i think everyone is different in degree, but i think accepted wisdom is to start out at some wattage below what you expect to average at and then raise power to some wattage above what you expect to average by the end..
e.g. i start our about 15-20 Watts below the average i'm going for and end up about 20-30 Watts above.. last couple mins can be more than 40 Watts above the average.. i think i'm more extreme than most though..
if you do as you describe what happens is that your red zone right away (after about 5-10mins) and after a certain point power drops off markedly, if you can even complete the interval and average power ends up being very low.. you will get a higher ave. power if you start out slowly and build. first third should feel easy, second third should feel a good sort of hurt and last third should start to hurt in a bad way.. i know for myself, i've tried it every which way and building as describe below always ends up with the higher average power. if i start out at my 20min power i always end up with an ave. lower than my 20min power...
i'll just add that it's better to be conservative in your tests and then raise the bar up..
also.. FTP is also approx. what you typically can do for a 20min effort in training and the adaptation one would get from a record 20min effort is only slightly more than a typical one, but the record will be a whole lot more painful.. so no need to go chasing records every day.. better to just get your training in and go home... every now and then you may get the feeling that you're on a good day and you may want to try to see what you can do but.. mentally and physically you don't want to be chasing records every time at bat..
also, also.. i don't know anyone that actually does 1hr FTP tests.. it's almost always extrapolated from 20min tests so i just dispatch with the *(.95) and just track 20min power directly...