It can be a bit misleading by the phrase "wavelength issue." Wavelength does not change, it is the spectrum of the bulb's output that changes, assume there is more than one type of phosphor being used in the bulb.
Anyway, if the electrity/bulb replacement cost is low, then one should get more, but older, bulbs in the hood if the space is available. Say a replacement bulb costs $50, then assumes the spectrum and intensity do not change much over the remaining life of an old bulb, you calculate the amount of time $50 can buy in electricty, assume the intensity/spectrum metric give you a 50% discount over the "utility function" of the bulb.
The concept can be a bit complicated, and I am sure some of you can do a better job in explaination than I did, but the notion that I want you to go away from is that a bulb burn up more cost in electrity over its lifetime than the cost of the bulb. This is certainly true for low-cost bulb (incandenscent, halogen, and other non-reef bulbs), and without actually going thru the calculation myself, I *think* it also hold true for PC bulb that we use in the hobby.
Maybe some of you who have more time can run thru the number using a spreadsheet or matlab, and see if my hypothesis is correct. AFAIK, nobody had done these type of empirical calculation on PC bulb used in the hobby, so I am sure many people will appreciate the effect.