Yes, I believe the bounding sphere's center is always at the center of the model's coordinate space (at coordinates (0,0,0) in Lightwave). The radius of the sphere will be calculated such, that the sphere will be big enough to include the point/vertex farthest away from the center (plus some margin for safety measures).
If you model is not properly centered around (0,0,0) in Lightwave, this'll produce a much larger radius and thus a much larger bounding sphere.
... at least that's what I think happens behind the scenes. Haven't really tested it, but it'd make the most sense computationally-wise and also considering your descriptions of what you're seeing with your models.
So as a rule of thumb you should always strive to position your larger models' as close to (0,0,0) as possible, to keep the sphere's radius as small as possible. This might make for some odd-looking maneuvers if your ship is turning to change course, though. Depends on how well the perceived center of mass of your model correlates with the (0,0,0) center of coordinate space.