Joe Romm has done a pretty thorough trashing of Matt Wald’s recent New York Times piece. Herein, I pile on. This is a shoddy enough piece of journalism to deserve it. Like Joe, I’ve talked to Matt Wald before, and generally I find him to be a good writer on energy. He’s capable of much better reporting than this.
That said, my larger beef is not with Wald nor the NYT per se, but rather with the analytical errors that are innate to his analysis, which are far too common in most journalism of this “what is the cost of competing power technology” type of piece.
The cost of electricity from any power generation technology is ultimately a function of five variables, some of which are innate to technologies and some of which are innate to specific uses of those technologies:
- Installed capital costs
- Cost of capital
- Capacity factor
- Non-fuel operating costs
- Fuel costs (the fuel price, divided by fuel efficiency)
Virtually everyone who compares power generation costs screws up one or more item on this list. Wald may well have screwed up all five.
Most of the mistakes that arise from these analyses comes from a near universal failure: no one other than engineers gives a damn about the cost of power generation at the plant gate. It has nothing to do with the price of tea in China, and only a little bit more to do with the full societal cost of delivered energy. But it’s ubiquitous in these analyses.
Roughly half of the capital cost of our electricity infrastructure comes from the transmission and distribution required to connect that power up to the load. It’s also the source of about 10 percent of the total energy losses in our electric grid, and nearly 100 percent of all our blackouts. It is also the most consistently subsidized part of the electric system, which has played a key role in the construction of remote, inefficient power plants that explain why the power industry today is only half as fuel efficient as it was in 1910. And of course, all those other societal externalities — from acid rain to global warming — associated with certain power technologies are necessarily ignored from any analysis that presumes regulatory stasis.
Installed capital costs
I have no idea what Wald assumed, but his prices are laughable. If it’s possible to build a modern coal plant with 7.8 cents/kWh costs as he suggests, I’d love to know where it’s being done. As I and others have noted, this is more like 12-13 cents/kWh by any reasonable analysis, and it keeps increasing.
A part of the reason for the high cost of coal is the cost of delivery since coal plants — for fairly obvious reasons — tend not to be built anywhere near where the people are. Transmission and distribution to connect that power to the load costs $1,300/kW, on average in the U.S., tacking 3-4 cents/kWh onto the cost of any remotely-sited power plant. Since the fuel costs alone for a coal plant are in the 2-3 cent/kWh range, I’m fairly certain Wald’s numbers simply ignored the T&D costs (and associated losses). They probably also ignored the fact that central power stations require a higher degree of redundancy than local ones, for the simple reason that more (small) generators make for a more reliable system than fewer (large) ones.
But hell, let’s give the NYT the benefit of the doubt. And let’s be conservative. Recent price increases notwithstanding, let’s assume that a coal plant that is only compliant with current environmental rules (e.g., pre-CO2) can be built for $2,500/kW, and that the 7.8 cents includes 2 cents for T&D and 2 cents for fuel. Throw in another penny for non-fuel operating costs and that leaves us with 2.8 cents/kWh for capital recovery. I will again be really generous and assume that this plant operates 24/7/365, without a single outage for maintenance or unplanned outages, so that 2.8 cents/kWh will pay off $0.028 x 8,760 = $245/kW per year. That’s a 10.2 year simple payback on the $2,500/kW investment, or a 7.5 percent annual rate of return (over 20 years of outage-free operation). Are you freakin’ kidding me? No one is building a power plant for those kind of economics. And that’s with hugely optimistic performance assumptions! Which brings us to the second assumption:
Cost of capital
Most power generation analyses look at the cost of capital (that is, how much do your investors and lenders expect you to pay back for every dollar they give you) based on current experience. People are building modern coal plants with an 11 percent cost of capital, so let’s use that. People are building wind turbines with a 15 percent cost of capital, so let’s use that.
But here’s the problem with that analysis: lenders don’t loan money based on fuel type. They loan it based on credit risk. And a regulated utility presents a massively lower credit risk than a small renewable developer. It is perhaps the most distortive subsidy in our electricity system, because it makes money flow to the most expensive, most inefficient power-generation technologies.
To be sure, just because this is “unfair” doesn’t make it unreal. Small, financially unsophisticated guys don’t have the kind of credibility with lenders that big utilities have, no matter how much we might want it to be otherwise. But when we are asking ourselves what the societal costs of various options are for the purposes of policy-making, Hippy Joe the Renewable Guy’s credit rating isn’t relevant — what matters is a comparison across levelized capital structures. I would be very surprised if the EPRI analysis Wald cites makes such an equivalent comparison.
Capacity factor is a measure of how many kWh a given kW of generation can produce in a year. It is a critical weakness of solar and wind, for all the usual reasons. (Rage, rage against the stillness of the night!)
It’s usually got a healthy dose of BS in it for traditional technologies as well. The whole U.S. coal fleet today runs at something like a 75 percent annual capacity factor, generating 6,750 kWh/kW-year, rather than the 8,760 in my way-too-generous assumption above. This isn’t because coal plants are particularly prone to outages. Rather, it’s because for much of the country, there are substantial pieces of the year when the nuke + hydro + coal fleet is capable of generating more power than the system needs during substantial fractions of the year. As the highest marginal-cost generation of those three, the coal “dials back” accordingly, running at less than full load for much of the year.
This matters because — as many a renewable developer knows — a plant that isn’t running is a plant that isn’t making money. This is a problem that no storage technology in the world will solve, but is often ignored in analyses that assume that coal and nuke plants will run at 90+ percent capacity factors even while others run much less. EPRI and Black & Veatch may assure you otherwise — but you can be quite certain that the equity investors who conspicuously chose not to invest in coal plants during the last two decades knew better.
Non-fuel operating costs
This is perhaps the least interesting of the numbers, for the simple reason that the data on O&M is pretty robust, and outside of equipment manufacturers, rarely overstated. It’s the cost of humdrum items like labor, water chemicals, insurance, etc. Costs that all have to be paid, to be sure, but they don’t swing wildly from one technology to the next. At the very low-end, you might see as little as 0.5 cents/kWh for natural gas-fired turbines or up to 2-3 cents/kWh for small-scale solid fuel plants (biomass, some coal). But it is worth noting that the cost of pollution clean up — from sulfur to CO2 — are borne predominantly by coal plants, and any sane investor is going to put a healthy cushion in their analysis to factor in coming CO2 regulation, either by demanding much higher returns or by assuming much larger non-fuel operating costs. Again, there’s no possible way this was included in Wald’s analysis.
This raises a final issue that all economic analyses struggle with: volatility.
The story of the last 15 years — from an energy, credit, or commodity perspective — is not a story of rising costs per se, but rather one of massive volatility. From 1995 to today, crude oil has gone from $20/bbl to $9 to $140 and back to $40. Natural gas has ranged from $2/MMBtu to $15. Coal hasn’t swung quite as strongly, but it’s been far from consistent.
How do you factor this volatility into your analysis? That turns out to be a really hard question. In theory, volatility = risk, and so you capture it all in the cost of capital. That’s what modern options-pricing is based on, and was used heavily by the folks at Long Term Capital Management to figure out how to make money in commodity markets. Say what you want about those guys, they were freakin’ smart. And pricing volatility is really freakin’ hard.
Moreover, the universal feature of all “alternative” energy sources, from solar to efficiency is that they reduce one’s exposure to fuel volatility. That clearly has value — and it clearly isn’t captured in any model that sets a fuel price, assumes a conversion efficiency, and extends that number ad infinitum to the future with nothing more than a known inflation factor.
Taking that all into account, one can maybe give Wald a bit of the benefit of the doubt, if only because he is repeating mistakes made by so many others, so consistently. But the Times gets cited a lot more often than Black & Veatch reports, and it ought to be held to a higher standard. Wald’s article simply isn’t varsity material.