Launching an item to low Earth orbit (LEO) takes a great deal of energy. Most of what is launched is the propellant needed to accelerate it to more than 27,000 kilometers per hour and get it the first few hundred kilometers above the surface of the Earth. The cost in dollars depends on many bookkeeping assumptions, but we know that only a small percentage of what leaves the pad ends up in orbit. Estimates of the launch costs of items via the Shuttle range from $6,000 to $12,000 per kilogram.
A kilogram of material orbiting the Earth is not going to do you any good at a lunar outpost; you need to get it from LEO to the Moon. Leaving Earth orbit and going to lunar orbit--the path Apollo took--gets only about one-third of the original mass to the vicinity of the Moon. From there, another propulsive maneuver will be required to get to the surface, meaning another mass reduction of about 50 percent. Alternatives to this route exist, such as going directly to the lunar surface or using other transportation "nodes," such as an Earth-Moon libration point. These have some advantages and some disadvantages involving mass ratios, safety, operations, landing site flexibility, and commonality with other missions; it is not a simple analysis.
When all of this is factored together, only about one-sixth of what makes it to LEO can be payload for the lunar outpost--the rest is the propellant to get it there. Thus, the cost of a kilogram of material on the surface is six times the cost of getting it to LEO, or many tens of thousands of dollars. Even if future launch vehicles can lower the cost by a factor of ten, we are still faced with a large bill for shipping and handling.
And these figures all assume the best propellant combination available today--hydrogen and oxygen. It is not a given that we will choose this propellant mixture for any or all of the spacecraft involved in the transportation system. Handling liquid hydrogen on orbit is a technology we have yet to demonstrate. Long stay times on the lunar surface can cause loss of the extremely low-boiling hydrogen. Furthermore, the required volume of liquid hydrogen is very large owing to its very low density. To alleviate these problems, we may choose a more storable fuel, such as a hydrocarbon (like methane), or a hydrazine derivative. This could remove the negative impacts of hydrogen on the overall system. Because of the lower performance of these propellant combinations, however, the ratio of payload delivered to the lunar surface to the initial mass in LEO will be lower than the 1:6 determined above. This translates to an even greater Earth launch mass to deliver that one kilogram to the Moon. During the Apollo missions, this ratio was around 1:10 because greater reliability dictated the choice of the less powerful hypergolic propellants. This difference would enter directly into any economic calculation. Indeed, making LOX on the Moon may make it more feasible to use these alternative fuels, benefiting from their favorable properties while ameliorating their effect on launch mass.
If materials could be produced on the Moon, rather than bringing them from Earth, all of the extra mass launched as propellant to get them there would no longer be required. For example, if the ascent of a lunar excursion vehicle from the lunar surface to lunar orbit requires, say, five tons of LOX, we would no longer need to take this propellant there at the cost of launching more than 30 tons to LEO. This is equivalent to the entire payload capacity of today's Shuttle. By making the material there, we are saving the cost, the logistics, and the risk of a launch.
The question then becomes: What does it take to produce useful products on the lunar surface? If the mass of a required item is small but the mass of the equipment to make it is large, then it clearly does not make sense to manufacture it there. If, on the other hand, the mass of the plant can be "paid back" in only a short operating time, we should consider making it there from those materials indigenous to the planet.