Such programs would do well to learn from Spain's mistakes, solar executives and analysts say. In just one year of boom, the country committed itself to solar payments estimated at $26.4 billion, which in turn led to taxpayer backlash and bust.
This is the contractually guaranteed subsidy, to be paid out over 25 years (source), for the 2,460 MW of solar power installed in 2008 (out of 3,000 MW cumulative - so basically all of it.) Based on this table of Spanish solar plant capacity factors (weighted average 18.9%), this would be equal to about 450 MW average output. But this is likely an overestimate, because a large (how large?) fraction of the solar power is residential rooftop PVs, not solar plants, and these get much lower capacity factors (they are not sun-following, they are at bad angles, they get less frequent cleaning).
To be clear: the subsidies, once granted, were guaranteed for 25 years at a rate indexed at 575% of market value. The subsidy changes affect new solar plants (hence the market collapse), but not existing ones: their subsidy will continue unaffected.
The production of electricity of photovoltaic origin is remunerated at a price equivalent to 575% of the Spanish reference rate (TRM) for the first 25 years and 460% from then on, according to Royal Decree 436/2004, which updates the legal and economic framework for the production of electricity under the Special Regime.
Incidentally, $26 billion is also the very infamous, much-maligned bid cost for two (2) ACR-1000 CANDUs in Ontario (recent discussion). That is, likewise, a lifetime cost, including all fuel and operations over 60 years. So let's compare: $26 billion for an average output of 450 MW over a 20-25 year lifespan, vs. $26 billion for (2 * 1,200 MWe * 0.95 capacity factor) = 2,280 MW over 60 years, plus some new highway exits (how did that get into the AECL bid anyway). That is, 80-100 TWh (over 20-25 years) vs. 1,200 TWh (over 60 years). Same cost (modulo interest rates and related financial quirks.)
(But there is a significant consistency problem with my numbers (a bias in favor of solar). If you divide the $26.4 billion, divided by the current subsidy rate of 0.44 €/kWh = 63c/kWh, is only 42 TWh, not 100 TWh. I do not know whether this reflects interest rate corrections, inflation corrections, variations in subsidy rate, or an overestimate on my part of the average capacity factor.)
There is one ray of hope. According to the NYT article, many of the solar plants receiving subsidies are under investigation for fraud. Perhaps if they are found ineligible, a few tens of billions will be saved.