Solar panels once on the house are paid for. However, future solar panels, thanks to research, will almost surely be cheaper, possibly a lot cheaper. In fact, as they get cheaper they will become a major source of grid power. What power company would run coal plants at 5 cents/kwh if they could put up solar panels at 2 cents? This happens if either solar gets cheaper or carbon fuels get more expensive — either way we switch to solar, or something even better.

Thus it would be wrong to do cost analysis assuming the cost of grid power will rise very far above your predicted future costs for solar. People would be disconnecting from the grid left and right if it were, at least during the day.

So your guess is almost surely wrong. Not that this is a bad thing. It just means that from a purely out-of-pocket-expense viewpoint, one should buy grid power until solar gets cheaper than grid power, and then one should consider putting in solar. However, after considering it, you will realize that soon the grid power you are buying will be solar so you don’t need to put it in. You’ll be able to buy it from them at a modest margin, and let them take the risk of putting in expensive panels. 2 years later, when panels cost 20% less, you can buy from the people putting in the cheaper panels, and so on, and so on.

If solar panels become like computers, dropping in price every year, it becomes an interesting race. Now to work out the cost, one must make a guess about how much less valuable today’s panels will be in 3 years. If, like a computer, they cost half, you’re in trouble. They still put out the same amount of power, but you are competing with a new installer who paid half as much.

If you believe Zeno’s paradox, you would never put in any panels, but in fact that’s not true. There is a number that justifies putting them in, it just requires depreciating them faster, just like computers, not because they stop working but because they can be out-competed. Which means you have to charge a bit more for the power. But soon that levels off.

PV panels can’t improve as well as computers do. First of all, if they’re 20% efficient today they can’t get more than 5 times more efficient. Secondly, while you can imagine cutting the cost of manufacture in half several times, soon the other costs — mounting boards, shipping, install, wiring and the biggest cost — real estate — start to dwarf the panel cost and so it doesn’t do any good to reduce them.

Many efforts today are actually to not increase efficiency but even to drop it, and make it cheaper to manufacture. In some places (poor areas) the real estate may be cheap but other things aren’t.

Anyway, this is all good news — for everybody but the early adopter who buys expensive 2006 technology panels. And even for him, since he did it because he wants a greener planet.