This assumes talent is geographically homogeneous. IMHO, Talent in top cities in the US is better than the one in the rest of the countries, and so it goes for other countries.

For top US firms, it may well be that there is a better pool for top talent, but not THAT much better. I believe it would be possible to do inferences from # of non-US citizen in FAANGs.

However, a company that was previously land-restricted to only the 98th percentile (top 2% minus top 1%) now has access to the top 1% and will be trying to compete, so there are actually more jobs available at the top.

You are forgetting that for each "land restricted" company that had 0% 99th percentile hires locally available there was x companies having more % somewhere else for it to eventually be an distribution.

There is no way you can filter out top 1/100 candidates anyway.

All that implies is that the number of top 1% candidates remains constant. Of course it does. Some companies had more than they needed, but if you take luxury away from a man, he will seek it out because he is used to it.