Fixing attempts:

So, two issues identified: one, the productivity of the Early Gears Transmission is way overtuned. Two, the giant gap in cars output between Early Gear Transmissions and Automobiles results in massive overbuilding of factories to satisfy early demand, then turns switching PMs from an attractive upgrade into a negative economic shock that takes years to fully recover from.

The first thing I tried was adding the 500 extra machinists from the two later PMs to Early Gear Transmissions as well, so the factories would have to pay more wages. I also added 5 oil and 1 engine as extra inputs, and increased the output from 12 to 18. This should make the PM less efficient, while at the same time needing fewer factories to satisfy demand, and narrowing the output gap between PMs a bit. After running this ingame for the few weeks the hiring processes took, cars went from -5% base price to -27% base price, and car industry productivity went from 60 to 36 (with some individual buildings as low as 22). Okay, better! However, that still makes this a top-5 industry in terms of productivity, despite their output goods sitting so far below base price. The private queue AI would likely still have overbuilt them significantly.

I then added another 200 machinists (1200 total), 100 engineers (600 total), 3 steel input (8 total), and 1 engine input (2 total). This caused productivity to drop to an average of 22, and for the first time, I saw some of the factories actually start dropping their cash reserves. Some even failed to hire to 100% employment. Alright! This is more like what I would expect from an industry whose output good is at nearly -30% under base price. Though I might add that oil was at +11% and engines at +22% at the same time. So with less factory overbuilding resulting in lower input prices and higher output prices, this would still be a highly productive industry even with the "inefficient" PM.

This also still leaves the Automobiles PM largely unusable for me. I would probably tune its output way down, from 52 to 36 or somesuch (so turning it on is only a doubling in car output instead of more than a quadrupling), with changed inputs to match. But given the huge overbuilding of car industries in my save, I can't actually test where a good balance of input goods and productivity would be. There simply is no way I could generate enough demand for even half of the output. I'd have to run a whole new game, first, to see how the private queue would respond to the changes in the Early Gear Transmissions productivity.

Thinking about it some more, employment numbers should probably go up sharply for all PMs. The game has an unspoken convention that every building level employs somehwere around 5000 people on average. The car industries with 1300 to 1800 violate this quite severely, which may well be part of the problem. But I can't really test that in my currently running game either.

And obviously this has ripple effects towards all of the other PMs of this building, including Pneumatic Tyres and the tank and aeroplane production ones, which all would need significant changes as well.

The impact of Shift Work:

I'll note that all of the numbers I presented here are on single shifts. Having double or triple shifts available is a neat little bit of flavor, but they're rarely ever actually usable. Because the private queue AI overbuilds car factories to such absurd levels early-on, switching to double shifts simply causes huge amounts of overproduction. I suppose that would be a way to reign in the extremely high productivity of car factories... but unfortunately, shifts cannot be used with the Early Gears Transmission PM. And even if they were - Shift Work is a technology you have to research separately. It's in a completely different tech tree, even. It would then feel a bit like... I'm forced to rush researching this tech early, just to fix the problematic car industries.

I do use extra shifts on my small number of military shipyards, but it works there chiefly because they're low productivity, so the private queue pretty much never builds them. That gives me the opportunity to look and say: hey, I could use some extra ironclads, let me add a shift here instead of putting a new building into the queue. That's neat! But in the ultra-high productivity automobile industries, the private queue will automagically make sure you always have more output than you need even on single shifts.

What makes it doubly weird is that the three industries that can massively swing in demand on short notice - arms, artillery, and munitions - don't actually have shift work PMs. Yes, they do have the mobilization/demobilization PMs, and they're massively useful in responding to the demand swings of war and peace!

But shift work is also fundamentally a tool to respond to sudden demand swings. And they're only implemented for... shipyards, which never have sudden demand swings at all... and car factories, which don't have demand swings at all until the Combustion Engine tech, and then only mildly. Until the absolute endgame, when tanks have been researched. With a good number of nations, you might never even get the Mobile Armor tech before 1936. And even when you do, the car factories have multiple mobilization levels, so you can respond to the swing like you can with the other military industries.

So yeah... I don't feel like this feature adds much besides flavor. The mobilization/demobilization PMs alone are easily enough of an improvement in those industries where it matters.

Thank you for all of this, it's a lot of detail and I really appreciate it - I've been away but I'll attempt to address this in the coming week.