Your linked paper is about correlating year of birth to a social security number, which I would reckon is much easier than correlating to zip code, given that social security numbers both predate zip codes and were geographically assigned based on the US State in which the applicant lived when the number was requested. My own SSN was assigned to me at age ten when I lived fifteen hundred miles from the place of my birth, and I have moved a further thousand miles from there in the meantime; there's basically no way to predict my ZIP code from it.

Anyway, according to that lookup table, I was born in 1927, a conclusion with which I strenuously disagree. Those methods may have been statistically useful in the year of publication but so many things have changed, and so many more people move between states than before the 80s, that I doubt its validity now. As a reference, the pandemic drastically slowed down American interstate household moves, but even so, 8% of the population moved states in 2020 -- including 18% of people between 20 and 29. (https://www.weforum.org/agenda/2021/12/american-relocation-h...)

In short, I think it may have been more feasible in the past to draw these correlations, and now it's less feasible. On the other hand, massive data leaks are de rigueur, so it's probably just easier to buy databases from criminals on the internet.