First off, space-based resources have a lot more potential than a lot of people seem to think. Outside of exotics like iridium (currently known to exist only in asteroids, as all known earthly deposits have been linked to impacts) even basics like iron would very profitable. Not only for shipment to Earth (if refined oribitally or on Luna, and shipped back as pure metal, the costs would probably be noticably lower than earth-based mines after the inital investment, which would admittedly be expensive), but would form the backbone of lunar or martian use, as there would be no need to escape Earth's gravity well.
Second, as long as nitrogen and water were supplied (ice-based asteroids are theorized to exist, and of course there's water known to exist on Mars, so Earth is not the only source) Luna could be used as a massive farm because the extreme amounts of free electricity available would allow deep layers of farmland. That alone could justify the expense, potentially producing as much food as Earth does today. Besides the obvious benefits for self-support, this would effectively eliminate Earth's food problems. Forever. (There is, of course, one limitation to that, which is that water and nitrates would have to regularly be returned to replace losses. If an elevator or nuclear-powered EM catapult were used ot launch form Earth, the cost of such shipments would be negligible.)
Third, these are dead worlds. There's no ecosystem to destroy. Thus, there is a significant ecological benefit to moving as much industry as possible there. In fact, pollution would actually be a good thing in some cases (Carbon dioxide, a prominent pollutant, would be of great agricultural use, and might even help in any martian transforming efforts. )
As for transshipment, it is better in all ways for a planet's moon to be used for shipments to and from the planet. The more massive an object is, the more energy is needed to land it safely. Because of a moon's lower gravity, less thrust is needed to overcome the acceleration and land safely. Thus, it's easier to land on the moon without crashing than it is on Earth. Besides this, if a crash does occur, there will be much less risk of disaster because thue bulk of the population will be deep underground and thus immune to harm. However, larger vessels move cargo more quickly and efficiently. Thus, it would be most practical, to, for example, use a 2000 ton freighter, land it on the moon, and difide it's cargo into 100 twenty-ton shipments. If one of those went ary, the danger would be far less than if the whole ship crashed.
In reverse, sending 200 individual shipments back up would only ever require you to generate enough push to launch twenty tons at one time (you'd have to do it 200 times, of course, well within current capabilities. These could then be gathered on the moon, loaded onto the freighter, and go onward to their destination.