Are they? Assuming an attacker needs 3 mtpa and 3 mwpa, with 50% crime/channel, 5% pop:
Halfling - 1.3 rtpa, 2 rwpa, 28.9 pop. 5 peasants/acre. 20.575 mil/acre. Assuming 120% ome/dme, and his 65%/35% split, I get 80 OPA assuming all ospec offense and 34.5 DPA.
Undead - 2 rtpa, 2 rwpa, 26.25 pop. 5 peasants/acre, 17.25 mil/acre. Same ome/dme, 65%/35% - 94.185 OPA, 28.98 DPA. Let's set the same DPA: 84.42 OPA and 34.5 DPA.
here's elf: 2 rtpa, 1.5 rwpa, 26.25 pop. 5 peasants/acre, 17.75 mil/acre. Same ome/dme, 65%/35% - 69.22 OPA, 37.275 DPA. If you set the same DPA: 72 OPA, 34.5 DPA.
Fae: 2 rtpa, 2 wpa, 26.25 pop. 5 peas/a, 17.25 mil. 53.82 opa, 36.225 DPA. 55.2 OPA if you set DPA constant.
Human: 1.9 rtpa, 1.9 rwpa, 25.175 pop. 5 peas/a. 16.375 mil/a. 76.6 opa, 26.51 dpa. 66.15 OPA if you set DPA constant.
This ignores the simple reality that the undead would really be running 1 rtpa with CS/WTs for defensive purposes, and have more military. Now you can say 'well gosh zauper, your calculation is so simplistic', but the reality is my calculation is both more realistic and more reasonable than his, while using his rules. For example, I'm not forcing the heavy attacker races to have 10+ DSPA (not clear why you would want that).
Edit: added elf/fae/human for the sake of argument.