Forecasting Technological Innovation

Aimee Gotway Bailey, Quan Minh Bui, J. Doyne Farmer, Robert M. Margolis, Ramamoorthy Ramesh

Research output: Contribution to conferencePaperpeer-review

5 Scopus Citations


Using a database of sixty-two different technologies, we study the issue of forecasting technological progress. We do so using the following methodology: pretending to be at a given time in the past, we forecast technology prices for years up to present day. Since our forecasts are in the past, we refer to it as hindcasting and analyze the predictions relative to what happened historically. We use hindcasting to evaluate a variety of different hypotheses for technological improvement. Our results indicate that forecasts using production are better than those using time. This conclusion is robust when analyzing randomly chosen subsets of our technology database. We then turn to investigating the interdependence of revenue and technological progress. We derive analytically an upper bound to the rate of technology improvement given the condition of increasing revenue and show empirically that all technologies fall within our derived bound. Our results suggest the observed advantage of using production models for forecasting is due in part to the direct relationship between production and revenue.

Original languageAmerican English
Number of pages6
StatePublished - 2012
Event2012 International Conference on Architecture of Computing Systems, ARCS 2012 - Munchen, Germany
Duration: 28 Feb 20122 Mar 2012


Conference2012 International Conference on Architecture of Computing Systems, ARCS 2012

NREL Publication Number

  • NREL/CP-6A20-56274


  • experience curve
  • innovation
  • learning curve
  • performance curve
  • technology evolution


Dive into the research topics of 'Forecasting Technological Innovation'. Together they form a unique fingerprint.

Cite this