The offset in primary winding power dissipation will be enough to take, eh, maybe 7% more from the secondary. (~30% reduction in VAs --> ~15% reduction in primary dissipation, ~7% increase in current because (7% increase)^2 ~= 15% increase.)
There is a nonzero gain available... but it's nothing important.
Reminder that, if you're connecting this to a FWB and cap-input filter, the power factor sucks (~0.6 at best) so for say 3A RMS current rating, you can only draw maybe 1.8A DC. (This partly adjusts for the fact that output DCV is higher than ACV RMS, but it's worse than a one-to-one trade so you are losing some capacity.)
Also consider using a buck transformer: get a ~40V transformer, wire it to subtract from 120V so you're supplying 80V to the primary, and you get 20V out of the 30V secondary. (Or 80V for 240V mains.) This transformer only needs to handle the bucking VAs, so it can be smaller than the main transformer is.
Though, not much smaller in this particular case. It's definitely a more useful strategy for small changes. Also needs an oddball 40V transformer, though 36 is probably common enough.
Note that, although the current rating is not changed in this configuration, you still have all the impedance (DC and AC resistance, and leakage inductance) of the full-voltage transformer, but its drop is a bigger fraction of the total: in other words, regulation will be about 30% poorer. If your load expects a stable voltage (say +/- 10% over full current range), you will be further restricted to a proportionally smaller current range as well (i.e. 20V 2A).
Upside to the bucking approach: core loss is much lower, so the transformer runs very cool and draws less idle power. This is just about mandatory to reuse highly-saturated designs like microwave oven transformers (which are still perfectly good for say 300 VA when rewound with a safer secondary voltage).
Tim