Non-resident population has to be included and divided to arrive the GDP per capita because they worked and contributed to GDP. In Singapore our resident population is only 4m, but we divide our GDP with our total population of 5.6m. Shanghai and Beijing include migrant workers to divide too.
I remember reading before that it's not easy to skew statistics for population beyond 500K, and that's why many rankings use that number as a minimum cut-off point for countries.
A small group of rich people won't distort GDP figures for a city of 20m people. Wealth figures maybe, but not GDP figures. Because they don't spend all of their wealth in a single year in the same city.
Look at Hong Kong. It has more than half as much UHNW of mainland China and more than UK/Italy/India/Russia/Brazil/SKorea in a population of 7m.
Yet HK's GDP per capita is 'only' $46K, around the same as Finland/Canada/Austria.
Yeah, median income usually reflect the ground more accurately.
I found only average though:
Found another website in Chinese. Even though they wrote average, from the presentation of the statistic and disparity from other sources' average I think it's median.
Beijing:
http://salarycalculator.sinaapp.com/report/北京
Shenzhen:
http://salarycalculator.sinaapp.com/report/深圳
Therefore it's unlikely that Shenzhen's GDP per capita is so much higher than Beijing when all sources point towards lower average/median income.