A while back I wrote about the notion of consumer surplus. It's the economic idea that while there may be a price that brings in more money than any other single price, a seller can often make more money by charging different prices to different people. This is because some people will simply not buy if the price is too high.
Some sellers simply use the consumer surplus to maximize their profits on that one item. Others will use the profit on one item to sell another below cost. Sometimes this is done to get customers' attentions, as with free giveaways. This is also known as a loss leader.
A similar strategy is when sellers see themselves as offering a certain class or suite of products. They know that their customers value the fact that they offer the complete line, because of one-stop shopping. Even if they lose money on one product, they will profit overall.
This is one of the reasons that even profitable transit providers will run "empty buses" or train cars (they usually have at least one passenger). As many of the commenters on the recent Streetsblog thread point out, having reasonably frequent off-peak service - for return trips, shopping, transfers, or even just to provide that extra wiggle room in case the passenger is running late - increases the ridership during peak times. Even if they lose money on one bus run, they will profit overall. This is true of other cost-benefit calculations, like energy use and greenhouse gas emissions.
Well, except that most of them don't actually profit. I'm reminded of the old joke, "We lose money on every sale, but we make it up on volume!" If the average farebox recovery is less than 1 (or if the average greenhouse gas emissions are too high), then you need to do something else to achieve your goals. But the point is that the not-quite-empty buses sometimes don't have as bad an effect on the overall averages as you might think.