It's really difficult to come up with good statistics, because of the number of things one can factor in. Tobacco industry in-country producing, or it's imported for example (total cost of cigarette smoking)? How many jobs directly, how many secondary (marketing, product chain, the news stand which sells them). How to factor in second-hand smoking exactly? Are smokers more likely to consume other drugs, if so, why? Because they have a tendency to try, or smoking causes them to try? How do you factor in that? How do you even "measure" that?
With alcohol: how to factor in violence due to alcohol? How much less police incidents Friday / Saturday evening due to less alcohol? But what about people maybe going out less than before, or spending less money when doing so (will they pay the same prize for drinks if there's no alcohol involved)? What about sports betting and gambling industry (where it's legal)? How will it affect illegal drug use (we've been through that, but it's been long enough for people to say "well, it was different than").
Even the death statistic is tough to do. "Pre-mature death due to smoking", maybe the person also was overweight and had diabetes, how do you estimate how long this person would have lived, if they didn't smoke?
In every step of the way is the possibility for variance. Social sciences, from sociology, to psychology, to health, is such a minefield. Medicine of course is half and half. You can do a lot of stuff on the level of biology, chemistry, hard science in a lab, but health benefits/damages are "observed" in practice (observational studies). And the number of problems there are just ... huge.
I could go on and on and on...