Curbing enthusiasm and anticipating global sideswipes: Predicting the future cost of electricity
The UK government wants to reduce emissions from the power sector without breaking the bank. But there are a lot of uncertainties involved in predicting the future cost of power from nuclear, gas, or renewable sources, with analysts often getting things wrong. So how can policymakers know which estimates to trust, and which technologies to back?
A report from the UK Energy Research Centre (UKERC) seeks to understand why estimates often get thrown off track. It looks back over old predictions to identify a number of pitfalls such analyses fall into, with the hope of improving estimates in the future.
Comparing estimates
The UKERC team surveyed a number of studies estimating the future cost of generating electricity. It found that in all cases, later estimates predicted electricity costing much more than previously predicted.
That’s a problem because policymakers want to know which energy sources represent value for money, and want reliable data to base their decisions on. The graph below shows how estimates for a range of technologies changed over the course of five years.
Each colour represents a different technology. The first bar shows analysts’ predictions in 2006, with estimates from 2011 next to them.
UKERC’s analysis shows 2011’s cost estimates are consistently higher than earlier predictions:
Image - UKERC electricity cost predictions (note)
It shows the estimated costs for electricity generated by efficient gas plants have risen by nearly 80 per cent, around 75 per cent from nuclear plants, and by over 40 per cent for onshore windfarms. Such results were “completely unanticipated”, the report’s lead author, Dr Rob Gross, said during a public lecture at Imperial College yesterday.
He says UKERC’s research shows just how much estimates can change in a short space of time – a cause for concern for policymakers who rely on them to make big, long term, decisions about the future of the UK’s energy sector.
Prediction difficulties
But that’s not a reason to totally disregard the analyses, Gross says. After all, policymakers’ hunger for best-guesses about future costs is unlikely to go away. Instead, analysts must try to understand why the estimates rocketed.
One explanation could simply be analysts’ over-optimism about future cost reductions, the report says.
Sometimes it’s in an industry’s interest to underestimate how much it will cost to generate electricity from a particular technology in in order to encourage investment, the report notes. It singles out nuclear and carbon capture and storage (CCS) analyses as potentially falling victim to such strategic analysis in particular.
That’s not always the case, however. Sometimes analysts’ estimates are overoptimistic simply because the projects are in their early stages, and the risks and costs are yet to be properly understood, the report claims.
Gross points to the Longanet CCS project – scrapped in 2011 – as an example of this. The estimated cost of building the project rose steadily over time, until it reached a point where the government and investors were no longer willing to support it.
Analysts also sometimes misjudge whether a technology has hit its most expensive point, or if costs are expected to continue to rise for a while, the report says. Technologies generally get more expensive before they get cheaper, Gross says, which means costs peak at some point. If analysts pick the wrong peak year, then costs come down later or less than expected, Gross argues.
This may be why previous estimates of electricity generated from offshore windfarms were low, Gross claims – analysts thought offshore wind technology was about to start getting cheaper, but costs actually continued to rise for a number of years.
But the problems with the predictions go beyond the analysts’ own errors or biases, Gross says. Often, estimates are thrown off track by big, real world changes – or “sideswipes”, as the report describes them.
For instance, the government may unexpectedly announce a commitment to a particular technology for ideological reasons, as has arguably been the case with nuclear power in recent years. Likewise, the government can sometimes go cold on technologies it once supported, reducing subsidy levels (as with biomass). Such decisions have a knock-on effect on how technologies develop, making it hard to predict how quickly costs will fall and by how much.
Global energy prices are also very hard to predict, throwing the models off course, the report says. For instance. coal and gas prices are very volatile, meaning the cost of electricity from fossil fuel plants also goes up and down. Likewise, the materials developers use to construct renewable energy technologies such as wind turbines and solar panels are also liable to unexpected price hikes and falls.
Both fuel and commodity prices escalated throughout the 2000s, meaning earlier predictions based on the prices of the day were liable to get out of date pretty quickly, the report says.
Getting better
UKERC’s report highlights many reasons why estimates of the cost of electricity may change. Some of the previous mistakes are easy to curb – being wary of seemingly over enthusiastic estimates, for example. Others – such as the government’s changeable preferences – are harder to factor in.
But analysts are improving their models as they learn from past difficulties, Gross says. That’s largely because, as time goes by, more data becomes available to help refine predictions, he argues.
So while predicting the precise future cost of electricity remains a near-impossible task, UKERC’s report suggests such analyses are still helpful – so long as analysts learn from past mistakes.