The US government underestimated solar energy installation in the US by 4,813%

The only thing certain in this life are death, taxes and the US department of energy’s massive underestimate of renewable energy capacity.

Every two years, the US Energy Information Administration (EIA), America’s official source for energy statistics, issues scenarios about how much solar, wind and conventional energy the future holds for the US. Every two years, since the mid-1990s, the EIA is wrong. Last year, it was spectacularly wrong.

The Natural Resources Defense Council and Statista recently teamed up to analyze the EIA’s predictions for energy usage and production. It found that the EIA’s ten-year estimates between 2006 to 2016 systematically understated the share of wind, solar and gas. Solar capacity, in particular, was a whopping 4,813% more in 2016 than the EIA had predicted it would be.

Meanwhile, EIA estimates regularly overstate US fossil fuel consumption, which some see as an attempt to boost the oil and gas industry.

These estimates matter because they form the basis for actions by the Environmental Protection Agency and other federal agencies (hence the outdated goals set in the Obama Administration’s Clean Power Plan, which may be reached without new policy intervention).

The agency’s “projections bear little resemblance to market realities” because they ignore publicly available evidence, argues the clean-energy non-profit Advanced Energy Economy. It cites the EIA’s 2015 expectation for solar capacity to double by 2026, despite a pipeline of projects that would—and ultimately did—exceed that benchmark by early 2017. Similarly, the EIA expected the installment of 6.5 GW of new wind capacity between 2017 and 2030—even though new US wind installations averaged 6.5 GW per year between 2007 and 2014. “They’re not just conservative about change,” Advanced Energy Economy vice-president Robert Keough told Politico. ” They’re ignoring the evidence of what’s actually happening in the market.”

But it may simply be a problem of methodology. Michael Grunwald at Politico reports the EIA is mostly using backwards-looking methodology to project the future. Creating baseline scenarios based on historical data works, but it works better when technology proceeds at a leisurely, incremental pace, unlike the fast-moving technologies of solar and wind energy capture.

In the agency’s defense, the pace of technological change is unpredictable. Conservative models are almost always wrong during times of breakneck technological or economic change (as with wind and solar), and the government is not in the business of rosy speculation. Politicians do the EIA no favors by promoting, then threatening, tax credits to encourage renewable investments.

Even as the EIA (slowly) adjusts its models, the US will likely stay the renewable course. Solar and wind have reached grid parity in many places, reports UtilityDive (pdf), and the majority of new electricity-generating capacity in the US in the last two years has come from renewables. That proportion is only expected to grow.