It is not pulled out of thin air. It's Stanford/Hoover, if they have access to the mean they have access to the median, probably more fine grained data too.
You would think that, but the article text says otherwise:
> the median income of taxpayers who left San Francisco would probably have been around $250,000.
I am as puzzled as you, why not just calculate the median or email the people with the data and ask for it? Otherwise it feels like reaching for anything to embellish a partisan talking point.
The silliest part is that it doesn't change the argument. $196k is a good income, $250k is better but not qualitatively different by any means.