There is no good reason to pay 200k a year to a guy who can make 200k a year in Sunnyvale - in Michigan, he'll have as much disposable income at, 120-140k a year as he did at 200k a year in California.
Before you say people wont work for less - I know plenty of people who would, including myself.
Michigan doesn't have that cycle happening (or, at least to the same extent). Telling somebody to take a hypothetical pay cut to move somewhere with less appeal isn't a sound strategy for attracting talent.
This ignores two facts of wealth in the bay area.
The value of land has risen considerably, so an engineer can grow wealthy just by owning (as opposed to renting) even without any "disposable" income (although I recognize that owning/renting can involve dipping into "disposable" income).
The promise of equity growth is compelling in a small-ish company. The likelihood that Ford employee equity might appreciate is _very_ tiny, is dependent upon so many factors out of the control of the tech employees (industry competition, regulations, supply of low-interest financing, labor unions, etc) that may not affect other unions.
As someone who has spent most of my career working for old line companies, I discount equity in the company as something akin to gambling - I'd rather take cash than options with a low strike price.
GM paid $500-1,000MM dollars for Cruise Automation, an autonomous vehicle startup with 40 employees. That means that GM thinks that three years of work by these employees are worth in the neighborhood of $12.5 to $25MM each.
Meaning a $200k salary is a serious low-ball. The fact that a web-dev in California can earn that, doesn't mean that's a good benchmark for a world-class AI researcher.
So yes, there is a good reason to pay these people well. The company knows that, if they don't, they face an existential threat. These potential employees are better off creating their own startup and auctioning off the results in three years.