This was primarily an experiment, not a pressing need.
I started in Photoshop, brought in an image rendered in ProPhoto, dropped a sampler point on a light area, and opened up the Exposure adjustment. Moved the exposure slider to -1 stop, all looked sane on the sampler point. Converted the image to ProPhotoGamma1 (profile constructed in Color Settings, same primaries and white point as ProPhoto, only change was gamma to 1.0), the sampler point RGB values go to exactly half for a one stop decrease. All is good, makes sense.
However in ACR I cannot reconcile values measured. I first dropped the exposure adjustment one stop, image looks sane, even though we've been "taught" that the exposure control doesn't really linearly control exposure. I set the output rendering to my ProPhotoGamma1, sampler point goes significantly lower that half its original value for a one stop decrease. Then I tried curves, Parametric, and pulled down the right side on a straight line RGB so that 255 in gives 128 out. My thought was this would roughly cause every rgb value in the image to be halved. Nope, a 222 white point goes to 96 with regular ProPhoto rendering, and 44 with ProPhotoGamma1. Obviously there is a lot more "magic" going on in ACR when it comes to Parametric curves adjustment.
My conclusion at this point is the only way to make an accurate exposure adjustment in in Photoshop, no apparent way to do it in ACR. Any opinions from the group?
An error in my original post, I was using the curves tool in Point mode, Custom settings.