Off the Rails

Monday 31 March 2025

In forecasting, sometimes the simple approach works best. Cold Eye Earth typically uses trailing five-year compound annual growth rates (CAGR) when making one-year growth forecasts for wind+solar generation, and total demand for global power. This avoids the granular, region by region approach often taken up by think-tanks and other professionals—which requires alot more labor—and which is quite superior for detecting short-term trend changes. The five year CAGR approach, on the other hand, is long enough to obtain a series of values that captures recent activity, but not so long to be unduly affected by older rates of growth that have likely gone stale. No special talent or expertise is required either, to apply a trailing CAGR to the near future. Indeed, the only discretionary decision made here by Cold Eye Earth is to prefer the five-year rate to the ten-year rate, or some other time period.

How is the method working out? Well, in 2023, Cold Eye Earth began to share the chart shown below with readers, indicating that combined generation of wind+solar would advance from 3427 to 3993 TWh, or 566 TWh, based on a five-year trailing CAGR of 16.6%. No polling, no news reports, no on the ground staff. The following year, when EI Statistical Review released the actual data for 2023, they arrived at this result: 537 TWh of generation growth for combined wind+solar. 566 vs 537? Pretty damn good.

Notice that there’s no expectation either that the five-year CAGR is perfect. The forecast got within 5%, which is an excellent result, and more than adequately meets the needs of professionals working in this area. Just to remind: the lengthier the time period of the forecast, the more unreliable it will be. So, while the five-year CAGR projection works great on a one-year forward basis, you can see that the above chart goes out to 2030—and during that time there is certain to be unexpected growth volatility (lower, or higher). To make this plain: that figure for 2030 of 11664 TWh is sure to be some distance from the actual result. Let’s hope it’s higher!