This thesis is concerned with the history of electricity generation costs, how they have changed over time, and the accuracy of forecasts of future costs. These costs are a critical input to policy, yet both estimates and forecasts have frequently proved to be wrong or have changed dramatically over relatively short timescales. The thesis presents evidence from three technology case studies (offshore wind, nuclear power and solar PV), supported by a review of the range of cost measures used in the economic, business and policy spheres, and the methodologies used to understand the factors that bear upon cost trajectories and approaches to forecasting future costs. Drawing upon the evidence from the case studies, the thesis examines how cost forecasts have changed over time, the (frequently wide) range of forecasts, the sources of errors, and how policy has responded to uncertainty and changes in both cost estimates and forecasts.

The findings address the limitations of commonly used cost metrics, challenge assumptions that costs will necessarily fall, discuss the meaning of regulatory certainty in the face of uncertain future costs, and emphasise the importance of context (why estimates are commissioned, and by whom, and also who they are undertaken by). The evidence suggests that the co-presentation and use of estimates and forecasts for technologies with very different technical and financial characteristics implies significantly more comparability between them than is wise, and can convey the message that the underlying uncertainties are similar, when in fact the reasons may be fundamentally different in character. This highlights how important an understanding of technology characteristics is when deriving estimates and forecasts, not simply because those characteristics bear upon the numerical values of the results, but because of the influence they have on the nature of the uncertainty of those results.