Oh come off it
The McKinsey report is produced to order, presumably to help privatise the NHS. That's what their reports are generally designed to do.
However, to suggest that you need to combat their economic voodoo with some modelling voodoo from 2008 (see the PLOS medicine paper referenced in the "pans out" link) is unfortunate, and does suggest you don't understand the situation. I'm reminded of economists whenever I think of this: http://www.xkcd.com/793/
Richard Doll did make the assumption that smoking was good for the economy; he carefully enunciated it as such and didn't lay it down as fact. I first came across this when reading some other modelling paper (think ~2009 in BMJ but can't now find) which contradicted this. It's a shame this is used as an argument from authority now; he'd be furious.
So here's a paper to read to gain some understanding about cost of illness studies: Cost-of-Illness Studies, by Larg and Moss, Pharmacoeconomics 2011. "While they attract much interest... inconsistencies in the way in which they are conducted and a lack of transparency in reporting have made interpretation difficult, and have ostensibly limited their usefulness." They proceed to demonstrate why you shouldn't believe this type of study, without some serious fact checking.
Here's some of the fact checking required (appendix A of PLOS medicine paper, available to everyone free of charge): "For our calculations, we did not take into account transitions between risk factor classes over time." They have (to make the modelling easier) assumed that your state (healthy/smoker/obese) at age 20 is constant for the next 100 years. I'd suggest that this is not a valid assumption. Most obesity occurs after the age of 20; there are far more smokers at 20 than at 30; changes in this classification makes the modelling almost impossible. An additional problem in their modelling is there is no feedback (at least hat I can see): they assume incidence of diseases is uncorrelated to risk factors. They assume that average health care costs for their diseases apply: usually those who are obese or smokers have significantly higher costs. 20 year olds have lower costs. If you die from the incidence of your illness, your costs are markedly different to if you continue to live.
TL;DR: don't base arguments on modelling papers, demand empirical data. Which would have been a good article, seeing as the McKinsey paper has (I think - haven't bothered to actually read it) much the same faults.