There are some number of jobs out there and most of them don't want remote workers. Job applicants that are willing to take jobs that forbid remote work can also apply to those that permit remote work, while workers that will only look at jobs that allow remote work can only apply to the small subset that allow it.
It's straightforward micro-economic exercise to see that the when you hold all other things equal the market clearing rate for a particular employee is higher if he is willing to take on-site work than if he is not.
It'd be the same thing if an employee was only willing to work at companies whose names started with a-m.
I don't think there is more aversion to remote work on the employer side than on the employee side, so that effect should not turn the tables.
Why? How does the value of the work I provide change based on where I'm located? Is an if statement somehow different if done in the Bay Area vs the Midwest?
Delivering more value can justify more compensation. But unless you're on contract for a percentage of the take, you don't necessarily get compensated based on output. Most software people get paid based on market rates which can be a lot lower than the actual value they deliver on the high-end--they're paid their "replacement cost" rather than per value added.
In my experience that is an incorrect argument to begin with: many large tech companies seem to have more flexible work hours and remote working capabilities just like startups. I would rather focus on work-life balance, working on products that matter, having equity in the company, being able to experiment with technologies and be part of the decision making process rather than a cog in the machinery as reasons to prefer a start-up.