In my town it increased traffic (and road accidents). Players would drive up to the nearest Pokestop, collect their loot, and drive to the next one. I saw more than a few drivers make sudden turns or stops when an interesting spawn would occur. I don't really miss the craze.

It did increase their activity level for 6 weeks, which is probably about how long it took most people to get bored with it.

More AR games with more fleshed out gameplay could raise activity levels more long term.

It just increased general phone usage by ~16% per day. Daily path length stayed the same. (Source: study on 2800 phone users that played Pokemon GO. Data collected for at least 4 weeks having the first gaming session in the middle. Data was collected through an app that monitors your phone usage, no self-reported data. Will be presented at MobileHCI '17)

Too small a sample size for physical activity (fine for HCI, though) and was probably orchestrated after the hype of it. Does this paper live in those parameters?

Initial sample size was 21k users but we've reduced it to the people that filled some questionnaire regarding demographics and personality traits. To which parameters are you referring?

Well, it did, but only for a short while. And that may just be because of how bad it was.