Energy forecasters consistently underestimate wind and solar. A critic explains why that’s a problem.

Renewable energy has grown faster than almost anyone predicted, for going on two decades now.

For the people whose job it is to accurately forecast renewable energy growth, that fact is ... awkward.

Why do professional energy analysts and agencies have such a tough time getting their arms around wind and solar? Why are they so consistently wrong about renewable energy, in the same direction, over and over again?

It’s a complicated issue. I once wrote a long, in-depth post on it, focusing on the (similarly pessimistic and incorrect) forecasts from the International Energy Agency. If you really want to dig in on these questions, you should read it.

The dispute, however, is still ongoing. Today I want to catch up on the latest developments.

The US Energy Information Administration (EIA) does forecasts too, which are contained in its “Annual Energy Outlook” (AEO). They have also consistently underestimated wind and solar.

Yesterday, EIA released something between a mea culpa and a defense of its work: “Wind and Solar Data and Projections from the U.S. Energy Information Administration: Past Performance and Ongoing Enhancements.“

The report reviewed the sources of EIA wind and solar data, discussed the accuracy (or inaccuracy) of past predictions, and detailed the changes that are being made to improve the forecasts going forward.

However, the tone of the report and some of its assertions struck many people as defensive and a little bit prickly. While the impetus behind the report is noble, and it’s a nice first step, there was definitely some ass covering and critic jabbing involved.

The report solicited a series of responses on Twitter from Alex Gilbert, an energy researcher and co-founder of the Spark Library, an energy research platform. Gilbert has published peer-reviewed critiques of EIA forecasts before, so he is well-versed in the issues.

I thought his response deserved a wider airing. I’ll add a few thoughts of my own below.

There are all sorts of juicy issues in here — enough to spawn a dozen posts. I’m particularly interested in how projections like EIA’s obscure the importance of risk (think fossil fuel price volatility) and risk hedging (think long-term, predictable power contracts) in wind and solar markets.

But the main lesson is not about the specifics of energy analysis at all — it’s a sociopolitical lesson, about how energy modeling is used by journalists and policymakers.

In a nutshell, the problem is this: EIA’s modeling is just that — modeling. It reflects the assumptions that go into it. Among them are assumptions about policy, technological innovation, and evolving costs.

Those things are extremely difficult to predict.

To take just one example: the federal tax credits (PTC and ITC) for renewable energy comprise one of the most important policy drivers in US renewable energy markets. They were renewed for another five years in December, at the very last minute, as part of a larger budget deal. It was a surprise to everyone.

And yet, the tax credits are crucial to accurately modeling near-future energy markets!

The best anyone can hope to do, in the context of an unpredictable and quickly evolving policy environment, is model a range of scenarios, based on a range of assumptions. And that’s more or less what EIA does.

The problem, as Gilbert notes, is that EIA’s public communications tend to emphasize its “reference case,” its baseline scenario, giving the impression that it constitutes a prediction.

It is not a prediction. It is not even meant to be taken as any more probable than any other scenario. It is just the center in a range of scenarios.

But journalists and pundits can’t help treating it as a prediction. What’s worse, utility executives and energy investors treat it that way. It distorts their decisions, causing them to underestimate wind and solar growth.

As EIA itself has said many times (often buried an a preface or a footnote), people should try to understand these projections and the assumptions that inform them; they shouldn’t take projections as predictions. Journalists, in particular, should resist the temptation to accept EIA projections uncritically, or to treat them like prophecies.

But Gilbert is right that EIA could do more to emphasize this fact up front. A range of projections is nothing but the reflection of a range of assumptions — assumptions that are open to debate and critique.

EIA’s self-reflective report on past projections is a good start. Now it just needs to relax a little, get a little less defensive, and keep the conversation going.