So while a 182 size city has a population of 2.13 billion, a size 183 city has -2.08 billion inhabitants. Size 184 -2.01 billion and so on. I don't fully understand the pattern, since cities with much higher sizes have a positive population again. E.g. a 364 city actually has about 2 billion again (positive), as I now found out. A size 1000 city likewise has around 2 billion. I didn't come across any city size though in which the population exceeded that of a 182 size city, namely 2.13 billion. I obviously didn't go through all the missing 800 city sizes up to 1000 though, so I can't exclude the possibility that maybe somewhere among them the population is indeed higher.
Err surely it's integer overflow; a signed 32-bit int will go from (2^31)-1 to -2^31, which is 2,147,483,647 to -2,147,483,648 [1], so obviously the population would never be higher (or lower).
[1] assuming two's complement, but it doesn't make a lot of difference.