It takes a fair bit of scientific elbow grease to prove anything is "X efficient".
One thing you can say is that efficiency does go down as the load goes up in a PM generator. That may be counter intuitive, so here's one thing to consider: The loss of energy in waste heat is calculated with I^2*R, a formula you might have seen before. The "I" is current and "R" is resistance. For a coil that has a resistance of (for example) 20 ohms, putting 2 Amps through it you lose this much heat:
(2A)^2 * (20Ohm) = 80 Watts (Heat loss)
At the same time, it's making useful power with that 2 Amps into your 28V battery:
(2A) * (28V) = 56 Watts (useful power)
My example (totally made up) makes this scenario:
Output power = 56 Watts
Input power = 56 Watts + 80 Watts = 136 Watts
Efficiency = 56 / 136 = 41%
Now that you have an example of the math, let's see what you can do with your starting point:
5 W / 28V = 0.18 A
(0.18A)^2 * (20Ohm) = 0.64 Watts (Heat loss)
Output power = 5 Watts
Input power = 5 Watts + 0.64 Watts = 5.64 Watts
Efficiency = 5 / 5.64 = 88%
So even though I started with a lucky guess, I think I get the same result, and it works out OK.
At higher power the output goes up, and the efficiency goes down. No free lunch, man.