I've been meaning to add the effect of tyre width to my program, and your question was just the kick I needed to get it done!

It's another counter-intuitive one: three tyres were tested by Jan Heine of (the rather excellent) Bicycle Magazine. The tyres were the same make/compound/tread pattern but in three different widths; 20, 23 and 25mm. Despite the extra width from the wider tyres probably causing more aerodynamic drag, they were actually faster than the thinner tyres when ridden on a typical tarmac surface (as opposed to a smooth velodrome surface or testing drum that has previously been used for tyre testing). The hypothesis being that the wider tyre is better able to absorb the small imperfections of the tamarc surface and reduce the amount of vertical motion they cause which absorbs power.

Even the pro TDF teams are switching to 25mm wide tyres from the previously "normal" 23mm wide ones.

I've added the effects to my program (bottom of the list):-

Marginal Gains Calculator with tyre width effects