ABSTRACT

McDonald and Schrattenholzer (2001) provided the ¥rst overview of experience curves for energy technology. Particular attention in their seminal publication was given to the distribution of learning rates for 48 cases of different energy technologies. While the statistical signi¥cance of a number of the included learning rates was not particularly high, several robust overall conclusions could be made on the basis of their distribution. One of the main ¥ndings was that their distribution had

Introduction ............................................................................................................ 425 Component Learning ............................................................................................. 428 Learning and Time ................................................................................................. 431 Cybernetic Approach ............................................................................................. 435 Conclusions ............................................................................................................440 Acknowledgment ................................................................................................... 441 References .............................................................................................................. 441

both commonalities and differences with respect to those published by Dutton and Thomas (1984) for (not exclusively energy-based) technologies manufactured by individual companies. Both of these studies found a median learning rate of close to 20%, but McDonald and Schrattenholzer (2001) found a higher frequency of smaller learning rates than Dutton and Thomas (1984) did.