There is far more to reciprocity failure than the point at which you need to begin correcting for it, what matters most is how much you need to correct. Draw a graph of total exposure time required against light level, it would be straight lines (pointing downwards at the right/bright end) with no reciprocity failure and lines at higher ISOs are lower (less time required). With long-exposure reciprocity failure, the left (dark) end of the graph curves upwards exponentially, i.e. more time is required. Different films have different exponents, so they curve up at different rates.
Originally Posted by StoneNYC
If you plot the graph for Fomapan (terrible reciprocity failure), it shoots upwards very steeply. Acros of course only has a very slight upward curve. TMY2 has a slightly steeper curve than Acros but for any exposure less than about 8 hours, the TMY2 and Acros curves do not intersect, i.e. TMY2 still has a shorter exposure. While it might be worse in terms of compensation required, the extra compensation is always less than 2 stops more than what Acros needs. You need days of exposure before Acros is better than TMY2.
If you shoot colour, I find that following the Acros tables for Provia 100F works very well (tested at 20 and 60 minutes) and the TMY2 table seems reasonable for Portra 400 though I haven't tested that much.