http://www.jetpletters.ac.ru/ps/1535/article_23489.pdf
This is a nice summary of an electrostatically-plugged cusp machine (albeit a toroidal cusp.) It seems like it would make a nice primer for polywell physics.
Page 2: "The escape of the electrons from the system along the magnetic field is blocked by the magnetic field between the aperatures and the reflectors, but there is a rapid transport of the electrons across the field lines to the walls of the aperatures. This loss must be balanced by an equal loss of ions. The ions cannot reach the walls as the electrons do, since the ions are in a deep potential well with respect to the walls, but they can escape freely along the magnetic field lines to the reflectors. The rate of this ion loss is adjusted by the plasma potential to match the electron loss. If the plasma potential were zero, the ions would pass freely through the entire width of the magnetic aperature, and this would exceed the electron loss. To reduce this loss, the plasma assumes a negative potential so that the ions can only escape along a narrow corridor at the center of the aperature. Elsewhere in the aperature, the escape of ions is blocked by the electrostatic barrier produced by the field of the grounded walls of the aperature."
The argument that the ion loss rate is controlled by the plasma potential makes sense; but it seems like the ion loss rate is going to balance the ion creation rate, not with the electron loss rate. The electron and ion rates (creation and loss should be equal for each species) are related, but probably not equal: each ion created by ionization introduces an electron, obviously, but also each ion produces some average number of electrons by bombarding the reflector when it is lost. And of course, injection of either species will alter things.