Hello Leonid,
I think your understanding is correct the mode expansion monitor simply calculates the overlap with the fields as measured by the monitor with modes of the guiding structure as calculated from the mode expansion monitor. You can specify different modes in the same expansion monitor as well to get the power to TE and TE1 for example. For multiple planes along a lossy waveguide you could see the overall transmission drop, and the T_fwd drop by a similar amount. Although T_fwd doesn't really tell you the propogation loss directly, you could get an idea be measuring the difference at multiple points along the propogation direction. This is unecessary in most cases because the transverse structure is invariant along the propogation direction. Furthermore, you can actually calculate the propogation loss directly from the propogation constant; ie the imagninary part of the refractive index. If this is small, the mode will propogate with effectively the same T_fwd for some distance.
In reality you could have other effects like coupling to other modes, or bending loss etc that are not contained in the propogation constant; however, for single mode waveguides you can usually safely assume that T_fwd is the ratio of power in the mode to input power. This is how you would extract S-params. Finally the difference between T and T_fwd is equal to the power that is scattered through the monitor.
I hope this was enlightening.
-Taylor