The continuation Pedersen suggests in his book 
5...Qxb6 6. Nf3 g6 7.Nc3 d6 8.e4 (8.Nd2!?) Bg7 (8...Bg4!?)  9.Nd2! 0-0 10.Be2 Nbd7 11.Nc4 Qc7 12.Bf4 Rb8 13.0-0 Ne5
should be dubious:
14.Nxe5 de5 15.Be3 Rxb2 16.Na4 Rxe2 17.Qxe2 Nxe4 (Seres- Nisipeanu, Eger Agria 1995) and now 18.Rac1! gives white a clear advantage.
I believe Black does not equalize after ...Qxb6, unless white commits himself to a2-a4. He should probably try delaying taking on b6, or play 5...e6, which is rather dull, without winning chances, but objectively equal. Anyway, White can probably avoid this using the Kiril Georgiev move order (4.Nf3), so I guess the question which way to take on b6 is still on.

Surely there are better alternatives to 5...e6? A quick survey of my meagre resources yields these recommendations:

Keene & Levy (An Opening Repertoire for the Attacking Player, Batsford, 1984): 5...Qxb6

Gufeld (Benko Gambit, Batsford, 1988): 5...d6 ("Intending the manoeuvre ...Nbd7-b6. Weaker is 5...Qxb6 since White carries out the knight tour Nf3-d2-c4 with gain of tempo...")

Pinski (The Benko Gambit, Quality Chess, 2005): No particular or clear recommendation (not a very useful book, IMO), but Black appears to fare better after 5...d6 or 5...Qxb6 than after 5...e6 or 5...a5 in the examples he gives.

Pedersen (Play the Benko Gambit, Everyman, 2011): 5...Qxb6

FWIW, I've had good results with 5...d6 (even if obliged to follow with 6...Qxb6 after 6.a4).

In the line
1. d4 Nf6 2. c4 c5 3. d5 b5 4. cxb5 a6 5. b6 e6 6. Nc3 exd5 7. Nxd5 Nxd5 8.Qxd5 Nc6 9. Nf3 Rb8 10. Ne5 Qf6 11. Nxc6 dxc6 12. Qf3 Be7 

Black cannot play for a win. Is there a better line in this variation?