doesn't really matter.

Sure it's sound to argue that wages would be higher with more constraint supply.

BUT: The network effect of all SWE talent from across the globe moving to the US is also huge.

Probably, you'd have a smaller overall tax base without H1B. Make no mistake most countries would like to keep their H1B expats :)

If you really wanted to grow US supply of engineers, you'd have to start by fixing the education system, making it cheaper, and then wait 5 years.