Abstract Learning by doing, or learning through market experience, reduces costs for energy productiontechnologies. This phenomenon is modelled by using experience curves which reflect the changes in thecost of the technology as it becomes increasingly used. This article calculates the Spanish photovoltaic(PV) learning curve over the period 2001–12 by using cost data from the PV sector itself (installers,distributors and even engineers) and determines the accuracy of the obtained progress ratio by using boththe coefficient of determination R 2 and also the error s PR , which is directly determined from fitting thedata. The results show a curve with a strong structural change in the speed of cost reduction in October2009. Keywords: learning curve; renewable energy policy; feed in tariff; Spanish photovoltaic learning rate Received 16 June 2013; revised 21 March 2014; accepted 25 July 2014 1 INTRODUCTION Experience curves have been widely used in the literature as amethod for cost forecasting in new technologies. Photovoltaictechnologies have been using this methodology in recent yearsnot only inorder toestimatefuturecosts,butalsoasakeydrivingforce for incorporating endogenous technological change intolarge-scaleenergy–environmental–economy models[1].Those models gave fundamental insights and analysis of theenergy mix under different scenarios and also gave an overviewof the effectiveness of environmental policies. Although themethodology is widely accepted, it is important to mention theintrinsic difficulty involved in the calculation methodology bynon-linear systems, as well as the problem of non-convexity ofthe system.Initially, most academics were studying the learning curve as aglobal trend in the market, as modules are marketed internation-ally, by using either total shipments or total production from thesector. Later when different countries started to publish the Feedin Tariffs (hereinafter FIT) policies, there was a shift to either cu-mulative capacity or gross electricity production as the studieswerefocusingonmodellingtheevolutionoftheenergy mix.The study of local learning rates (hereinafter LRs) is ratherrecent. For example, looking into the relevant literature we canfind the calculations for Germany, the Netherlands or the USAover different time periods, for the complete system so SystemLR (modules, inverters, wiring, labour cost and project and ad-ministrative cost) or for individual components such asmodules, inverters or BOS (BOS: balance of system: a namegiven to the system without the panels) obtaining a partial LR(BOS LR, modules LRor inverter LR).After 10 years of solar PV promotion in Spain it is interestingto look at the Spanish figures, in terms of cost of the installationsfor end purchasers together with the achieved cumulative cap-acity and calculate a sector experience curve during this periodof time. This figure will give us an evaluation of the cost-effectiveness of the policy measures over the time and it will alsobe a base for calculating the social benefit obtained with the ap-plication of FIT policies; the most widely used subsidy whenpromoting photovoltaic installations.This article collects the installation system costs from theperiod 2001 to 2012 for calculating the Spanish learning curveand its LR and determines the accuracy of the obtained rates byusing both the coefficient of determination R